基于置信椭球的V/M故障检测与隔离算法改进

动态系统的故障诊断技术在组合导航系统中有着重要意义.首先介绍了一类经典状态一致性故障检测/隔离方法-Kerr的基于置信椭球的V/M处理方法的原理与处理过程.Kerr的V/M方法以偏差方向的累计和进行隔离判断,可能发生两种情况下的隔离模糊问题.为此,对Kerr的经典V/M算法进行了改造,在其投票函数中采用偏差距离的累计代替偏差方向的累计,使得故障隔离的敏感性大幅度增加,解决了经典V/M算法在实用中存在的第一种情况下的隔离模糊问题.采用算例证明了新方法的有效性.

Improvement on V/M Fault Detection and Isolation Algorithm based on Confidence Ellipsoids

This paper firstly introduced Kerr's classic V/M method based on confidence ellipsoids,which belongs to a type of mostly used FDI methods based on state consistency.Kerr's V/M method uses the sum of bias directions to determine the isolation,as may lead to two kinds of isolation ambiguities.To solve this dificiency,we innovated Kerr's V/M algorithm.The new V/M algorithm replaces bias directions in the vote function by bias ranges,so that greatly improved the sensitivity of fault isolation,and then solved the first kind of problem of isolation ambiguity that the classical V/M algorithm often suffers from in practice.