一种新的基于参数估计的故障诊断方法

传统的参数估计方法利用历史数据对参数进行估计，形成一次估计值序列，基于参数估计的故障诊断方法就是利用这个序列对故障进行诊断和分离的，但是，一次估计值序列跟踪真实参数变化存在明显的滞后，本文提出一种边估边修正参数序列的方法，从而克服了参数估计滞后，提高了参数估计的准确度，以此我们为故障检测和分离设计了一种新的补偿最小二乘算法，仿真表明这种算法用于故障诊断是有效的。

A New Fault Diagnosis Method Based on Parameter Estimation

Traditional parameter estimation methods use the history data to estimate system parameters and generate a recent parameter estimation series. The fault diagnosis method based on parameter estimation utilizes this series to diagnosis and isolate fault information. The drawback is that the recent parameter series has significant delay from the true value of time varying parameter. This paper suggests that using future data to compensate the recent parameter series will overcome the delay, then improve the accuracy of the parameter series estimation. We present a new parameter series compensation method based on LS estimation for fault detection and isolation. The efficiency of this method is demonstrated.