Analysing agricultural information through machine learning and artificial intelligence for SMART IRRIGATION

Abstract

Traditional agriculture has been the global foundation for development for centuries. However, to meet this demand and the exponential growth of the population, farmers need water to irrigate their property. Farmers require a fix that modifies their business practices because of the scarcity of using this resource. To keep up with and satisfy Demand, Agriculture 4.0 has become a reality thanks to new technologies. Automated Irrigation: Smart irrigation systems can automatically adjust schedules based on real-time data, such as weather conditions, soil moisture levels, and evapotranspiration rates. This helps farmers apply water more efficiently, reducing water waste and improving crop health. By collecting and analysing agricultural information through a combination of the Internet of Things and artificial intelligence, decisions have become increasingly precise to make decision-making easier. A cost-effective, intelligent, and adaptable irrigation strategy that can be used in a variety of settings is presented in this paper. For smart agriculture, machine learning algorithms are the foundation for this strategy. MongoDB and the Node-RED platform were used. We created an acquisition map from a collection of sensors (soil moisture, temperature, and rain) in a setting that guaranteed improved plant development for months. Based on our data, we used many different models: SVM, Simple Bayes, KNN, and Regression using logit. The outcomes showed that K-Nearest Neighbours outperformed other models (LR, SVM, NB) with a 98.6% identification rate and a 0.12 root mean square error (RMSE). In addition, we finally made available the online tool that combines the predictions made by our models with the different data released by the sensors for improved environmental visualisation and control.