基于小波包和距离判别法的滚动轴承故障诊断

Fault Diagnosis for Roller Bearing Based on Wavelet Packet and Distance Discriminant Analysis Method

将小波包分析与距离判别分析法相结合的方法应用于滚动轴承故障诊断问题中。利用小波包分析技术提取了滚动轴承典型故障的振动加速度信号的状态特征向量,选用此特征向量作为距离判别分析模型的判别因子,以滚动轴承故障实测模拟数据作为学习样本进行训练,通过分析计算,建立了相应线性判别函数,并利用回代估计方法进行检验。研究结果表明:这种新模型判别能力强,交叉确认估计的误判率为0,不需要优化网络结构,是解决滚动轴承故障诊断的一种有效方法。

The wavelet packet and distance discrimination analysis method is applied to fault for diagnosis roller bearing. The state features of typical vibration signal are extracted by wavelet packet technology, and those features are selected as discriminant factor of distance discriminant analysis model,the linear distance discrimination model is established through training a large set of learning samples from a series of roller bearing fault projects. It has been proved that proposed new method has excellent discriminant potency and a low cross-validated misjudgment ratio, hasn＇t need optimize structure and can be used the fault diagnosis for roller bearing.