Research on Wind Turbine Composite Fault Decoupling and Slight Fault Extraction Based on Continuous Spectral Kurtosis Deconvolution

• Published in: Journal of Vibration Engineering & Technologies Vol. 12; no. 3; pp. 2975 - 2986
• Main Authors: Li, Pei-hang, Dang, Jian, Jia, Rong, Li, Jiang-feng, Ma, Xi-ping
• Format: Journal Article
• Language: English
• Published: Singapore Springer Nature Singapore 01.03.2024
• Subjects: Acoustics; Control; Dynamical Systems; Engineering; Engineering Acoustics; Original Paper; Vibration; Fault diagnosis; Continuous spectral kurtosis deconvolution; Wind turbine bearing; Feature extraction; Composite fault decouple
• ISSN: 2523-3920; 2523-3939
• DOI: 10.1007/s42417-023-01026-z

Purpose Fault diagnosis for wind turbines can effectively reduce the impact of failures, ensuring the safe and reliable operation of machinery equipment significantly and avoiding significant maintenance costs. However, the features of composite faults in wind turbines are coupled under the influence of multiple vibration sources. Particularly, slight fault signals are generally covered by noise and strong fault signals, making them difficult to identify effectively Methods An early composite fault diagnosis algorithm named Continuous Spectral Kurtosis Deconvolution (CSKD) is proposed in this paper by analyzing the phenomenon of multiple spectral kurtosis maxima in the fast kurtosis graph under the composite fault condition. According to the maximum spectral kurtosis in the fast kurtosis graph and its corresponding frequency and frequency resolution, a band-pass FIR filter model is established for deconvolution processing to obtain significant fault signals. The obtained fault signals are removed by the narrow-band band-stop filter to achieve the purpose of significantly decoupling the composite fault. Results The proposed algorithm is verified by a typical model simulation and engineering application, with the composite fault of the inner and outer rings of the large-scale wind turbine bearing. Results show that the proposed algorithm can realize the decoupling of large-scale rotating machinery composite fault signals and extract slighter signal features in the early composite faults effectively. Conclusion Results show that the proposed algorithm is able to provide a solution for composite fault diagnosis of wind turbines. The method can be extended to further consider more complex and practical cases.