Rolling bearing fault diagnosis method based on deep adversarial transfer learning

Abstract

The invention relates to a rolling bearing fault diagnosis method based on deep adversarial transfer learning, and belongs to the field of intelligent diagnosis of mechanical systems. A rolling bearing fault is diagnosed by adopting a network model of deep adversarial transfer learning, and the model comprises a feature extraction layer, a fault classification layer, a global domain adversarial layer and a local domain adversarial layer. The model adopts the sum of classification loss, global domain adversarial loss and local fault class adversarial loss as a loss function for training, and it is ensured that accurate classification of target domain faults is achieved through source domain data with fault labels in the training process. According to the method, common characteristic attributes of the source domain fault and the target domain fault are ensured to be in the same distribution through domain confrontation and category confrontation learning while the fault classifier is effectively constructed, so that the characteristic distribution difference of the source domain and the target domain is reduced, and the fault classification accuracy of the rolling bearing is improved.

Description

technical field [0001] The invention relates to a rolling bearing fault diagnosis method based on deep anti-migration learning, which belongs to the field of intelligent diagnosis of mechanical systems. Background technique [0002] With the improvement of the automation and intelligence level of mechanical equipment and the expansion of production scale, the data volume of sensors monitoring the operation status of mechanical equipment is growing explosively, and fault diagnosis in the industrial big data environment has become a trend. Rolling bearings, as the key basic parts in mechanical equipment, are widely used in various fields. Its health status is closely related to the working performance of the whole machine. Ensuring the safe operation of rolling bearings is of great significance to the production capacity and safety guarantee of enterprises. Rolling bearings usually operate under complex and variable working conditions. How to accurately and quickly diagnose r...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to provide a deep anti-migration learning rolling bearing fault diagnosis method to solve the problem of low accuracy of rolling bearing fault diagnosis in actual engineering sites

Images