A model predictive control approach for cotton farm microgrid operation under uncertainties

This study addresses the problem of optimizing cotton farm operating costs under uncertainties. A model predictive control (MPC) optimization approach is adopted to maximize the usage of renewable energy and minimize the overall water pumping cost during the cotton growth and irrigation period. To deal with the uncertainties in renewable generation, water demand, precipitation and evaporation, this paper formulates the operation problem of the cotton farm pumping system as a stochastic programming problem, and then an MPC approach is adopted to solve the stochastic programming iteratively to cater to real-time changes in uncertain weather conditions and irrigation demand. Scenario generation and reduction techniques are applied to obtain typical scenarios with their probability. Then they will be used to facilitate the obtained stochastic optimization problem in all the MPC iteration steps. A case study is conducted using the historical irrigation data and local meteorological data of an actual cotton farm. The results show that MPC under uncertainty to optimize water pump operation can obtain the expected value of AU$36,386, which saves A$3,193 compared with the manual mode of operation.