Jixie chuandong (Jan 2018)
Application of Prior-knowledge for Intelligent Gear Fault Diagnosis Under Variable Speed Condition
Abstract
The training data of an intelligent gear fault diagnosis model usually comes from vibration experiments which run at specific speeds,when the actual running speed of the gear transmission system is not consistent with these vibration experiments,the diagnosis model may be invalidated. The vibration mechanism analysis of gear transmission system shows that,the vibration response signal of a gear transmission system contains interference components that vary greatly with the speed and cannot reflect the fault information,the fault characteristics in amplitude spectrum of the vibration response signal have less sensitivity to the variation of running speed. These interference components can be removed with the help of prior-knowledge which is composed of structure parameters,operating parameters and the vibration mechanism of gear transmission system. Using the amplitude spectrum of the vibration response signal whose interference components are removed as input of the diagnosis model,it can reduce the feature difference between the samples of same class which correspond to different running speeds,and improve the generalization ability of the diagnosis model effectively,so that it can adapt to speed variation in a certain extent.
