Computationally Efficient Model Predictive Control for Enhanced Primary-Level Management of Standalone Microgrids With Hybrid Storage Systems

This paper proposes a highly efficient model predictive control (MPC) technique for primary-level control of a hybrid storage system (HSS) in a microgrid with renewables. Generally, active set (AS)-based MPCs are used to control HSS-based microgrid systems, which are computationally complex. The high-complexity prevents making control decisions within the sampling time, which negatively affects the robustness of the control system. Moreover, high complexity necessitates powerful controllers, which increase the system's overall cost and energy consumption. On the other hand, less complex controls, such as supervisory control, reduce the storage system's stability, reliability, and lifetime. The proposed method is a closed-form AS-based MPC that shares the mutual advantages of both AS-based MPC and supervisory controllers with additional robustness. It thus significantly reduces the computational complexity while being stable and reliable, improving storage lifetime, and having higher robustness compared to MPC and supervisory controllers. The proposed method is compared to modified AS based MPC, Hildreth quadratic programming (HQP), and sequential quadratic programming (SQP) based MPC recently employed for an HSS microgrid application. The analysis showed that the proposed method could significantly outperform the conventional MPC methods in computational complexity and robustness, as well as the supervisory controller in terms of reliability, stability, enhanced battery lifetime, and robustness.