检测与诊断齿轮裂纹故障的一种方法

Gearbox fault diagnosis of sequential probability ratio based on radial basis function optimized particle filter

基于径向基函数(RBF)网络优化的粒子滤波降噪与序贯概率比检验相结合的原理,提出了一种检测与诊断齿轮裂纹故障的方法,并采集一种无裂纹与另外两种存在差异裂纹齿轮的水平方向振动信号,对该方法进行验证.首先,运用RBF网络优化的粒子滤波程序对原始振动信号进行降噪预处理,将振动真实值从中提出;然后,利用时域分析法提取振动真实值的特征参数(峭度值)序列;最后,将特征值序列输入序贯概率比检验程序,根据结果图综合分析对不同齿轮故障进行区分.结果表明建立的优化粒子滤波程序对原始振动信号降噪处理效果良好,获得了细致、准确和稳定的振动信号;序贯概率比检验能比较与区分齿轮不同的故障,改进了齿轮箱故障检测与诊断效果.

A method for the detection and diagnosis of gear crack was put forward based on the principles of radial basis function （RBF） network optimized particle filter and sequential probability ratio test, The horizontal vibration signals of a crack-free and two different crack gears were collected to verify the method. Firstly, the true vibration value was extracted from the original signals after the RBF opti mized particle filter operating. Then, the time domain analysis was used to extract characteristic param- eter sequence （kurtosis value）. Finally, failure mode was determined according to the result map of which the kurtosis value sequence was put into the sequential probability ratio test procedures. The re- suits show that the established procedure for optimized particle filter has a good effect on noise reduc- tion with detailed, accurate and stable vibration signals; gears＇ different failures can be compared and distinguished by sequential probability ratio test, which improves the effect of gearbox fault detection and diagnosis.