滚动轴承故障诊断优化仿真研究

Fault Diagnosis of Rolling Element Bearing Based on Wavelet Package and LSSVM

研究提高滚动轴承故障诊断准确率问题,滚动轴承故障振动信号具有非平稳,造成系统不稳定,针对传统方法难以提取故障信息的不足,提出一种小波包和最小二乘支持向量机的滚动轴承故障诊断方法(WP-LSSVM)。首先采用小波包对滚动轴承振动信号进行降噪处理,消除背景和噪声信息,然后小波包对去噪后振动信号分解并计算能量特征值,最后采用最小二乘支持向量机对能量特征值进行学习,建立滚动轴承故障诊断模型。仿真结果表明,滚动轴承故障诊断训练和测试时间减少,且故障诊断准确率得到提高。

To improve the accuracy of fault diagnosis of rolling beating, the paper put forward a fault diagnosis method of roiling bearings based on wavelet package and least squares support vector machine （WP - LSSVM）. First- ly, wavelet packet was used for rolling bearing vibration signal de - noising, elimination of background and noise in- formation. Then wavelet packet was used for the decomposition of the denoised vibration signals and calculation of the energy eigenvalue. Finally, the least squares support vector machine was training with the energy eigenvalues to es- tablish the rolling bearing fault diagnosis model. Simulation experimental results show that, compared with other fault diagnosis methods of rolling bearing fault diagnosis, WP - LSSVM reduces the training time and testing time, and im- proves the accuracy of fault diagnosis of rolling bearing.