中文摘要：

为精确提取滚动轴承振动信号的故障特征,提出了一种基于参数优化多尺度排列熵与模糊C均值聚类的故障诊断方法。首先,针对多尺度排列熵算法的参数确定问题,综合考虑参数之间的交互影响,基于遗传算法与微粒群算法对参数进行优化;然后,利用参数优化多尺度排列熵对滚动轴承振动信号进行特征提取,并通过模糊C均值聚类确定标准聚类中心;最后,采用Euclid贴近度对故障样本进行分类。通过分类系数与平均模糊熵检验聚类效果,证明了多尺度排列熵参数优化的有效性;与单一尺度排列熵、样本熵结合模糊C均值聚类方法的对比分析表明,基于参数优化多尺度排列熵与模糊C均值聚类的故障诊断方法具有更高的故障识别率和更广阔的适用范围。

英文摘要：

To extract fault features of rolling bearing vibration signals precisely, a fault diagnosis method based on parameter optimized multi - scale permutation entropy （MPE） and fuzzy C - means clustering （FCM） is proposed. Firstly, aiming at the problem of parameter determination of MPE and considering the interaction among parameters comprehensively, the parameters are optimized respectively by genetic algorithm and particle swarm optimization algorithm. Then the fault features of rolling bearing are extracted by parameter optimized MPE, and FCM is used to obtain the standard clustering centers. Finally, the fault samples are clustered by a Euclid nearness degree. The validity of the parameter optimization is proved by calculating the classification coefficient and average fuzzy entropy. Compared with single scale permutation entropy and sample entropy, which are combined with FCM separately, the results demonstrate that the diagnosis method based on parameter optimized MPE and FCM has a higher rate of fault recognition and a wider range of application.