Bearing health state identification method based on multi-view attention network

Abstract

The invention provides a bearing health state identification method based on a multi-view attention network, and the method comprises the following steps: S1, obtaining a data set, and converting the data set into different representation forms through a signal processing method; s2, carrying out dimension reduction processing on the signals in each representation form and combining the signals into multichannel data; and S3, inputting the multi-channel data into a deep residual network with an attention module for network training until the network converges. According to the scheme, the effectiveness of fault diagnosis is ensured in breadth and depth by modeling the signal from point to surface and from surface to point. The comprehensiveness of the fault information is ensured by various forms of data; the deep network containing the attention module ensures the effectiveness and accuracy of the extracted features. The method is not only suitable for bearing fault diagnosis under a conventional working condition, but also suitable for bearing fault diagnosis under a low signal-to-noise ratio condition.

Description

technical field [0001] The invention belongs to the technical field of bearing fault diagnosis, and in particular relates to a bearing health state identification method based on a multi-view attention network. Background technique [0002] With the development and progress of science and technology, bearings and their intelligent applications are more and more widely used in social production and life. Its health status is not only related to economic benefits, but also related to the safety of people and the whole society. The health diagnosis of mechanical equipment is of great significance. Timely digging out the faults in mechanical equipment can prevent potential accidents and avoid unnecessary casualties and economic losses. [0003] However, the working environment of bearings in actual engineering does not necessarily have very good working conditions. In fact, in many cases, bearings operate in noisy environments. At the same time, due to the complexity of the m...

AI Technical Summary

Problems solved by technology

However, these attention networks are all attention networks based on time-domain signals or spectrum, and the attention weights of these networks are distributed on the shape dimension of the feature rather than its channel dimension, so although the fault diagnosis ability has been improved, but Due to the insufficient expressiveness of the single-channel data used, the diagnostic method still has certain limitations.

Images