A Time-Frequency 2D Reassigning Transform Method and Its Application in Rolling Bearing Fault Diagnosis

Abstract Current time-frequency analysis methods are mainly post-processed for either the time or frequency dimension, while time-frequency methods that jointly process the time and frequency dimensions also suffer from problems such as unsatisfactory processing results. Therefore, a new method called time-frequency 2D reassigning transform is proposed for simultaneously processing signal data in both time and frequency dimensions. The proposed new method classifies the signal into two parts, time dimension processing and frequency dimension components, by classifying the signal into two parts based on the Gaussian window chirp rate discrimination criterion, tendency selection of the single dimension method with weight adjustment operator, extraction of the local maximum of the signal with the post-processing method of redistribution, and introduction of time-frequency redistribution operator to reduce the noise interference highlighting the instantaneous frequency. Comparative analyses using hybrid analogue signals containing transient and non-linear FM signals are carried out to illustrate the effectiveness of the method. The analysis using three different sets of bearing fault signals shows that the proposed method can accurately extract the fault characteristics of bearings while ensuring high resolution, and is suitable for bearing fault diagnosis.