Optimization, Modeling and Implementation of Plant Water Consumption Control Using Genetic Algorithm and Artificial Neural Network in a Hybrid Structure

Abstract

Smart and water-saving systems recommended for agricultural irrigation are of great importance for countries and regions with limited water resources. Thus, it is aimed to develop a cost-efficient smart irrigation system that can be monitored by smart devices and smart phones for walnut plants using an artificial neural network. Accordingly, three different artificial neural network models based on temperature (HM1), temperature-soil moisture value (HM2), and temperature-soil moisture-plant life cycle (HM3) are formed to estimate the water needs of the plant considering climatic and environmental factors. The optimization of parameters used in artificial neural networks is realized with the genetic algorithm by building a hybrid structure to acquire the maximum efficiency of smart irrigation systems in three different models formed. In the estimation of irrigation program formed according to the reference evapotranspiration (ETo) estimated by the FAO Penman-Monteith equation and the relevant evapotranspiration (ETc) value of the walnut plant, HM2 model provides more efficient results than HM1 by 16.6% and HM3 by 6.5%. The optimal smart irrigation system presented by the hybrid model in line with HM2, the most successful model, is analyzed, and an application is created using data obtained. Additionally, which input parameters should be used to obtain more efficient results in the smart irrigation system is optimized with the developed hybrid structures. The presented smart irrigation system consumes less water and provides savings regarding water consumption measurements and manpower compared to classical methods such as wild irrigation, which cannot be measured and consume too much water.