Fault Diagnosis of Rolling Bearings Based on GAF-MTF-CNN

Aiming at the problems of low test accuracy and poor model generalization ability of traditional image coding method and neural network method for bearing fault diagnosis, a rolling bearing fault diagnosis method based on Gramian angular field (GAF), Markov transition field (MTF) and convolutional neural network (CNN) was proposed. GAF and MTF transformation were performed on the sampled vibration data of each section of bearing to generate two-dimensional images, and the data set was constructed by horizontal stitching method, and then it was imported into the multi-layer CNN with batch normalization and random inactivation operation for diagnostic testing. The results show that, compared with the data processing method using only GAF and MTF, the data set constructed by the data processing method in this paper has a high accuracy of model test trained in the built CNN, up to 99.87%. The built CNN has good generalization ability and high accuracy of network model, which proves the feasibility of the data set construction method in bearing fault diagnosis.