基于参数估计的传感器故障诊断的改进方法  

AN IMPROVED METHOD FOR SENSOR FAULT DIAGNOSIS BASED ON PARAMETER ESTIMATION

作　　者：刘志成[1] 刘春涛[2] 

机构地区：[1]太原工业学院电气工程系,山西太原030008 [2]北京理工大学化工与环境学院,北京100081

出　　处：《计算机应用与软件》2009年第9期183-185,205,共4页Computer Applications and Software

摘　　要：针对基于参数估计的非线性系统传感器故障诊断方法中存在的滤波稳定性差、估计精度低的缺点,提出了卡尔曼滤波与小波滤波相结合的方法。在将传感器故障参数都等效为偏差型故障参数的基础上,通过增大强跟踪滤波器算法中的量测噪声方差和系统噪声方差,使其大于实际噪声方差,以提高滤波器的稳定性和故障检测的快速性,同时引入小波滤波以提高对故障参数的估计精度。仿真实验表明,该方法较好地兼顾了滤波稳定性、估计精度及速度。Aiming at the deficiencies the parameter estimation-based nonlinear system sensor fault diagnosis has in poor stability of filtering and low precision of estimation,this paper puts forward a method combining Kalman filtering with wavelet filtering. On the basis of making all sensor fault parameters equivalent to parameter bias faults, by enlarging measurement noise variance and system noise variance in strong tracking filter method, they exceed real noise variances, so as to improve the stability of filter and the rapidity of fault detection, at the same time wavelet filtering is introduced to improve the estimation precision of fault parameters. The results of simulation experiments show that this method considers to stability, estimation precision and speed of the filtering well.

关 键 词：故障参数估计 强跟踪滤波器 小波滤波 估计精度 跟踪速度 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置] O212.1[自动化与计算机技术—控制科学与工程]