Rolling Bearings Fault Diagnosis under Variable Speed Conditions Based on Multitime‐Frequency Ridge Extraction

The vibration signals of rolling bearings under variable speed conditions exhibit time‐varying, nonstationary, and nonlinear characteristics, making it challenging to extract fault features. A fast path optimization‐based multitime‐frequency ridge extraction (FPOMRE) method is proposed in this paper to address the problem. First, the time‐frequency representation of the fault signal is obtained through the adaptive short‐time Fourier transform (STFT), and then the FPOMRE algorithm is used to extract additional ridges below the lowest time‐frequency ridge multiple times. The frequency ratio of each ridge to the additional ridge, the average value of the ratio, and the variance of the ratio are calculated, respectively, and finally, the matching degree between the average ratio and the fault feature coefficient is compared with the diagnosis of the bearing fault. The simulation and experimental results show that the FPOMRE method can effectively diagnose the bearing fault under constant speed and variable speed conditions, and the extraction accuracy of time‐frequency ridges under constant speed is higher than that under variable speed, and the extraction accuracy under high speed is higher than that under low speed. The relative error of extracting rotation frequency under constant speed and variable speed is 0.5% and 1.3%, respectively. In addition, comparing the FPOMRE with the maximum amplitude method (MAM), both FPOMRE and MAM can effectively extract the instantaneous shaft rotation frequency at a constant speed. However, under variable speed conditions, the ridge extracted by FPOMRE has a higher accuracy and lower average processing time than that extracted by MAM.