Data-Driven Adaptive Control for Islanded Microgrid With Time Delay

Time delays have a significant and unavoidable stability impact on hierarchical control in microgrid systems. To address this challenge with a generalizable approach, this paper proposes a novel adaptive control method to mitigate the adverse effects of time delay on system stability. Firstly, linearized state-space equations are derived to conduct small-signal analysis, incorporating the effects of delays. The Chebyshev method is employed to handle the delay terms, enabling the computation of the system's characteristic root distribution. Then, a cost function based on these characteristic roots is formulated to optimize the control gain combinations, thereby establishing a mapping between communication delays and optimal control gains. Deep learning techniques are utilized to fit this complex nonlinear mapping, resulting in a high-precision predictive model. Finally, the trained model is tested by a four-DG microgrid system in MATLAB/Simulink for online application, demonstrating the superiority of the proposed adaptive control parameter method. The results highlight its effectiveness in maintaining system stability and enhancing dynamic performance under varying delay conditions.