A reinforcement learning-based control system for higher resonance frequency conditions of grid-integrated LCL-filtered BESS

D. Khan, M. Qais, P. Hu

Research output: Contribution to journalArticlepeer-review

Abstract

The battery energy storage system (BESS) is utilized to improve the stability of the power system that includes renewable energy resources. BESSs are integrated into the grid via an inverter and an LCL filter. However, the LCL filter introduces a resonance challenge due to the variations of the equivalent grid inductance. This issue is adeptly addressed through the application of complex single-loop and multi-loop active damping strategies. However, these strategies involve the use of a variety of dampers, which subsequently increase the complexity and cost of the overall system. This paper designs a deep reinforcement learning-based proportional-resonant (RL-PR) controller to improve grid stability under different operating conditions. Firstly, the proposed RL-PR controller is trained by maximizing a reward function and minimizing the total harmonic distortion of the injected current from BESS to the grid and the absolute error between it and the reference current. After that, the trained RL-PR controller is applied in the control system of the LCL-filtered grid-connected BESS that is modeled using MATLAB/SIMULINK. The obtained results demonstrate that the RL-PR controller operates effectively within a resonance frequency range from 0 to 1/6th of the system sampling frequency and exhibits robustness to grid impedance variations of up to 100 % of its nominal value. Furthermore, the controller maintains robust performance across varying levels of injected current and even in the face of short-circuit conditions. Finally, to validate the efficacy of the proposed method, an experimental system based on the controller hardware in the loop approach is developed, and the results from this setup align with MATLAB/SIMULINK simulations.
Original languageUndefined/Unknown
Article number1121373
JournalJournal of Energy Storage
Volume93
Early online date4 Jun 2024
DOIs
Publication statusPublished - 15 Jul 2024

Cite this