Deep learning fault diagnosis method integrated with prior knowledge

Abstract

The invention relates to the technical field of fault diagnosis, in particular to a deep learning fault diagnosis method integrated with priori knowledge. The method comprises the following steps: step S1, carrying out data processing, wherein in the step, a fault diagnosis data set X is processed based on sliding window processing, and then a picture-like sample data set and an attention matrix A of the picture-like sample data set are obtained; s2, constructing a model architecture, wherein in the step, a 2D-CNN model is constructed to process a similar picture sample data set so as to obtain a corresponding feature map F, and the feature map F is processed based on channel-oriented average pooling and channel-oriented maximum pooling so as to obtain an average pooling output P1 and a maximum pooling output P2. a weight matrix W is obtained according to the attention matrix A, the average pooling output P1 and the maximum pooling output P2, so that the model output is a feature map based on an attention mechanism. The method can better integrate priori knowledge into a deep learning technology.

Description

technical field [0001] The invention relates to the technical field of fault diagnosis, in particular to a deep learning fault diagnosis method incorporating prior knowledge. Background technique [0002] The development of sensor technology enables enterprises to collect fault data conveniently, economically and quickly. Due to the superiority of deep learning technology in extracting predictable features from fault data, its application in fault diagnosis has achieved remarkable results. [0003] The fault diagnosis method based on deep learning mainly has the following steps: [0004] 1) Data collection: collect fault data in the actual production process through sensor technology; [0005] 2) Data preprocessing: Supplement, enhance, clean, transform and other operations on the collected raw data to improve data quality and make it suitable for the input requirements of deep learning methods; [0006] 3) Model architecture construction: Design the deep learning model a...

AI Technical Summary

Problems solved by technology

However, deep learning is a kind of "black box" technology, and its explainability is very low, which also restricts the application of deep learning technology in fault diagnosis.

Images