Study On Fault Diagnosis Method Of Gear Fracture Based On Auditory Model

Gear transmission is one of the most common forms of transmission,but the proportion of tooth fracture is the highest in the gear failure forms,it is about 41%.Broken teeth make gears cannot mesh properly,and induce periodic impulse response components.If the impulse response component can be extracted and the cycle of impulse response can be identified,or identify the fault status of the mechanical equipment,it can provide an effective basis for fault diagnosis.In this paper,gear fracture fault as the research object,the main contents include:(1)A two-stage helical gear reducer test rig and man-made gear fracture fault were set up,using the acceleration sensor and the DASP data acquisition device to collect vibration signals,which provide data support for the fault diagnosis method and verify the feasibility and reliability of the method.(2)A fault diagnosis method based on Variational Mode Decomposition and wavelet soft threshold denoising is proposed.Firstly,using VMD algorithm to decompose vibration signal,consider that the transient impulse response component is high in energy and occupies a wide frequency band,therefore,the decomposition of the high frequency signal to be superimposed reconstruction.Finally,the reconstructed signal is denoised by wavelet soft threshold denoising.The result shows that the transient impulse response component is extracted accurately.(3)Considering the human auditory system has an instinctive reaction for sudden transient acoustic signals,a method extracting feature of gear fracture fault based on auditory model and signal probability density of extreme points is proposed.The extreme point of transient shock component is extracted by using Gammatone band-pass filter bank,extreme point amplitude probability density and its derivative.Meanwhile,due to the system vibration,it will produce extreme points which are irrelevant to impact of the gear fracture,in order to accurately extract the impact of the gear fracture,according to the transient signal frequency band continuity and multi-band distribution characteristics,design appropriate extraction method.It has been verified by the measured signals show that the proposed method can accurately depict feature of gear fracture fault,and can extract impact component which only induced by gear fracture fault in a variety of types of transient impulse response component,and accuracy of extraction results is well.(4)EA auditory model was established by simulating the human auditory system,and use EA model to process the vibration signal under the four states of broken tooth,pitting,wear,normal state and to be identified signal,set up the auditory spectrum feature sets,and the auditory perceptual feature parameters are introduced into the state recognition,using Principal Component Analysis and projection to achieve the purpose of state identification,the time-frequency domain characteristics of different fault states are also analyzed.