An Extended Hybrid Regression and Neural Network Framework for Real-Time Smart Irrigation Management

Abstract

To improve the accuracy and reliability of smart irrigation systems, this work extends the predictive framework by integrating a hybrid ensemble regression model with web-based deployment. The proposed extension combines Gradient Boosting, XGBoost, and AdaBoost using a Voting Regressor to accurately forecast environmental parameters such as soil moisture, temperature, and humidity. These predicted features are utilized by a two-stage ANN with the proposed TANELU activation function to determine irrigation necessity and optimal watering time. A Flask-based web interface enables real-time data input, processing, and irrigation recommendations, making the system scalable, user-friendly, and adaptable to diverse agricultural conditions while effectively reducing water wastage.