基于模糊信息粒化与小波支持向量机的滚动轴承性能退化趋势预测

Degradation Trend Prediction of Rolling Bearings Based on Fuzzy Information Granulation and Wavelet Support Vector Machine

针对滚动轴承的性能退化指标及其波动范围难以有效预测的问题,提出了一种基于模糊信息粒化与小波支持向量机的滚动轴承性能退化趋势预测方法。首先以一定的时间间隔采集滚动轴承运行过程中的振动信号序列,提取各个振动信号序列的特征指标,对特征指标序列进行模糊信息粒化,进而提取各个粒化窗口的有效分量信息;随后通过构建小波支持向量机对各个指标分量分别建立预测模型,实现对滚动轴承性能退化指标的退化趋势及波动范围的预测。实验结果表明,该预测方法可以有效跟踪滚动轴承性能衰退指标的变化趋势,并对其指标的波动范围进行有效预测。

Aiming at the problems where the performance degradation index and its fluctuation ranges of the roller bearings were difficult to forecast effectively,a method was proposed herein based on fuzzy information granulation and WSVM for roller bearing performance degradation trend prediction.Firstly,the vibration signal sequences of rolling bearings in the operation process were acquired at a certain time interval,then,those feature indexes were extracted from those vibration signals.In order to acquire the effective component information,the process of the fuzzy information granulation for those feature indexes might be performed.Subsequently,aprediction model was established for each feature index by constructing the WSVM,and the degradation trend and the fluctuation ranges of the performance feature indexes of the rolling bearings were predicted.The experimental results show that the proposed method can track the change tendency of the rolling bearing performance effectively,and the degradation trend and the fluctuation ranges of the performance feature indexes may be predicted effectively.