基于Kriging的仿真优化方法，是利用试验设计所产生的采样点构造近似Kriging模型，以代替复杂的仿真模型进行优化分析。但存在Kriging建模效率低、优化精度不佳等问题。为此，本项目研究基于Kriging的仿真优化方法关键技术。建模方面，通过分析Kriging参数，形成基于增广梯度的快速构造方法，以提高建模效率。优化方面，运用对偶理论研究基于Kriging的无约束优化，旨在平衡全局与局部搜索的情况下改善优化效率和精度；其次，为大幅提升优化效率，结合多目标优化策略与多点加点准则形成基于Kriging的并行优化；最后，针对初始采样不存在可行点的约束问题，分析最大约束概率和Kriging参数对可行点的影响，研究基于Kriging的约束优化。以此研制仿真优化原型系统，并以燃料电池汽车设计为例进行验证。本研究有望提高基于Kriging仿真优化中的效率和精度，促进优化设计理论体系的完善与发展。

In the Kriging-based simulation optimization method, the approximate Kriging model, which is constructed by sampling points generated by the experimental design, can replace the complex simulation model to perform optimization analysis. However, it has low modeling efficiency and poor optimization accuracy. In view of this, by analyzing Kriging parameters, a gradient‑enhanced fast construction method is built so as to improve the modeling efficiency in modeling. For optimization aspect, duality theory is used to research the Kriging-based unconstrained optimization in order to improve optimization efficiency and accuracy under the circumstances of balancing between exploration and exploitation. In addition, the multi-points parallel sampling optimization method is proposed by multi objective optimization strategy and multi-point sampling criterion so as to further improve optimization efficiency. Finally, for the constrained optimization problem in which there is no any feasible point in initial experimental design, a Kriging-based constraint global optimization method is researched by analyzing maximum probability constraint and Kriging parameters how to influence feasible point. To this end, the corresponding simulation optimization prototype system is developed, and the proposed methods are verified by fuel cell vehicle. This research is expected to enhance efficiency and accuracy of the Kriging-based optimization problems and promote the perfection and development of the theory system of optimization design.