事件驱动的非线性网络控制系统自适应优化控制策略研究

The integration of network communication and control science has brought new developments to automation technology. However, the challenges in the stability analysis and controller design of the networked control system (NCS) are followed due to the uncertainty of the network. Considering the limited network communication resources, network-induced delay, packet dropout, and other uncertainties, this project aims at developing an event-driven adaptive optimal control strategy for nonlinear networked control system (NNCS), after realizing system modeling, adaptive optimal control, and low energy consumption control. First of all, we fully analyze the effect of network-induced delay and packet dropout on the control performance of the NNCS and study the working mode of each node in the NNCS. On the base of these, we will model the NNCS with network-induced delay and packet dropout as a stochastic nonlinear discrete-time system. Secondly, by combining the dynamic programming and reinforcement learning methods and utilizing the function approximation of neural networks, the design of adaptive optimal control can be realized. Finally, to address the issue of limited network communication resources, an event-driven NNCS control strategy is put forward by using event-driven mechanism and Lyapunov stability theory. The research results of this project will contribute to further promoting the application of NCS in the complex industrial processes.

网络通信与控制科学的融合给自动化技术带来新的发展。然而，网络不确定性问题给网络控制系统（NCS）的稳定性分析和控制器设计带来的挑战也随之而来。本项目考虑网络通信中面临的资源受限问题以及时延、丢包等不确定性因素，依次从系统建模、自适应优化控制以及低能耗控制三个方面开展事件驱动的非线性网络控制系统（NNCS）自适应优化控制策略的研究。首先，充分考虑时延、丢包对NNCS控制性能的影响，研究NNCS中各节点的工作方式，将具有时延和丢包问题的NNCS建模为随机非线性离散时间系统；其次，结合动态规划和强化学习的思想，利用神经网络的函数逼近功能，实现自适应优化控制器设计；最后针对网络通信资源有限的问题，利用事件驱动机理以及Lyapunov稳定性理论，提出NNCS的事件驱动控制策略。本项目的研究成果将为进一步推动NCS在复杂工业过程中的应用做出贡献。

本项目针对复杂非线性系统建模、基于强化学习的优化控制策略、事件触发控制策略开展研究，为解决不同问题提出了相应的控制策略，并且在机器人、直升机、波能量转换控制系统等应用领域中进行了研究。同时，本项目将单智能体强化学习算法扩展到多智能体系统中，进一步研究了多智能体一致性、任务分解等问题。.项目在研究过程中，共发表国内外SCI期刊论文10篇，EI会议论文5篇，申请国家发明专利2项（其中已获授权1项），软件著作权1项，获中国自动化学会科技进步奖一等奖1项，获江苏省优秀博士学位论文，圆满完成了项目结题的各项研究计划。