基于分散化预测滤波的故障诊断方法研究

针对多故障源系统的故障诊断问题，提出一种采用分散化预测滤波与两种不同类型残差相结合的诊断方法。该方法针对不同类型故障敏感的特点，利用不同类型残差，区分系统组件／执行机构故障和传感器故障；利用模型误差估计与故障各分量的对应关系，辨识和隔离系统组件／执行机构故障；利用分散化滤波的结构，识别和隔离不同传感器故障。以卫星姿态估计系统为例，仿真验证该故障诊断方法，结果表明，分散化预测滤波与集中预测滤波相比，对多类型故障讲轩枪测识别以及系统舌柏古白体力事程．

Fault Diagnosis Method Based on a Decentralized Predictive Filter

A new fault diagnosis method for a complex system with more than one type of faults is proposed in this paper.To distinguish different faults,two kinds of residuals generated from only one decentralized predictive filter are used,because they have different sensitivities to component /actuator faults and sensor faults.Even more,for a component or actuator fault,the source of it can be found exactly depending on the relationship between residuals and faults.And for a sensor fault,it can be detected and isola...