嵌套拉丁超立方设计的优化
Optimization of Nested Latin Hypercube Designs
由于嵌套拉丁超立方设计~(nested Latin hypercube design,~NLHD) 在整个设计区域中并不总具有很好的空间填充性, 文章提出了一个分层加强随机进化算法对~NLHD 进行优化, 提高它的空间填充性. 该算法优先考虑在实际中更重要的底层设计, 逐层优化后得到最终的~NLHD. 提出了三类基本操作来搜索更好的设计, 并保证了每步更新后的设计依然保持~NLHD 的结构. 算例表明该算法速度快, 效率高.
The nested Latin hypercube design (NLHD) does not always have good space-filling property over the experimental region. This paper introduces a multi-layer enhanced stochastic evolutionary algorithm to improve that property of NLHDs. The proposed algorithm first considers the sub-designs in the lower layers, and optimizes the NLHD layer by layer. Three basic element-exchanging operations are proposed to search better NLHDs. The whole design always keeps the structure of the NLHD after each operation. Numerical examples indicate that the proposed algorithm is fast and efficient.
空间填充设计 / 随机进化算法 / 计算机试验. {{custom_keyword}} /
/
〈 | 〉 |