基于极限学习机的滚动轴承故障诊断方法

Rolling Bearing Fault Diagnosis Method Based on the Extreme Learning Machine

针对传统智能故障诊断方法所需调整参数多且难以确定、训练速度慢,致使滚动轴承故障分类精度、效率差的问题,提出了一种基于极限学习机的滚动轴承故障诊断方法。首先,将采集的信号经EMD后,提取与原信号相关度较大的IMF能量指标。其次,建立滚动轴承的极限学习机故障分类模型;最后,将能量指标组成的特征向量作为模型输入进行滚动轴承不同故障状态的分类识别。实验结果表明:与基于BP、SVM、PSO-SVM与GA-SVM故障分类方法相比,基于极限学习机的滚动轴承故障诊断方法具有更快的运行速度、更高的分类精度。

Because of the traditional intelligent fault diagnosis methods is needed to adjust many paramaters that is difficult to determine and has slow training speed,the rolling bearing fault classification accuracy and efficiency is not satisfied. In this paper,a rolling bearing fault diagnosis method based on extreme learning machine is put forw ard. First of all,it extracts the energy of the IMF that has larger correlation with the original signal. Then,fault classification model based on extreme learning machine of rolling bearing is established. Finally,the feature vectors of energy index is inputed to the model to identify the different failure states. The experimental results show that compared with the BP,SVM,PSO-SVM,GA-SVM,rolling bearing fault diagnosis method based on extreme learning machine has faster speed and higher classification accuracy.