冗余传感器故障诊断的最优奇偶向量法与广义似然比检验法的等效性

Is OPT Better than GLT in Fault Diagnosis of Redundant Sensor System?

通过分析用于冗余传感器量测系统故障检测与隔离(FDI)的广义似然比检验(GLT)方法和最优奇偶向量检验(OPT)方法,指出用OPT方法进行FDI所存在的问题,并进行修正。证明了GLT方法和修正后的OPT方法进行FDI是完全等效(一致)的。

Some researchers believe that, compared with generalized likelihood ratio test (GLT), using optimal parity-vector test (OPT) can increase the probability of fault detection and isolation. In fact, using OPT generally gives false warnings several times those given by GLT. We first correct OPT in such a way that it gives no more false warnings than GLT. We prove mathematically that, for redundant system configured by single axis sensors of the same precision, the two methods-GLT and the corrected OPT-of fault detection and isolation lead to completely the same results. It is worth pointing out that the core of our mathematical proof consists of two theorems-Theorems 2 and 4. So the corrected OPT is not better than GLT in fault diagnosis of redundant sensor system.