Adaptive Neural Network-Based Predefined-Time Nonlinear Control for Frequency Regulation of Heterogeneous Power Systems

This study develops an adaptive neural network-based predefined-time nonlinear sliding mode control (ANNBPTSMC) to achieve frequency regulation in heterogeneous power systems supported by grid-forming (GFM) energy storage. Neural network is taken to tackle the nonlinear disturbance resulting from uncertainty of grid-following (GFL) wind turbines and randomness of the load. Consequently, the proposed ANNBPTSMC scheme achieves the inherent advantages of sliding mode control (SMC), including strong robustness and rapid transient performance. Meanwhile, it exhibits enhanced capability in addressing complex nonlinear disturbances. In addition, the designed ANNBPTSMC ensures that the frequency of the heterogeneous power system reaches stability region within a predefined time. Notably, the convergence time is independent of the system’s initial states and is explicitly determined by a tunable parameter. The predefined-time frequency stabilization is rigorously verified through Lyapunov-based analytical methods. Simulations conducted in MATLAB/Simulink demonstrate that the proposed control strategy significantly outperforms existing methods in enhancing the dynamic response of heterogeneous power systems under varying operational conditions.