NEW DIAGNOSIS METHOD FOR WIND TURBINE FAILURE PREDICTION BY COMPREHENSIVE VIBRATION ANALYSIS

Abstract

New diagnosis method by comprehensive vibration analysis combined with latest machine learning technology is developed. In this method, overlapped signals are handled comprehensively as evaluation signals rather than noise. It can be applied to low-speed moving parts such as main rotor bearings of wind turbines, which are difficult to detect damage early, and as a result, the damage detection of wind turbine bearings is made possible three months before repair is required.