Fault Diagnosis of Rolling Bearing in Wind Turbine Using Vibration Signal Processing and Machine Learning

Rolling bearings, as the main equipment of wind turbines, affect the normal operation of wind generators. If the bearings are damaged, it leads to huge and significant economic losses. Most defects are not readily apparent due to the small vibrational effects of bearings operating under severe conditions. Envelope analysis is one of the effective methods in signal processing. The envelope analysis covers the extremes or minimum and maximum points of the signal. Radial basis neural network is also used in classification problems. The radial basal neural network has three layers. In this research, by using vibration signal processing by envelope analysis method, the features of vibration signals are extracted in the time domain and the radial basis neural network of the vibration data set of wind turbine bearings in real operation conditions in terms of health and defectiveness is identified and classified. The simulation results show the very successful performance of signal processing and neural network learning.

کلیدواژه ها:

fault detection ، rolling bearings ، wind turbine ، vibration signal processing ، envelope analysis ، neural network ، machine learning