A fault feature extraction method for flexible thin-wall bearings based on TMSST and correlation kurtosis

For flexible thin-wall bearings，a fault feature extraction method based on time-reassigned multi-synchro-squeezing transform（TMSST）and correlation kurtosis is proposed. In this method，the compression operator is used to improve the time-frequency aggregation of STFT results. The pulse characteristics at the best frequency point are selected in combination with correlation kurtosis. The impulse frequency characteristics are obtained by analyzing the pulse characteristics. The method is applied to the feature extraction of fault signals of inner and outer rings of flexible thin-walled bearings，and compared with the correlation kurtosis of S transform. The results show that the method in this paper can successfully extract fault features and better reflect the time varying frequency of fault features，which provides a new perspective of time-frequency analysis for bearing fault diagnosis.