Scada data-driven wind turbine main bearing fault prognosis based on one-class support vector machines

Abstract

This work proposes a fault prognosis methodology to predict the main bearing fault several months in advance and let turbine operators plan ahead. Reducing downtime is of paramount importance in wind energy industry to address its energy loss impact. The main advantages of the proposed methodology are the following ones. It is an unsupervised approach, thus it does not require faulty data to be trained; ii) it is based only on exogenous data and one representative temperature close to the subsystem to diagnose, thus avoiding data contamination; iii) it accomplishes the prognosis (various months in advance) of the main bearing fault; and iv) the validity and performance of the established methodology is demonstrated on a real underproduction wind turbine.