Reinforcement Learning-Based Adaptive Finite-Time Performance Constraint Control for Nonlinear Systems
本文针对非线性系统，利用强化学习的评论家-演员神经网络，结合加幂积分器和预设性能技术，设计了一种自适应有限时间最优控制器，使系统在有限时间内稳定且状态误差满足预设约束，同时最小化成本函数。
This article focuses on the issue of reinforcement learning (RL)-based adaptive optimal finite-time performance constraint control for nonlinear systems. By the aid of RL-based critic-actor neural networks (NNs) construction, an optimal finite-time adaptive performance constraint controller is constructed. Via the adding a power integrator and prescribed performance techniques, a performance constraint-based adaptive finite-time optimal control strategy is developed, which demonstrates the considered system is semi-global practical finite-time stability (SGPFS), and all state errors can remain within a preset error constraint in finite time. Meanwhile, the proposed optimal control strategy can minimum the corresponding cost function. Finally, a numerical example is implemented to verify the feasibility of the developed control strategy and theory.