Integral inference method for ε-N and P-ε-N curves

As the strain fatigue data scatter in lgΔε_e-lgN or lgΔε_p-lgN coordinates, and their standard deviation increases with reducing of elasticity and plasticity components, ε-N and P-ε-N curves cannot be obtained by group experiment method and homoscedasticity regression analysis method. The mean value method and Wissching method are commonly used in engineering. But they are both approximation methods with significant error. Therefore, an integral inference method for ε-N and P-ε-N curves is presented based on the heteroscedasticity regression analysis theory. It can analyze scattered and heteroscedastic data through treating all the experiment data in different strain levels as an integration. Compared with traditional methods, the presented method can not only save specimens, but also increase precision in ε-N and P-ε-N curve tests effectively.