A weighted multi-scale dictionary learning framework for fault identification of planetary gear bearings

Abstract

The invention discloses a planetary gear bearing fault identification method with a weighted multi-dimensioned dictionary learning frame. The method comprises the following steps that a sub-block operator is constructed based on a planetary gear bearing vibration signal; the weighted multi-dimensioned dictionary learning frame is constructed based on the sub-block operator, the weighted multi-dimensioned dictionary learning frame is solved in an optimized way, and a fault characteristic signal is obtained; a weighted multi-dimensioned dictionary special case is constructed based on Q-switchingwavelet and L0 regularity, and a planetary gear bearing fault characteristic signal is extracted via the special case; and a fault type is identified by envelope analysis based on the extracted faultcharacteristic signal.

Description

technical field [0001] The invention belongs to the technical field of fault diagnosis methods, in particular to a planetary wheel bearing fault identification method based on a weighted multi-scale dictionary learning framework. Background technique [0002] Electromechanical equipment is an important part of modern industry, and its operation safety is particularly important. The traditional way of regular maintenance consumes a lot of manpower, material and financial resources, and can no longer adapt to the trend of modern industrial development. Condition-based maintenance has the obvious advantages of small scale, high efficiency, good economic affordability and the ability to diagnose and predict major disasters. An important prerequisite for effective condition-based maintenance is to build a complex prediction and health management system. Vibration signal analysis and fault diagnosis are one of the important components of the health management system. The vibrati...

AI Technical Summary

Problems solved by technology

However, the dictionary learning algorithm also has corresponding shortcomings: 1. Before learning the dictionary, we need to divide the original signal into blocks, so that our model is only limited to the learning of local information; Wave interference is very sensitive, however, there are often a lot of harmonic interference in the vibration signal; three, structured dictionary learning needs to establish a structured mathematical description of the characteristic signal, but often many physical characteristics cannot establish an effective mathematical description; four, The dictionary learning model often has a large amount of calculation due to the limitation of its sample dimension

Images