Defect Type Recognition System for Wind Turbine by Subtractive Clustering

Abstract

This paper aims to provide a theoretical procedure for defect type recognition of gearboxes and blades on wind turbine. Firstly, a test environment and pre-defect wind turbines were set up according to most usually happened defect types in wind turbine. The measurement of output current signals from these pre-defect wind turbines under operating is then operated for every defect type. Secondly, Hilbert-Huang Transform (HHT) is applied to covert those current signals into time-frequency domain. The HHT reveals that those output current signals have three basic features: time, frequency and energy distribution components. Through observe these features, it could recognize different physical characteristic of each defect type. Finally, using fractal theory to extract the pattern features from HHT time-frequency spectrum and then combining it with subtractive clustering identification method for defect type recognition. To show the efficiency of the proposed approach, simulated works have been conducted. The results show that the proposed approach that using HHT analysis and combined with fractal theory to extract defect features can effectively identify which defect type the wind turbine is and also reduce recognition time to find out the defect of wind turbines more efficient.