滚动轴承表面损伤故障智能诊断新方法

New intelligent diagnosis method for ball bearing faults due to surface damage

本文针对目前基于小波变换的滚动轴承故障诊断研究中普遍存在小波变换参数选取和故障特征计算无法自动完成的问题,提出了一种基于小波包变换的滚动轴承故障特征自动提取技术,实现了小波函数参数的自动选取和故障特征的自动提取。最后,基于结构自适应神经网络方法建立了滚动轴承的集成神经网络智能诊断模型,利用实际的滚动轴承实验数据进行了验证,结果表明了本文方法的有效性。

At present, in the study of ball bearing fault diagnosis based on wavelet transform, the parameter selection of wavelet transform and computation of fault features can not be accomplished automatically. In this paper, a new method based on wavelet packet transform for auto-extracting ball bearing fault features is put forward, which can select the wavelet function parameters and extract the fault features automatically. An integrated neural network based on structure self-adaptive neural network model was established to implement the intelligent diagnosis of ball bearing faults, practical ball bearing experiment data were used to verify the new method, and the results fully show that the new method is correct and effective.