基于强化学习的非线性输入受限系统最优控制  

OPTIMAL CONTROL OF NONLINEAR INPUT-CONSTRAINTED SYSTEMS BASED ON REINFORCEMENT LEARNING

作　　者：高晓格[1] 韩淑云[2] Gao Xiaoge;Han Shuyun(Automotive College,Anyang Vocational and Technical College,Anyang 455000,Henan,China;Faculty of Artificial Intelligence in Education,Central China Normal University,Wuhan 430079,Hubei,China)

机构地区：[1]安阳职业技术学院汽车学院,河南安阳455000 [2]华中师范大学人工智能教育学部,湖北武汉430079

出　　处：《计算机应用与软件》2025年第2期287-291,298,共6页Computer Applications and Software

基　　金：国家自然科学基金项目(61972173)。

摘　　要：针对一类输入受限的非线性系统最优跟踪控制问题,提出一种基于强化学习的自适应动态规划的控制策略。通过设计一种合适的性能指标函数解决控制系统输入受限问题;通过设计评价神经网络来估计系统的最优性能指标函数,从而求解控制系统HJB(Hamilton-Jacobi-Bellman)方程,获得最优控制输入;利用Lyapunov方法获得评价网络的权重更新率,并证明系统的跟踪误差和评价网络的权重估计误差为最终一致有界(UUB);通过数值仿真实验验证该控制策略的有效性。An adaptive dynamic programming control strategy based on reinforcement learning is designed for optimal tracking control of a class of nonlinear systems with input constraints.An appropriate performance index function was designed to solve the input limitation problem of the control system.An evaluation neural network was designed to estimate the optimal performance index function of the system,and the HJB equation of the control system was solved to obtain the optimal control input.The weight update rate of the evaluation network was obtained by using Lyapunov method,and it was proved that the tracking error of the system and the weight estimation error of the evaluation network were ultimately uniformly bounded(UUB).The optimal utilization numerical simulation results show the effectiveness of the proposed control strategy.

关 键 词：非线性系统 输入受限 强化学习 自适应动态规划 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]