Study On The Dynamic Characteristics Of A Running-in Attractor In Tribosystem

A system tends to develop towards a steady state,which is called an attractor.The chaotic attractor is a kind of attractor,which is a moving form of the disorder steady state of the chaotic system.The tribosystem is a chaotic system,existing a chaotic attractor.The chaotic attractor of a tribosystem forming in the running-in process is a running-in attractor,symbolizing the stable wear state of a tribosystem.It is of great significance to study the chaotic characteristics of the tribosystem.In order to reveal the dynamic characteristics of the running-in attractor,enrich the characteristic parameters of the running-in attractor,and realize on-line identification and prediction of wear states of mechanical system,the characteristics and experimental model of the running-in attractor are studied based on the chaos and fractal theory and BP neural network.To systematically explore the dynamic characteristics of the running-in attractor from multiple dynamic perspectives,moving cut data-approximate entropy?MC-ApEn?is introduced to recognize the state abrupt points of the running-in attractor.Additionally,3D histogram of phase points is proposed to describe the space distribution features of the phase points constitute the phase trajectory.And the quantitative characteristic parameters of 3D histogram of phase points,maximum number of phase points and the number of bins containing phase points,are used to quantify the spatial distribution characteristics of the phase points.Based on the distance matrix,the convergence?CON?and standard deviance of distance matrix?STD?are proposed to characterize the degree of convergence and the dispersion degree of the running-in attractor,respectively,so as to reflect the stability of the running-in attractor in a tribosystem.According to the Stribeck curve and the selection criteria of the surface roughness,the ranges of the load,speed and the initial surface roughness of the disk are determined.A series of wear experiments are conducted on a pin-on-disk tribometer.Then,calculate the correlation dimension D,MC-ApEn,the maximum phase points number and bin number storing phase points of 3D histogram,CON and STD of the friction coefficient?COF?signals extracting from the wear experiments,and describe the variation curves of these characteristic parameters in the figures.Then analyze the variations of the running-in times,COF values in the stable wear process,and the aforementioned characteristic parameters as the wear process continuing.Experimental results show that with the increasing load,the running-in time continuously decreases;the COF value in stable process firstly keeps stable at a minimum,then suddenly increases,and finally keeps relatively stable at amaximum;all the characteristic parameters follow the “bath curve” or “inverted bath curve” evolution rules,also corresponding to the running-in,stable wear,and rapidly wear processes.However,the evolution rules of all the characteristic parameters disappear when the load increases to a certain value.That is to say,the tribosystem has no the forming condition of the running-in attractor under this load.With the increase of the rotating speed,the running-in time continuously decreases;the COF value in stable process firstly increases and then decreases;all the characteristic parameters follow the “bath curve” or“inverted bath curve” evolution rules,agreeing with the running-in,stable wear,rapid wear processes.However,the variations of all the characteristic parameters have no obvious rules when the speed increases to a certain value.That is to say,the running-in attractor in the tribosystem can not form under this rotating speed.With the increasing initial surface roughness of the disk,the running-in time and the COF value in stable process gradually increase;All the characteristic parameters follow the “bath curve” or “inverted bath curve”evolution rules,also in accord to the running-in,stable wear,and rapid wear processes.Furthermore,the tribosystem can faster enters into the stable wear process when the initial surface roughness equals to a certain balanced roughness.Consequently,the system parameters have an influence on the formation of a running-in attractor.The system parameters of the medium or low loads and velocities as well as relatively smooth surfaces,are conductive to improve the reliability of the mechanical parts and prolong the service lifespan of the mechanical parts.Phased wear experiments are performed on a ring-on-disk tribometer.Prior to testing,the masses of the specimens are measured.After testing,the masses of the specimens are also measured,and the wear particles are collected.Then,the variation of the wear mass of the friction pair with the wear process is studied.The geometrical and fractal features of the wear particle group are studied and used to compare with the characteristic parameters of the running-in attractor.Experimental results show that the variation curve of the wear mass of the friction pair is consistent with the theoretical wear mass curve of the mechanical part.The counts of the wear particles with different chord lengths obey the normal distribution.With the wear process continuing,the equivalent mean chord length of the wear particle group follows the evolution rule of “decrease-stability-increase”,and the fractal dimension of the wear particle distribution follows the evolution rule of“increase-stability-decrease”.The variation processes of the equivalent mean chord length and fractal dimension of the wear particle group,are compared with the variation processes of D,MC-ApEn and STD of the running-in attractor.It is found that the evolution of the fractal features of the wear particle group is consistent with the chaotic characteristics ofthe running-in attractor.Therefore,the fractal features of the wear particle group can be used to characterize the dynamic characteristics of the running-in attractor.On the basis of the aforementioned phased wear experiments,prior to and after testing,the surface roughness values and 3D topographies of the ring and disk are measured.The2 D and 3D fractal features of the surface topographies of the specimens are studied,and they are used to compare with the characteristic parameters of the running-in attractor.Experimental results show that with the wear process continuing,the surface roughness of the ring gradually increases,and the surface roughness of the disk firstly increases and then remains relatively steady,and finally increases once again.For the disk,the 2D profile fractal dimension,characteristic roughness,and the 3D topography fractal dimension,obey the evolution law of “increase-stability-decrease”.The variation processes of the three fractal parameters of the disk surface are compared with the variation processes of D,MC-ApEn and STD of the running-in attractor.It is found that the fractal features and the chaotic features of the tribosystem agree with each other.Therefore,the 2D and 3D fractal features of the worn surface topography can be applied to characterize the dynamic characteristics of the running-in attractor.The scheme of the orthogonal experiments with different loads,rotating speeds and initial surface roughness values of disk are designed by the Design_expert₈ software.The running-in orthogonal experiments are performed on a pin-on-disk tribometer.After testing,the characteristic parameters D,CON,MC-ApEn and STD values of the COF signals in the stale wear process are calculated,the running-in times and the COF values in the stable wear process are summed.Then models of the running-in time,COF values,D,MC-ApEn and STD in the stable wear process are established based on the BP neural network.Through training and testing,the prediction accuracy of BP models reaches 80%,so the BP models can be used to predict the running-in attractor.Based the running-in attractor model,the output parameters of the tribosystem under different system parameters are predicted and constructed to a prediction database.Then apply the LabVIEW and MATLB to develop an on-line wear state recognition system.The function of this system is to compare the predicted and real values of the output parameters,so as to identify and monitor the current running-in states of a tribosystem.This paper has 132 figures,28 tables and 191 references.