Research On Calculation And Analysis Method Of Fatigue Life Of Automobile Parts Under Two-Dimension Random Load

When researching automobile fatigue life,there are two methods of applying load:recycling load and random load.Random load is more fittable to actual driving condition of a vehicle comparing to recycling load.As a vehicle works under random load,the load is a two-dimension random load including two random variable quantities:mean value and amplitude.So,based on load spectrums of automobile parts,this paper did researching on probability density functions of equivalent loads of two-dimension random loads,and calculation and analysis methods of fatigue life under two-dimension random load.The main work and creation points are as following:1.Calculation theory and methods of fatigue life under two-dimension random load were studied considering respective random characteristics of mean value and amplitude.A calculation model of fatigue life under two-dimension random load was established.Result of the researching showed that,when fatigue resistance coefficient C_f and theoretic fatigue limit(S_eq)_c that presented by equivalent stress amplitude are definite values,the key factor influencing fatigue life is probability density function of equivalent load.2.Numerical integration calculation method of fatigue life under two-dimension random load was studied,which calculates fatigue life with numerical integration to probability density function of equivalent load of two-dimension random load.Adopted experimental design method,the correctness and availability of the method were verified.Result of comparison analysis showed that,as to continuous or approximate continuous load spectrum,numerical integration method is a fatigue life calculation method superior to traditional 8-levels load segmentation method.3.Two-dimension random characteristics of load spectrums of bearing system parts of a city SUV on test ground were studied,distribution inspections and parameter estimations were performed,the results showed that,amplitudes of the loads followed Weibull distribution and mean values followed Gauss distribution.Fitting accuracies of the load spectrums of four coil springs were studied,the result showed that,as distribution inspections and parameter estimations were performed after deleting small load data,the estimated parameters could realize good fitting accuracies to original load spectrums.Correlation analyses were performed,the results showed that the amplitudes were uncorrelated or weak correlated to the mean values,so we must consider their respective random characteristics when performing fatigue life analysis.4.Of equivalent loads of two-dimension random loads whose mean values and amplitudes were independent each other,probability density functions were derived with algebra method on two conditions:（1）Mean value and amplitude followed Gauss distribution also;（2）Amplitude followed Weibull distribution and dimension parameter=1,position parameter=0,mean value followed Gauss distribution.Thus,probability density functions of equivalent loads were acquired.The function graphs of different distribution parameters were also studied,the variation regularities were acquired between probability density functions and distribution parameters.Based on the probability density functions,fatigue lives were calculated with Gauss-Legendre numerical integration method,so the feasibility、rationality and availability of fatigue life calculation with the derived probability density functions were verified.Influences of different distribution parameters to fatigue lives were analyzed,variation regulations between distribution parameters and fatigue lives were acquired.5.Random data of amplitudes and mean values of two-dimension random loads were generated with Monto-Carlo method,equivalent loads of the generated random data were transferred with Goodman formula,distribution inspections and parameter estimations of the equivalent loads were performed,distribution forms and parameters of probability density functions were acquired.Relationships were studied between parameters of the equivalent loads and the mean values and amplitudes,functions were matched with least-squares polynomial,fitting functions of the distribution parameters were acquired.Fitting accuracies of the functions were inspected,the result showed that the fitting functions have good accuracies.6.Took four coil springs as the researching targets,two-dimension random load spectrums were processed with Monto-Carlo probability density function fitting method,fatigue lives were calculated with numerical integration method based on established calculation model of fatigue life.The results were compared with which calculated and analyzed with three methods used in engineering practice:load segmentation method、equivalent load conversion method and linear extrapolation method,so rationality and availability of established calculation model and Monto-Carlo probability density function fitting method was verified,as to fatigue life calculation and analysis of automobile parts.A comparison of advantages and disadvantages was analyzed about the four methods,as the result,linear extrapolation method is a unreasonable method,it has two obvious problems:（1）If small probability loads were collected in sample,the result of fatigue life would be smaller;（2）If small probability loads were not collected in sample,the result of fatigue life would be larger.The superiority and inferiority ranking of the four methods is:Monto-Carlo probability density function fitting method（equivalent load conversion method）、load segmentation method、linear extrapolation method,as to Monto-Carlo probability density function fitting method and equivalent load conversion method,the selection of the two methods is decided by correlation between amplitudes and mean values,Monto-Carlo probability density function fitting method will be selected on low correlation condition,equivalent load conversion method will be selected on high correlation condition.