基于进化粒子滤波器的混合系统故障诊断  

Fault diagnosis of hybrid systems based on the evolutionary particle filter

作　　者：莫以为[1] 萧德云[1] 

机构地区：[1]清华大学自动化系,北京100084

出　　处：《控制与决策》2004年第6期611-615,共5页Control and Decision

基　　金：国家863计划基金项目(2002AA412510;2002AA412420).

摘　　要：在混合系统中,需要同时估计系统的混合状态和系统参数.针对存在这类故障的混合系统的混合状态和参数估计问题提出一种解决方案,对混合系统使用混合自动机建模,并使用粒子滤波算法对其混合状态进行估计,对可能发生变化(故障)的参数使用进化粒子滤波算法进行估计,从而实现了混合系统的故障诊断.将这二者结合起来构成混合系统故障诊断的应用框架,实现了混合系统的混合状态和系统参数的在线估计.仿真结果表明所提出的应用框架和方法是可行的.It is necessary to estimate hybrid states and parameter simultaneously for the fault diagnosis of the hybrid system. A solution to cope with hybrid state and parameter estimations of hybrid system with time-varying parameters is proposed. Based on modeling the hybrid system with hybrid automation, an approach is presented to estimate hybrid states with particle filter. The evolutionary particle filter is used to estimate the time-varying parameter. An application frame for the fault diagnosis of the hybrid system is constructed, in which two type estimators are combined to complete the hybrid states and parameter estimations on-line. Simulation results demonstrate the feasibility of the proposed approach.

关 键 词：粒子滤波 混合系统 状态估计 参数估计 故障诊断 

分 类 号：TP206[自动化与计算机技术—检测技术与自动化装置]