Rolling Bearing Fault Classification based on CEEMDAN-Transformer

Rolling bearings are critical components in rotating mechanical equipment. They are usually prone to frequent faults due to prolonged operation at high speeds under complex and variable working conditions. This paper proposes a fault classification method to enhance the system’s safety and reliability. Firstly, Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) method is utilized to extract vibration features from signals, effectively eliminating the interference of white noise during signal analysis. Subsequently, the extracted features are input into a Transformer model for fault classification, while an internal multi-head attention mechanism of the Transformer model is leveraged to capture both global dependencies and local patterns within input sequences for further optimizing feature extraction and representation. Finally, experimental results demonstrate that the proposed fault classification method achieves a higher accuracy.