A scenario-based stochastic model predictive control approach for microgrid operation at an Australian cotton farm under uncertainties

Abstract

This study presents a scenario-based model predictive control (MPC) approach to minimize the cotton farm microgrid operational cost under uncertainties. Uncertainties in cotton farms may come from renewable energy generation, water demand, precipitation, and evaporation, so the cotton field pumping system operation can be formulated as a stochastic MPC problem to accommodate uncertain climate conditions and real-time changes in irrigation demand. Scenario generation and reduction techniques can obtain typical scenarios and their probabilities. The typical scenarios can be used in the MPC iterative step to facilitate modelling the proposed stochastic optimization problem. This study discusses static and dynamic uncertainty modelling techniques used for MPC, and each technique is analysed separately in grid-connected and islanded microgrids through case studies. In the grid-connected dynamic scenario-based MPC, the operational cost is AU$ 18,797 over the entire irrigation period, which is AU$ 8759 lower than that of the standard MPC. Furthermore, for the islanded dynamic scenario-based MPC, the operational cost is AU$ 24,443 over the entire irrigation period, which is AU$ 6721 lower than the standard MPC.