Energy management of the grid-connected residential photovoltaic-battery system using model predictive control coupled with dynamic programming

This study developed a method coupling model predictive control (MPC) with dynamic programming (DP) for consecutive operation scheduling of the PVB system. The actual power data of a household were collected for optimization. Three strategies, including the economic optimization strategy (OP_C), the grid-power optimization strategy (OP_P), and the maximizing self-consumption strategy (MSC) were proposed, and compared comprehensively. Comparative experiments were conducted by reproducing the historical PV and load conditions in the lab. It was found that the developed method has the advantage of considering interactive effects of adjacent days. All the developed strategies could be implemented well in experiment, with the maximum relative deviation of 8.89 % to the simulated results. The OP_C strategy reduced the operational costs at the expense of weakening grid stability and lowering SCR and SSR, while the OP_P strategy achieved the grid-friendliness at the expense of increasing both the operational cost and battery aging. The MSC strategy has the compromise performance in both the operational economy and the grid-power stability. Limited by the economic constraints, the battery capacity cannot be fully used for the OP_C strategy. As for the OP_P strategy, there is an optimal PV capacity for any battery capacity to minimize the grid power fluctuation. When the economy and the grid-power stability were treated equally (λ₁ = λ₂ = 0.5), the operational cost, SCR and SSR were similar to that of the MSC strategy, while the grid power fluctuation was much smaller. The findings in this study can be used as guidance for optimal design and operational management of residential PVB systems.