混杂系统粒子滤波混合状态估计及故障诊断算法

混杂系统同时包含连续动态特性和离散动态特性,并且两种动态相互作用,使其故障诊断变得更加困难。针对此问题，提出了一种混合系统粒子滤波混合状态估计及故障诊断算法,提高了现有方法的适用范围和诊断效率。针对混杂系统受控迁移、自治迁移和随机迁移等特点,首先利用随机混杂自动机对系统离散状态(包括故障)和连续状态进行统一建模,重点对现有基于扩展卡尔曼粒子滤波的连续估计算法进行改进,支持利用在线监测数据来估计混杂系统各类迁移产生的各种离散和连续状态,最后根据离散状态估计结果快速实现故障诊断。通过对典型非线性混杂系统的故障诊断,证明了该方法的有效性。

Hybrid state estimation and fault diagnosis algorithm of hybrid systems using particle filter

Hybrid systems are composed of discrete event dynamic systems and continuous time dynamic systems, which interact with each other. It leads to that the fault diagnosis of hybrid systems is particularly difficult. In order to expand the scope of application and improve the diagnosis efficiency, a hybrid state estimation based hybrid systems fault diagnosis method is proposed. Considering the controlled migration, the autonomous migration and the stochastic migration of hybrid systems, the discrete states (including fault states) and continuous states of the system are modeled based on the stochastic hybrid automaton. The common extended Kalman particle filter based hybrid estimation algorithm is developed so as to be applied in the hybrid estimation of discrete and continuous states produced by the migrations of hybrid systems. Finally, the fault diagnosis can be achieved rapidly according to the estimated result of discrete states. A simulation experiment is employed to conduct the fault diagnosis on a typical nonlinear hybrid system, and the results indicate that this method is effective.