The Application of Statistical Techniques to the Analysis of Fatigue Endurance Data

摘要

The maximum likelihood analysis method is a powerful tool in the statistical analysis of fatigue endurance data which includes run-out or censored values. A FORTRAN program has been developed for maximum likelihood analysis, as well as for calculating the likelihood ratio test statistic. Families of stress/life fatigue data from laboratory coupon tests at different mean stresses for different types of specimen (i.e. low load transfer and high load transfer) are analysed. The characteristic parameters of the underlying statistical distribution of fatigue data are estimated. Best fit lines and lower confidence /reliability limits for each data set are estimated over a range of stress levels. In addition, Walker's equivalencing approach and an RMS error method have been implemented for comparison purposes, and the results are discussed.