Smart Farming Using Machine Learning Algorithms

1,2,3,4

5

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056

Volume: 10 Issue: 01 | Jan 2023 www.irjet.net p-ISSN: 2395-0072

Smart Farming Using Machine Learning Algorithms

Vibha Pawar1 , Tejaswini More2 , Pradnya Gaikwad3 , Anuradha Patil4 , D.J. Dattawadkar5

Abstract –

The majority of India's agricultural products have been negatively impacted by climate change in terms of performance over the past 20 years. Prior to harvest, crop output predictions would aid farmers and policymakers in deciding on the best course of action for marketing and storage. Before cultivating on the agricultural field, this project will assist the farmers in learning the yield of their crop, enabling them to make the best choices. By creating a working prototype of an interactive predictionsystem,ittries to find a solution. It will be put into practice to implement such a system with a user-friendly web-based graphic user interface and the machine learning algorithm. The results will be made available to the farmers of the prediction. Therefore, there are various ways or algorithms for this type of data analytics in crop prediction, and we can anticipate crop production with the aid of those algorithms. It employs the random forest algorithm. There are no suitable solutions or technologies to deal with the scenario we are in, despite the analysis of all these concerns and problems, including weather, temperature, humidity, rainfall, and moisture. In India, there are numerous approaches to boost agricultural economic growth. Machine Learning is the technique most widely used in today’s world. ANN is the most widely used algorithm for prediction. It is based on a collection of nodes. These nodes are called neurons. Neurons work in a way similar to the human brain. Hence, it gives more appropriate results.

1. INTRODUCTION

About70%ofIndiansworkintheagriculturalsector,which iswhyitwasimportanttoincludeitinouranalysisofthe economy of the nation. Crop yield prediction is a huge problem in the agricultural sector. Crop prediction is the processoffiguringoutwhatthefarmercangrow.Buildinga system that would operate with maximum accuracy and takeintoaccountallsignificantvariablesthatcanaffectthe outcome of the crop prediction is imperative. Numerous studies have been conducted to forecast the crop that a farmer can grow. Most of the farmers try to know crop yieldandwhetheritmeetstheirexpectations.Theyevaluate the previous experience of the farmer on a specific crop

yield.Thedecisionofafarmeronwhichcroptocultivateis typicallyinfluencedbyhisintuitionandotherunimportant variables, such as the desire to make quick money, ignoranceofmarketdemand,exaggerationofasoil'sability to support a particular crop, and so forth. The farmer's family'sfinancialsituationcouldbeseverelystrainedbya choicehemade. Maybethisisoneofthenumerousfactors contributingtotheinnumerablefarmersuicidecasesthat the media reports on every day. we suggest a system, an intelligentsystem,which,beforeadvisingtheuseronthe best crop to plant, will take into account environmental factors (such as temperature and rainfall) and soil characteristics such as pH value, soil type, and nutrient concentration.

A website created as part of the "Smart Farming using MachineLearning"projecthelpsfarmersbypredictingthe crop that will be grown. This calls for specific conditions includingtemperature,precipitation,andsoilmoisture.The suggestedsystemspecifiesthekindofcropsafarmermay raiseonhisproperty.Asuitabledatasetthatdescribesthe best crop is required for the crop prediction process in ordertoreducethelikelihoodofcropfailure.Anotherthing to keep in mind is that it uses technology for prediction moreanddoesnotrequirealotofhumanresources.

1.1 Introduction to ANN

ArtificialNeuralNetworks,alsoknownasANNisthe machine learning algorithm which works similar to human brain. The human nervous system contains neurons.Similartoneurons,ANNalgorithmworks.Just liketheworkingoftheseneuronsfromthepastdata, ANNlearnsfromdata.

Itprovidestheoutputinthe formofclassificationsor predictions.ANNisfeed-forwardapproach thatsends datainstraightforwardway.ANNcontainsthreelayers input layer, hidden layer and output layer. In input layerittakesdataandsendtothehiddenlayer.Hidden layer applies various activation function, preprocess that data and send to the output layer. Output layer givesdesiredoutput.

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056

Volume: 10 Issue: 01 | Jan 2023 www.irjet.net p-ISSN: 2395-0072

1.2 Introduction to IOT

IOT stands for Internet Of Things. It is collection of variousembeddeddevicessuchassensors,software andotherdifferenttechnologies.IOTisbasicallyused for the purpose of connecting and exchanging the information or data through the devices over the internet. IOT applications use the machine learning algorithms to analyze large amount of data. Machine learning technology is used to predict the possible outputonthebasisofpastandfuturedata.IOTsensors collect the data through the sensors and send to the server. The model is trained using this data and predicts the output using ANN machine learning algorithm.

2. Related Work

[1]Thelimitsofcurrenttechnologiesandtheirusefulness in yield prediction are highlighted in this research. This methodassistsauserinexploringpotentialcropsandtheir yieldinordertomakemoreinformedjudgements.Onthe provided datasets from the states of Maharashtra and Karnataka, many machine learning algorithms, including RandomForest,ANN,SVM,MLR,andKNN,weredeployed and assessed for yield to accuracy. The results show that RandomForestRegression,whichhas95%accuracy,isthe best standard algorithm when applied to the given datasets.[2] In this paper, the author has discussed the effect of weather conditions on crop yield. The paper focuses on artificial neural network technology. The parametersusedaresensorparameterssuchastypeofsoil, Ph value, N, P, K values, etc. The multilayer perceptron modelisdevelopedbyusinganeuralnetwork.Theaccuracy ofthemodelisvalidatedusingcross-validation.Theweka toolisusedforexecution.Theaccuracyobtainedis97.5%. Thepaperfocusesonthedescriptionofdifferentnumbers of agronomic-based models. Models have used artificial neuralnetworkalgorithm.Cropyieldpredictionusingaerial pictures have been utilized for taking decision-related harvesting.ThispaperreportsontheuseofArtificialNeural Networks to predict the rice crop yield for Maharashtra state,India.[3]Theauthorsofthisarticlesuggestedamodel thatusesmeteorologicaldatarecords fromthepastasthe trainingset.Theselectionofacropbasedonthepredicted productionrate,whichisaffectedbyanumberofcriteria,is demonstratedbytheuseofnumerousmethods,including Artificial Neural Networks, K-Nearest Neighbours, and RegularizedGreedyForest.Anothercleverapproach,shown in, enables the prediction of soil characteristics like phosphorus concentration. To obtain high prediction accuracy, the authors apply a variety of classification algorithms,includingNaiveBayes,C4.5,LinearRegression, andLeastMedianSquare.[4]Theproject'sgoalistoforecast the crop. In this research, an intelligent farming strategy basedontwoemergingtechnologies:machinelearningand the Internet of Things is provided. This method aims to

supportfarmersinmakingwell-informedcropforecasting decisions.

3. Role of IOT

Proper placement of sensors is necessary. This point includesthe properlocation ofsensors.For prediction of cropdifferentparametersliketemperature,humidity,soil moisture,rainfallofthatparticularregionareresponsible.

For more accuracy soil moisture sensor are placed at multiplepointsinthesoilinthefarm.Throughtemperature andsoilmoisturesensorlivedataiscollected.Rainfalland moisturelevelaretakenmanually.

In this paper raspberry-pi model 4B+ is used as a computingunit.Thetemperatureandsoilmoisturesensors send datatoraspberry-piunit.Herealldataiscollectedand ANNalgorithmisappliedtoproduceoutput.

Python3.5versionisusedtodocodingsothatsensordata isprocessedandresultisproduced.

Alldatafromsensorsiscollectedintocomputingunit.

Sensordata andtheresultthatisproducedisstoredasa primarystorage.

4. System Architecture

Fig -1:SystemArchitecture

Soil-moisturesensorandtemperaturesensorcollectthelive dataandwillsendtotheRaspberrypi.Raspberrypistore andpreprocessthatdataandthensenddatatothemodel. Thisdataissentbythemodeltotheserverforvisualization. Modelistrainedusingthisdataforpredictionofcrop.ANN algorithm is used to train the model. This trained ANN modelwillgivethecroptobegrown.

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056

Volume: 10 Issue: 01 | Jan 2023 www.irjet.net p-ISSN: 2395-0072

5. Cause and error diagram

Fig

-3

:Homepage

After clicking on register user can see registration page. User can register to website by filling the details in the registrationformandthenclickonsubmit.Thenclickonthe returntologinbuttonwhichredirectyoutologinpage.

Fig -2: FishboneDiagram

TheclimateandIOTarethefactorsthatareconsideredfor the predicton of crop. The factors affecting climate are rainfallandtemperature.Thefactors thatareaffectingIoT areitcan havelessspeedforthetransferofdata.UsingIoT can lead to high cost. These are the factors affecting the prediction.

6. Proposed System

We proposed a smart farming crop recommendations systemwhichtakesintoconsiderationalltheappropriate parameters including temperature, rainfall, and soil moisture to predict crop suitability. The proposed model provides crop selection based on economic and environmentalconditionsandbenefittomaximizethecrop yieldthatwillbesubsequentlyhelptomeettheincreasing demand for the country’s food supplies. We also provide realtimedatacollectionsbysensorsRaspberrypiandIOT technology. This system is fundamental concern with performingtheprimaryfunctionAgroconsultant,whichis providingcroprecommendationstofarmers.

7. Results

Thetrainingdatathatisusedcontainsinformationabout ph,soil-moisture,rainfall,humidity.

Farmersuseagroconsultant websitefortheprediction of crop.

Fig -4:Registrationpage

After clicking register option the user will redirect to the login page. By verifying register credentials admin will activatetheuseraccount.Usercanloginwiththeverified credentials.

Fig -5:Login page

Selectthecroppredictionoptionfromthehomepagemenu. Fillalltherequiredinformationandclickonsubmitbutton.

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056

Volume: 10 Issue: 01 | Jan 2023 www.irjet.net p-ISSN: 2395-0072

Fig -6:Cropprediction

Weatherinformationgivesthedataaboutthetemperature alongwithdateandtimefromusingpublicAPI.

Fig -7:WeatherInformation

This shows the live data which is gathered by the IoT device.

Fig -9:PasswordChange

Thepasswordrecoveryisaspecialthinginwhichweuse email id to get our new password using forget password button.

Fig -8:IoTInformation

This module is used for changing the password with old password.

Fig -10:PasswordRecovery

Theresulttabshowswhateverpredictionofthecropforthe givendatavalues.

Fig -11:Results

8. CONCLUSION

We get the crop prediction for the specific environment using different parameters. The ANN model gives the accuracyof95%approximately.Thisisusefulforfarmersto increasetheprofitandproductivityofthecrop.

International Research Journal of Engineering and Technology (IRJET) e-ISSN: 2395-0056

Volume: 10 Issue: 01 | Jan 2023 www.irjet.net p-ISSN: 2395-0072

9. Future Work

Building this particular application in the regional languages, so that it would be more comfortable for farmers. Crop disease detection and prevention. A generalized prediction model for various crops by considering other parameters like humidity and solar radiationcanbedeveloped.Alsogivinginformationabout micronutrients

10. REFERENCES

[1] Shilpa Mangesh Pande, Dr. Prem Kumar Ramesh, Anmol,B.RAishwarya,KarunaRohilla,KumarShaurya, "CropRecommenderSystem'UsingMachineLearning Approach",IEEEConference2021

[2] Shivani S. Kale, RRC, VTU, Belgaum, Karnataka,"A MachineLearningApproachtoPredictCropYieldand SuccessRate",IEEE2019.

[3] Zeel Doshi, Subhash Nadkarni,"Agro Consultant: Intelligent Crop Recommendation System Using MachineLearningAlgorithm”,IEEE2018

[4] AkshataWani,VirajYadav,RitikaTodankar,Pradnyan Wade,Prof.ShikhaMalik.”CropPredictionusingIoT& MachineLearningAlgorithm”.

2022, IRJET | Impact Factor value: 7.529 | ISO 9001:2008 Certified Journal