Aero-engine rolling bearing fault diagnosis method based on twin network metric learning

Abstract

The invention belongs to the field of rolling bearing fault diagnosis, provides an aero-engine rolling bearing fault diagnosis method based on twin network metric learning, combines a twin network with metric learning, and provides an aero-engine bearing fault diagnosis method based on a twin network structure and adopting a metric learning strategy. The original learning strategy of the twin network is changed, and the difference between samples is measured by adopting the distance of sample features in metric learning. Under the condition that the sample size is seriously insufficient, a multi-classification task with the insufficient sample size is converted into a plurality of dichotomy tasks by adopting a form of constructing a learning task, so that available model training tasks are greatly expanded. A powerful support is provided for the aero-engine bearing fault diagnosis problem under the condition that samples are lacked, and meanwhile the method has high engineering practical application value.

Description

technical field [0001] The invention relates to the field of fault diagnosis of rolling bearings, in particular to a fault diagnosis method of aero-engine rolling bearings based on twin network metric learning. Background technique [0002] The aircraft engine is regarded as the heart of the aircraft and must have extremely high reliability. Any minor failure may result in engine damage or even serious casualties. Maintenance of aero-engines is a huge expense. Bearings, as key components of aero-engines, determine the performance of aero-engines and directly affect their service life. In order to ensure that aero-engines can work safely and save maintenance costs, bearing fault diagnosis is very important. An efficient and reliable aero-engine bearing fault diagnosis system is of great significance to the safe flight of aircraft. [0003] The existing data-driven bearing fault diagnosis method does not rely on the prior knowledge of experts, and directly uses the data colle...

AI Technical Summary

Problems solved by technology

The current fault diagnosis method based on deep learning considers the situation of sufficient data when constructing the network model. Software simulation, so it is difficult to collect a large amount of valid data, and it is even more difficult to collect enough labeled data to meet the requirements of data-driven methods

[0004] In the field of small-shot learning fault diagnosis developed by insufficient sample size, poor model generalization ability is a difficult problem

Solving the problem of insufficient samples usually starts with increasing the data volume. This method is effective when the data volume has a certain scale, but in the case of extreme shortage of data samples, it cannot effectively expand the data volume through resampling.

How to solve the problem of ineffective model training caused by insufficient data volume is the direction that needs to be explored in the existing aero-engine bearing fault diagnosis

Images