Rolling bearing fault diagnosis method based on manifold preserving transfer learning

Abstract

The invention provides a rolling bearing fault diagnosis method based on manifold preserving transfer learning. Local linear space identification (LFDA) and transfer component analysis (TCA) are combined, a semi-supervised transfer component analysis method (TCAPLMS) which retains a local manifold structure is designed, data distribution difference between different domain data sets is reduced, and a local manifold structure capable of retaining sample label information and state feature information is obtained. A preferred feature selection method (PSFFC) based on fault sensitivity and feature correlation is put forward to be embedded into the framework, so that redundant information of a time-frequency statistical feature parameter space is reduced. According to the method provided by the invention, the diagnosis accuracy can be obviously improved, and the method has strong adaptability and generalization ability to actual industrial scenes.

Description

technical field [0001] The invention relates to the field of fault diagnosis, in particular to a rolling bearing fault diagnosis method based on popularity-keeping transfer learning. Background technique [0002] When rotating machinery operates in harsh working environments, the failure probability of rolling bearings (REBs) is generally higher than other components of rotating machinery. At the same time, bearings play a very important role in industrial applications. In order to ensure the reliability of their work and reduce the economic loss caused by damage, the fault diagnosis of bearings has attracted more and more attention. With the advent of the big data era, signal processing and data mining technologies are developing rapidly, and data-driven fault diagnosis methods are also developing rapidly. However, traditional data-driven intelligent diagnosis methods have the following disadvantages in terms of applicability in practical industrial applications. (1) In t...

AI Technical Summary

Problems solved by technology

(1) In the traditional mechanical equipment fault diagnosis model, the feature extraction and fault classification steps assume that the training data and test data have the same distribution, but when the working conditions in the actual industrial scene are inconsistent, this common premise does not hold, resulting in the intelligent diagnosis method The usefulness of the

(2) Due to the changeable working conditions of rotating machinery and various types of faults, but limited to the situation where the experimental site and equipment faults cannot be completely simulated, only a small part of the actual measurable data for bearing faults makes the marking target fault data insufficient

Therefore, traditional data-driven intelligent diagnosis methods cannot establish accurate target bearing fault diagnosis models in real diagnosis scenarios

Images