Reliability Analysis Of Response Surface Method Research

The response surface method, the basic idea is to use an explicit simple function instead of structure and function of the implicit function reliability analysis, in other words, use a small number of sample points fitting out a simple response surface approximate the real limit state surface. Obviously, the selection of sample points and the response surface form is the key to the response surface method.In terms of the selection of sample points, the traditional response surface method usually adopts the factor design method or similar method, including choosing center using the check point in the process of iteration. As is known to all, the checking point area of limit state surface accuracy of reliability calculation accuracy has important influence. Although the checking points centered factor design method can ensure the sample points are located around the checking point, but most were not approximate limit state surface, thus to some extent influence the response surface fitting precision. Refer to this paper, the interpolation mapping thought, in view of the commonly used do not contain cross terms of quadratic response surface, the introduction of the interpolation mapping point set, put forward a kind of can make effective sample points are situated near the limit state surface, in the focus on the checking point area can be the method for choosing a bit far areas at the same time, thus developing a kind of improved response surface method.The coefficient of response surface form and fitting, the traditional form of the response surface is quadratic response surface, but for the strongly nonlinear limit state surface is a complex and corresponding quadratic response surface approximation precision is not enough, that cause reliability estimates the larger error. Coefficient is proposed to fit in, in order to further improve the accuracy, a response surface function of the commonly used has a high precision is containing the cross terms of quadra tic response surface and high response surface, but these two forms of the response surface in the process of solving the undetermined coefficient solution involves large matrix calculations, to pay more at the expense of workload in order to improve the p recision and efficiency is low. To solve above problems, this article third chapter will be in the checking point for dimension reduction approximate combined with high-order correction, construct a new reliability analysis method, based on dimension reduc tion and high order approximate model of the response surface method. This method is good to avoid the solving large complex matrix, improve the efficiency; And to discriminate between the variables, reflects the differences of the variables; Introduction of high order correction, in guarantee efficiency at the same time, both the accuracy of reliability.Finally, a practical engineering, the author of this paper space truss structure, ANSYS finite element modeling and static structure analysis. On some of the typical bar strength reliability calculation, reliability analysis was carried out on the rack midpoint displacement, the second chapter is based on interpolation mapping were used respectively to the response surface method and the third chapter is based on dimension reduction approximation and the high order correction of the response surface method to calculate the failure probability, and compared with the classical response surface method and monte carlo method efficiency and precision. Can be seen from the results of the analysis, this paper puts forward two methods are feasible, in both accuracy and efficiency has a good guarantee.