Variational Mode Decomposition Guided by Time-Frequency Domain Difference Information

To address the difficulty of extracting fault features of rolling bearings in noisy environments, and the issues where the main effects of Variational Mode Decomposition (VMD) are influenced by the decomposition level and penalty factor, a method based on time-frequency domain difference information-guided Variational Mode Decomposition for rolling bearing fault feature extraction is proposed. Initially, the constraints in VMD decomposition are utilized as discrimination conditions, constructing similarity coefficient differences and energy difference ratios as convergence criteria to find the optimal parameter combination. The IMF component with the maximum envelope peak factor is selected for envelope demodulation to determine the fault type. Simulation and experimental results demonstrate that this method can extract fault features from noise interference, confirming its effectiveness.