Hybrid Decision Model for Weather Dependent Farm Irrigation Using Resilient Backpropagation Based Neural Network Pattern Classification and Fuzzy Logic

Abstract

Irrigation in agricultural lands plays a crucial role in water and soil conservation. Real-time prediction of soil moisture content using wireless sensor network (WSN) based soil and environmental parameters sensing may provide an efficient platform to meet the irrigation requirement of agriculture land. In this research article, we have proposed Resilient Back-propagation optimization technique to train neural network pattern classification algorithm for the prediction of soil moisture content. Finally, the predicted soil moisture content is used by fuzzy weather model for generating adequate suggestions regarding irrigation requirement. The fuzzy model is developed by considering different weather parameters like sun light intensity, wind speed, environment humidity and environment temperature. Different weather conditions like cloudy situation, low pressure, cyclone and storm conditions are simulated in the fuzzy model. The soil moisture content prediction algorithm is tested with soil moisture content in each 1 h advance by considering eleven different soil and environmental parameters collected during a field test. The prediction errors are analysed using MSE (Mean Square Error), RMSE (Root Mean Square Error), and R-squared error.