Multi-axis Industrial Robot Fault Diagnosis Model Based on Improved One-Dimensional Convolutional Neural Network

Abstract

Industrial robots have become indispensable equipment in the automated manufacturing process. However, there are currently few deep learning fault diagnosis methods based on industrial robot operation. Aiming at the problems of low fault diagnosis accuracy and slow speed during the operation of industrial robots, a fault diagnosis model based on an improved one-dimensional convolutional neural network is proposed. To solve the problem of lack of industrial robot fault datasets, this paper uses the method based on random sampling and Mixup data augmentation to enhance data. Then, the model based on the original operation data of industrial robot are trained end-to-end by orthogonal regularization (SRIP) that combines with a one-dimensional convolutional neural network (CNN-1D). The experiment tests the diagnostic accuracy based on 3 million pieces of industrial robot operating data, which includes torque, speed, position, and current. Compared with the WDCNN and CNN-1D models, SRIPCNN-1D method can diagnose industrial robot faults effectively.