This paper proposes an adaptive optimized intelligent control strategy for nonlinear strict-feedback systems with asymmetric hysteretic actuators by integrating reinforcement learning with backstepping. Although inverse hysteresis compensation is commonly employed, its performance is inherently limited by modeling inaccuracies, leading to non-negligible residual error. To address this issue, a dual-stage control framework is developed. First, an inverse asymmetric shifted Prandtl-Ishlinskii hysteresis compensator is applied to counteract the dominant hysteresis nonlinearity. Subsequently, an optimized backstepping controller is designed using a reinforcement learning-based identifier-critic-actor architecture with the dynamic surface technique to further suppress the residual error, thereby ensuring system stability and tracking performance. The main contributions of this work are threefold: 1) A simplified reinforcement learning mechanism is established, where the weight update laws for the actor and critic networks are designed to relax the persistent excitation condition while reducing computational complexity; 2) The dynamic surface technique is introduced to effectively circumvent the “differential explosion” problem inherent in conventional backstepping; 3) Adaptive parameters are incorporated to compensate for the residual error following inverse compensation. A rigorous Lyapunov-based stability analysis demonstrates that all closed-loop signals are semiglobally uniformly ultimately bounded. Simulation results confirm the effectiveness and robustness of the proposed controller.
扫码关注我们
求助内容:
应助结果提醒方式:
