自适应GA-WAF盲源分离用于滚动轴承故障诊断

Adaptive GA-WAF Blind Bource Separation for the Bearing Fault Giagnosis

针对CICA方法用于轴承早期故障诊断中,滤波参数选择困难的问题,提出了一种基于遗传算法改进的小波滤波、CICA方法相结合的故障诊断方法,小波滤波器的参数通过遗传算法优化得到。在对采集到的多路信号进行滤波时,提出以峰值因数为目标函数,借助遗传算法构建带通滤波器,这样避免了盲目或者凭经验选择滤波器参数,经过滤波后,提高了轴承故障信号信噪比。接下来,输入到以负熵为目标函数,乘子算法为优化算法的CICA方法中,可以将有用的信号从混合的信号中分离出来,利用TEO方法进行解调识别故障特征。

For solving the problem that it is difficult to select filtering parameters whenCICA method is used in bearing early fault diagnosis,a fault diagnosis method based on genetic algorithm improved wavelet filter and CICA method combined is proposed. The parameters of wavelet filter are optimized by genetic algorithm.When filtering the collected multichannel signals,a band-pass filter is constructed by genetic algorithm with the peak factor as the objective function,which avoids blind or empirical selection of filter parameters. After filtering,the signal-to-noise ratio(SNR)of bearing fault signals is improved.Next,input into the CICA method,which takes negative entropy as the objective function and multiplicative algorithm as the optimization algorithm,can separate useful signals from mixed signals,and use TEO method to demodulate and identify fault features.