Model based predictive control strategy for water saving drip irrigation

Abstract

Traditional irrigation control systems is characterized with inefficient management of water and often results in low water productivity index and reduced cultivation yield. In addition, insufficient water supply and high rate of water loss due to evapotranspiration increases plant stress which often affects its growth and development. Therefore, to address this issues, this paper is aimed at developing a model predictive control (MPC) strategy for water saving drip irrigation experiment that will regulate the soil moisture content within the desired field capacity and above the wilting point, while scheduling irrigation to replace the loss of water from soil and plant due to evapotranspiration in the greenhouse environment. The controller design involves a data driven predictive model identified and integrated with the MPC designer in MATLAB and thereafter exported in Simulink for simulation. The generate controller code was modified and deployed on a Raspberry Pi 4 controller to generate a pulse width modulated signal to drive the pump for the control water mixed with fertilizer. To achieve enhancement of controller an Internet of Things (IoT) integration was used for easy soil, weather, and plant monitoring which are used to update the MPC model for the irrigation control. The performance of the proposed MPC controller deployed drip irrigated Greenhouse(GH1) is benchmarked against an existing automatic evapotranspiration (ETo) model based controller in Greenhouse(GH2), with each greenhouse containing 80 poly bags of Cantaloupe plant with similar growth stage. The results obtained shows that, the proposed MPC-based irrigation system has higher water productivity index of 36.8 g/liters, good quality of fruit with average sweetness level of 13.5 Brix compared to automatic ETo-based irrigation system with 25.6 g/liters and 10.5 Brix, respectively. However, the total mass of harvested fruit for ETo-based irrigation system is higher than MPC-based irrigation system by 21.7%. The performance of the proposed MPC controller was achieved through the integration of event based scheduling with IoT monitoring as well as inclusion of evapotranspiration effect in the plant dynamics.