Application of Cluster Quantization in Wind Turbine Bearing Degradation Assessment
In order to solve the problems of poor monotony and insufficient explanability of wind turbine bearing degradation in-dexes,a degradation assessment method for wind turbine bearing was proposed based on t-SNE and cluster quantization.Firstly,the time-domain,frequency-domain and time-frequency domain features were extracted from the vibration signals of the health reference state and the monitoring state at any time,and then the referential features were fused.Secondly,t-SNE is used to re-duce the dimensionality of high-dimensional data.Finally,the clustering quantization factor was selected to characterize the degradation degree of the wind turbine bearing,and the adaptive threshold was set to realize the degradation assessment.Through the comparison of other algorithms and the verification of actual signals,the degradation index established in this paper can timely warn the early fault of the wind turbine bearing with strong monotony and can effectively reduce the false alarm rate.
T-SNECluster Quantization FactorAdaptive ThresholdDegradation Index Monotony
万方数据
维普
