IoT-Based Smart Irrigation System Using Hybrid Ensemble Models for Water Usage Prediction

Background: According to Earth.org’s report 600 million people in India faced an acute water shortage, which also significantly affects agricultural productivity as the country’s agricultural sector heavily relies on groundwater for irrigation. According to UN News in Breadbasket of India Punjab, groundwater levels have dropped significantly. India’s annual groundwater consumption is 90%. Global Citizen reported groundwater loss in India threatens millions of farmer’s ability to grow. Also, Environmental changes have strained India’s water resources. IoT-based smart irrigation systems can help minimize these challenges by optimizing water usage in agriculture.

Objective: This System’s objective is to optimize the irrigation system for maximum water efficiency and less water wastage and to enhance crop yields. This system helps to save costs by lowering water bills and energy consumption. This system provides Remote monitoring and control for farmers’ convenience. This system is scalable, configurable, and adaptive to another system by real-time data.

Methods/Materials:Soil Moisture Sensors, Temperature and Humidity Sensors, Rain Sensors, and Other Weather data collection sensors are used to collect the data so the system will provide the water when needed. Arduino or Raspberry Pi-like microcontrollers are used to process the data from sensors and control the irrigation valves based on parameters. We have used different machine learning models like SVM, Linear Regression, Decision Tree, Random Forest, Boosting, Bagging, and two hybrid ensemble models LRBoost(linear regression and boosting) and LR²F (linear regression and random forest )to predict the water usage.

Results:So after using different kinds of regression models, we have found that among all the models that we have used Ensemble Liner regression and Random Forest model outperformed the other models by acquiring an accuracy of 96.34% MSE score of 0.0016 and RMSE score of 0.040. So we have chosen the respected model.