Optimal fractional fuzzy sliding-mode control for fractional-order fuzzy systems based on actor-critic reinforcement learning scheme

초록

This study presents a novel optimal fractional-order sliding mode control (FSMC) strategy for fractional-order systems, formulated using the Takagi-Sugeno fuzzy model and integral reinforcement learning (RL). The proposed FSMC framework ensures asymptotic stability by applying fractional-order calculus and effectively reducing chattering effects. An equivalent integer-order auxiliary system and a global performance index derive the fractional-order Hamilton-Jacobi-Bellman equation, which provides the optimal control policy. The control solution is obtained using a neural network-based actor-critic RL architecture, which adaptively updates the controller parameters by adjusting the input and output weights. This adaptive learning mechanism improves the robustness of the system and the tracking performance. The effectiveness of the proposed method is validated through simulation studies and control engineering applications, demonstrating its superior performance in terms of disturbance rejection and optimal tracking capability.

키워드