分布式非线性网络化系统的状态估计与故障诊断及其应用

The past decades have witnessed the rapid increasing of system size and complexity, motivated by the development of communication technology and microelectronic technique. State estimation and fault diagnosis are two important research areas that focus on system safety. However, most of the available literature on model-based state estimation and fault diagnosis are devoted to centralized systems, where the filter or fault diagnosis scheme has access to all the available measurements. Motivated by the background of safety formation flying of an unmanned helicopter swarm, we propose the state estimation and fault diagnosis issues for distributed nonlinear networked systems. Distributed filter or fault diagnosis is more suitable than centralized for large-scale interconnected dynamical systems due to its lower complexity and less use of network resources. Traditional filtering and fault diagnosis schemes may not be applied to distributed networked systems, since not all measurements are available in every node. ..In this project, some new theories, new techniques and new methodologies will be proposed on the scientific problems of consensus filtering and fault diagnosis for distributed nonlinear networked systems. We consider a class of nonlinear networked systems, whose sensors are deployed dispersedly and a sensor node can communicate with its neighbour nodes. Due to the limited capability of network cable, data may suffer from randomly occurring missing, randomly occurring communication delay, as well as randomly occurring quantization, resulting the stochastic incomplete measurements. ..In this project, we will deal with a more general class of nonlinear systems, and extended Kalman filtering, consensus filtering, as well as fault diagnosis techniques will be obtained. We will try to make both theoretical breakthrough as well as important technical achievements and some results will be tested experimentally based on a real unmanned helicopter swarm platform. The obtained results should provide theoretical support to guarantee the safe formation flying of an unmanned helicopter swarm.

集成电路与通讯技术的飞速发展使得当今系统的规模越来越庞大、系统复杂度也逐渐增加，同时，系统的安全性要求也日益提高。现有较为成熟的状态估计与故障诊断问题的理论研究成果多是在集中式框架下针对线性系统或特殊的非线性系统给出的，其应用到大规模网络化系统中时会造成通讯与计算的严重负担，且无法实现对各分布式元件故障的单独诊断。本项目从无人直升机安全编队飞行的实际需求出发，考虑具有分布式传感器的一类非线性网络化系统，每个传感器可与邻近传感器通过通信网络进行数据交互，网络的引入会给数据的传输过程带来随机的不完整性。基于此类系统，我们研究分布式非线性网络化系统的状态估计及故障诊断问题，给出分布式一致滤波及故障诊断的若干理论成果，并在无人直升机群的实验平台上进行实际验证。本课题的研究成果不但具有重要的理论意义，而且对提高分布式非线性网络化系统的安全性具备潜在的工程应用价值。

集成电路与通讯技术的飞速发展使得当今系统的规模越来越庞大、系统复杂度也逐渐增加，同时，系统的安全性要求也日益提高。现有较为成熟的状态估计与故障诊断问题的理论研究成果多是在集中式框架下针对线性系统或特殊的非线性系统给出的，其应用到大规模网络化系统时会造成通讯与计算的严重负担，且无法实现对各分布式元件故障的单独诊断。本项目解决了非线性网络化系统的分布式滤波、状态估计、故障检测与诊断、控制等问题。具体地，给出了新的非线性网络化系统的多指标滤波算法（H-infinity滤波、分布式H-infinity滤波、递推强跟踪滤波、最优滤波、增益受限滤波等），得到了新的非线性网络化系统的故障估计与诊断策略（多项式近似故障检测、H-infinity故障估计、最小二乘故障检测与诊断等），提出了新的非线性网络化系统的多指标控制方法（事件触发一致性控制、H-infinity控制、可靠控制、滑模控制等）；此外，将提出的理论结果应用于解决多智能体系统、传感器网络、复杂网络的分析与设计问题，给出了二阶多智能体系统的分布式故障检测方法、基于事件触发机制的时滞复杂网络状态估计策略、基于无线传感器网络的事件触发分布式滤波算法等。本项目推动了网络化控制理论和非线性系统理论的进一步发展，研究成果为解决非线性网络化系统的分析和综合问题提供了有效的理论与方法。