基于改进层次全局模糊熵和MCFS的滚动轴承损伤识别

Damage identification of rolling bearing based on improved hierarchical global fuzzy entropy and MCFS

针对传统的多尺度特征提取方法无法捕捉振动信号高频故障信息的问题,提出了一种基于改进层次全局模糊熵(IHGFE)全局全频段特征提取、多聚类特征选择(MCFS)特征降维和支持向量机分类的滚动轴承故障诊断方法。首先,提出了能够捕捉振动信号低频到高频的全局特征的IHGFE非线性动力学方法,并将其用于滚动轴承的故障特征提取;然后,利用MCFS对初始特征向量进行了维数约简和优化,构建了低维且对故障敏感的故障特征向量;最后,建立了基于支持向量机的多故障分类器,实现了滚动轴承损伤的智能化识别,并通过两个滚动轴承实验进行了对比分析。研究结果表明:IHGFE的分类准确率和识别稳定性均优于对比方法,证明了其在特征提取中能够在一定程度上解决现有方法无法同时考虑信号的高频特征和全局特征的问题,可为进一步扩展模糊熵方法在滚动轴承损伤识别中的应用提供参考。

Aiming at the problem that the conventional multiscale feature extraction method could not capture the high frequency fault information of vibration signals,a rolling bearing fault diagnosis method based on improved hierarchical global fuzzy entropy(IHGFE)and multi cluster feature selection(MCFS)was proposed.First of all,an IHGFE nonlinear dynamics method that could capture the global features of the vibration signal from low frequency to high frequency was proposed and used for fault feature extraction of rolling bearing.Then,MCFS was used to reduce and optimize the initial feature vectors to build low dimensional and fault sensitive fault feature vectors.Finally,a multi fault classifier based on support vector machine was established to realize intelligent identification of rolling bearing damage,and two rolling bearing experiments were carried out for comparative analysis.The research results show that the IHGFE has better classification accuracy and recognition stability than the comparison method,which proves that IHGFE can solve the defect that the existing methods can not consider both the high frequency features and the global features of signals to a certain extent in feature extraction,and can provide a reference for further expanding the application of fuzzy entropy method in rolling bearing damage identification.