Irrigation System Automation Using Finite State Machine Model and Machine Learning Techniques

Abstract

Irrigation practices can be upgraded by the aid of finite state machines and machine learning techniques. The low water use efficiency (WUE) is the universal problem encountered by the existing irrigation systems. The finite automata model provides an efficient irrigation system with input features such as soil properties, crop coefficient, and weather data. The K-Nearest Neighbor (KNN) algorithm predicts crop water requirement based on crop growth stage with accuracy of 97.35% and for soil texture classification with accuracy of 93.65%. The proposed irrigation automation model improves water productivity.