A novel hybrid strategy for damage detection of wind turbine yaw bearing
Xiaolong Wang,Hanwei Jin,Zixiang Fu,Jiangtao Xiong,Bowen Zhang,Haichao Shi,Huilong Han
DOI: https://doi.org/10.1177/14759217241295334
2024-11-29
Structural Health Monitoring
Abstract:Structural Health Monitoring, Ahead of Print. Compared with the strong background noise, the impact characteristics of wind turbine yaw bearing are weak in the initial damage stage. In order to completely eliminate unnecessary noise and detect damage characteristics, the useful component should be extracted from the complex vibration signal. To solve the problem that existing methods cannot effectively extract the useful component, an innovative hybrid strategy fusing weighted dual tree complex wavelet packet transform (WDTCWPT) with optimized minimum noise amplitude deconvolution (OMNAD) is developed. First, the envelope variable Rényi entropy indicator is defined, which reflects the impact strength. Meanwhile, this indicator is used to determine the optimal number of WDTCWPT decomposition level, and the weighted wavelet coefficient reconstruction is executed to get the enhanced signal. Subsequently, a square envelope correlation L-Kurtosis (SECLK) indicator is further defined and SECLK spectrum is utilized for estimating the deconvolution period. Ulteriorly, the multiple grid search algorithm is designed to automatically determine the optimal filter length as well as the noise ratio. And the optimal deconvolution signal with obvious damage characteristics can be obtained by OMNAD. Finally, the damage type of yaw bearing can be judged by performing envelope spectrum analysis on the deconvolution signal. The detection results of simulation and experiment signals demonstrate this developed hybrid strategy can accurately detect the local damage of yaw bearing under strong interference environment, which has certain engineering application value.
engineering, multidisciplinary,instruments & instrumentation