中文摘要：

1引言在产品的研发、设计过程中,往往需要寻找适当的设计参数,使得在同样的投入下获得更高性能的产品.产品的性能可看作是设计参数的函数,这种函数往往没有解析表达式,可称为黑箱函数.对黑箱函数每调用一次,就需要做一次耗费大量时间、财力的试验,即便是用计算机模拟,也是非常耗时的过程,期间消耗的资源和时间占整个设计过程的绝大多数,因此对产品的优化设计,即黑箱函数的全局优化,应当尽可能地以少量的函数值调用,获得高效益的设计参数.也就是说,对这类问题,函数调用次数的多少而不

英文摘要：

The author proposes a combinational response surface method （CRSM） for solving the global optimization of expensive black box functions. We use the approximation of a sequence of the optimum points of response surface model to the optimum point of the black box function for reducing the number of evaluations of the black box function, and at the same time, use the sampling points and their values to generate an the interpolation function and introduce a func- tion called ＂ function of closeness＂ for describing the distribution of the sampling points. We then combine the two function to form the combinational response sur- face, optimize the response surface model and use the optimum point to renew the sampling points. Repeating the above process will generate a sequence of points which will be proved to be converged to the global optimum point of the black box function. Numerical tests are compared the CRSM with some outstanding methods. The results illuminate the high emciency of the CRSM.