基于事件触发机制的时变系统非脆弱故障诊断问题研究

在技术过程的可靠性与安全性领域，由于系统本质上所具有的时变因素，对时变系统故障诊断问题的研究十分必要；而现代控制系统受数字芯片有限精度存储、量化效应、计算误差及环境突变等因素影响，易导致诊断单元参数偏移，产生脆弱性；同时，随着通讯技术的普及，事件触发机制成为一定拓扑和资源约束下实现低功耗、高效率信息综合的有效手段。本项目针对几类典型的离散时变系统，在不同事件触发模式下，开展非脆弱故障诊断研究。内容涵盖：1)静态触发模式下的最优故障检测；2)动态触发模式下的故障诊断单元集成设计；3)自触发模式下的故障估计。力图结合系统特性和非脆弱设计的建模特征，分别从“确定性描述-指标优化”和“随机性描述-指标优化”两方面，提出若干适用于时变系统故障诊断问题的新概念和新方法，并将相关方法在液压驱动四足仿生机器人系统中进行验证，丰富故障诊断理论知识体系和应用技术基础。

In the field of reliability and safety of technical processes, due to the time-varying factors inherently in plants, fault diagnosis for time-varying systems is of significant importance. It is noticed that modern control system are affected by limited precision storage, quantization effects, calculation errors of digital chips as well as environmental abrupt changes, which may lead to deviation of the parameters of diagnostic units and induce the so-called nonfragility. Meanwhile, with the popularization of communication technology, event-triggering mechanism becomes an effective tool to achieve low-power cost but high-efficiency information integration subject to certain topologies and limited resources. This proposal aims to investigate the problem of non-fragile fault diagnosis for several typical discrete time-varying systems under different event-triggering modes. The content covers: 1) optimal fault detection using static event-triggered strategy; 2) integrated design of fault diagnosis unit using dynamic event-triggered strategy; 3) fault estimation using self-triggered strategy. The main purpose of this proposal lies in that, by combining the characteristics of the concerned systems and the ways of modelling the non-fragility, some novel concepts and methodolgy of fault diagnosis are proposed for time-varying systems, both from the manners of ‘deterministic description-performance index optimization’ and ‘stochastic description-performance index optimization’. With the aid of applying the addressed methodolody to a hydraulically actuated quadruped bionic robot system, not only the theoretical knowledge of fault diagnosis is enriched, but the foundation for practical utilization of the methods is laid.