Power management in isolated microgrids using machine learning-based robust model predictive control
Chou‐Yi Hsu, Amit Ved, Hannah Jessie Rani R, Zayd Ajsan Balsem, Nora Rashid Najem, A. C. Singh, P. Sasi Kiran, Ankita Aggarwal, Satish Kumar Samal, Alireza Kamranfar
Abstract
Microgrids are small power generation units that can generate power using renewable energy. Given their variable nature, it is important to use an effective power control and management strategy. This paper presents a new control strategy based on a type-2 neural fuzzy robust model predictive control (MPC) for an isolated microgrid system that include various types of renewable energy sources (RES) such as photovoltaics (PVs), wind turbines (WTs), fuel cells (FCs) and battery energy storage systems (BESSs). In the presented microgrid structure in this paper, the FC and BESSs are used in parallel to increase lifespan and efficiency of the system. In the proposed fuzzy system, settings are adjusted using a restricted Boltzmann machine (RBM) and contrastive divergence (CD) algorithm, and are then used for MPC optimization. Simulations conducted in MATLAB/Simulink under three different scenarios (normal, 30% uncertainty, and 50% uncertainty) demonstrate the effectiveness of the proposed controller. The proposed method achieves a frequency deviation error of 0.02 p.u. under normal conditions, 0.03 p.u. with 30% uncertainty, and 0.05 p.u. with 50% uncertainty. The simulation results of the proposed method have been compared with type-1 fuzzy MPC and conventional droop controller, demonstrating that the proposed controller has better and faster performance than other methods.