An LSTM-based mixed-integer model predictive control for irrigation scheduling

The development of well-devised irrigation scheduling methods is desirable from the perspectives of plant quality and water conservation. Accordingly, in this article, a mixed-integer model predictive control system is proposed to address the daily irrigation scheduling problem. In this framework, a long short-term memory (LSTM) model of the soil-crop-atmosphere system is employed to evaluate the objective of ensuring optimal water uptake in crops while minimizing total water consumption and irrigation costs. To enhance the computational efficiency of the proposed method, a heuristic method involving the logistic sigmoid function is used to approximate the binary variable that arises in the mixed-integer formulation. Through computer simulations, the proposed scheduler is applied to homogeneous and spatially variable fields. The results of these simulation experiments reveal that the proposed method can prescribe optimal/near-optimal irrigation schedules that are typical of irrigation practice within practical computational budgets.