A state diagnosis method for rolling bearings based on self-attention neural network

Abstract

The invention discloses a method for diagnosing a state of a rolling bearing based on a self-attention neural network, comprising the steps of: using a vibration acceleration sensor to acquire timingsequence signals of vibration acceleration of the rolling bearing in different states under different motor loads to obtain vibration acceleration data to be used as sample data; performing data enhancement processing on the acquired sample data; attaching corresponding labels to the sample data subjected to the data enhancement processing according to the type of the state of the rolling bearing;extending characteristic vectors of each sample data by using multilayer mapping, so as to change one-dimensional characteristics into multi-dimensional characteristics; establishing a self-attentionnetwork diagnosis model; training the self-attention network diagnosis model by using the processed sample data, evaluating the trained self-attention network diagnosis model, and applying the trained self-attention network diagnosis model to monitoring the to-be-diagnosed rolling bearing. According to the method of the invention, a data enhancement technology and a characteristic extension method are adopted, so that a requirement of the method on data volume can be reduced, and the capacity of the method for coping with an extreme environment is improved.

Description

Technical field[0001]The invention relates to the technical field, in particular to a method for diagnosing the state of a rolling bearing based on a self-attention neural network.Background technique[0002]In modern industry, fault diagnosis of mechanical devices is an essential part of safe production. An excellent fault monitoring system can usually effectively avoid the occurrence of dangerous situations. Therefore, health monitoring technology is the basic demand of modern industry. Rolling bearings are the most commonly used components in rotating mechanical devices. Once the bearing fails, it may cause serious accidents and even cause huge economic losses and casualties. The fault diagnosis method is very important to improve the safety and reliability of machinery. In the past two decades, bearing fault diagnosis technology has been a hot research topic. The current mainstream method is to perform vibration analysis and monitoring on the bearing vibration acceleration time se...

AI Technical Summary

Problems solved by technology

[0005] In view of this, the purpose of the present invention is to provide a rolling bearing state diagnosis method based on self-attention neural network, which uses data enhancement technology and feature expansion operation to solve the problem that the current mainstream method relies on a large amount of data

Images