A vibration data fault classification method based on depth domain adaptation

Abstract

The invention discloses a vibration data fault classification method based on depth domain adaptation. The vibration data fault classification method comprises the steps of constructing a source domain containing a large amount of sample data and a target domain containing a small amount of sample data; constructing a classifier by using the sample data in the source domain; constructing paired samples according to the sample data in different domains in the source domain and the target domain; constructing a twin network, inputting the paired samples as training samples into the twin networkfor domain adaptation, and obtaining a final loss function of the paired samples; and optimizing the final loss function, and obtaining a trained fault classification model. According to the present invention, the problem that an existing deep network model is poor in diagnosis effect under the condition that fault sample data are insufficient is solved, a domain self-adaption method in deep learning and transfer learning is combined, the existing data are utilized to the maximum extent, the generalization capacity of the model is improved, and therefore better classification accuracy is obtained.

Description

technical field [0001] The invention relates to the technical field of vibration signal processing and fault classification, in particular to a vibration data fault classification method based on depth field self-adaptation. Background technique [0002] Major mechanical equipment such as wind power generation equipment, aeroengines, and high-end CNC machine tools are developing in the direction of large-scale, complex, high-speed and high-precision equipment. Once an accident occurs in the equipment, it will bring huge economic losses and casualties. Therefore, it is of great significance to study advanced mechanical fault diagnosis methods to ensure the safe operation of equipment. [0003] Traditional bearing fault diagnosis methods can usually be divided into two parts: feature extraction and classifier classification. Among them, feature extraction based on signal processing technology is often aimed at specific problems, requiring diagnostic experts to deeply understa...

AI Technical Summary

Problems solved by technology

[0008] In order to overcome the defects in the above-mentioned prior art, the present invention provides a vibration data fault classification method based on depth field self-adaptation, which solves the problem of poor diagnosis effect of the existing deep network model under the condition of insufficient fault sample data. The combination of learning and domain adaptive methods in transfer learning maximizes the use of existing data, improves the generalization ability of the model, and obtains better classification accuracy

Images