A new intelligent design method for building material fatigue S-N curve

摘要

It is the basis of material fatigue reliability analysis to obtain fatigue S-N curve, The curve is usually obtained from experiment or by fitting many data of loading test basis on three parameters empirical formula, but those methods are complex, high expense and imprecision. In accordance with these disadvantages above methods, a new intelligent method based on common machine learning algorithm (support vector machine) is presented to obtain S-N curve of material fatigue economically and effectively. Complicated and strong nonlinear S-N curve was simulated by design and conformation of support vector machine learning algorithm, compared the errors with output value of the intelligent model, test value and output value from fitting values of three parameters power function, support vector machine learning algorithm had excellent ability of nonlinear modeling and generalization. It gained high precision under limited learning samples and mean relative error is 0.008954%, it provided an economical, practical and reliable approach for material fatigue design.