An improved Lempel-Ziv complexity for bearing fault diagnosis based on the time-frequency encoding method

Lempel-Ziv complexity (LZC) is extensively utilized in the identification of bearing faults. The present enhancement in LZC encoding relies on time domain information. When the magnitude of fluctuation is minimal, LZC utilizing time domain information encoding is unable to properly differentiate signals with varying frequency components. Thus, an improved Lempel-Ziv complexity based on the time-frequency encoding method is proposed. Initially, the time domain encoding is obtained according to the quartile of the amplitude. Then, the frequency domain encoding is calculated based on the statistic value at each frequency along the frequency direction of the Wigner Trispectrum. Finally, the ultimate encoding is derived from both time domain encoding and frequency domain encoding. The proposed method is validated through the actual data of bearing. The time-frequency encoding technique can significantly augment the capacity of encoding sequences to depict signal variations and boost the LZC representation of signal complexity.