Logistic regression and feature extraction based fault diagnosis of main bearing of wind turbines

The total installed capacity of wind turbines is continuously increasing with focus on Renewable Energy Sources (RES). The reliability and efficiency of wind turbines has become a major issue. The occurrence of fault in various components of wind turbine needs to be addressed for improved performance. Among these issues, wind turbine bearing fault is the most significant. Different techniques such as online monitoring using Artificial Intelligence (AI), have been proposed and still research is being carried out in this domain. This paper presents a fault diagnosis analysis of main shaft bearings of wind turbines. The goal is to monitor the condition of wind turbines for early fault prediction so that the turbine can immediately be adjusted for improved performance and extended service life. Different techniques such as Fast Fourier transform (FFT), Hilbert Huang Transformation (HHT), Feature extraction, Logistic Regression (LR) are applied on a real data set of 18 wind turbines to accurately evaluate the health of the turbine. The results highlight the superior and reliable performance of these techniques for bearing fault detection for cost effective operation and maintenance (O&M).