Small sample fault diagnosis method and system for rolling bearing

Abstract

The invention discloses a small sample fault diagnosis method and system for a rolling bearing. The method comprises the following steps: constructing a fault diagnosis model based on an output module and three residual modules with the same structure; acquiring data of a rolling bearing to be detected, inputting the data into the fault diagnosis model, and processing the data through a kaconv convolutional layer in the residual module to obtain extraction features; sequentially inputting the extracted features to a batch normalization layer and a first activation layer to obtain output features; inputting the output features to a double-attention module to obtain enhanced features; inputting the enhanced features and to-be-detected rolling bearing data into a first fusion layer for fusion, and then inputting the fused enhanced features and the to-be-detected rolling bearing data into a first maximum pooling layer to obtain fused enhanced features; inputting the fusion enhancement feature into two residual modules in sequence to obtain a final fusion enhancement feature; and inputting the final fusion enhancement feature into an output module to obtain a final diagnosis result. And the fault diagnosis precision of the rolling bearing under the small sample is improved.

Images