Instantaneous Frequency Estimation-Based Order Tracking for Bearing Fault Diagnosis Under Strong Noise

Under strong noise, bearing fault-related instantaneous frequency (IF) is difficult to extract by time-frequency analysis (TFA)-based ridge extraction method; hence, the tacholess order tracking is unsuitable for characterizing bearing fault characteristic frequency (FCF). To address the above problem, an IF estimation-based order tracking is developed in this article. The fundamental principle of the developed technique is to obtain the IF through the defined instantaneous frequency estimation operator (IFEO) and recovery factor, and then the initial signal is resampled using the IF to achieve bearing fault diagnosis. Specifically, the IFEO is first defined based on the normalization theory, and then the pseudo signal is obtained by resampling the original signal through the IFEO that can match the frequency-modulated (FM) law of the original signal. Second, the spectra concentration index is constructed to calculate the optimal IFEO. Third, the recovery factor corresponding to the optimal IFEO is calculated by searching the highest peak from the envelope spectrogram of the pseudo signal, and then the IF of the maximum amplitude component is calculated. Finally, based on the IF, the bearing signal is resampled, and the fault characteristic order (FCO) spectrum is obtained to detect the bearing fault type. Analysis results of the simulated and measured bearing signals indicate that the developed technique can accurately predict the IF and detect the bearing fault and has better effectiveness in calculating IF and identifying bearing fault type than the traditional ridge extraction method under strong noise.