modeling of corrosionfatigue crack growth rate based on least square support vector machine technique

Understanding crack growth behavior in engineering components subjected to cyclic fatigue loadings is necessary for design and maintenance purpose. fatigue crack growth (fcg) rate strongly depends on the applied loading characteristics in a nonlinear manner, and when the mechanical loadings combine with environmental attacks, this dependency will be more complicated. since, the experimental investigation of fcg behavior under various loading and environmental conditions is timeconsuming and expensive, applying a reliable methodology for prediction of this property is essential. in this regard, a modeling technique based on least square support vector machine (lssvm) framework is employed for prediction of fcg behavior of three different alloys including, ti6al4v alloy and two custrengthened high strength low alloy (hsla) steels in the air and corrosive media. the parameters of the developed model were calculated employing the coupled simulated annealing optimization technique. the performance and accuracy of the developed models were tested and validated by their ability to predict the experimental data. statistical error analyses indicated that the developed model can satisfactorily represent the experimental data with high accuracy.