Adaptive Extraction Of Acoustic Emission Feature Of Gear Fault

Gears as the core components of rotating equipment,their own structure and operating environment is more complex,so in the process of operation often appear wear and tear or even fracture conditions,resulting in the failure of the entire machinery and equipment.Not only do they suffer economically,but they also threaten personal safety in serious cases.Therefore,the fault diagnosis of gears has important practical significance.Acoustic emission is a method of detecting kinetic energy emitted from the material defect itself and is highly sensitive to faults and can be detected in real time,so the application of acoustic emission technology to fault diagnosis is receiving increasing attention.In fault diagnosis,the operating background noise of the mechanical system will interfere with the acoustic emission signal,which makes it challenging to extract the fault-related features.Selecting the appropriate parameters representing the gear state in the feature extraction is also a fundamental problem,which will lead to significant interference in the fault feature diagnosis.This paper studies the adaptive feature extraction in fault diagnosis based on acoustic emission technology.The main work are as follows:1.Gears fault diagnosis simulation experiment was conducted based on the acoustic emission signal.For the background noise contained in the acoustic emission signal in this paper.Disposition method of acoustic transmission signal of gear wear fault are determined by using empirical mode decomposition,Ensemble Empirical Mode Decomposition,and Variational mode decomposition.2.Analyze the influence of the number of modes and the penalty factor in Variational mode decomposition on signal decomposition,The grey Wolf optimization algorithm is also used to improve the parameter selection problem.Using the correlation coefficient principle,we can improve the validity of the reconstructed signal.3.To address the problem of gear acoustic emission signal feature extraction,a support vector machine recursive feature elimination method is used to adaptively rank the importance of feature parameters,comparing the recognition accuracy of the selected feature vector set with the randomly selected feature vector set and all the feature vector sets by the support vector machine.4.Support vector machine is used to analyze the experimental data and optimize the centralized kernel parameters and penalty factor selection problem for the gear fault classification problem.