Bearing Early Fault Detection and Diagnosis Method and System Based on Singular Value and Graph Theory Feature Fusion

Abstract

The invention provides a bearing early fault detection and diagnosis method and system based on singular value and graph theory feature fusion. The method comprises the steps of extracting a feature graph of an operation state of a mechanical bearing; detecting a bearing early fault; diagnosing a bearing fault; and learning fault types possibly occurring in an offline state, establishing a corresponding graph model set, classifying a feature graph and state graph set after a fault symptom point occurs, and diagnosing a type of a current fault state in real time. A symptom point at which the early fault occurs can be monitored and identified in real time online in a mechanical operation process, and moreover, the type of the fault can be detected timely. The method can be directly applied to related application of early fault detection and diagnosis of an industrial mechanical bearing.

Description

technical field [0001] The present disclosure relates to the technical field of bearing fault diagnosis, in particular to a bearing early fault detection and diagnosis method and system based on fusion of singular value and graph theory features. Background technique [0002] At present, the mechanical system in actual use often works in the environment of high speed, heavy load, high and low temperature alternating, and the mechanical system may fail during the continuous working process. Among them, bearings are an important part of rotating machinery, and 40% of mechanical system failures are caused by bearing failures. Therefore, early fault detection and diagnosis of bearings can reduce the time and cost of downtime maintenance, as well as product failures caused by downtime maintenance. economic loss, or even avoid personal injury. The early fault detection and diagnosis of bearings can evaluate the running state of the machine in real time, find out the symptoms of e...

AI Technical Summary

Problems solved by technology

[0003] The inventor found in the research that the current commonly used bearing fault diagnosis method is based on the analysis method in the frequency domain, but this method requires professional knowledge to identify the fault type, which is an offline method; the method based on artificial intelligence requires a large amount of history. When the data is used to train the classifier, when the historical data is relatively small, the detection accuracy will be reduced, and the calculation amount will be large, so the real-time and online diagnosis performance will be low

Images