Home

Weather Forecast using Kalman Filter Algorithm with Warning

1. Forecast Time Accuweather Forecast System Forecast Percent C C Difference 2 00 PM 29 26 86 7 66 3 00 PM 29 27 22 6 33 4 00 PM 29 27 18 6 48 5 00 PM 31 27 18 13 13 6 00 PM 29 27 08 6 85 7 00 PM 28 26 86 4 16 8 00 PM 26 26 78 2 96 9 00 PM 24 26 71 10 69 10 00 PM 26 26 60 2 28 11 00 PM 26 26 54 2 06 Average Percentage Difference 6 26 The data under the system forecast show little changes on its temperature forecast because the estimation of values depends on the actual measurement of the temperature on the area of testing and historical data The system forecast increases its value if and only if the present estimated forecast value is lower than the actual temperature of the area Otherwise the system forecast decreases A value of 6 26 was computed for the average percentage difference of the 10 trials in which it does not affect the capability of KF as an algorithm for weather forecasting since the 41 Accuweather forecast considered the whole area of Quezon City while in the system forecast it was based on the specific location in the area Table No 3 6 Comparison of System Temperature Forecast to Actual Temperature Measurement Forecast Time Actual Temperature System Forecast Percent Measurement C C Difference 2 00 PM 27 26 86 0 52 3 00 PM 29 27 22 6 33 4 00 PM 27 27 18 0 66 5 00 PM 27 27 18 0 66 6 00 PM 21 27 08 0 30 7 00 PM 27 26
2. HELP Figure No 3 16 Driver Setup x64 4 Click the Install button 5 A prompt will be displayed telling that the drive is successfully completed 62 Installing the MATLAB software 1 Insert the CD for installing MATLAB 2 Double click setup 3 A window will be displayed similar to the figure below See Figure 3 17 A MathWorks Installer 20 X M ATLAB Install MathWorks Software we SIMULINK This program will install MathWorks products on your computer You may also be required to R2009a activate your software Install automatically using the Internet recommended Connection Settings Install manually without using the Internet MathWorks products are protected by patents see www mathworks com patents and copyright laws By entering into the Software License Agreement that follows you will also agree to additional restrictions on your use of these programs Any unauthorized use reproduction or distribution may result in civil and criminal penalties MATLAB and Simulink are registered trademarks of The MathWorks Inc Please see www mathworks com trademarks for a list of additional trademarks Other product or brand names may be trademarks or registered trademarks of their respective holders The MathWorks Figure No 3 17 MathWorks Installer 4 Choose Install manually without using the Internet Then click Next 5 Inthe License Agreement
3. drawnow set handles btnGetData Enable on AcqFlag 0 sendDataF 0 while 1 oras clock myDate datestr oras mmmm dd yyyy HH MM SS AM myTitle sprintf Weather Forecast Using Kalman Filter Algorithm s myDate set handles figure1 name myTitle if AcqFlag if sendDataF set handles btnSendData String Sending Data set handles btnSendData Enable off drawnow 90 btString sprintf US_ 08 2f_ 08 2f_ 08 2f n PredT 10 PredP 10 PredH 10 fprintf s btString pause 10 sendDataF 0 set handles btnSendData String Send Data set handles btnSendData Enable on drawnow end fprintf s UD dummy count fscanf s s if count gt 19 stri dummy 5 8 str2 dummy 9 12 str3 dummy 13 16 tempVal sampCtr sscanf str1 d 4 88 10 presVal sampCtr sscanf str2 d 4 89 5000 0 095 0 009 2 humiVal sampCtr sscanf str3 od if humiVal sampCtr gt 7351 RH lt 0 out of range humBase 101 elseif humiVal sampCtr gt 7224 base 0 hUpper 7351 hLower 7224 humBase 0 elseif humiVal sampCtr gt 7100 base 10 hUpper 7224 91 hLower 7110 humBase 10 elseif humiVal sampCtr gt 6976 base 20 hUpper 7110 hLower 6976 humBase 20 elseif humiVal sampCtr gt 6853 base 30 hUpper 6976 hLower 6853 humBase 30 elseif humiVal sampCtr gt 6728 base 4
4. 70 70 71 71 12 72 13 73 74 75 75 76 76 viii Figure 3 42 fig File for Kalman Filter Figure 3 43 Unhandled Internal Error Figure 3 44 Error on Opening Serial Port 11 79 79 ABSTRACT The Kalman Filter Algorithm filters the sensor s data and process data which are used for predicting the next state or value The algorithm is used in this study to predict the weather condition on an hourly basis for a given location The goal is to implement the processes and the equations of the basic Kalman Filter Algorithm in order to produce a forecast based on the given set of available data used to train the system The device uses a microcontroller specifically PIC16F877A and is integrated with different sensors for each parameter and wireless technology for the data transfer The sensors include LM35 for temperature the capacitive humidity kit for humidity MPX114AP for measuring the pressure and Zigbee technology for wireless data transfer The system includes methods for alerting people by using GSM and Android application through Bluetooth The tests conducted show that the sensors integrated in the device for measuring temperature pressure and relative humidity produced almost the same measurements as the instruments used for measuring the actual value of the said parameters Furthermore statistical tests show that the system forecast is accurate as evidenced by the small per cent difference between th
5. 33 2 Visit the website of Accuweather or go to the link below and select the Las Pinas link button Click the Hourly Forecast link button on the left side of the window http www accuweather com en ph philippines weather 3 In the Accuweather site the forecast measurements of relative humidity per hour are displayed Record 10 measurement samples and its corresponding time for the true value which will be used as the reference for conducting the measurement using the device 4 Compute the percentage difference of the measured value to the true value using the equation used in the previous test 3 4 1 3 Pressure 1 The trial was conducted in Mapua Institute of Technology wherein the device was positioned on each floor after each trial from higher to lower altitude 2 For every step record the change of value in pressure which is reflected in the data logger This data will be recorded under measured value In assessing the true value of pressure an electronic barometer was applied 3 Calculate the percentage difference of the measured value to the true value using the equation used in the previous test 3 4 2 Procedures to be followed in examining the system for the prediction of the three weather parameters using a Kalman Filter algorithm 1 To begin install the two devices and the laptop Plug the device bigger one and plug the other device smaller one into the laptop using USB to serial port cable 2 Run the M
6. Conclusion References CONCLUSION RECOMMENDATIONS REFERENCES APPENDICES 41 43 47 50 50 51 53 55 57 58 60 61 62 LIST OF TABLES Table 3 1 List of Components Table 3 2 Tabulated data for Temperature Table 3 3 Tabulated data for Relative Humidity Table 3 4 Tabulated data for Pressure Table 3 5 Comparison of Temperature Forecast between the System and Accuweather Table 3 6 Comparison of System Temperature Forecast to Actual Temperature Measurement Table 3 7 Comparison of Humidity Forecast between the System and Accuweather Table 3 8 Comparison of System Humidity Forecast to Actual Humidity Measurement Table 3 9 Comparison of Pressure Forecast between the System and Accuweather Table 3 10 Comparison of System Pressure Forecast to Actual Pressure Measurement Table 3 11 T Test for Temperature Table 3 12 T Test for Relative Humidity Table 3 13 T Test for Pressure 20 37 38 39 41 42 44 45 47 48 50 52 53 vi LIST OF FIGURES Figure 1 1 Conceptual Framework Figure 2 1 Weather Instruments Figure 2 2 GSM GPRS Module Framework Figure 2 3 Circuit of the Kalman Filter Figure 3 1 Block Diagram of the Device Figure 3 2 Zigbee Schematic Diagram Figure 3 3 Temperature Sensor Schematic Diagram Figure 3 4 Pressure Sensor Schematic Diagram Figure 3 5 Humidity Sensor Schematic Diagram Figure 3 6 Schematic Diagram PIC16F877A Microcontroller Figu
7. Send Warning Text GSM No Yes Continue Monitoring No Click Stop Data Acquisition Figure No 3 8 Program Flowchart ED 31 3 4 Validating Testing and Determining the Accuracy of the System The validation of the device ensured that the system meets the requirements and specifications stated in Chapter 1 It includes the following parts 1 the device can measure and monitor the actual basic weather parameters and 2 the prediction of the three weather parameters temperature pressure and relative humidity using the Kalman Filter algorithm A test will be conducted in each part to validate the method 3 4 1 Procedures to be followed in measuring the accuracy of measurements of the device The test required only the data logger part of the program which is responsible for gathering monitoring and displaying the measurements received by the laptop from the sensors integrated in the device Each parameter was subjected to a scenario wherein two types of measurements were acquired true value using the right instrument for measuring a certain weather parameter and measured value using the measurements of the device displayed in the data logger Instruments used in the test for measuring the true value are the digital thermometer for the temperature electronic barometer for pressure and the Accuweather site for the relative humidity The differences between the measured value to the true value were express
8. laptop or PC Add a Device Allow a Device to Connect Show Bluetooth Devices Send a File Receive a File Join a Personal Area Network Remove Icon OOF a amp Customize ip 6 Figure No 3 28 Opening the Bluetooth Settings 69 2 The Bluetooth Settings will appear as shown in Figure 3 29 Click the COM ports tab Bluetooth Settings Options COM Ports Hardware Discovery F Allow Bluetooth devices to find this computer To protect your privacy select this check box only when you want a Bluetooth device to find this computer Connections F Allow Bluetooth devices to connect to this computer 7 Alert me when a new Bluetooth device wants to connect F Show the Bluetooth icon in the notification area Change settings for a Bluetooth enabled device Figure No 3 29 Bluetooth Settings Window 3 Inthe COM ports tab click the Add button Rem sists Choose a COM port for a Bluetooth enabled device Figure No 3 30 COM Ports 4 The Add COM port window will be displayed Verify that the radio button for Incoming device initiates the connection is selected Then click OK button 70 Bluetooth Settings 8 ng Add COM Port mm Select the type of COM serial port that you want to add Incoming device initiates the connection Outgoing computer initiates the connection Device that will use
9. large number of peripherals which was approximately 40 pins For the humidity sensor a capacitive humidity sensor kit was used Taranovich stated in 2011 that a capacitive humidity sensor is more appropriate to use than a resistive humidity sensor kit because the capacitive kit 20 ranges its percent relative humidity from 0 to 100 while in a resistive sensor kit it only ranges from 20 to 90 percent only The schematic diagram for the Zigbee connection is shown in Figure3 2 It has been said that ZigBee is more advantageous to use than other wireless technologies because it requires low consumption of power and low data rates Scheneider Electric Industries SAS 2011 Basically two Zigbee modules were used in the design wherein one will serve as the transmitter and the other will serve as the receiver The transmitter is used to transfer data coming from the sensors into the data logger which is implemented in MATLAB The receiver is used to read the data coming from the microcontroller which are parameter measurements iV vec ADO DOUT ADL DIN ADZ ct ch ak DOSA AD3 ee RESET RTS ails PWMO ASSOCIATE RESERVE D VREF PWM ON DUR crs gi END AD4 7 XBEE Figure No 3 2 Zigbee Schematic Diagram For the temperature sensor LM35 is used which is depicted in the schematic diagram shown in Figure3 3 In this study PIC16F877A is used as the microcontroller which has 8 channels AQ A5 and EO E2 of
10. 10 bit resolution ADC analog to digital converter module Microchip 2012 Since the input voltage is 5V which is the maximum value the range of voltages starting at OV and ended by 5V needs to be separated into equal steps starting from 000 up to 1023 21 vee our RAI CHD TEMPERATURE Figure No 3 3 Temperature Sensor Schematic Diagram The ADC step can be calculated using equation 3 1 Vref ADC Step PEF Wi 3 1 Solving for the ADC step a value of 4 883 mV is calculated wherein it is the minimum voltage that the ADC can read To change or convert the ADC reading to Celsius degrees the input voltage is divided by 10mV C which is the sensor s sensitivity Thus the equation used is shown below Temperature in C Reading 4 883 10w 3 2 Figure 3 4 shows the schematic diagram for the pressure sensor In addition to the sensing element a capacitor of 0 1 F is connected Connecting a capacitor to the ground will filter unwanted frequencies from the input voltage PRESSURE Hy SENSOR a Figure No 3 4 Pressure Sensor Schematic Diagram 22 As observed in the schematic diagram of humidity sensor in Figure3 5 a 22KQ resistor is connected to the sensor element This sensor will produce an analog voltage which will go into an analog to digital converter or ADC of the microcontroller Then it will generate a digital voltage output for humidity Almost same process will be performed on the other se
11. 100 9 0 0297 5 100 9381 100 9 0 0378 6 100 9381 100 9 0 0378 7 100 9381 100 9 0 0378 8 100 9382 100 9 0 0379 9 1009382 100 9 0 0379 10 100 9382 100 9 0 0379 39 Figure 3 12 proved that the measured value and true value are almost equal and the percentage acquired in the percent difference does not contribute significantly to the trend of the measurement of pressure A constant value of about 100 9 for air pressure was noted during the time of the testing which indicates that precipitation is less likely Otherwise if the measured pressure is lower there is an indication that a low pressure system or front is approaching Pressure True Value vs Measured Value Measured Value kPa True Value kPa Ss Au x zZ v fo Dn n o Sami u 1 2 3 4 5 6 7 8 9 10 Trials Figure No 3 12 Pressure True Value vs Measured Value 40 3 4 2 Determining the Accuracy of System Forecasts to Accuweather Forecasts and Actual Measurements Hourly Forecasts of Temperature Table 3 5 illustrates the collected data for temperature of the system forecast and Accuweather forecast As shown in the table the forecast temperatures in the Accuweather range from 24 C 31 C during the testing which reflect the actual weather condition at that time in Quezon City which is a cloudy day Table No 3 5 Comparison of Temperature Forecast between the System and Accuweather
12. IstMode Callback hObject eventdata handles global grpFlag contents cellstr get handles IstMode String 113 selected contents get handles IstMode Value if stremp selected Temperature 1 grpFlag 1 elseif strcmp selected Pressure 1 grpFlag 2 elseif strcmp selected Humidity 1 grpFlag 3 end message sprintf s d selected grpFlag set handles pnIGrapher Title message function IstMode_CreateFcn hObject eventdata handles if ispc amp amp isequal get hObject BackgroundColor get 0O defaultUicontrolBackgroundColor set hObject BackgroundColor white end function btnSendData_Callback hObject eventdata handles global sendDataF if sendDataF sendDataF 1 end function pniGrapher_ResizeFcn hsObject eventdata handles 114 ANDROID Main Activity Region Module Attributes FullScreen False IncludeTitle True ApplicationLabel Kalman VersionCode 1 VersionName SupportedOrientations unspecified CanInstallToExternalStorage False End Region Activity module Sub Process_Globals Dim admin As BluetoothAdmin Dim seriall As Serial Dim foundDevices As List Type NameAndMac Name As String Mac As String Dim connectedDevice As NameAndMac Dim strBuffer As String Dim tempData As String Dim i As Int Dim preTemp As String Dim prePress As String Dim preHumid As String Dim Timer1 As Timer End Sub Sub Globals Dim sf As Strin
13. Lynn S Thesis Adviser Dr Felicito S Caluyo 5 a Pressure Humidity 3 Start Data Acquisition Send Warning Text Figure No 3 38 fig file 6 To open the M file right click the Start Data Acquisition button then select View Callbacks then Callback mE Cut Ctrl X Copy Ctrl C Paste Ctrl V Clear Duplicate Ctrl D E IE BE Bring to Front Ctrl F Callback Send to Back Ctrl B CreateFcn Object Browser DeleteFcn M file Editor ButtonDownFcn KeyPressFcn View Callbacks gt Fe ee l Property Inspector Push Button Prope Figure No 3 39 M file 7 Click the Run button g Desktop Wind elp ee fy SRK Kh Figure No 3 40 Run Button 8 Select a Raw Data CSV in order for the program to run D Name p Places Gl KalmanTest Desktop Libraries amp Network Figure No 3 41 Selecting Raw Data string most commonly a literal comma or tab Babylons English Dictionary 9 The GUI is shown below Look in Ji WirelessWeather aae Date modified Type 8 31 2013 5 08 PM Microsoft Note A CSV file consists of any number of records separated by line breaks of any kind each record consists of fields separated by some other character or 76 F WWM ig me ao ea JE WE Ee File Edit View Layout Tools Help Oom amp Ov 4B ha PR gt Weather Forecast using Kal
14. PIC16F877A Device 16F877A Declare Xtal 20 Declare Adin_Res 10 Declare Adin_Tad FRC Declare Adin_Stime 50 Declare Hserial_Baud 9600 Declare Hserial_RCSTA 10010000 Declare Hserial_TXSTA 00100100 Declare Hserial_Clear True Symbol gsmPW PORTD 5 Symbol gsmIn PORTD 7 Symbol gsmOut PORTD 6 Symbol LEDO1 PORTD O Dim aRead1 As Word Dim aRead2 As Word Dim aRead3 As Word Dim aReadTemp As Dword Dim gCtr As Word Dim uNum1 12 As Byte Dim myCom As Byte Dim pTemp 9 As Byte pPres 9 As Byte pHum 9 As Byte ADCON1 80 TRISA FF TRISB FF TRISC BF TRISD 9E TRISE 07 DelayMS 50 80 pTemp 8 0 pPres 8 0 pHum 8 0 GoSub subs_SMS_Init HSerOut U While 1 1 HSerIn Wait U myCom Select Case myCom Case D aReadTemp 0 For gCtr 1 To 1000 aRead1 ADIn 0 aReadTemp aReadTemp aReadi DelayUS 5 Next gCtr aReadTemp aReadTemp 1000 aReadi aReadTemp aReadTemp 0 For gCtr 1 To 1000 aRead3 ADIn 2 aReadTemp aReadTemp aRead3 DelayUS 5 Next gCtr aReadTemp aReadTemp 1000 aRead3 aReadTemp aRead2 Counter PORTD 2 900 HSerOut ADC Dec4 aRead3 32 Dec4 aRead1 32 Dec4 aRead2 13 81 Case S HSerIn Wait _ Str pTemp 8 Wait _ Str pPres 8 Wait _ Str pHum 8 GoSub subs_SMS_Send HSerOut U EndSelect DelayMS 1000 Wend subs SMS Init gsmPW 1 DelayMS 2500 gsmPW 0 DelayMS 5000 LEDO1 1 SerOut gsmOut
15. Unhandled internal error in guidefunc The error below simply shows that the inputted COM port number for the serial connection is not available 78 J cume gt 2 Unhandled internal error in guidefunc Error using gt lt a href matlab opentoline C Program Files MATLAB R2009a toolbox matlab iofun serial fopen m 72 0 gt serial fop en at 72 lt a gt Port COM14 is not available Available ports COM15 Use INSTRFIND to determine if other instrument objects are connected to the requested device Error in gt fopen at 72 Error in gt WWM gt WWM_ peningFen at 86 Error in gt gui_mainfen at 221 Error in gt WWM at 42 Error in gt guidefunc gt layoutActivate at 1140 Error in gt guidefunc at 14 Ca eee Figure No 3 43 Unhandled Internal Error 2 Error using serial fopen line 72 the COM port used for the Bluetooth connection is not available The error shown in the figure below displays same port number for the available and not available port gt gt WWM Error using serial fopen line 72 Open failed Port COM30 is not available Available ports COM30 Use INSIRFIND to determine if other instrument objects are connected to the requested device Figure No 3 44 Error on opening serial port 3 Error the Bluetooth connection of the laptop or PC to the android application in mobile phone is disconnected 79 APPENDIX B M Files Codes
16. applied by the researchers to provide weather forecast It is applied and implemented using MATLAB based on its equations to perform the correction and prediction state This algorithm uses two data historical and present data The combination of the past and present information in Kalman filter produced possible estimate of the parameters for the prediction of the next state of weather condition The program developed for Kalman Filter Algorithm required the system to be trained by inputting the 80 available data sets or the historical data These data are gathered in Accuweather and presented in a CSV file format The remaining 20 available data are used in the validation of the system A high percentage of available data are allocated in the historical data in order to fully develop the system in the given set of measurements for each parameter The testing of the system is divided into two parts determining the accuracy of measurements of the device and the forecast produced by the system using KF algorithm The results of the first part of the test show that the sensors used for measuring temperature pressure 58 and relative humidity produced almost the same measurements as the instruments used for measuring the actual value of the said parameters Small differences are recorded in the percentage difference of the actual value to the experimental value for each trial in each parameter which leads to a conclusion that the device is capable o
17. are applicable in the system These insights are very useful and are used to improve our thesis Finally we would like to thank our Almighty Father for giving us high hopes and being our source of strength To God be the glory TABLE OF CONTENTS TITLE PAGE APPROVAL PAGE ACKNOWLEDGEMENT TABLE OF CONTENTS LIST OF TABLES LIST OF FIGURES ABSTRACT Chapter 1 INTRODUCTION Chapter 2 REVIEW OF LITERATURE Sensors Weather Instruments GSM Weather Parameters Predictive Algorithm Kalman Filter PIC16F877A Microcontroller Summary Chapter 3 METHODOLOGY Abstract Introduction Methodology Design and Implementation of the Device Applying Algorithm for the Prediction of Weather Conditions Write a Program and Training of the System Validating Testing and Determining the Accuracy of the System Results and Discussion Determining the Accuracy of the Measurement of the Device Temperature Relative Humidity Pressure Determining the Accuracy of System Forecast to Accuweather Forecasts ii iii iv vi 10 11 12 13 13 15 16 17 17 17 19 19 24 29 32 37 37 37 38 39 41 Chapter 4 Chapter 5 and Actual Measurements Hourly Forecasts of Temperature Hourly Forecasts of Relative Humidity Hourly Forecasts of Pressure Statistical Tool Temperature System Forecast and Actual Measurement Relative Humidity System Forecast and Actual Measurement Pressure System Forecast and Actual Measurement
18. are one of the main aspects that may help save lives or prevent damage which could be avoided by preventive arrangements The warning will be based on the predicted weather due to different parameters at a specific date or time Furthermore the device will use an algorithm in predicting the weather by analyzing vital parameters such as probabilities of temperature atmospheric pressure and relative humidity Data regarding weather conditions on a specific locality from PAGASA will be monitored in order to produce accurate and valid weather forecast Since the weather is continuous multidimensional dynamic and chaotic and hence difficult to predict the results will not be very accurate Due to these circumstances weather is observed to be changing simultaneously which leads to a conclusion that forecast can be out of date easily Hence the algorithm will determine the limits of the predicted weather condition rather than its accuracy Moreover Quezon City will be the chosen locality wherein data measured by the sensors located in the said locality will be processed using an algorithm that is not being used by PAGASA such as the Kalman filter The wireless device would basically send the data measured using the sensors into the data logger After that measured parameters will be used in the application of the algorithm for prediction purpose Using the algorithm parameters used by PAGASA can also be predicted These include rainfall maximum and m
19. electronic signal containing the trends of the pressure is reported for a computer which is then plotted in a computer monitor instead of using special graph paper in tracking the change in pressure 2 3 GSM Suriya Shopna stated that GSM which stands for Global System for Mobile Communications is developed by the European Telecommunications Standards Institute ETSD It is also a set which is usually used to describe the protocols used by mobile phones for the second generation digital cellular networks Also it was developed in order to replace first generation 1G analog cellular networks 2013 gt Computer RS 232 _ Processor Controller based system Communication Interface GSM GPRS Module Power Supply Circuit Figure No 2 2 GSM GPRS Module Framework Image from http www engineersgarage com articles gsm gprs modules GSM can be connected to control panels to provide the following facilities report system events via text messaging to mobile telephones remotely arm disarm and obtain the 11 current status of the alarm system via text messaging high speed modem communication for upload download and backup signalling path for digital communicator Telexecom 2 4 Weather Parameters The PAGASA deals with the measurement of almost fifteen 15 weather parameters Weather parameters are determined directly by human observation by instruments or by a TEMPERATURE GUSTINESS TSLTG
20. ende DAY MAX MIN MEAN DRY WET DEW RHH VP STN PRESS MSLP RAINFLL CLD SUNDUR PAEWOIRAVESPD SPD DIR TIME ra ra PC BULB C euw PTC mbs mbs mbs AMT eri lokt min dept mps mps Lp gmt combination of both These parameters are shown below PAGASA described the parameters as follows maximum temperature is the maximum temperature in C recorded for the day which occurs in the early afternoon while the minimum temperature is the minimum temperature in C recorded for the day usually occurring during the early hours of the morning before sunrise To compute for the mean temperature get the average of the maximum and minimum temperature The dry bulb temperature gave the air temperature in C at the time of observation and the wet bulb temperature gives the temperature in C that an air parcel would have if cooled to saturation at constant pressure by evaporating water in it The dew point temperature is the temperature in C at a given pressure wherein the air must cool down for it to saturate Also this parameter is the temperature when the moisture begins to condense causing the formation of dew on objects 12 Relative humidity is the ratio of the amount of water vapor actually in the air to the maximum amount the air can hold at that temperature Schlatter T 2010 Vapor pressure must also be included that denotes the partial pressure of water vapor in the atmosphere As the water evaporates additional water vapor i
21. file of the Kalman Filter program using a MATLAB software 34 Note The version of MATLAB to be used in executing the M file must support the table property 3 Select the CSV file that contains the 80 available data that will be used for training the system See Appendix A for more information on a CSV file as to i nl SO Figure No 3 9 Selection of CSV file 4 Open the Android Application and establish the Bluetooth connection between the laptop and cell phone 5 Click the Start Data Acquisition button found in the bottom right portion of the program Note The recording time of data for system forecast must be an hour before the forecast time stated in the table to have consistency in the time of two forecasts 6 Collect and tabulate the system forecast by having a column name in order such as Forecast Time Accuweather Forecast unit System Forecast unit Actual Measurement unit and Percent Difference Make at least ten hours of continuous simulation of the system 35 Note The collection of measurements for the two forecasts system forecast and Accuweather forecast must be every hour and the actual measurement is collected in the Accuweather site by clicking the Now tab in the same window for Hourly Forecasts during the time stated in the forecast time Plot the system and Accuweather forecast and actual measurements to visually represent the results of the three sets of parameters Compute th
22. kRep PredH zeros kRep lastPT 0 lastPP 0 lastPH 0 aVar 0 pVar 1 fori 1 kRep kVar pVar pVar sDevT aVar aVar kVar DataT i aVar pVar 1 kVar pVar PredT i aVar end lastPT pVar aVar 0 pVar 1 88 fori 1 kRep kVar pVar pVar sDevP aVar aVar kVar DataP i aVar pVar 1 kVar pVar PredP i aVar end lastPP pVar aVar 0 pVar 1 fori 1 kRep kVar pVar pVar sDevH aVar aVar kVar DataH i aVar pVar 1 kVar pVar PredH i aVar end lastPH pVar h subplot 1 1 1 Parent handles pniGrapher hold on subplot 1 1 plot DataT b plot PredT r legend Actual Data Predicted Data xlabel Samples ylabel Temperature c drawnow axes h VeDAlAACH BOCK eas Ee OE ereen eeen s serial COM31 BaudRate 9600 TimeOut 5 89 Terminator 13 fopen s s ReadAsyncMode manual s2 serial COM33 BaudRate 9600 TimeOut 5 OutputBufferSize 1024 Terminator 13 fopen s2 s2 ReadAsyncMode manual format the table colNames Temperature Pressure Humidity Predicted T Predicted P Predicted H Heat Index colData 0 0 0 0 0 0 0 set handles tbIDataLogs data colData ColumnName colNames set handles tbIDataLogs ColumnWidth 80 80 80 80 80 80 120
23. of forecasting and involves average statistics that are gathered in many years The Analog Method is a forecasting method that is more complicated compared to the Climatology Method This method basically examines the present forecast and compares it to the past forecasts which actually look familiar or form an analogy Basically the forecaster would assume that the forecast would behave in the same as the weather did in the past On the other hand Numerical Weather Prediction NWP is an algorithm the that uses a computer for forecasting Forecast models run on supercomputers which show the prediction on weather parameters such as temperature humidity pressure rainfall and wind The features predicted by the supercomputers which interact with each other are examined by the forecaster and used to produce the present weather There are some downsides in the existing prediction method The said method is only effective when most of the data gathered are the same with the expected data in a specific time of the year Thus if the values were different at that time climatology method will not succeed which will result in a difference between the previous and current time In addition NWP s equation in predicting certain parameters leads to a low precision and low accuracy especially when the previous data are completely unknown From the dissertation of Hester Gerbrecht Marx in 2008 Forecasts generated by Numerical Weather Prediction NWP model
24. presents the differences between the measured and true values of humidity These differences are within the acceptable range of values since small differences are calculated This proves that the humidity sensor used to measure the parameter in the device produced almost same performance in measuring humidity as that of the Accuweather The Accuweather website stated that it provides a commercially accepted measurement of certain parameters which can be used for testing purposes 38 Relative Humidity True Value vs Measured Value 120 100 oo oO Measured Value H True Value gt oO 9 gt 60 E N oo 123 4 5 67 8 9 10 Trials Figure No 3 11 Relative Humidity True Value vs Measured Value Pressure Atmospheric pressure is defined as the force per unit area exerted against a surface by the weight of the air molecules above that surface WW2010 2010 The measured and true value of pressure change only in small part due to the fact that this parameter does not easily alter its value unlike other parameters The percentage difference in each trial is close to zero percent suggesting that the device can accurately assess the pressure in a given location Table No 3 4 Tabulated Data for Pressure TRIAL Measured Value True Value Percentage Difference kPa kPa kPa 1 100 9295 100 9 0 02923 2 100 9297 100 9 0 02943 3 100 9297 100 9 0 02943 4 100 9300
25. same for the two measurements system forecast and actual measurements Alternative Hypothesis The mean temperature of the weather condition differs for the two measurements system forecast and actual measurements The computation of important parameters are shown below 50 2 2 2 2 Standard Error of the Difference se sos tE EES re 0288 Ny na 10 10 2 ny n2 ia Degrees of Freedom DF df 3 737 10 2 2 st 3 n nz Po Lai Kar n 1 1 n2 1 95 Confidence Interval for the Difference 0 6427 0 6407 T Value T 0 0035 7 3 ny N2 P Value p 0 9972 Observation A two tailed unequal variance test was conducted to determine the difference of the temperature forecast produced by the system and actual measurement The result showed that there was not a significant difference between the system forecast M 26 901 SD 0 2505 and the actual measurement M 26 9 SD 0 8756 two sample t 10 0 0035 p 0 9972 The direction of the difference in the sample means is reflected by the sign positive or negative of the t value At the 0 05 significance level and 95 confidence level it turned out that the null hypothesis is true so fail to reject the null hypothesis The mean temperature of the weather condition is same for the two measurements system forecast and actual measurements Test 2 Relative Humidity System Forecast and Actual Measurement The value for m
26. system forecast on pressure were almost equal to the Accuweather forecast as observed in the table 3 9 The present measurement of the pressure affects the prognosis for the next hour Thus there are small difference between the Accuweather and System Forecast Also the test showed that the algorithm may reach a certain point wherein it will be difficult to converge or estimates the present estimate value due to the consistent values in the measured parameters 47 Table No 3 10 Comparison of System Pressure Forecast to Actual Pressure Measurement Forecast Time Actual Pressure System Forecast Percent Measurement Difference 2 00 PM 101 2 100 79 0 41 3 00 PM 101 1 100 82 0 28 4 00 PM 101 0 100 80 0 20 5 00 PM 101 0 100 80 0 20 6 00 PM 101 1 100 84 0 26 7 00 PM 101 1 100 85 0 25 8 00 PM 101 2 100 89 0 31 9 00 PM 101 3 100 92 0 38 10 00 PM 101 3 100 92 0 38 11 00 PM 101 3 100 94 0 36 Average Percentage Difference 0 303 Similar to the previous table the system forecasts on pressure were almost similar to the actual pressure for any given time during the testing period In this test it is expected that relatively minor differences will be calculated pressure does not change its value as quickly as other parameters do As a result this visual representation of the differences between the system forecast to the Accuweather forecast and actual measurements were almost equal to
27. the COM port Figure No 3 31 Add COM Port 5 A verification message will be displayed in the Taskbar informing that the device driver software was installed successfully and the COM port for Bluetooth is displayed Then click OK in the Bluetooth Settings window _ Standard Serial over Bluetooth link COM13 gt X Device driver software installed successfully Figure No 3 32 Standard Serial over Bluetooth link 71 Note This COM port will be used and declared in the MATLAB program for Weather Prediction using Kalman Filter Assign this port number in the COM port declaration of the code VTS s serial COM13 176 BaudRate 9600 177 TimeOut 5 178 Terminator 13 319 fopenis 180 s ReadAsyncMode manual Figure No 3 33 Modified COM Port 3 User s Manual 1 Connect the USB to serial port in the serial port of the device the small one See figure below Figure No 3 34 Zigbee Serial Connection 2 In the same device connect the two USB cable to the USB port of the PC or laptop Be sure that the driver is installed properly The red LED will blink to signify that it is ready to use 72 SSK ten 2 Figure No 3 35 Complete Set up of Zigbee 3 Plug the other device the larger one A red LED will light up first and then the green LED will start to blink Wait until the light of the green LED is steady which signifies that the device is ready
28. the measured temperatures which is essential in calculating the forecast as discussed earlier in this chapter However the Accuweather forecast showed fluctuations in its temperature forecast resulting to an observation that during the test the system forecast values were closer to the actual measurements Hourly Forecasts of Relative Humidity Relative humidity is an important weather parameter used in forecasting weather Humidity indicates the possibility of dew fog or precipitation When the recorded measurement 43 of humidity is high it makes people feel hotter outside in the summer because it reduces the effectiveness of sweating to cool the body by preventing the evaporation of perspiration from the skin This effect is calculated in a heat index table McEntee et al 2009 Table No 3 7 Comparison of Humidity Forecast between the System and Accuweather Forecast Time Accuweather Forecast System Forecast Percent Difference 2 00 PM 68 88 11 25 76 3 00 PM 66 73 23 10 39 4 00 PM 71 90 09 23 70 5 00 PM 64 89 25 32 95 6 00 PM 69 85 58 21 45 7 00 PM 75 92 87 21 29 8 00 PM 79 91 42 14 58 9 00 PM 93 96 89 4 10 10 00 PM 90 97 17 7 66 11 00 PM 84 98 45 15 84 Average Percent Difference 17 712 Table 3 7 shows the forecasts from two different sources the Accuweather site and the system using KF algorithm The average difference calculated is equal to 17 772
29. to use otherwise it is in OFF mode To turn it ON switch the toggle switch at the back of the device Figure No 3 36 Device for Measuring Weather Parameters 73 4 Open the MATLAB gt Click File menu gt Click New gt Select GUI gt Select Open Existing GUI then browse the Fig File of the program The program is provided on the CD in the Wireless Weather folder A MATLAB 7 8 0 R2009a Fite Edit View Debug Parallel Desktop Window Help N ew gt Blank M File Ctrl N_ s User Doc Open Ctrl 0 Function M File Close Current Directory Ctrl W Class M File Figure Import Data Variable Save Workspace As Ctrl S Model Set Path GUI Preferences Deployment Project Print Ctrl P 1 C relessWeather WWM m 2 Ci b guide guidefunc m 3 C relessWeather WWM m 4 C Rar D100 381 WWM m Exit MATLAB Ctrl Q U New to MATLAB Watch this Video see Demos or read Getting Started z Figure No 3 37 Opening an Existing GUI file 5 A window will appear like the figure below 74 of wwMiig En 7 2 6 File Edit View Layout Tools Help Aii Eel ERAAI e Oe gt T T T T T T Weather Forecast using Kalman Filter Algorithm with Warning System via GSM and e Android Application Panel Mapua Institute of Technology School of EECE BS Computer Engineering Thesis Members Capapas Trexia D Paule Jezzelle Joyce Clarice M Vista Rochelle
30. 0 kVar pVar pVar sDevT aVar aVar kVar presVal i aVar pVar 1 kVar pVar PredP i aVar rowData i 5 cellstr num2str aVar 2f end lastPP pVar 94 sRW sCL size PredH aVar PredH sRW pVar lastPH PredH zeros 10 fori 1 10 kVar pVar pVar sDevT aVar aVar kVar humiVal i aVar pVar 1 kVar pVar PredH i aVar rowData i 6 cellstr num2str aVar 2f end lastPH pVar get Heat Index fori 1 10 tRounded round PredT i hRounded round PredH i if tRounded 27 if hRounded 40 rowData i 7 cellstr sprintf CAUTION 27 elseif hRounded 45 rowData i 7 cellstr sprintf CAUTION 27 elseif hRounded 50 rowData i 7 cellstr sprintf CAUTION 27 elseif hRounded 55 rowData i 7 cellstr sprintf CAUTION 28 elseif hRounded 60 95 rowData i 7 cellstr sprintf CAUTION 28 elseif hRounded 65 rowData i 7 cellstr sprintf CAUTION 28 elseif hRounded 70 rowData i 7 cellstr sprintf CAUTION 29 elseif hRounded 75 rowData i 7 cellstr sprintf CAUTION 29 elseif hRounded 80 rowData i 7 cellstr sprintf CAUTION 30 elseif hRounded 85 rowData i 7 cellstr sprintf CAUTION 30 elseif hRounded 90 rowData i 7 cellstr sprintf CAUTION 31 elseif hRounded 95 rowData i 7 cellstr sprintf CAUTION 31 elsei
31. 0 hUpper 6853 hLower 6728 humBase 40 elseif humiVal sampCtr gt 6600 base 50 hUpper 6728 hLower 6600 humBase 50 elseif humiVal sampCtr gt 6468 base 60 hUpper 6600 hLower 6468 humBase 60 elseif humiVal sampCtr gt 6330 base 70 hUpper 6468 hLower 6330 humBase 70 92 elseif humiVal sampCtr gt 6186 base 80 hUpper 6330 hLower 6186 humBase 80 elseif humiVal sampCtr gt 6033 base 90 hUpper 6186 hLower 6033 humBase 90 else humBase 100 end switch humBase case 101 humiVal sampCtr 0 case 100 humiVal sampCtr 100 otherwise freqH 6660 freqDiff hUpper freqH freqInterval hUpper hLower humiVal sampCtr freqDiff freqInterval 10 humBase end 93 rowData sampCtr 1 cellstr num2str tempVal sampCtr 2f rowData sampCtr 2 cellstr num2str presVal sampCtr 2f rowData sampCtr 3 cellstr num2str humiVal sampCtr 2f if sampCtr lt sampTotal sampCtr sampCtr 1 else compute Kalman Predictions sRW SCL size PredT aVar PredT sRW pVar lastPT PredT zeros 1 fori 1 10 kVar pVar pVar sDevT aVar aVar kVar tempVal i aVar pVar 1 kVar pVar PredT i aVar rowData i 4 cellstr num2str aVar 2f end lastPT pVar sRW sCL size PredP aVar PredP sRW pVar lastPP PredP zeros 1 fori 1 1
32. 2011 Zigbee Setting Standards for Energy Efficient Control Networks Volume 1 1 11 12 Taranovich S 2011 Humidity Sensors and Signal Conditioning Choices Online Available http www digikey com us en techzone sensors resources articles humidity sensors and signal conditioning choices html Accessed September 25 2013 The MathWorks Inc 2013 MATLAB Product Description The Language of Technical Computing Online Available http www mathworks com help matlab learn_matlab product description html WW2010 Department of Atmospheric Sciences University of Illinois at Urbana Champaign Atmospheric Pressure Online Available http ww2010 atmos uiuc edu Gh guides mtr prs def rxml 57 Chapter 4 CONCLUSIONS A measuring and monitoring device for weather parameters such as temperature pressure and relative humidity is designed and implemented in this study The objective is achieved by using a sensor for each parameter and is integrated to the microcontroller Such sensors are the LM35 for temperature capacitive humidity kit for humidity and MPX114AP for measuring the pressure Zigbee technology is also included which allows the transmission of measurements wirelessly from the device to the laptop or PC Also PIC16F877A microcontroller is used for collecting and processing of the gathered measurements Since the device is real time it sends measurements every 10 seconds The Kalman Filter Algorithm is proposed and
33. 43 elseif hRounded 50 rowData i 7 cellstr sprintf DANGER 46 elseif hRounded 55 107 rowData i 7 cellstr sprintf DANGER 48 elseif hRounded 60 rowData i 7 cellstr sprintf DANGER 51 elseif hRounded 65 rowData i 7 cellstr sprintf EXTREME DANGER 55 elseif hRounded 70 rowData i 7 cellstr sprintf EXTREME DANGER 58 else rowData i 7 cellstr sprintf EXTREME DANGER end elseif tRounded 38 if hRounded 40 rowData i 7 cellstr sprintf DANGER 43 elseif hRounded 45 rowData i 7 cellstr sprintf DANGER 46 elseif hRounded 50 rowData i 7 cellstr sprintf DANGER 49 elseif hRounded 55 rowData i 7 cellstr sprintf DANGER 52 elseif hRounded 60 rowData i 7 cellstr sprintf EXTREME DANGER 55 elseif hRounded 65 108 rowData i 7 cellstr sprintf EXTREME DANGER 59 else rowData i 7 cellstr sprintf EXTREME DANGER end elseif tRounded 39 if hRounded 40 rowData i 7 cellstr sprintf DANGER 46 elseif hRounded 45 rowData i 7 cellstr sprintf DANGER 49 elseif hRounded 50 rowData i 7 cellstr sprintf DANGER 52 elseif hRounded 55 rowData i 7 cellstr sprintf EXTREME DANGER 55 elseif hRounded 60 rowData i 7 cellstr sprintf EXTREME DANGER 59 else rowData i 7 cellstr sprintf EXTREME DANGER end elseif tRoun
34. 84 AT 13 DelayMS 2000 LEDO1 0 DelayMS 500 LEDO1 1 SerOut gsmOut 84 AT CIURC 0 13 SerIn gsmIn 84 1500 subs_SMS_InitFailed Wait OK DelayMS 500 LEDO1 0 DelayMS 500 LEDO1 1 SerOut gsmOut 84 AT CNMI 3 1 0 0 0 13 SerIn gsmIn 84 1500 subs_SMS_InitFailed Wait OK DelayMS 500 LEDO1 0 DelayMS 500 LEDO1 1 82 SerOut gsmOut 84 AT CFUN 1 13 SerIn gsmIn 84 1500 subs_SMS_InitFailed Wait OK DelayMS 500 LEDO1 0 DelayMS 500 LEDO1 1 SerOut gsmOut 84 AT CMGF 1 13 SerIn gsmIn 84 1500 subs_SMS_InitFailed Wait OK DelayMS 500 LEDO1 0 DelayMS 500 LEDO1 1 SerOut gsmOut 84 AT CMEE 0 13 SerIn gsmIn 84 1500 subs_SMS_InitFailed Wait OK DelayMS 500 LEDO1 0 DelayMS 500 LEDO1 1 SerOut gsmOut 84 AT CSDH 0 13 SerIn gsmIn 84 1500 subs_SMS_InitFailed Wait OK DelayMS 500 LEDO1 0 DelayMS 500 LEDO1 1 SerOut gsmOut 84 ATE1 13 SerIn gsmIn 84 1500 subs_SMS_InitFailed Wait OK DelayMS 500 LEDO1 0 DelayMS 500 LEDO1 0 For gCtr 10 To 1 Step 1 LEDO1 LEDO1 DelayMS 500 Next gCtr For gCtr 0 To 10 uNum1 gCtr 0 Next gCtr 83 uNum1 11 0 retrieve admin number SerOut gsmOut 84 AT CPBF 34 ADMIN 34 13 SerIn gsmIn 84 3500 subs_SMS_InitFailed Wait CPBF Wait 34 Str uNum1 11 Wait OK DelayMS 2000 LEDO1 1 Return subs SMS Send sj SS Start LED
35. 86 0 52 8 00 PM 27 26 78 0 82 9 00 PM 26 26 71 2 69 10 00 PM 26 26 60 2 28 11 00 PM 26 26 54 2 06 Average Percentage Difference 1 684 It is observed in Table 3 6 that the actual temperature has a range within 26 29 Celsius during the test which was accepted in the range of measurements for a cloudy day The computed average difference between the two variables system forecast and actual measurement is equal to 1 684 which is lower compared to the average difference of the system forecast and Accuweather forecast These slight differences were acceptable since the location of the test was not on the same location of the weather station of Accuweather 42 Temperature N ui Accuweather Forecast C e N nN O E System Forecast C r oO pen o 8 z v 3 Im 2 E He te Actual Temperature Measurement C K X S S INT SS FS ESE S ew Forecast Time Figure No 3 13 Temperature Figure 3 13 presents the visual representation of the system forecast and the Accuweather forecast and actual measurement It shows that the system forecast is close to the actual measurement at any given time during the time of testing compared to the Accuweather forecast It is observed that the time between 2 00 PM to 11 00 PM the system forecast temperature red line showed an increasing change in its value considering its previous value from the previous hour It is caused by
36. DANGER 41 elseif hRounded 95 rowData i 7 cellstr sprintf DANGER 42 elseif hRounded 100 rowData i 7 cellstr sprintf DANGER 44 end elseif tRounded 31 if hRounded 40 rowData i 7 cellstr sprintf CAUTION 31 elseif hRounded 45 rowData i 7 cellstr sprintf CAUTION 32 elseif hRounded 50 rowData i 7 cellstr sprintf EXTREME CAUTION 33 elseif hRounded 55 100 rowData i 7 cellstr sprintf EXTREME CAUTION 34 elseif hRounded 60 rowData i 7 cellstr sprintf EXTREME CAUTION 35 elseif hRounded 65 rowData i 7 cellstr sprintf EXTREME CAUTION 36 elseif hRounded 70 rowData i 7 cellstr sprintf EXTREME CAUTION 38 elseif hRounded 75 rowData i 7 cellstr sprintf EXTREME CAUTION 39 elseif hRounded 80 rowData i 7 cellstr sprintf DANGER 41 elseif hRounded 85 rowData i 7 cellstr sprintf DANGER 43 elseif hRounded 90 rowData i 7 cellstr sprintf DANGER 45 elseif hRounded 95 rowData i 7 cellstr sprintf DANGER 47 elseif hRounded 100 rowData i 7 cellstr sprintf DANGER 49 end elseif tRounded 32 if hRounded 40 101 rowData i 7 cellstr sprintf CAUTION 32 elseif hRounded 45 rowData i 7 cellstr sprintf EXTREME CAUTION 33 elseif hRounded 50 rowData i 7 cellstr sprintf EXTREME CAUTION 34 elseif hRounded 55 rowD
37. Data Predicted Data xlabel Samples ylabel Pressure kPa case 3 subplot 1 1 1 cla plot humiVal b plot PredH r legend Actual Data Predicted Data xlabel Samples ylabel Humidity end grid on drawnow set handles btnSendData Enable on btString sprintf DATA_ 08 2f_ 08 2f_ 08 2f n PredT 10 PredP 10 PredH 10 fprintf s2 btString colNames Temperature Pressure Humidity Predicted T Predicted P Predicted H Heat Index colData tempVal 20 presVal 20 humiVal 20 colData rowData set handles tbIDataLogs data rowData ColumnName colNames tempVal zeros 1 presVal zeros 1 humiVal zeros 1 sampCtr 1 112 end end else tempVal zeros 1 presVal zeros 1 humiVal zeros 1 subplot 1 1 1 cla colNames Temperature Pressure Humidity Predicted T Predicted P Predicted H Heat Index colData 0 0 0 0 0 0 0 set handles tbIDataLogs data colData ColumnName colNames end axes h end YoDataAcd Bock He function varargout WWM_OutputFcn hObject eventdata handles varargout 1 handles output function btnGetData_Callback hObject eventdata handles global AcqFlag if AcqFlag AcqFlag 1 set handles btnGetData String Stop Data Acquisition else AcqFlag 0 set handles btnGetData String Start Data Acquisition end function
38. HT Arrows show the wind direction A Ss kmph 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 b Cloud Cover on Saturday 02 Mar at 8pm PHT Surface Wind on Sunday 03 Mar at 2am PHT Mean Sea Level Pressure Isobars kmph 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 c Surface Wind on Sunday 03 Mar at 2am PHT Image a b and c from http www weather forecast com 5 30554 0673 E R 580 D Ba Ce BEN Er G 586 5 5673 a5 12Z 02 Mar 2013 500 hPa d Upper air map of Southeast asia Image from http weather uwyo edu upperair uamap html 2 1 Sensors From an article What forces affect our weather in Annenberg Learner and Foundation 2013 it has been stated that devices are required in order to view large weather system specifically on a worldwide scale From this point satellites are considered as invaluable which shows large weather events and other global weather systems For every satellite two types of sensors were used such as the imager and the sender which is a visible light sensor and an infrared sensor that reads temperature respectively 2 2 Weather Instruments In the Climate Education for K12 by McKemy weather instruments are used to measure weather parameters and to describe the local weather Thus in measuring temperature a thermometer is used An electronic temperature sensor is used to measure the outside air temperature Devices that use
39. O1 0 For gCtr 1 To 3 LEDO1 LEDO1 DelayMS 500 Next gCtr LEDO1 1 SerOut gsmOut 84 AT 13 DelayMS 500 LEDO1 0 DelayMS 500 LEDO1 1 SerOut gsmOut 84 AT 13 DelayMS 500 LEDO1 0 DelayMS 500 LEDO1 1 SerOut gsmOut 84 AT CMGS 34 Str uNum1 11 34 13 84 DelayMS 500 LEDO1 0 DelayMS 500 LEDO1 1 SerOut gsmOut 84 Data Results 13 SerOut gsmOut 84 P Temperature Str pTemp 8 13 SerOut gsmOut 84 P Pressure Str pPres 8 13 SerOut gsmOut 84 P Humidity Str pHum 8 13 LEDO1 0 DelayMS 500 LEDO1 1 SerOut gsmOut 84 26 DelayMS 500 LEDO1 1 sj SMS Out Return subs SMS InitFailed LEDO1 1 While 1 1 LEDO1 LEDO1 DelayMS 250 Wend Return End In programming the microcontroller the researchers first define XTAL which serves as the crystal oscillator of PIC16F877A It is the oscillator that produces electrical oscillations at a frequency determined by the physical characteristics of a piezoelectric quartz crystal The 85 FreeDictionary After that they declared the following parameters Adin_Res 10 Adin_Tad FRC and Adin_Stime 50 which signifies a 10 bit result required FRC as a chosen oscillator and allows 50us sample time respectively The next thing that must be implemented is to configure the following input ports declared as TRISA TRISB TRISC TRISD and TRISE After that the microcontroller would be able to read
40. Weather Forecast using Kalman Filter Algorithm with Warning System via GSM and Android Application by Trexia D Capapas Jezzelle Joyce Clarice M Paule Rochelle Lynn S Vista A Thesis Report Submitted to the School of EE ECE CoE In Partial Fulfilment of the Requirements for the Degree Bachelor of Science in Computer Engineering Mapua Institute of Technology March 2014 Bachelor of Science in Computer Engincering u Joshua Panci Meoiber 1 Panel 2 he Committee Chat and accepted by the School of Wheetrical Dlecmonicn ani hh Gat papar ie be eas of the thests requirement for the degree in Bachelor of Computer Enginscring as fulfillment Science in Computer Engineering Fikah S eae Schoed of FECT ACKNOWLEDGEMENT First and foremost we would like to take this opportunity to express our deep sense of gratitude to our advisor Dr Felicito S Caluyo for his exemplary guidance constant encouragement throughout our thesis for his patience and immense knowledge We would also like to thank Engr Noel B Linsangan chairman of CpE course for the intuitive comments and guidance throughout the checking and editing of documentation We would also take this opportunity to thank the Accuweather company for providing online forecasts which helps us to pursue and to complete this study especially on the testing part of the system We would also like to thank our panels who have given us recommendations that
41. able MsgReceived from ServiceAStream strBuffer strBuffer amp BTService MsgReceived If strBuffer Contains DATA AND strBuffer StartsWith DATA False Then prepare prefix data strBuffer DATA_ BTService MsgReceived Log strBuffer Else If strBuffer Contains DATA_ AND strBuffer StartsWith DATA_ True Then Get Complete Data If strBuffer Contains DATA_ AND strBuffer EndsWith Chr 13 True Then Log strBuffer ParseData strBuffer End If End If End Sub Sub TestData_Click DummyTest End Sub Sub DummyTest preTemp Rnd 1 1000 119 prePress Rnd 1 1000 preHumid Rnd 1 1000 strBuffer DATA_ amp preTemp amp _ amp prePress amp _ amp preHumid amp _ amp Chr 13 ParseData End Sub Sub ParseData tempData Regex Split _ strBuffer preTemp tempData 1 prePress tempData 2 preHumid tempData 3 IbITemperature Text preTemp IbIPressure Text prePress IblHumidity Text preHumid strBuffer End Sub Sub Timer1_Tick IbIDateTime Text Date Time amp DateTime Date DateTime Now amp amp DateTime Time DateTime Now End Sub 120
42. ata i 7 cellstr sprintf EXTREME CAUTION 36 elseif hRounded 60 rowData i 7 cellstr sprintf EXTREME CAUTION 37 elseif hRounded 65 rowData i 7 cellstr sprintf EXTREME CAUTION 39 elseif hRounded 70 rowData i 7 cellstr sprintf DANGER 40 elseif hRounded 75 rowData i 7 cellstr sprintf DANGER 42 elseif hRounded 80 rowData i 7 cellstr sprintf DANGER 44 elseif hRounded 85 rowData i 7 cellstr sprintf DANGER 47 elseif hRounded 90 rowData i 7 cellstr sprintf DANGER 49 elseif hRounded 95 102 rowData i 7 cellstr sprintf EXTREME DANGER 51 elseif hRounded 100 rowData i 7 cellstr sprintf EXTREME DANGER 54 end elseif tRounded 33 if hRounded 40 rowData i 7 cellstr sprintf EXTREME CAUTION 34 elseif hRounded 45 rowData i 7 cellstr sprintf EXTREME CAUTION 35 elseif hRounded 50 rowData i 7 cellstr sprintf EXTREME CAUTION 36 elseif hRounded 55 rowData i 7 cellstr sprintf EXTREME CAUTION 38 elseif hRounded 60 rowData i 7 cellstr sprintf DANGER 40 elseif hRounded 65 rowData i 7 cellstr sprintf DANGER 41 elseif hRounded 70 rowData i 7 cellstr sprintf DANGER 43 elseif hRounded 75 rowData i 7 cellstr sprintf DANGER 46 elseif hRounded 80 103 rowData i 7 cellstr sprintf DANGER 48 elseif hRound
43. choose Yes Then click Next 6 Inthe File Installation Key window choose I have the File Installation for my license 4 File Installation Key lele M ATLAB Provide File Installation Key SIMULINK R2009a Ido not have the File Installation Key Help me with the next steps You may have received the File Installation Key from the Administrator of the license or from the MathWorks Web site Figure No 3 18 File Installation Key Window 63 7 In order to get the installation key open the crack folder in the CD Open the notepad file named install T install Notepad U Erm File Edit Format View Help we offer you two ways to license matlab r2009a standalone 1 choose install manually without using the internet 2 enter the file installation key 11111 11111 02011 44270 3 use icense_standalone dat when asked for license file network choose install manually without using the internet enter the file installation key 11111 11111 02011 06717 if neccessary install license manager use license_server dat when asked for license file Figure No 3 19 File Installation Key Copy the file installation key on the space provided Then click Next 9 Choose Typical Click Next Specify the folder where you want to install MATLAB Click Next M ATLAB Specify installation folder SIMULINK Enter the full path to the installation
44. ded 40 if hRounded 40 rowData i 7 cellstr sprintf DANGER 48 109 elseif hRounded 45 rowData i 7 cellstr sprintf DANGER 51 elseif hRounded 50 rowData i 7 cellstr sprintf EXTREME DANGER 55 elseif hRounded 55 rowData i 7 cellstr sprintf EXTREME DANGER 59 else rowData i 7 cellstr sprintf EXTREME DANGER end elseif tRounded 41 if hRounded 40 rowData i 7 cellstr sprintf DANGER 51 elseif hRounded 45 rowData i 7 cellstr sprintf EXTREME DANGER 54 elseif hRounded 50 rowData i 7 cellstr sprintf EXTREME DANGER 58 else rowData i 7 cellstr sprintf EXTREME DANGER end elseif tRounded 42 if hRounded 40 110 rowData i 7 cellstr sprintf EXTREME DANGER 54 elseif hRounded 45 rowData i 7 cellstr sprintf EXTREME DANGER 57 else rowData i 7 cellstr sprintf EXTREME DANGER end elseif tRounded 43 if hRounded 40 rowData i 7 cellstr sprintf EXTREME DANGER 57 else rowData i 7 cellstr sprintf EXTREME DANGER end else rowData i 7 cellstr sprintf UNDEFINED end end switch grpFlag case 1 subplot 1 1 1 cla plot tempVal b plot PredT r 111 legend Actual Data Predicted Data xlabel Samples ylabel Temperature c case 2 subplot 1 1 1 cla plot presVal b plot PredP r legend Actual
45. e percentage difference to determine the accuracy of system forecast to the Accuweather forecast and system forecast to actual measurements Sample1i Sample2 Sample1 Sample2 2 PD x 100 3 17 Compare the means of the two forecasts using a statistical tool the calculation of t test must be performed Below are the steps in dealing with t test a State the null and alternative hypotheses b Determine the significance level c Compute for T value d Calculate the Degrees of Freedom e Determine the p value 36 Results and Discussion 3 4 1 Determining the Accuracy of the Measurement of the Device Temperature Table No 3 2 Tabulated Data for Temperature TRIAL Measured Value True Value Percentage Difference CO CO 1 43 4320 43 5 0 1564 2 42 8421 42 7 0 3322 3 40 5400 40 5 0 0987 4 37 8000 38 0 0 0528 5 36 1126 35 9 0 5905 6 34 6480 34 7 0 1500 7 33 1840 32 9 0 8595 8 32 6960 32 6 0 2940 9 31 7290 31 5 0 7244 10 29 2800 29 0 0 9609 Table 3 2 shows the collected data in testing the accuracy of the temperature measurement of the device to the digital thermometer The use of blower allowed the temperature in the box to be controlled Based on Table 3 2 the two measurements produced almost the same value for every trial resulting in small percentage difference This indicates that the device can produce accurate measurements of the temperature like t
46. e predicted and the actual values obtained from measurements Keywords Kalman filter weather forecasting Chapter I INTRODUCTION The weather is basically a condition of the atmosphere at a certain time and place and is measured by several parameters that contribute to it Some of the weather parameters are temperature dry and wet bulb amount of rainfall dew point humidity pressure and wind This study will focus on forecasting a weather condition at a certain time in the future wherein a warning system will be used to inform the people of incoming weather condition in the specific area Forecasting deals with the prediction of the upcoming weather condition and varies from different time and location Weather forecasts can be made using statistical dependencies which use predictor variables to extrapolate the weather situation or simulations to gain information about the possible state of the weather in the future The prediction of weather conditions will be based on the measurements gathered from the sensors of different parameters that affect weather The device containing the sensors will be placed in one location in the chosen locality Weather forecasting is considered as one of the objectives in a research dealing with the atmosphere K Lerner and B Lerner stated in the year 2003 that some methods used in forecasting are Climatology Method Analog Method and Numerical Weather Prediction NWP The Climatology Method is a simple way
47. e to have the capability to measure these three basic weather parameters approximately every 10 seconds for the real time purposes The measurements from the sensors will then be sent wirelessly using Zigbee technology The data logger will be implemented using MATLAB to read collect and gather data that will be received from the sensors The data logger will display 10 samples of measurement from each parameter 19 Temperature Sensor Pressure Sensor Rel Humidity Sensor Figure No 3 1 Block Diagram of the Device Y Z 5 m SR En EE oo gx a Sa Data Logger The block diagram of the device is shown in Figure3 1 Sensors for each parameter and the Zigbee module for wireless data transfer were connected in the microcontroller The specific sensors and components listed in Table 3 1 are used to design and implement the device The sensors used were affordable and light in size thus making the device cheap and handy Table No 3 1 List of Components Component PIC 16F877A Temperature Sensor LM35 Humidity Sensor Capacitive Humidity Kit Atmospheric Pressure Sensor MPX114AP GSM Module As shown in Figure3 1 the PIC16F877A microcontroller was used to connect the three sensors and the standard Zigbee for the wireless data transfer This microcontroller was used because it has an advantage over other microcontrollers which is easy to use attributable to its
48. e variable P does notequated to a value of 0 because this would mean that there 28 is no noise in the environment which is not true and impractical Letting Po 1 would lead all the consequent X to be zero Step 2 Compute the Kalman Filter estimate and covariance A Compute the Kalman Gain Pe K ER 3 12 B Compute the estimate u X Kilze r 3 13 C Compute the estimate covariance P 1 K PF 3 14 Step 3 Update t t 1 Step 4 Set value for the evolution model prior estimate and prior error covariance P and repeat step 2 A Set 87 2 4 3 15 B Set P7 Pi 3 16 3 3 Write a program and train the system Write a program for the Algorithm The program for weather prediction using Kalman Filter is implemented in MATLAB It is a high level language and interactive environment for numerical computation visualization 29 and programming The MATLAB allows analysis of data developing algorithms and creating models and applications The MathWorks Inc 2013 The step by step procedures in dealing with Kalman Filter that were discussed earlier are implemented in MATLAB The program flowchart implemented for the system is presented in Figure 3 8 Training of the System Accuweather is a company that produces forecasts for places everywhere in the US and other locations worldwide including the Philippines Thus Accuweather is considered as the World s Weather Authorit
49. each other as shown in Figure 3 15 These data proved that using the Kalman filter algorithm the forecast for pressure are similar to what Accuweather forecasts produced and the measurement of the actual data 48 Pressure 101 4 101 3 101 2 101 1 101 100 9 100 8 100 7 100 6 100 5 DD NEE EE ES so D aS Ro NS ES gt AS A oS Accuweather Forecast kPa System Forecast kPa a s wn u pe a Actual Pressure Measurement kPa D ny Forecast Time Figure No 3 15 Pressure 49 Statistical Tool Test 1 Temperature System Forecast and Actual Measurement Using the collected data presented in Table 3 5 the value for mean variance standard deviation and standard error are as follows Table No 3 11 T Test for Temperature Forecast Time Actual Temperature System Forecast Measurement C C 2 00 PM 2 26 86 3 00 PM 29 2122 4 00 PM 27 27 18 5 00 PM 27 27 18 6 00 PM 27 27 08 7 00 PM 27 26 86 8 00 PM 27 26 78 9 00 PM 26 26 71 10 00 PM 26 26 60 11 00 PM 26 26 54 Mean 26 9 26 901 Standard Deviation 0 8756 0 2505 Standard Error Mean 0 277 0 079 A two sided t test is appropriate to use in determining the difference in the average temperature measurements of system forecasts and Accuweather forecasts The hypotheses for this test are as follows Null Hypothesis The mean temperature of the weather condition is
50. ean variance standard deviation and standard error for the two sources are as follows 51 Table No 3 12 T Test for Relative Humidity Forecast Time Actual Humidity System Forecast Measurement 2 00 PM 88 88 11 3 00 PM 69 73 23 4 00 PM 94 90 09 5 00 PM 88 89 25 6 00 PM 83 85 58 7 00 PM 88 92 87 8 00 PM 88 91 42 9 00 PM 94 96 89 10 00 PM 94 97 17 11 00 PM 94 98 45 Mean 88 90 306 Standard Deviation 7 6739 7 3372 Standard Error Mean 2 427 2 32 The hypotheses for this test are as follows Null Hypothesis The mean relative humidity of the weather condition is same for the two measurements system forecast and actual measurements Alternative Hypothesis The mean relative humidity of the weather condition differs for the two measurements system forecast and actual measurements The computation of important parameters are shown below 2 2 2 2 Standard Error of the Difference se se 2 26722 27 3 3574 n N2 10 10 2 2 2 Ss S ny n2 Degrees of Freedom DF df z 17 i i n 1 1 n2 1 95 Confidence Interval for the Difference 9 3894 4 7774 X Y T Value T 0 6868 2 22 5 S 21 22 ny Nz 52 P Value p 0 5014 Observation A two tailed unequal variance test was conducted to determine the difference of the relative humidity forecast produced by the system and actual measuremen
51. ed 85 rowData i 7 cellstr sprintf DANGER 51 elseif hRounded 90 rowData i 7 cellstr sprintf EXTREME DANGER 54 elseif hRounded 95 rowData i 7 cellstr sprintf EXTREME DANGER 57 elseif hRounded 100 rowData i 7 cellstr sprintf EXTREME DANGER end elseif tRounded 34 if hRounded 40 rowData i 7 cellstr sprintf EXTREME CAUTION 35 elseif hRounded 45 rowData i 7 cellstr sprintf EXTREME CAUTION 37 elseif hRounded 50 rowData i 7 cellstr sprintf EXTREME CAUTION 38 elseif hRounded 55 rowData i 7 cellstr sprintf DANGER 40 elseif hRounded 60 rowData i 7 cellstr sprintf DANGER 42 elseif hRounded 65 104 rowData i 7 cellstr sprintf DANGER 44 elseif hRounded 70 rowData i 7 cellstr sprintf DANGER 47 elseif hRounded 75 rowData i 7 cellstr sprintf DANGER 49 elseif hRounded 80 rowData i 7 cellstr sprintf EXTREME DANGER 52 elseif hRounded 85 rowData i 7 cellstr sprintf EXTREME DANGER 55 elseif hRounded 90 rowData i 7 cellstr sprintf EXTREME DANGER elseif hRounded 95 rowData i 7 cellstr sprintf EXTREME DANGER elseif hRounded 100 rowData i 7 cellstr sprintf EXTREME DANGER 54 end elseif tRounded 35 if hRounded 40 rowData i 7 cellstr sprintf EXTREME CAUTION 37 elseif hRounded 45 rowData
52. ed by the calculating percent difference 3 4 1 1 Temperature The steps in the procedure for temperature measurements are listed below 32 1 Set up the devices and the laptop in a room Plug in the device bigger one and connect the other device smaller one in the laptop using USB to serial port cable 2 Place the digital thermometer and the device bigger one within a box 3 Run the M file of the Kalman Filter algorithm using MATLAB 4 Using a blower heat up the temperature inside the box Set the measurement of the true value to 43 Celsius 5 Click the Start Data Acquisition button found in the GUI of the program 6 Record the measured value of the temperature which is exhibited in the data logger of the system Since the program will output 10 samples compute its mean and record under the Measured Value column At the same time record the measurement displayed in the digital thermometer under the True Value column 7 Make 10 trials of the test and record the collected data In the succeeding trials let the temperature inside the box cool down Set the 10 trial to have a measurement of 29 Celsius for the True Value 8 Compute the percentage difference using the equation True Value Measured Value True Value Measured Value 2 100 PD 3 4 1 2 Relative Humidity 1 A similar procedure as in the previous test is used to install the devices and laptop in this test which was taken in Las Pinas City
53. edu edu k12 instruments Accessed March 2 2013 Nilofer Mehta Kalman Filtering of Sensor Data Online Available http web cs dal ca reilly CSC16609 seminars sensors_kalman_filters pdf Noordin K Oon C and M Ismail A Low Cost Microcontroller based Weather Monitoring System Department of Electrical Engineering Faculty of Engineering University of Malaya 50603 Kuala Lumpur Malaysia Segaran Toby Programming Collective Intelligence Building Smart Web 2 0 Applications O Reilly 2007 Simeonov I Killifarev H and Ilarionov R Embedded System for Short term Weather Forecasting CompSysTech 2006 The Daily Weather Forecast Online Available http kidlat pagasa dost gov ph genmet pwf html Yu X Efe M and Kaynak O A Backpropagation Learning Framework for Feedforward Neural Networks Faculty of Informatics and Communication Central Queensland University Australia 2001 61 APPENDIX A Operation s Manual 1 System Requirement gt Operating System Windows 7 64 bit 2 Installation Procedure Installing the software for the drive 1 Open the Other Software folder in the provided CD 2 Double click the CH341SER64Bit 3 A window will be displayed like as shown below P SE DriverSetup X64 amp Device Driver Install Uninstall Select INF File CH341SER INF v nor WCH CN INSTALL __ USB SERIAL CH340 __ 11704 2011 3 3 2011 11 UNINSTALL
54. f hRounded 100 rowData i 7 cellstr sprintf CAUTION 32 end elseif tRounded 28 if hRounded 40 rowData i 7 cellstr sprintf CAUTION 28 elseif hRounded 45 96 rowData i 7 cellstr sprintf CAUTION 28 elseif hRounded 50 rowData i 7 cellstr sprintf CAUTION 28 elseif hRounded 55 rowData i 7 cellstr sprintf CAUTION 29 elseif hRounded 60 rowData i 7 cellstr sprintf CAUTION 29 elseif hRounded 65 rowData i 7 cellstr sprintf CAUTION 30 elseif hRounded 70 rowData i 7 cellstr sprintf CAUTION 31 elseif hRounded 75 rowData i 7 cellstr sprintf CAUTION 31 elseif hRounded 80 rowData i 7 cellstr sprintf CAUTION 32 elseif hRounded 85 rowData i 7 cellstr sprintf EXTREME CAUTION 33 elseif hRounded 90 rowData i 7 cellstr sprintf EXTREME CAUTION 34 elseif hRounded 95 rowData i 7 cellstr sprintf EXTREME CAUTION 35 elseif hRounded 100 97 rowData i 7 cellstr sprintf EXTREME CAUTION 36 end elseif tRounded 29 if hRounded 40 rowData i 7 cellstr sprintf CAUTION 29 elseif hRounded 45 rowData i 7 cellstr sprintf CAUTION 29 elseif hRounded 50 rowData i 7 cellstr sprintf CAUTION 30 elseif hRounded 55 rowData i 7 cellstr sprintf CAUTION 30 elseif hRounded 60 rowData i 7 cellstr sprintf CAUTION 31 elseif hR
55. f measuring and monitoring weather parameters in a specific location The second part of the testing focused on determining the accuracy of Kalman filter in forecasting The calculation of percentage difference and use of statistical tool such as t test allowed the determination of the difference of the system forecast to the Accuweather forecast and to the actual measurement Based on the results it showed that both system forecasts and actual measurements are almost equal which implies that Kalman Filter can be used as an algorithm for weather forecasting An application program is developed to provide access to the general public on the predicted values by incorporating an Android application and text using GSM technology on the system The Android application allowed the person to monitor the changes on the weather forecasts on their mobile phone The Bluetooth has successfully allowed the transmission of data wirelessly from the computer to mobile phone On the other hand in case of emergency a warning message through GSM technology is used in the system to inform the person A command button is displayed in the graphical user interface GUI of the MATLAB program which allows the operator to control the sending of text messages for emergency purposes only 59 Chapter 5 RECOMMENDATIONS The study focused on the implementation and testing of Kalman filter as an algorithm for predicting weather condition It is proposed that the testing of t
56. f the previous time step as initial value Therefore the Kalman Filter is called as a recursive filter 2013 Prediction of the state at time step i and the corresponding covariance Previous state at time step i 1 Correction of the state at time step i and the corresponding covariance Observations at time stepi Figure No 2 3 Circuit of the Kalman Filter Kleinbauer stated in 2004 that the basic components of the Kalman Filter are the state vector the dynamic model and the observation model State vector contains the variables of interest It describes the state of the dynamic system and represents its degrees of freedom The variables in the state vector cannot be measured directly but they can be inferred from the values that are measurable Subash J 2012 The state vector has two values such as the priori value x which represents the predicted value before the update and the posteriori value which is the corrected value after the update It is said that the significant advantages of Kalman filter are that it combines measured data as well as previous knowledge about the system and measuring devices to produce an 14 optimal estimate of the desired variables Also compared to other filters it minimizes the error significantly Vij V and Mehra R 2011 s However Graham Hesketh stated that Kalman Filter is computationally complex especially for large numbers of sensor channe
57. folder R2009a C Program Files MATLAB R2009a Restore Default Folder Space available 424711 MB Maximum space required 0 MB Figure No 3 20 Folder Selection 64 11 Click Install The installation procedure will start M Confirmation re ae ww i lelo MATLAB Confirm your installation settings SIMULINK Installation folder R2009a C Program Files MATLAB2 Products MATLAB Distributed Computing Server 4 1 MATLAB 7 8 Simulink 7 3 Aerospace Blockset 3 3 Aerospace Toolbox 2 3 Bioinformatics Toolbox 3 3 Communications Blockset 4 2 Communications Toolbox 4 3 Control System Toolbox 8 3 Curve Fitting Toolbox 2 0 Database Toolbox 3 5 1 Datafeed Toolbox 3 3 Econometrics Toolbox 1 1 Filter Dacian HNI Coder A The MathWorks Figure No 3 21 Installation Confirmation 12 After installing activate MATLAB and click Next After that click Finish To activate your software leave this selected Installation Comphste M ATLAB Installation is complete SIMULINK F activate MATLAB R2009a Note You will not be able to use MATLAB until you activate the software See the Help to learn more about activation The MathWorks Next gt Click Next to proceed to activation If you cleared the check box button label changes to Finish Figure No 3 22 Installation Complete 65 Installing the Android application of Kalman Filter 1 Con
58. form the person A command button is displayed in the graphical user interface GUI of the MATLAB program which allows the operator to control the sending of text messages for emergency purposes only 56 References Dhar S et al A Complete Simulation Of Intra Vehicle Link Through Best PossibleWireless Network IJCEE vol 2 no 4 pp 674 Aug 2010 Esme B 2009 Kalman Filter for Dummies Online Available http bilgin esme org BitsB ytes KalmanFilterforDummies aspx Faragher R Understanding the Basis of the Kalman Filter via a Simple and Intuitive Derivation Lecture Notes Signal Processing Magazine IEEE vol 29 no 5 pp 128 132 Sept 2012 doi 10 1109 MSP 2012 2203621 Microchip PIC16F87XA 28 40 44 Pin Enhanced Flash Microcontrollers Microchip Technology Inc USA 2012 Mastro M 2013 Financial Derivative and Energy Market Valuation John Wiley amp Sons Inc DOI 10 1002 9781118501788 ch7 NASA Official 2013 Data Enhanced Investigations for Climate Change Education DICCE Giovanni Help Page Online Available http disc sci gsfc nasa gov giovanni additional users manual DICCE_Help Rountree D M 2010 Paranormal Technology Understanding the Science of Ghost Hunting iUniverse 1663 Liberty Drive Bloomington IN 4703 Scaringe R P 2007 Indoor Air Quality and Mold Remediation Mainstream Engineering Corporation 200 Yellow Place Rockledge Florida 3295 Scheneider Electric Industries SAS
59. gFunctions sf Initialize Dim Paneli As Panel Dim Labeli As Label Dim Label2 As Label Dim Label4 As Label Dim Label6 As Label Dim IblDateTime As Label Dim IblHumidity As Label Dim IbIPressure As Label 115 Dim IblTemperature As Label End Sub Sub Activity_Create FirstTime As Boolean Activity Title kalman Activity LoadLayout kalman If FirstTime Then serial1 Initialize serial1 admin Initialize admin End If Timer1 Initialize Timer1 1000 Timeri Enabled True Activity AddMenulItem Connect ConnectBT Activity AddMenultem Test Data TestData Activity AddMenultem Quit QuitApp DateTime TimeFormat hh mm ss End Sub Sub Activity_Resume If admin IsEnabled False Then If admin Enable False Then ToastMessageShow Error enabling Bluetooth adapter True Else ToastMessageShow Enabling Bluetooth adapter False the StateChanged event will be soon raised 116 End If Else Admin_StateChanged admin STATE_ON 0 End If End Sub Sub Activity_Pause UserClosed As Boolean End Sub Sub Admin_StateChanged NewState As Int OldState As Int btnConnectBT Enabled NewState admin STATE_ON btnAllowConnection Enabled btnSearchForDevices Enabled Log NewState End Sub Sub Seriall_Connected Success As Boolean ProgressDialogHide Log connected amp Success If Success False Then Log LastException Message ToastMessageShow Error connecting amp Las
60. gt Unknown sources 7 A prompt will ask you to choose from the two options displayed in your mobile phone to complete the action Such options include Package installer and Verify and Install Choose Package Installer Complete action using ey Package installer Verify and install Use by default for this action Figure No 3 25 Package Installer 67 8 After that the system will verify the installation of the application Click the Install button and wait until the installation is completed M al 8 26 PM Do you want to install this application Allow this application to Network communication create Bluetooth connections System tools Bluetooth administration Figure No 3 26 Kalman Installation 9 Two options will then be showed on your phone s screen Open or Done Open button must be selected to open the android application and Done button to exit Note Once the android application is opened in your mobile phone it will turn on your Bluetooth connection to let the other devices laptop or PC to connect Figure 3 27 shows the GUI of the android application 68 N all 8 27PM Activity Predicted Temperature Predicted Pressure Predicted Humidity Figure No 3 27 Kalman GUI Creating COM port for Bluetooth 1 Once your mobile phone is paired on the PC or laptop find and click the Bluetooth icon Select the Open Settings to display the current settings of the Bluetooth on your
61. he digital thermometer Temperature True Value vs Measured Value ui O O Measured Value CO True Value C oO 4 30 2 1 oo ZA Q v gt Ss fm Q Qa v m 123 4 5 6 7 8 9 10 Trials Figure No 3 10 Temperature True Value vs Measured Value 37 Relative Humidity As shown in Table 3 3 most of the readings showed a 100 relative humidity The relative humidity is based on the location and on the condition of the atmosphere A high percentage relative humidity indicates that the air is holding the maximum water it can hold at the current temperature and any additional moisture will result in condensation If the measured relative humidity is low it indicates that the atmosphere in the area is dry and can hold more moisture at that value of temperature Results show that when the temperature decreases the relative humidity goes up due to the small amount of moisture that cooler air can hold for the current temperature and vice versa NASA 2013 Table No 3 3 Tabulated Data for Relative Humidity TRIAL Date amp Time Measured Value True Value Percentage Aug 19 2013 Difference 1 12 00 AM 100 99 1 0050 2 3 00 AM 100 99 1 0050 3 5 00 AM 100 100 0 4 11 00 AM 100 100 0 5 2 00 PM 100 100 0 6 5 00 PM 100 100 0 7 6 00 PM 100 99 1 0050 8 7 00 PM 100 96 4 0816 9 10 00 PM 100 100 0 10 11 00 PM 100 100 0 Figure 3 11
62. he same measurements as the instruments used for measuring the actual value of the said parameters Furthermore statistical tests show that the system forecast is accurate as evidenced by the small per cent difference between the predicted and the actual values obtained from measurements Keywords Kalman filter weather forecasting Introduction The objective of this study is to design a system for weather prediction with a warning system via GSM and Android Application Specifically it aims to achieve the following i to design and implement a device that will measure and monitor actual basic weather parameters such as temperature pressure and relative humidity 11 to develop and apply an algorithm for predicting weather conditions iii to write a program for the algorithm and use 80 available data sets to train the system iv to use 20 of the available data sets to validate the system v to simulate the system by applying available initial conditions such as those provided by the device for monitoring actual weather conditions vi to test the system and determine in terms of the limits of its capabilities in predicting weather conditions vii to develop an application program to provide access by the general public to the predicted values 17 This study will be significant in informing an individual or group of people with the information on the anticipated weather over forthcoming one to three days for sites in the areas s
63. he system must be performed and exercised on the exact location where the weather station of the data to be compared is located This is to have a more appropriate observation and comparisons on the system forecast to the forecast of the chosen weather station In Chapter 3 of this study it is noted that the KF algorithm is a recursive algorithm which denotes that the calculation of the present value is dependent upon the previous calculated value of it Thus the system must have the capability to store the previous value for each parameter used It is reached solely by throwing the whole system turned on for the continuous simulation For the future researchers who will continue to improve this study it is recommended that a database must be implemented in MATLAB to achieve ease of usage of the system Furthermore the device capability to send information efficiently and faster can be achieved by embracing new technologies as a substitute for the components employed in this study Some technologies for wireless data communication can be used as a substitute for the GSM technology and Bluetooth such as WI Fi The standard Zigbee used in the device can be upgraded to a Pro Zigbee which has more advantages like faster transmission of measured data for the data logger purposes This will allow the user to observe the changes easily occurred on each parameter since the system exhibits real time 60 REFERENCES Annenberg Learner and Foundatio
64. he weather condition differs for the two measurements system forecast and actual measurements The computation of important parameters are shown below 2 2 2 2 Standard Error of the Difference se se _ EE 0 061 1 2 2 212 Degrees of Freedom DF df s 18 2 si 33 ny nz n 1 1 n2 1 95 Confidence Interval for the Difference 0 0190 0 2390 X Y _ 0 799 T Value 1 7918 2 2 0 6722 1 5 ny Nz P Value p 0 0900 Observation A two tailed unequal variance test was conducted to determine the difference of the pressure forecast produced by the system and actual measurement The result showed that there was not a significant difference between the system forecast M 101 05 SD 0 1546 and the actual measurement M 101 16 SD 0 1174 two sample t 18 1 7918 p 0 0900 At the 0 05 significance level and 95 confidence level it turned out that the null hypothesis is true so fail to reject the null hypothesis The mean pressure of the weather condition is same for the two measurements system forecast and actual measurements 54 Conclusions A measuring and monitoring device for weather parameters such as temperature pressure and relative humidity is designed and implemented in this study The objective is achieved by using a sensor for each parameter and is integrated to the microcontroller Such sensors are the LM35 for temperature capacitive humidity kit fo
65. i 7 cellstr sprintf EXTREME CAUTION 39 elseif hRounded 50 105 rowData i 7 cellstr sprintf DANGER 41 elseif hRounded 55 rowData i 7 cellstr sprintf DANGER 43 elseif hRounded 60 rowData i 7 cellstr sprintf DANGER 45 elseif hRounded 65 rowData i 7 cellstr sprintf DANGER 48 elseif hRounded 70 rowData i 7 cellstr sprintf DANGER 50 elseif hRounded 75 rowData i 7 cellstr sprintf EXTREME DANGER 53 elseif hRounded 80 rowData i 7 cellstr sprintf EXTREME DANGER 57 else rowData i 7 cellstr sprintf EXTREME DANGER end elseif tRounded 36 if hRounded 40 rowData i 7 cellstr sprintf EXTREME CAUTION 39 elseif hRounded 45 rowData i 7 cellstr sprintf DANGER 41 elseif hRounded 50 106 rowData i 7 cellstr sprintf DANGER 43 elseif hRounded 55 rowData i 7 cellstr sprintf DANGER 46 elseif hRounded 60 rowData i 7 cellstr sprintf DANGER 48 elseif hRounded 65 rowData i 7 cellstr sprintf DANGER 51 elseif hRounded 70 rowData i 7 cellstr sprintf EXTREME DANGER 54 elseif hRounded 75 rowData i 7 cellstr sprintf EXTREME DANGER 58 else rowData i 7 cellstr sprintf EXTREME DANGER end elseif tRounded 37 if hRounded 40 rowData i 7 cellstr sprintf DANGER 41 elseif hRounded 45 rowData i 7 cellstr sprintf DANGER
66. imple one dimensional signal The information that will be used in this study are accessible from the historical data based on the last known measurements of temperature relative humidity and pressure and current reading of measurements of the three parameters in the area The data from the two sources such as current measurements and historical data are combined and processed together to make the best possible estimate of the temperature relative humidity and pressure The Kalman Filter algorithm converges the estimation of a value into correct estimations Thus assuming a poor estimated Gaussian noise parameter present in a signal will be corrected The better you estimate the noise parameters the better estimates you get Esme B 2009 Figure3 7 shows the step by step process in Kalman Filter which will be discussed further in the various steps in dealing with Kalman Filter 24 Basic Linear Kalman Filter Build the model Select initial guess for x and Po Compute the Kalman Filter estimate and covanance Set value for x and P Figure No 3 7 Process in Kalman Filter Ramsey Faragher stated in 2012 that there are two models associated with the Kalman filter Equations 3 3 to 3 9 and the corresponding function or significance are presented in the 25 said IEEE magazine The Kalman filter model assumes that the state of the system during time t evolved at time t 1 because of the prior state which is e
67. inimum temperature dew point vapor pressure relative humidity mean sea level pressure prevailing winds and cloud Selected parameters such as temperature atmospheric pressure and relative humidity will utilize the historical data and sensor data for prediction in the application of the algorithm Developing an application program to provide access by the general public to the predicted values will be limited only in sending text messages to selected people such as the barangay officials and through an Android Application using Bluetooth CONCEPTUAL FRAMEWORK Temperature Atmospheric pressure and Relative humidity sensors Machine Intelligence System Data basedalgorithm for weather prediction Data Logger Device Predicted Weather Android App Za Warning Message Person GSM Module Figure No 1 1 Conceptual Framework Historical Data Chapter 2 REVIEW OF RELATED LITERATURE Statistical post processing techniques are used to remove the systematic bias in modelled data Marxs H G 2008 There are actually common features that characterize several methods for weather prediction The said features are sensors data acquisition assimilation processing of received data and post processing the result and finally representing the obtained results and forecasts In sensors data acquisition there are specialized sensors and sensor modules in collecting da
68. ls it requires conditional independence of the measurement errors sample to sample It also requires linear models for state dynamics and observation processes and getting a suitable value for Q a k a tuning can be something of a black art 2000 2 6 PIC16F877A Microcontroller In the study of Ibrahim Al Adwan and Munaf S N Al D entitled The Use of ZigBee Wireless Network for Monitoring and Controlling Greenhouse Climate they used PIC16F877A microcontroller for the purpose of storing the instantaneous values of the environmental parameters they focused on and the Zigbee for wireless communication The environmental parameters include the temperature humidity and light or solar radiation According to them sensors provide input information for the automation system by measuring the climate variables of the greenhouse Sensor generated signals are acquired and conditioned by a PIC16F877A microcontroller These sensors are connected to a PIC16F877A microcontroller which consists of embedded ADCs They also used a ZigBee transceiver which is directly connected to the microcontroller to provide a wireless connection with the central station From the study A Low Cost Microcontroller based Weather Monitoring System of Kamarul Noordin Chow Onn and Mohamad Isamail they measure temperature atmospheric pressure and relative humidity remotely by using the appropriate sensors that are not only important in environmental or weather moni
69. man Filter Algorithm with Warning System via GSM and Android Application Mapua Institute of Technology School of EECE BS Computer Engineering Thesis Members Capapas Trexia D Paule Jezzelle Joyce Clarice M Vista Rochelle Lynn S Thesis Adviser Dr Felicito S Caluyo 1 2 Temperature 2 Pressure Humidity x Start Data Acquisition Send Warning Text 4 Figure No 3 42 fig File for Kalman Filter 10 You can choose any parameter you want to be displayed in the graph provided Click the Start Data Acquisition After 15 seconds the measurements and the predicted value for the weather parameters temperature pressure and humidity will be displayed Each parameter will have ten 10 samples 4 Troubleshooting Guides and Procedures 1 Clicking the Start Data Acquisition button in MATLAB doesn t work a b C First make sure that the device is plugged in Check if the serial cable is properly connected to the laptop and to the device Open the MATLAB and the m file again 2 The COM port used in serial connection is not available a Make sure that the serial cable is properly connected to your PC or laptop and the device Verify the COM port number in the Device Manager by following the steps provided in the installation of Bluetooth In the m file locate the COM port number used Change it to the COM port number displayed in the pop up window in the Taskbar Run
70. n 2013 Weather What forces affect our weather Online Available http www learner org interactives weather forecasting2 html Accessed March 2 2013 Graham Hesketh Kalman Filters for Fun Information Engineering Rolls Royce Strategic Research Center Group 2000 Greg Welch Gary Bishop An Introduction to the Kalman Filter University of North Carolina at Chapel Hill Department of Computer Science 2001 Grewal M and Andrews A P Kalman Filtering Theory and Practice using MATLAB 3 edition John Wiley amp Sons Inc 2008 Jean de DieuNiyigaba VariationalKalman Filter Online Available http personal lut fi 98COC55 1 E3C9 4CB2 A0A0 78B5374C87C8 FinalDownload DownloadId 1868387BF6482526F6A 1B6A1DECFFC3 1 98C0CS5 1 E3C9 4CB2 A0A0 78B5374C87C8 wiki lib exe fetch php en technomathematics time_series_research start variational_kalm an_filter pdf Kleinbauer R Kalman Filtering Implementation with MatLab Field of Study Geodesy and Geoinformatics at Universitat Stuttgart Helsinki 2004 Lee Lerner K and Wilmoth Lerner B Other Forecasting Methods in World of Earth Science The Gale Group Inc 2003 Manolis Anadranistakis et al Correcting Temperature and Humidity Forecasts using Kalman Filtering Potential for Agricultural Protection in Northern Greece Nevrokopi Greece 2004 McKemy D et al Climate Education for K 12 Instruments Online Available http www nc climate ncsu
71. n state equations are given by K P7 HT HPZ HT RT 3 7 EF Xx K 2 HX 3 8 P 1 KA Pe 3 9 Where X estimate of x at time t X the prior estimate P error covariance P the prior error covariance 27 K Kalman gain Ramsey Faragher 2012 Steps in using the Kalman Filter Algorithm Step 0 Build the model The derivations will consider a simple one dimensional signal Thus the entities in the model are represented in numerical value and not in matrix form The original equation is shown in equation 3 3 and the reduced equation is given in equation 3 10 Xt AxX t Btt we 3 3 Xt Xt Wt 3 10 The value of A is equated to a value of 1 since the next value will be the same as the previous one due to its recursive part which is the nature of Kalman filter Then the u parameter equates to a value of O and eliminated from the equation since no control signal was involved in this study Zt Hix Vt 3 4 Zt Xt H Ve 3 11 The equation 3 4 shows the linear combination of the measurement value and the signal value This equation is then brought down to simpler equation which is shown in equation 3 11 by having the parameter H equated to a value of 1 because in real life state value and some noise are present in measurable value Esme B 2009 Step 1 Set t 0 then select the initial guess for x9 and Py For the initial guess let X 0 and Py 1 Th
72. nect your cellphone to the PC or laptop via Bluetooth Note Most laptops and desktops do have Bluetooth If yours doesn t a USB Bluetooth adapter can be used 2 Enable Bluetooth on your cell phone 3 Go to Control Panel under the Start Menu and then go to the Hardware and Sound From there choose Devices and Printers and click Add a Bluetooth device as shown in the Figure 3 23 below Devices and Printers Zj Add a device Add a printer Add a Bluetooth device Mouse W Device Manager Figure No 3 23 Add a Bluetooth Device 4 Scan for devices on your PC or laptop and choose your device name Note Make sure that your Bluetooth status is discoverable in order for the computer to detect your device 5 Once your device is paired with your PC or laptop transfer the Kalman Ape file provided to your cell phone Note Open the Apps folder and then go to Objects Right click on the Kalman apk file and then send to Bluetooth device See Figure 3 24 66 amp Kalman apk Open File 151 KB Share Dropbox link View on Dropbox com View previous versions Share with gt Add to archive Add to Kalman rar Compress and email Compress to Kalman rar and email Restore previous versions Sendto Bluetooth device Figure No 3 24 Kalman apk 6 Open the Kalman apk file on your cellphone Note Be sure to allow your mobile phone in installing non Market applications Go to Settings gt Applications
73. nsors temperature sensor and pressure sensor 117 RES gt HUMIDITY SENSOR PRESSURE SENSOR TEMPERATURE 5 vl ove gi Jg GSM MODULE i mene EV o If m 5 L PWR H j CRYSTAL o HEE ca o XBEERX 3 o HUMIDITY Figure No 3 6 Schematic Diagram of PIC16F877A Microcontroller Figure3 6 shows the schematic diagram for the PIC16F877A microcontroller As illustrated in the previous schematic diagrams the temperature pressure and relative humidity sensors are directly connected to the RAO RAI and RCS pin of the PIC16F877A 23 microcontroller respectively The transmitter Zigbee and receiver Zigbee are connected to pin 25 and pin 24 of the microcontroller respectively A crystal oscillator is connected to PIC16F877A microcontroller because it has no internal oscillator which is one of its disadvantages 3 2 Applying the Algorithm for the Prediction of Weather Conditions Kalman Filter is commonly referred as an optimal recursive computation of the least squares algorithm It is a subset of a Bayes Filter where the premise of a Gaussian distribution and that the current state is linearly dependent upon the previous state are imposed Dhar S et al 2010 It is optimized in a way that the Kalman filter minimizes the mean square error of the estimated parameters of all noise that is Gaussian Mastro 2013 In this study the Kalman filter algorithm equations will be applied by using a s
74. o that they may take necessary precautions from a coming danger in the chosen locality Thus in these cases forecasts are one of the main aspects that may help save lives or prevent damage which could be avoided by preventive arrangements The warning will be based on the predicted weather due to different parameters at a specific date or time Furthermore the device will use an algorithm in predicting the weather by analyzing vital parameters such as probabilities of temperature atmospheric pressure and relative humidity Data regarding weather conditions on a specific locality from PAGASA will be monitored in order to produce accurate and valid weather forecast Since the weather is continuous multidimensional dynamic and chaotic and hence difficult to predict the results will not be very accurate Due to these circumstances weather is observed to be changing simultaneously which leads to a conclusion that forecast can be out of date easily Hence the algorithm will determine the limits of the predicted weather condition rather than its accuracy Moreover Quezon City will be the chosen locality wherein data measured by the sensors located in the said locality will be processed using an algorithm that is not being used by PAGASA such as the Kalman filter The wireless device would basically send the data measured using the sensors into the data logger After that measured parameters will be used in the application of the algorithm for predicti
75. on purpose Using the algorithm parameters used by PAGASA can also be predicted These include rainfall maximum and minimum temperature dew point vapor pressure relative humidity mean sea level pressure prevailing winds and cloud Selected parameters such as temperature atmospheric pressure and relative humidity will utilize the historical data and sensor data for prediction in the application of the 18 algorithm Developing an application program to provide access by the general public to the predicted values will be limited only in sending text messages to selected people such as the barangay officials and through an Android Application using Bluetooth This chapter discusses the methods that are used to gather analyze interpret and report data It defines the scope and limitations of the research design used Furthermore this chapter is divided into three sections The first section includes the data gathering to be performed which discussed the processes used to come up with the data The next section is about the processes performed and analysis of the collected data and lastly the validation of the results Methodology 3 1 Design and Implementation of the Device The basic weather parameters such as temperature atmospheric pressure and relative humidity will be measured using a device designed to gather these parameters It uses specific sensors for each parameter to collect data The sensors will be integrated into one devic
76. ounded 65 rowData i 7 cellstr sprintf CAUTION 32 elseif hRounded 70 rowData i 7 cellstr sprintf EXTREME CAUTION 33 elseif hRounded 75 rowData i 7 cellstr sprintf EXTREME CAUTION 34 elseif hRounded 80 rowData i 7 cellstr sprintf EXTREME CAUTION 35 elseif hRounded 85 98 rowData i 7 cellstr sprintf EXTREME CAUTION 36 elseif hRounded 90 rowData i 7 cellstr sprintf EXTREME CAUTION 37 elseif hRounded 95 rowData i 7 cellstr sprintf EXTREME CAUTION 38 elseif hRounded 100 rowData i 7 cellstr sprintf DANGER 40 end elseif tRounded 30 if hRounded 40 rowData i 7 cellstr sprintf CAUTION 30 elseif hRounded 45 rowData i 7 cellstr sprintf CAUTION 30 elseif hRounded 50 rowData i 7 cellstr sprintf CAUTION 31 elseif hRounded 55 rowData i 7 cellstr sprintf CAUTION 32 elseif hRounded 60 rowData i 7 cellstr sprintf EXTREME CAUTION 33 elseif hRounded 65 rowData i 7 cellstr sprintf EXTREME CAUTION 34 elseif hRounded 70 99 rowData i 7 cellstr sprintf EXTREME CAUTION 35 elseif hRounded 75 rowData i 7 cellstr sprintf EXTREME CAUTION 36 elseif hRounded 80 rowData i 7 cellstr sprintf EXTREME CAUTION 38 elseif hRounded 85 rowData i 7 cellstr sprintf EXTREME CAUTION 39 elseif hRounded 90 rowData i 7 cellstr sprintf
77. r humidity and MPX114AP for measuring the pressure Zigbee technology is also included which allows the transmission of measurements wirelessly from the device to the laptop or PC Also PIC16F877A microcontroller is used for collecting and processing of the gathered measurements Since the device is real time it sends measurements every 10 seconds The Kalman Filter Algorithm is proposed and applied by the researchers to provide weather forecast It is applied and implemented using MATLAB based on its equations to perform the correction and prediction state This algorithm uses two data historical and present data The combination of the past and present information in Kalman filter produced possible estimate of the parameters for the prediction of the next state of weather condition The program developed for Kalman Filter Algorithm required the system to be trained by inputting the 80 available data sets or the historical data These data are gathered in Accuweather and presented in a CSV file format The remaining 20 available data are used in the validation of the system A high percentage of available data are allocated in the historical data in order to fully develop the system in the given set of measurements for each parameter The testing of the system is divided into two parts determining the accuracy of measurements of the device and the forecast produced by the system using KF algorithm The results of the first part of the test show tha
78. r to the actual value compared to the Accuweather forecast at any given of time during the testing period 45 Relative Humidity Accuweather Forecast E System Forecast Ss gt E gt I v 2 p a v a te Actual Humidity Measurement NE SF PF GF F GS a oF of LDP HP HP HP HP HF HM H Forecast Time Figure No 3 14 Relative Humidity 46 Hourly Forecasts of Pressure Atmospheric pressure is an important parameter that is essential in monitoring the weather condition Table 3 9 presents that the differences in hour to hour changes in the atmospheric pressure are caused by the movement and development of pressure systems affecting the country very often MET EIREANN These forecasts proved that the pressure did not change its value quickly compared to the two previous parameters Thus there is no available forecast on pressure for all weather forecast companies like Accuweather Table 3 9 Comparison of Pressure Forecast between the System and Accuweather Forecast Time Accuweather Forecast System Forecast Percent kPa kPa Difference 2 00 PM 101 3 100 79 0 50 3 00 PM 101 2 100 82 0 38 4 00 PM 101 1 100 80 0 30 5 00 PM 101 0 100 80 0 20 6 00 PM 101 0 100 84 0 16 7 00 PM 101 1 100 85 0 25 8 00 PM 101 1 100 89 0 21 9 00 PM 101 2 100 92 0 28 10 00 PM 101 3 100 92 0 38 11 00 PM 101 3 100 94 0 36 Average Percentage Difference 0 302 The
79. re 3 7 Process in Kalman Filter Figure 3 8 Program Flowchart Figure 3 9 Selection of CSV file Figure 3 10 Temperature True Value vs Measured Value Figure 3 11 Relative Humidity True Value vs Measured Value Figure 3 12 Pressure True Value vs Measured Value Figure 3 13 Temperature Figure 3 14 Relative Humidity Figure 3 15 Pressure Figure 3 16 Driver Setup x64 Figure 3 17 MathWorks installer Figure 3 18 File Installation Key Window 10 11 14 20 21 22 22 23 23 25 31 35 37 39 40 43 46 49 62 63 63 vii Figure 3 19 Figure 3 20 Figure 3 21 Figure 3 22 Figure 3 23 Figure 3 24 Figure 3 25 Figure 3 26 Figure 3 27 Figure 3 28 Figure 3 29 Figure 3 30 Figure 3 31 Figure 3 32 Figure 3 33 Figure 3 34 Figure 3 35 Figure 3 36 Figure 3 37 Figure 3 38 Figure 3 39 Figure 3 40 Figure 3 41 File Installation Key Folder Selection Installation Confirmation Installation Complete Add a Bluetooth Device Kalaman apk Package Installer Kalman Installation Kalman GUI Opening the Bluetooth Settings Bluetooth Settings Window COM Ports Add COM Port Standard Serial over Bluetooth Modified COM Port Zigbee Serial Connection Complete Set up of Zigbee Device for Measuring Weather Parameters Opening an Existing GUI File fig File M file Run Button Selecting Raw Data 64 64 65 65 66 67 67 68 69 69
80. s Kostas Lagouvardos Vassilik Kotroni and Helias Elefteriadis has been motivated by the importance of meteorological conditions for the protection of potato cultivation from mildew On the other hand the proposed device has a warning system to inform the people of the possible damage that the forthcoming weather might bring GSM for emergency text and an Android Application are used as a warning system for easy transmission of data 16 Chapter 3 Abstract The Kalman Filter Algorithm filters the sensor s data and process data which are used for predicting the next state or value The algorithm is used in this study to predict the weather condition on an hourly basis for a given location The goal is to implement the processes and the equations of the basic Kalman Filter Algorithm in order to produce a forecast based on the given set of available data used to train the system The device uses a microcontroller specifically PIC16F877A and is integrated with different sensors for each parameter and wireless technology for the data transfer The sensors include LM35 for temperature the capacitive humidity kit for humidity MPX114AP for measuring the pressure and Zigbee technology for wireless data transfer The system includes methods for alerting people by using GSM and Android application through Bluetooth The tests conducted show that the sensors integrated in the device for measuring temperature pressure and relative humidity produced almost t
81. s are imperfect due to errors in the initial and boundary conditions that are fed into the model Therefore there is a need to apply statistical post processing techniques in modeled forecast fields to improve forecast quality and value The objective of this study is to design a system for weather prediction with a warning system via GSM and Android Application Specifically it aims to achieve the following i to design and implement a device that will measure and monitor actual basic weather parameters such as temperature pressure and relative humidity 11 to develop and apply an algorithm for predicting weather conditions iii to write a program for the algorithm and use 80 available data sets to train the system iv to use 20 of the available data sets to validate the system v to simulate the system by applying available initial conditions such as those provided by the device for monitoring actual weather conditions vi to test the system and determine in terms of the limits of its capabilities in predicting weather conditions vii to develop an application program to provide access by the general public to the predicted values This study will be significant in informing an individual or group of people with the information on the anticipated weather over forthcoming one to three days for sites in the areas so that they may take necessary precautions from a coming danger in the chosen locality Thus in these cases forecasts
82. s introduced into the space above and pressure increases slightly as the new vapor is added The increasing pressure is due to an increase in the partial pressure of water vapor Mean Sea Level Pressure or MSLP is also considered the atmospheric pressure at mean sea level and is the force which is exerted per unit area at the mean sea level through the weight of the atmosphere The rainfall describes the amount of precipitation and is usually expressed in millimeters The cloud can also be a weather variable that depicts the amount of cloud present in the sky expressed in oktas of the sky cover Prevailing wind direction is most frequently observed during a given period while the average wind speed in meters per second is the arithmetic average of the observed wind speed 2 5 Predictive Algorithm Kalman Filter Theoretically the Kalman filter is an estimator for what is called the linear quadratic problem which is the problem of estimating the instantaneous state a linear dynamic system perturbed by white noise by using measurements linearly related to the state but corrupted by white noise The resulting estimator is statistically optimal with respect to any quadratic function of estimation error M Grewal and A P Andrews 2008 13 Thus according to K Sreedhar A Venu and A Hriprasad the Kalman Filter consists of two steps the prediction and the correction This procedure is repeated for each time step with the state o
83. t The result showed that there was not a significant difference between the system forecast M 90 306 SD 7 3372 and the actual measurement M 88 SD 7 6739 two sample t 17 0 6868 p 0 5014 The direction of the difference in the sample means is reflected by the sign positive or negative of the t value At the 0 05 significance level and 95 confidence level it turned out that the null hypothesis is true so fail to reject the null hypothesis The mean relative humidity of the weather condition is same for the two measurements system forecast and actual measurements Test 3 Pressure System Forecast and Actual Measurement The value for mean variance standard deviation and standard error for the two sources are as follows Table No 3 14 T Test for Pressure Forecast Time Actual Pressure System Forecast Measurement 2 00 PM 101 2 100 86 3 00 PM 101 1 100 97 4 00 PM 101 0 100 90 5 00 PM 101 0 100 90 6 00 PM 101 1 100 97 7 00 PM 101 1 101 01 8 00 PM 101 2 101 20 9 00 PM 101 3 101 23 10 00 PM 101 3 101 23 11 00 PM 101 3 101 23 Mean 101 16 101 05 Standard Deviation 0 1174 0 1546 Standard Error Mean 0 037 0 049 53 The hypotheses for this test are as follows Null Hypothesis The mean pressure of the weather condition is same for the two measurements system forecast and actual measurements Alternative Hypothesis The mean pressure of t
84. t the sensors used for measuring temperature pressure and relative humidity produced almost the same measurements as the instruments used for 55 measuring the actual value of the said parameters Small differences are recorded in the percentage difference of the actual value to the experimental value for each trial in each parameter which leads to a conclusion that the device is capable of measuring and monitoring weather parameters in a specific location The second part of the testing focused on determining the accuracy of Kalman filter in forecasting The calculation of percentage difference and use of statistical tool such as t test allowed the determination of the difference of the system forecast to the Accuweather forecast and to the actual measurement Based on the results it showed that both system forecasts and actual measurements are almost equal which implies that Kalman Filter can be used as an algorithm for weather forecasting An application program is developed to provide access to the general public on the predicted values by incorporating an Android application and text using GSM technology on the system The Android application allowed the person to monitor the changes on the weather forecasts on their mobile phone The Bluetooth has successfully allowed the transmission of data wirelessly from the computer to mobile phone On the other hand in case of emergency a warning message through GSM technology is used in the system to in
85. tException Message True Else StartService BTService End If End Sub Sub Admin_DeviceFound Name As String MacAddress As String Log Name amp amp MacAddress Dim nm As NameAndMac nm Name Name nm Mac MacAddress 117 foundDevices Add nm ProgressDialogShow Searching for devices device found Replace foundDevices Size ProgressDialogShow Connecting device wait End Sub Sub Admin_DiscoveryFinished ProgressDialogHide If foundDevices Size 0 Then ToastMessageShow No device found True Else Dim As List Initialize For i 0 To foundDevices Size 1 Dim nm As NameAndMac nm foundDevices Get i Add nm Name Next Dim res As Int res InputList l Choose device to connect 1 If res lt gt DialogResponse CANCEL Then connectedDevice foundDevices Get res ProgressDialogShow Trying to connect to amp connectedDevice Name amp amp connectedDevice Mac amp seriall Connect connectedDevice Mac End If connectedDevice Mac 00 12 07 12 47 31 seriall Connect connectedDevice Mac End If End Sub Sub ConnectBT_Click foundDevices Initialize If admin StartDiscovery False Then ToastMessageShow Error starting discovery process True Else 118 ProgressDialogShow Searching for devices End If End Sub Sub QuitApp_Click ExitApplication End Sub Sub NewMsgFromServiceAStream The received message comes from the global vari
86. ta for typical meteorological parameters of the atmosphere It has been said that processing received data is made with the help of a previously created mathematical model of a combination of thermodynamic equations which describe the state of the atmosphere in a given location in any moment of time and analysis of numerical statistical data I Simeonov H Killifarev and R Ilarionov 2006 From the Daily Weather Forecast of Kidlat Pagasa in the Philippines the following weather parameters have interest to the users of the forecast cloudiness rainfall and wind Basically weather maps such as surface maps and upper air maps are used by weather forecasters since they depict the distribution patterns of temperature pressure humidity and wind at different levels of the atmosphere Furthermore there are five standard levels of the upper air maps that are constructed twice daily at twelve hourly interval The surface maps are made four times daily at six hourly intervals On the surface maps the distribution patterns of rain or other forms of precipitation and cloudiness can also be delineated The figures below show the different weather maps a the surface temperature b cloud cover c surface wind and the d upper air map Philippines Weather Map Surface Temperature on Saturday 02 Mar at 2pm PHT Freezing evel contours a Surface Temperature on Saturday 02 Mar at 2pm PHT Cloud Cover on Saturday 02 Mar at 8pm P
87. the M file 77 3 No COM port number displayed in the Device Manager a Make sure that the serial cable is properly connected to your laptop and device Reinstall the driver See the Installation Procedure Reopen the MATLAB and rerun the m file If same problem was encountered the serial cable is defective Try using another serial cable then rerun the m file 4 The android application cannot connect to the program via Bluetooth connection or there is an error occurred while making a connection a b C d f 8 Quit the android application of Kalman Filter Restart the mobile phone Open the android application and reconnect it to the laptop or PC If the android application is still cannot connect to the laptop or PC go to the Bluetooth settings and to the COM Ports tab Remove the COM port number of the incoming device and add again another COM port number of the incoming device Modify the COM port number in the MATLAB program for Kalman filter Reconnect again the android application to the laptop or PC 5 The COM port number for Bluetooth connection is not available a Check the Bluetooth Settings for the COM port number of the incoming device Check the COM port number declared in the MATLAB program If the COM port number is the same for the previous steps rewrite again the COM port number in the MATLAB program and save it Click the Run button in MATLAB 5 Error Definitions 1
88. the measured data of the following sensors temperature pressure and humidity Then it will output the measured values which will then be displayed in the data logger of the MATLAB GUI 86 MATLAB function varargout WWM varargin gui_Singleton 1 gui_State struct gui_Name mfilename gui_Singleton gui_Singleton gui_OpeningFen WWM_OpeningF cn gui_OutputFcn WWM_OutputFcn gui_LayoutFen gui_Callback if nargin amp amp ischar varargin 1 gui_State gui_Callback str2func varargin 1 end if nargout varargout 1 nargout gui_mainfcn gui_State varargin else gui_mainfcn gui_State varargin end handles output hObject delete instrfindall movegui hObject center if isdeployed cd fileparts which mfilename From Brett end set handles figurei name Weather Forecast Using Kalman Filter Algorithm guidata hObject handles global AcqFlag global grpFlag global sendDataF sampTotal 10 tempVal zeros 1 presVal zeros 1 humiVal zeros 1 rowData zeros sampTotal 3 sampCtr 1 87 grpFlag 1 CSVPath folder uigetfile CSV Select Raw Data CSV CSVPath fullfile folder CSVPath rawData csvread CSVPath 1 0 sDevT 0 1 sDevP 0 1 sDevH 0 1 DataT rawData 1 DataP rawData 2 DataH rawData 3 kRep size DataT PredT zeros kRep PredP zeros
89. the said sensor is contained within a vented unit since it allows air to flow across the sensor After that the temperature is measured while the thermometer is protected from the direct heat of the sun hihi nn u oe Sling Psychrameter a Image from NOAA b Image from Wikimedia Commons c Image from NASA Figure No 2 1 Weather Instruments a Electronic Temperature Sensor b Modern Aneroid Barometer and c Sling Psychrometer Image from http www nc climate ncsu edu edu k 12 instruments Moreover in order to measure relative humidity a hygrometer is used Another common instrument that meteorologists used to determine relative humidity is the Psychrometer which is whirling around while being held Thus after whirling the said instrument the dew point and the relative humidity can be obtained by using the Psychrometer chart based on the wet and dry bulb 10 For atmospheric pressure barometers are used in order to measure the said parameter One of the most common types of barometer is an aneroid barometer which uses a sealed can of air to detect changes in atmospheric pressure The concept of using this instrument is that as the atmospheric pressure goes up it pushes in on the can and the can is slightly reduced in volume moving an indicator needle towards higher pressure If the atmospheric pressure goes down the can expands slightly and the needle indicates lower pressure Nowadays the
90. toring but also crucial for many industrial processes 15 The analogue outputs of the sensors are connected to a microcontroller specifically PIC16F877A through an ADC for digital signal conversion and data logging 2 7 Summary According to the research paper entitled Correcting Temperature and Humidity Forecasts using Kalman Filtering Potential for Agricultural Protection in Northern Greece of Manolis Anadranistakis Kostas Lagouvardos Vassilik Kotroni and Helias Elefteriadis the application of the Kalman theory filtering can substantially reduce errors of the near surface temperature and humidity forecasts provided for 2 3 days ahead in time Based on the corrected forecasts farmers can then schedule their fungicide spraying programs according to the expected weather thus reducing the cost and the ecological impact of frequent preventive spraying interventions Also the success of the method is also supported by the fact that after correction and for both parameters the mean error decreases to values close to zero showing that the method is able to provide almost unbiased corrected forecasts while the standard deviation of the error also decreases by 50 Similar to the researchers proposed device it mainly focused on forecasting weather parameters such as temperature and humidity using Kalman Filter Algorithm An additional weather parameter pressure is also being measured by the device The work of Manolis Anadranistaki
91. which is relatively high However it did not affect the capability of the KF algorithm since the study focused more on the determination of the accuracy between system forecast and actual measurement 44 Table No 3 8 Comparison of System Humidity Forecast to Actual Humidity Measurement Forecast Time Actual Humidity System Forecast Percent Measurement Difference 2 00 PM 88 88 11 0 12 3 00 PM 69 73 23 5 95 4 00 PM 94 90 09 4 25 5 00 PM 88 89 25 1 41 6 00 PM 83 85 58 3 06 7 00 PM 88 92 87 5 39 8 00 PM 88 91 42 3 81 9 00 PM 94 96 89 3 03 10 00 PM 94 97 17 3 32 11 00 PM 94 98 45 4 62 Average Percent Difference 3 496 The system forecast and actual measurement for relative humidity were recorded and tabulated as shown in Table 3 8 The percentage difference included in the table depicts how much the system forecast differs from the actual data Based on the calculation small values of differences were recorded which means that the system forecasts are almost equal to the actual measurements at any given time Moreover the average percent difference recorded is equal to 3 496 which was acceptable and the results proved that the Kalman filter can be applied as an algorithm for weather forecasting Figure 3 14 shows the differences between the system forecast to the Accuweather forecast and actual measurements It was easily observed that the system forecast were almost close
92. xpressed in equation 3 3 where Xt AxX t Brut wy 3 3 x is the state vector containing the terms of interest for the system temperature relative humidity and pressure at time t Uz is the vector containing any control inputs A is the state transition matrix which applies the effect of each system state parameter at time t 1 on the system state at time t B is the control input matrix which applies the effect of each control input parameter in the vector u on the state vector w is the vector containing the process noise terms for each parameter in the state vector The process noise is assumed to be drawn from a zero mean multivariate normal distribution with covariance given by the covariance matrix Q The measurements of the system can be performed by using the equation 3 4 which represents the second model where Zt Hix Vt 3 4 Zz is the vector of measurements 26 e H is the transformation matrix that maps the state vector parameters into the measurement domain e vis the vector containing the measurement noise terms for each observation in the measurement vector Like the process noise the measurement noise is assumed to be zero mean Gaussian white noise with covariance R The Kalman filter has two states the prediction and correction state or measurement update The basic Kalman filter equations for the prediction state are RT Akt Brus 3 5 PE A PA Q 3 6 The correctio
93. y In this study the historical data needed to train the system are obtained from the Accuweather site 80 of the available data is used to train the system while the remaining 20 data is used to validate the system The gathering of data is based on the short range hourly forecast which will be used in testing the system For the short range the gathering of data was based on an hourly forecast of the Accuweather for Quezon City which was the target location for the testing of the system The collected data include only the parameters used in this study such as the temperature pressure and relative humidity The data will be arranged in order by time and tabulated in a CSV file then saved in the folder where the Kalman Filter Fig and m file are located In executing the program MATLAB is used which contains the processes and equations of Kalman Filter the program will first ask for a CSV file After selecting the specific CSV file the system will then calculate for the parameters such as the prior estimate prior error covariance Kalman gain estimate and estimate covariance which will be essential in forecasting the next state of weather condition 30 Software Development Click Start Data Acquisition Select csv file Process the Historical Data Receive measurements for temperature humidity and pressure Tabulate and plot the estimated measurement Sends a command to PIC16F877A microcontroller for

Weather Forecast using Kalman Filter Algorithm with Warning

Contents

Download Pdf Manuals

Related Search

Related Contents