Discrimination of failure condition for wind turbines by Subtractive Clustering

Abstract

This study proposes a novel method for common failure conditions classification of wind turbine based on the Hilbert-Huang transform (HHT) with fractal feature enhancement. First, this study establishes four common defect types and then the current of generators from these pre-failure wind turbines under operating are measured. Secondly, the HHT can represent instantaneous frequency components through empirical mode decomposition, and then transform to a 3D Hilbert energy spectrum. Finally, this study extracts the fractal theory feature parameters from the 3D energy spectrum by using a Subtractive Clustering for failure condition discrimination. To demonstrate the effectiveness of the proposed method, this study investigates its identification ability using 120 sets of field-tested patterns of wind turbines. The simulation results indicate that the classification rate of the proposed approach is suitable for practical uses even in 15% noise interference condition. Therefore, the proposed methodology can help detection personnel to find the failure type by current signal of wind generator with great reduced of wrong judgment.