Intelligent Diagnosis of Gearbox Based on Spatial Attention Convolutional Neural Network

Convolutional neural network (CNN) can diagnose faults with equipment in a timely manner via their powerful learning ability. To improve the diagnosis accuracy, the typical method is to increase the number of network layers or preprocess the raw data, which often leads to the increase of network parameters and reduces the efficiency. This paper presents a one-dimensional (1D) CNN model based on spatial attention (SA-CNN) for fault diagnosis of the gearbox. Spatial attention uses spatial global pooling and convolution operations to calculate the importance of features in the feature map, without increase a lot of network parameters. It makes the network pay more attention to the valuable part of the feature map. Meanwhile, the use of 1D-CNN effectively reduces the number of calculations and finally achieves high efficiency and accurate diagnosis. In order to verify the effectiveness of the presented diagnosis model, the wind turbine gearbox test-rig is used for experimental verification, and the results show that the presented diagnosis model can improve the accuracy of diagnosis effectively.