Multi-Stage Energy Management System Based On Stochastic Optimization and Extremum-Seeking Adaptation

Abstract

This paper proposes a Multi-stage Home Energy Management System (MS-HEMS) for power demand distribution among the Photovoltaic system (PV), the Energy Storage System (ESS), and the Electrical Power Grid (EPG). MS-HEMS consists of two layers: the Anticipative layer (AL) and the reactive layer (RL). The AL employs Particle Swarm Optimization (PSO) for day-ahead energy management based on weather and energy consumption forecasts; the RL includes an Extremum-Seeking Controller (ESC) that determines the ideal power setpoint of each source in real-time, compensating for prediction uncertainties and calculation time horizons. The optimization problem considers the energy bill, Peak to Average Ratio (PAR), and battery degradation cost. The proposed MS-HEMS is highlighted using predicted and actual measurements and increased the energy bill gain by 10.8% while reducing the PAR by 56.1% compared to the offline approach (OF-HEMS).