A moment plotting method for characterizing aircraft fatigue data population distributions

Abstract

A method was developed where, by statistically analyzing samples of aircraft fatigue data from a population, the population's probability distribution function is determined. The method uses plots of a sample's coefficient of variation, and measures of skewness and kurtosis, superimposed on templates of curves created from the moment functions of Normal, Lognormal, Weibull and Exponential distributions. The moment functions are derived for each distribution. The plotting technique is combined with a comparison of goodness of fit statistics to form the complete distribution selection method. The method is first tested on random numbers generated from known distributions to see if the correct distributions are selected; then the method is applied to helicopter loads data, fatigue life characterization test data, and counting accelerometer data