Comparison evolutionary algorithms with Metropolis-Hastings method exemplified by high-fatigue Wohler curve parameter identification

In this work, evolutionary algorithms together with the Metropolis-Hastings sampling technique have been used for parameter identification of the Wohler curve of duraluminum alloy 2024-T3. An evolutionary algorithm is a subset of evolutionary computation, a generic population-based metaheuristic optimization algorithm. The Metropolis-Hasting algorithm is one of the most widespread Markov chain Monte Carlo methods for posterior distribution estimation. In this contribution, both algorithms have been presented to estimate the probability density functions using Wohler parameters as a case study. Results were shown in terms of distribution shape and parameter correlations and the differences, arising from applied algorithms, have been compared. The information about parameter distributions of Wohler equation is useful to prepare risk analyses based on statistical safe life approach. The safe life approach can be met, for instance, in assessing the reliability of an aircraft.