强噪声背景下的多精度传感器故障诊断

Fault Diagnosis of Multi-Precision Sensors in Large Noise Background

针对背景噪声变化很大、多精度冗余传感器故障难以诊断的问题,提出了基于二次卡尔曼滤波的故障诊断方法。该方法首先通过小波噪声估计预测观测值噪声强度,接着对传感器数据进行卡尔曼滤波预处理,降低观测值的不确定性,并将故障信息最大化,然后利用冗余特性,轮流使用一个传感器测量值作为输入,另一个作为输出建立循环卡尔曼滤波方程组,通过决策函数对所得到的新息进行故障诊断。实验分析了故障检测率与噪声强度的关系,结果表明,该方法能提高故障诊断的准确性,具有较好的鲁棒性。

When background noise changes large,it is difficult to diagnose the failure of multi-precision redundancy sensors system.In order to address this problem,a method based on two stage Kalman filter（TSKF） is proposed.Firstly,the noise estimated by wavelets,and the sensor data is preprocessed by strong tracking Kalman filter,which is to reduce the uncertainty of observation and maximize fault information,and then it takes advantage of the redundancy to using a sensor as input,and the other sensors as output to establish the Kalman filter equations,by which the innovation obtained is used to the sensors fault diagnosis.The relationship between fault detection rate and the strength of noise is discussed by experiments,which shows that the method is robust and improve the accuracy of fault diagnosis.