Zhang Lulu, Zhang Huaguang, Yue Xiaohui, Wang Tianbiao
College of Information Science and Engineering, Northeastern University, China.
College of Information Science and Engineering, Northeastern University, China; State Key Laboratory of Synthetical Automation for Process Industries (Northeastern University), China.
Neural Netw. 2025 Nov;191:107852. doi: 10.1016/j.neunet.2025.107852. Epub 2025 Jul 9.
This paper studies the optimal tracking problem for an unknown nonlinear systems subject to input and performance constraints. A data-driven constrained optimal tracking control scheme is designed to make the system states pursue the desired trajectory while minimizing the cost and strictly limiting the tracking error within thepredefined zones. Specifically, a finite-time performance function is deployed to ensure that errors converge to steady-state regions within a user-defined time. Furthermore, by employing a nonquadratic cost function, a modified Hamilton-Jacobi-Bellman equation is constructed to ensure input limitations are satisfied. Subsequently, the adaptive dynamic programming algorithm, implemented with neural networks (NNs) in an actor-critic structure, is employed to learn the optimal control policy without relying on any prior information about the system dynamics. Meanwhile, the weights of the actor-critic NNs are tuned using the least-squares method based on the collected dataset. Finally, simulations on Chua's circuit demonstrate the effectiveness and benefits of the designed algorithm.
本文研究了受输入和性能约束的未知非线性系统的最优跟踪问题。设计了一种数据驱动的约束最优跟踪控制方案，以使系统状态跟踪期望轨迹，同时最小化成本并将跟踪误差严格限制在预定义区域内。具体而言，采用有限时间性能函数确保误差在用户定义的时间内收敛到稳态区域。此外，通过使用非二次成本函数，构建了修正的哈密顿 - 雅可比 - 贝尔曼方程以确保满足输入限制。随后，采用在演员 - 评论家结构中使用神经网络（NN）实现的自适应动态规划算法，在不依赖于任何关于系统动力学的先验信息的情况下学习最优控制策略。同时，基于收集的数据集使用最小二乘法调整演员 - 评论家神经网络的权重。最后，对蔡氏电路的仿真证明了所设计算法的有效性和优势。
