Vibration-based Condition Monitoring in Wind Turbine Gearbox Using Convolutional Neural Network

A significantly increased production of wind energy offers a path to achieve the goals of green energy policies in the United States and other countries. However failures in wind turbines and specifically their gear boxes are higher due to their operation in unpredictable wind conditions that result in downtime and losses. Improved and early detection of faults in wind turbines will greatly increase their reliability and commercial feasibility. Faults in the rotating components in the gear box can leave their signature in the vibrations that can be measured by accelerometers. This paper presents a machine learning framework and results in vibration based condition monitoring of wind turbines. Time series acceleration data from several bearings in the wind turbine drive train are used to calculate the bivariate cyclic spectral coherence and spectral kurtosis that produce two dimensional images. These images are used to train convolutional neural networks that then identify faults, including those of small magnitude or in early stages, in test data with a high accuracy. Benchmark test cases inspired by an NREL study are tested and successfully detected.