切换系统的迭代学习观测器设计及其主动容错控制研究

Switched systems could model practical systems with the multi-mode property, which has been one of the hot research topics in the control theory area. However, due to the complex dynamic characteristics in switched systems, the estimated errors of iterative learning observers can only converge to a compact set, which is determined by the estimated errors of the previous step and the disturbances. As a result, the precision of the iterative learning observers will be affected. Then, when we consider the controller design after the integration and optimization of the multi-observer estimations in the iterative processes, the traditional integrated design of the observers and fault tolerant controllers will not be suited. On the other hand, in the most existing results, the fault reconstruction problems for switches systems have been considered with the additive faults, while different types of failures with coupling relationships may occur simultaneously in practice. The aforementioned problems have brought new challenges for the iterative learning observer design and the active fault tolerant control for switched systems. This project will investigate and analyze the iterative learning observer design problem under the persistent dwell time switching signals. Then, combined with the linear transformation, the fault reconstruction problem with simultaneous multiplicative and additive faults will be investigated. Further, this project will develop the active fault tolerant control approach based on the integration of the multiple observers in the iterative processes. Parts of the theoretical results will be employed to a ship maneuvering control system. The aim of this project is to build a complete theoretical framework for iterative learning observer design and active fault tolerant control methods for switched systems, and the applicability of the fault tolerant control theories for switched systems will be improved in engineering projects.

切换系统可以建模具有多模态性质的实际系统，是控制领域的研究热点之一。然而，由于切换系统复杂的动态特性，迭代学习观测器的估计误差只能收敛到由上一步估计误差与扰动决定的集合内，估计精度易受影响。而考虑对迭代过程的多观测器估计值整合优化后再设计控制器，传统的观测器与控制器集成设计方法将不再适用。另外，目前切换系统的故障重构方法主要针对的是加性故障，实际系统中相互耦合的不同类型故障可能同时发生。这些问题给切换系统的迭代学习观测器设计与主动容错控制带来了新的挑战。本项目将在持续驻留时间切换信号下，深入研究迭代学习观测器的设计问题，并结合线性变换等策略同时重构乘性与加性多类型故障，进而研究基于多观测器迭代整合技术的主动容错控制方法。本项目部分理论成果将尝试应用在一类船舶操纵控制系统。本项目拟形成一套较为完善的切换系统迭代学习观测器设计与主动容错控制方法，提高切换系统容错控制理论在工程实践中的应用能力。

本项目研究了切换系统的迭代学习观测器设计与容错控制问题。项目首先给出了平均驻留时间切换信号下的自适应学习观测器设计方法；进而，解决了持续驻留时间切换信号下的乘性和加性多类型故障重构问题；接下来，设计了基于比例微分结构的切换系统迭代学习观测器，提高了迭代学习观测器的估计精度；最后，提出了基于自适应学习观测器的多路协同主动容错控制策略，实现对估计误差与未知故障的实时补偿。本项目形成了复杂切换律下切换系统迭代学习估计与容错控制及应用的一个较为完备的理论研究体系，相关研究成果在无人艇切换控制等领域具有潜在的应用价值，为切换系统控制理论的实际工程实践打下了坚实的基础。.项目组在控制领域高水平学术期刊与会议发表学术论文10篇，授权国家发明专利4项。其中，项目负责人以第一作者在控制领域权威期刊IEEE Transactions系列汇刊上发表长文(Regular paper) 5篇。1篇论文获得IEEE-ICUS2021会议最佳论文奖。基于本项目的研究成果，项目负责人获批了国家自然科学基金面上项目、中国博士后科学基金特别资助、黑龙江省自然科学基金联合引导项目等，培养(正在培养)硕士研究生7名。