Optimal Multitimescale Model Predictive Control-Based Power Allocation of Hybrid Energy Storage Systems via Reference Curve Tracking

Abstract

The intermittency of renewable energy has made stable power supply a focal and challenging research topic in today’s power system. However, most existing intelligence-enabled control protocols pose significant computational complexity, hindering their practical applications in multitimescale power control scenarios. This article proposes an optimal power allocation method for hybrid energy storage systems (HESSs) based on complementary power reference curves. The approach leverages batteries’ dynamic characteristics and output properties and algorithmic features to construct complementary power reference curves. Through real-time tracking of these reference curves, the method achieves coordinated and stable power output across multiple timescales. The improved model predictive control (MPC) is integrated into the control strategy to efficiently convert power tracking into current prediction. This approach enhances system dynamic response without needing excessive computational resources. Furthermore, the proposed method is integrated into a dynamic voltage restorer (DVR) for power quality control. Finally, the feasibility and effectiveness of the proposed approach is validated through hardware experiments.