Integration of Day-Ahead HEMS Planning with MPC for Real-Time Operation of Smart Homes with PV and V2G

This paper developed a new Home Energy Management System (HEMS) tool that integrates multi-objective day-ahead optimization with a real-time HEMS operation. The approach begins by applying the Long Short-Term Memory neural network to perform day-ahead predictions. With these results, a day-ahead multi-objective HEMS optimization aimed at minimizing energy costs and enhancing user comfort is solved by the Non-dominated Sorting Genetic Algorithm III, integrating an optimal Load Scheduling (LS) and Electric Vehicle (EV) energy management. Finally, with the LS defined by the NSGA-III, the real-time SH operation using a Model Predictive Controller (MPC) is the main contribution of the paper, which utilizes the flexibility of the EV battery via a bidirectional charger to address prediction errors and minimize energy costs. The integration of multi-objective day-ahead optimization with real time HEMS operation results in a substantial cost reduction of 19.91% compared to the baseline scenario. Furthermore, the annual case study demonstrates an additional 1.78% cost reduction when this integrated approach is compared to the use of day-ahead optimization alone. Therefore, the inclusion of the proposed MPC is justified by its ability to generate further savings without necessitating additional investments or compromising user comfort.