Bearing fault detection method based on convolution multi-head self-attention mechanism

Abstract

The invention discloses a bearing fault detection method based on a convolution multi-head self-attention mechanism. The detection method comprises the following steps: acquiring and preprocessing a fault bearing vibration signal; generating a bearing fault data set; and constructing a convolutional multi-head self-attention mechanism network, and performing training to obtain a bearing fault detection result. The convolution multi-head self-attention mechanism network comprises a convolution layer, a position encoder, a multi-head self-attention module, a global average pooling layer and a full connection layer; the convolution layer extracts bearing signal initial features; the position encoder performs position encoding on the initial characteristics of the bearing; the multi-head self-attention module learns the initial features; the global average pooling layer regularizes the network to prevent overfitting; and the full connection layer outputs different fault types of the bearing. An efficient and accurate method is provided for bearing fault detection, so that normal operation of mechanical equipment is effectively maintained.

Description

technical field [0001] The invention relates to the field of equipment health management, in particular to a bearing fault diagnosis method. Background technique [0002] As the core of rotating mechanical components, bearings are related to the normal operation of the entire mechanical equipment. During the operation of the equipment, the stator, rotor and other parts of the bearing are easily damaged due to overload, friction, corrosion and gluing. These failures can lead to the failure of the entire mechanical equipment, affecting the performance of the production equipment and possibly causing personal injury. In order to maintain the high performance of the machine while avoiding casualties and economic loss due to bearing failure, the best solution is to carry out fault detection and health status monitoring of bearings. [0003] With the rapid development of sensor technology, computer technology, and information processing technology, equipment health management me...

AI Technical Summary

Problems solved by technology

[0004] The published invention patent with the application number 201910728620.X discloses "a rolling bearing fault diagnosis based on self-attention neural network", which learns vibration signals through self-attention mechanism; this method attempts to use a single self-attention method to learn bearing fault characteristics , failed to learn rich bearing fault features

At the same time, the direct use of the self-attention mechanism failed to effectively learn the local features of the bearing

Images