基于IITD和FCM聚类的滚动轴承故障诊断

基于Akima插值和固有时间尺度分解（ITD）中的线性变换，提出了一种改进的固有时间尺度分解（IITD），在此基础上，进一步提出基于IITD近似熵（AE）和模糊C均值聚类（FCM）相结合的滚动轴承故障的诊断方法。采用IITD方法对滚动轴承的振动信号进行分解，通过互信息分析，筛选出前3个含主要特征信息的固有旋转分量（PR），计算其近似熵值作为信号的特征向量，将得到的特征向量输入到FCM分类器中分析并得到分类结果。实验分析表明:分别与基于EMD、ITD近似熵和FCM聚类相结合的方法比较，该方法的分类系数更接近1，平均模糊熵更接近0，即此方法对滚动轴承的正常、内圈故障、外圈故障、滚动体故障信号以及对内圈的不同损伤程度信号均能更有效更准确地进行分类。 

Rolling bearing fault diagnosis based on IITD and FCM clustering

An improved intrinsic time-scale decomposition (IITD) method was proposed based on Akima interpolation and linear transformation of intrinsic time-scale decomposition (ITD). Furthermore, based on approximate entropy (AE) and fuzzy C-means clustering (FCM), a new analysis method of using IITD for fault vibration signal of rolling bearing was proposed as well. The vibration signal was decomposed with IITD to obtain a certain number of proper rotation (PR) and a trend. By using mutual information analysis, three PR components were sifted out and the AE was calculated as the eigenvectors. The constructed eigenvectors were put into FCM classifier to recognize different fault types. These results were compared with the methods based on empirical mode decomposition (EMD) and ITD approximate entropy and FCM respectively. The classification coefficient with use of this method was calculated more closer to 1 and average fuzzy entropy was calculated more closer to 0. Accurate fault identification was presented for roller bearings normal, inner faults, outer faults, rolling body fault signals and different damage degree signals of rolling body faults.