ML Based Forest Fire Prediction and Forest Enviroment Monito

Every year we lost a large number trees , herbs , shrubs ,and many animal species due to forest fire also we the forest size getting smaller due to illegal logging of trees in forest . It takes a hundreds of years to naturally grow a forest protect the forest but every year we lost a large number natural resources and forest due to forest fire . Most of the forest fires either cause due to high dryness and temperature rise in forest. As we know for sustainable and to reduce our carbon footprint we need to protect the forest because it contain a diverse amount of natural resources from herbs to medical plants to woods and animals. So we need to to reduce the loss f this important natural resource with technology. We protect a large number of accidents and losses of forest with technology and predicting the forest fire before it actually takes place

Here we use the sensors available in Nordic Thingy collect the data of forest environment like the sounds , temperature rise and fall and humidity etc and we create the ML model using this data as we know main cause of forest fire is high dryness and high temperature so our devices connected to each other captures these data the humidity ,sound , temperature , sun light and other data and predicts if the environment is very dry and temperature is very high that might cause forest fire once the prediction and possible hot spot area where the forest fire might cause the drone and plane with a water tank moisture those area with water spray so the dry leafs and other highly dry area with very high temperature which possible cause forest fire get wet and stop the forest fire before it cause.

First you need the Nordic Thingy Edge Impulse firmware . The Nordic thingy 53 device that I got have pre configured with Edge Impulse firm ware if you have not that firmware on Nordic Thingy 53 device then you need to first upload the firmware for Edge Impulse . This helps you to collect the data prepare the ML model on Edge Impulse and then deploy that on your Nordic device . For this you have two methods first one is upload the firmware using the nRF Programmer app using ios phone or nRF connect app using windows so what ever looks good to you can update the firmware using that .

Updating Edge Impulse Firmware

download the nRF programmer app on ios/android device from app store now download the EDGE Impulse firmware . Next connect the nRF thingy to the programmer app using Bluetooth and then update the firmware.

Preparing Edge Impulse

Now create the account on edge impulse https://www.edgeimpulse.com/ and then create a new project we are making AI based forest fire prediction system so

we named it forest fire detection Now we are ready to make ML model and collect data for our device.

Collecting Datasheet

Now you can either collect the data using the laptop by connecting the nRF to USB and then you can add the nRF thingy device using the serial port. Login into Edge Impulse and then open the forest fired detection project we have created earlier now you can connect the nRF device to the project using connect device option and then select the serial port and give permission for serial port in browser. Now go to data accusation and then select the sensor and port and collect the data for the cases like data for normal climate , data for humid rainy environment , data for very hot dry summer environment when the fire likely to cause. But the best way for collection the datasheet for our ML model is using the app . Download the nRF Edge Impulse app in ios device then login into your Edge Impulse Account and then connect the Nordic Thingy device and then go to data acquisition option in app and then select the sensor and the frequency and samples number for datasheet collection.for forst firs detection and predection system we need the light and humidity , temperature and air data so select the light plus environment for data accusation

No using the above method collect the datasets for ML model for all king of situation and environment and label them aquesation .Place the Nordic device in forest and the collect the data for Humid Rainy environment, Extreme Hot Summer dry environment , Hot but humid environment , Extremely Hot but night environment so in way collect the data for all kind of environment in forest and also collect the data where the Forest fire is caused and then label them correctly .

Now open the Edge Impulse and you can see the data graphs when there is extreme hot and very dry environment when forest fire likely to cause , then you can see the graph of rainy , cold and all other environment data and graphs

Training ML Model

Now open the app and then select the deploy option in navigation bay and then do the configuration like eon tuner and other and then and tap on build now our ML model get trained and based on data sheet and learn to predict and classify the environment and help us in alerting the forest fire situation before it occur

Updating The Firmware

Now download the firmware in app app and then upload the firmware to nRF Thingy device . If you are using Laptop then you can open the EDGE IMPULSE Project and then go to deploy option then select the thingy board and then you got the .bin file for firmware upload the firmware to nRF using nRF connect and programmer on laptop by connecting the nRF using Bluetooth or USB

Deploying and Predicting Forest Fire

Now you can put the nRF Nordic Thingy device in deep forest and attach it to any tree and it start showing the output based on its ML model and prediction. When ever it detect the very high bright sun light and very dery and high temperature nearly the 49 to 50 degree centigrate it shows you the forest firs likely to couse and notify you regarding that hotspot area on the app now you use the drones with water tank to wet those area so the dry grasses and leafs do not cat fire and they get wet due to humidity cause due to spray of water droplets