Unified linear regression method for the analysis of seven commonly used statistical distributions

作者：Zhao Y; Sun Y; Gao Q

来源：Journal of Mechanical Strength, 2001, 23(1): 102-106+116.

A unified linear regression method for the analysis of seven commonly used statistical distributions, namely Weibull (two-and three-parameter), normal, lognormal, extreme minimum value, extreme maximum value, and exponential, is developed. In the method, statistical parameters and their confidence bounds, and probability confidence bounds of the seven distributions are evaluated by a linear regression method. Statistical tests of the distributions are conveniently made by the Pearson statistical parameter-linear relationship coefficient r. Critical value of the r for an accepted distribution is determined by a transformation of r-to-t-distribution function This method provides a basis of determining an appropriate assumed distribution from the seven distributions in analysis of random data. Availability of this method has been verified by an analysis of the cyclic stress amplitude data of a engineering material, 1Cr18Ni9Ti stainless steel pipe-weld metal, and previous analyses of the fatigue life and crack growth rate data of engineering materials.

• 出版日期2001