International Journal of Mechanical Engineering

This paper proposes a digital signal processing-based method for composite fault feature extraction in bearings. By conducting multi-scale time-frequency analysis and feature decomposition on bearing vibration signals, the method achieves effective separation and identification of composite fault characteristics. Utilizing wavelet packet decomposition, empirical mode decomposition, and instantaneous frequency extraction techniques, the approach enhances the representation of weak fault features in signals. Combined with statistical feature quantization, it optimizes features to improve diagnostic accuracy and robustness. Simulation and experimental results demonstrate that this method can stably capture multiple types of fault information under complex operating conditions, providing reliable data support for bearing condition monitoring.

Digital signal processing; Bearing; Composite fault; Feature extraction; Time-frequency analysis