滚动轴承早期故障的特征提取与智能诊断

Feature Extraction and Intelligent Diagnosis for Ball Bearing Early Faults

在基于小波变换的滚动轴承故障诊断研究中,目前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。基于此,提出了一种基于二进离散小波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。同时,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。最后,利用实际的滚动轴承实验数据验证了所提方法的有效性。

In the study on ball bearing fault diagnosis based on wavelet transform, the parameter selection of wavelet transform and computation of fault features cannot be carried out automatically at present. Aiming at these problems, a new ball bearing fault feature auto extracting method based on binary discrete wavelet trans form is proposed in this article, which can select automatically wavelet function parameters and extract the fault features. In addition, an intelligent diagnosis model based on the neural network with self-adaptive structure is established to implement the intelligent diagnosis of ball bearing faults. Finally, practical ball bearing experiment data is used to verify the new method put forward in this article, and the results fully validate its application.