基于时间-小波能量谱样本熵的滚动轴承智能诊断方法

An intelligent method for rolling element bearing fault diagnosis based on time-wavelet energy spectrum sample entropy

为了解决滚动轴承故障模式智能识别与运行状态检测问题,提出了时间-小波能量谱样本熵的计算方法,并将其作为特征参数用于滚动轴承智能诊断的研究。采用Hermitian小波对轴承信号进行连续小波变换,得到蕴含故障信息的时间-小波能量谱序列,再通过计算其样本熵值,量化提取信号中的故障特征信息。轴承不同故障模式下的时间-小波能量谱样本熵区分明显,以此作为特征向量输入支持向量机,实现了对轴承不同故障模式的智能识别。之后计算轴承全寿命周期实验数据的时间-小波能量谱样本熵,按照时间顺序排列,绘制出了轴承运行状态曲线,通过判断曲线走势可有效诊断出轴承早期故障的发生。实验结果表明,时间-小波能量谱样本熵可以有效用于滚动轴承智能诊断的研究。

In order to solve the problem of fault mode intelligent recognition and running state detection of rolling element bearing, a new method called time-wavelet energy spectrum sample entropy as the characteristic parameter was proposed for bearing fault intelligent diagnosis. Time-wavelet energy spectrum which contained fault information of bearing was obtained through the Hermitian wavelet continuous wavelet transform, and fault feature was quantitatively extracted by calculating the sample entropy of the energy spectrum. The time-wavelet energy spectrum sample entropies of bearings under different fault modes could be distinguished clearly, which could be treated as input characteristic vectors of a support vector machine (SVM) in order to complete the intelligent recognition of different fault modes of bearings. Next, the trend of running state of bearing was acquired through calculating the time-wavelet energy spectrum sample entropy of data from the whole life cycle test rig of bearing and arranging them chronologically. The early damage occurring in bearing could be effectively detected by judging the running state trend. Practical examples show the proposed method can be applied to the research for intelligent diagnosis of rolling element bearing efficiently. © 2017, Editorial Office of Journal of Vibration and Shock. All right reserved.