A New Method for Parameters Estimation of Probabilistic Fatigue Crack Growth Rate Model

摘要

A new method is proposed to estimate the parameters of probabilistic fatigue crack growth rate models, including the Paris equation and its' improved type. To take the statistical characteristics of whole test data into account, the method inherits the idea from the general maximum likelihood method which is widely used in parameters estimation of fatigue S-N curves, ε-N curves, and da/dN-△K curves, and extends the conventional correlation coefficient optimization method into parameters evaluation not only for mean curve, but also for standard deviation curve and probabilistic curve. Analysis on the test data of 16MnR steel indicates that present method is available and feasible. Comparing to the general maximum likelihood method, present method has simpler algorithm, and can avoid constructing and solving the likelihood function, so it is speedier in calculation.