［Purposes］ A response surface method is proposed for the optimization of expensive black box functions, which can better balance local search and global search. ［Methods］ The CORS method is improved, introducing the mechanism of judging the local optimum solution to CORS, then jump out of it using filled function method. ［Findings］ Numerical experiments on 37 open global optimization algorithm test problems show that the adaptive response surface algorithm based on the local jump out local optimal mechanism of filled functions can converge to a better approximate global optimal solution. ［Conclusions］ The local optimal solution jumping mechanism based on the filled function method can improve the global search ability of the algorithm.