基于新型事件触发的非线性随机系统容错控制与优化方法研究

Nonlinear stochastic system is becoming a research focus in automatic control fields as it can describe complex industrial systems more accurately. However, the problems of industrial network, such as data transmission delay, packet dropout and communication constraints, make the integration and application of network technique and nonlinear stochastic systems be a new challenge. This project focuses on the fault-tolerant control and optimization problem of nonlinear stochastic systems for practical industrial application scenarios. (1)For the real industrial network with limited communication resources, a novel event-triggered data transmission mechanism is proposed. Combining with fuzzy modeling theory, a novel event-triggered fuzzy fault estimation observer is designed, which can achieve desired fault estimation performance. (2)For the fault-tolerant tracking control problem of nonlinear stochastic systems, an adaptive fuzzy fault-tolerant tracking control approach based on fault estimation information and prescribed performance function is proposed to improve the tracking control performance. (3) For the fault-tolerant control optimization problem of nonlinear stochastic systems, the reinforcement learning method is applied to online adjust the parameters and structure of controller for optimizing fault-tolerant controller, which can further improve fault compensation and tracking control performance. The research results will provide an important theoretical foundation and key technology for the application of nonlinear networked stochastic systems in real industrial fields.

非线性随机系统能够精确描述复杂工业系统，成为自动控制领域的研究热点。然而，工业网络中存在数据传输时延、丢包和通信约束等，使得网络技术与非线性随机系统的融合应用成为新的挑战。本项目面向实际工业应用场景，研究非线性随机系统的容错控制与优化问题。(1)针对实际工业网络通信资源受限问题，拟提出基于新型事件触发的数据传输机制，结合模糊建模理论设计故障估计观测器，实现期望的故障估计性能。(2)针对非线性随机系统的容错跟踪控制问题，根据故障估计信息，拟提出基于预设性能函数的自适应模糊容错跟踪控制方法，提升系统的跟踪控制性能。(3)针对非线性随机系统的容错控制优化问题，拟引入强化学习在线实时调整控制器参数与结构，优化容错控制器，进一步提升系统的故障补偿效果与跟踪控制性能。研究成果将为非线性网络化随机系统的实际工业应用提供重要理论基础与关键技术。

本项目主要围绕复杂工业网络环境下非线性随机系统的容错控制与优化问题展开研究工作。首先，提出了基于动态事件触发机制的数据传输方案来节约有限的网络通信资源。针对控制系统传感器存在非线性，提出了基于动态事件触发机制的故障检测方法。其次，针对非线性系统存在信号量化和网络攻击，提出了基于动态事件触发机制的故障检测与控制方法。并进一步针对通信网络存在欺骗攻击，提出了基于动态事件触发机制和滚动优化思想的的模型预测控制方法。研究成果为网络化非线性随机系统的研究和发展提供理论基础和关键技术支撑。在本项目的执行过程中，共发表学术论文6篇，授权专利6项，培养在读硕士研究生3人，协助培养在读博士研究生2人。