A method of solving the non-probabilistic reliability index based on support vector machine

Since the limit state function of the mechanical structure is an implicit function of the uncertain variables, it is difficult to analyze structure reliability by means of using conventional methods. A method obtained the non-probability reliability index the basis of support vector machines is proposed for solving more complexity or implicitly of the limit state function. The uncertain variables are described by the convex model, and a method to get sample is rationally constructed. A numerical example is given to assess the reliability of structure with a number of uncertain variables and to elaborate the process of structural reliability on support vector machine. The results show that the method of fitting nonlinear implicit limit state function can not only meet the actual requirement, but also easy to calculate. The method provides a new way for further study about the multi-failure mode reliability.