Fault diagnosis of wind turbine gearbox based on convolutional neural network

Abstract

Gearboxes are crucial components of wind turbines, which is costly, hard to repair, and contribute the longest downtime to wind turbines. Fault diagnosis is a feasible and promising technique to analysis and diagnosis the failure properties of the rotational machinery for instance gearboxes. In this paper, The main research is to use CNN to identify the fault of wind turbine gearbox. Initially, on the basis of vibration signals, STFT method is used for obtaining the two-dimensional time-frequency domain signal. Subsequently, Convolution Neural Network is applied to extract failure features. The feasibility of this method is verified by comparing the results of a convolutional neural network powered by one-dimensional vibration signals.