Variable-Gain Robust Exact Differentiator-Based Neuro-Adaptive Control Design for Dynamic Wind Power Optimization

Abstract

Introducing an innovative approach, this study presents the Neuro-Adaptive Terminal Sliding Mode Control (NATSMC) for achieving Maximum Power Point Tracking (MPPT) within Permanent Magnet Synchronous Generator (PMSG)-based Wind Energy Conversion Systems (WECS). The proposed strategy effectively addresses performance challenges in the presence of uncertain disturbances, aligning seamlessly with the inherent characteristics of WECS. To realize this approach, we integrate an enhanced Finite-Time Performance Function (FTPF) with a hyperbolic tangent function, forming a Fast Terminal Sliding Mode (FTSM) surface. This configuration ensures rapid convergence with minimal overshoot. Notably, the advanced control mechanisms embedded in this technique contribute to the optimization of power generation and system stability in the dynamic context of wind dynamics. The methodology unfolds through several key steps. Initially, we transform the system model into an input-output format through coordinate transformation, enhancing its suitability for control applications. Next, we introduce a Self-Recurrent Wavelet Neural Network (SRWNN) for Lie derivative estimation, offering a robust and dynamic approach to harness wind energy effectively. Further enhancing precision and robustness, the proposed control law integrates a Variable-Gain Robust Exact Differentiator (VG-RED) to estimate derivatives under uncertain bounded disturbances. This observer minimizes high-frequency chattering, thereby improving control stability. A rigorous Lyapunov stability analysis confirms the uniform boundedness of closed-loop signals and strengthens the overall dynamic control system. Finally, extensive simulations are conducted within the WECS framework to validate the proposed approach. These simulations demonstrate significantly improved MPPT accuracy, reduced response times, and enhanced control efficiency when compared favorably with conventional control schemes found in existing literature.

Keywords

• Finite-time mechanism
• wind energy conversion systems
• neural sliding mode control
• maximum power point tracking
• self-recurrent wavelet neural network
• wind-power extraction