Kriging metamodel based optimization
Kriging metamodel based optimization
复制标题
基于克里格元模型的优化
DOI：
10.1142/9789812779670_0016
复制
发表时间：
2007
期刊：
影响因子：
--
通讯作者：
J. Jung
中科院分区：
文献类型：
--
作者：
T. Lee;J. Jung
许多设计优化问题可能由目标函数和约束方程组成，需要进行复杂且耗时的数值分析。在这种情况下，传统的优化技术的基础上直接集成计算昂贵的高保真度模拟可能是不切实际的，因为许多迭代函数调用的巨大的计算成本。解决这一问题的方法之一是采用目标和约束的近似响应。近似的目标是提供一个连续的函数，在可接受的保真度内进行评估是廉价的。这种近似模型通常被称为元模型，即，“模型的模型”。1
Many design optimization problems can consist of objective functions and constraint equations that require complicated and time-consuming numerical analyses. In this case, conventional optimization technique based on direct integration with computationally expensive high-fidelity simulations may be impractical because of enormous computational cost of many iterative function calls. One of solutions to overcome the problem is to use approximate responses of the objectives and constraints. The goal of approximation is to provide a continuous function that is inexpensive to evaluate within acceptable fidelity. This approximation model is often referred to metamodel, i.e., ‘models of the model’. 1