详细阐述构造最优实验设计的原始随机进化算法,并在原始算法的基础上,拓展广度搜索,改进深度搜索,以提高最优实验设计的计算速度。通过不同规模和不同优化准则的拉丁超立方体最优实验设计,验证改进算法的应用效果。算例分析表明,改进算法能够比原始算法节省约30%~60%的机时完成最优实验设计,而且改进算法对应于优化准则的最优值与原始算法最优值的差别仅为1%~3%。可见,改进算法能够兼顾最优实验设计的计算时间和优化质量,明显提高最优实验设计的构造效率。
The stochastic evolutionary algorithm is demonstrated in detail for the construction of optimal experiment design.Based on the original algorithm,breadth search is expanded and depth search is improved,to increase the calculation speed of optimal experimental design.The application results of the improved algorithm are verified by searching Latin hypercube optimal design of varying scales under different optimization criteria.The example analysis shows that the improved algorithm will be better than the original algorithm for reducing 30% to 60% computing time to find the optimum experimental design and the global approxima- tion optimization value of improved algorithm is different from that of original algorithm for a mere 1% to 3%.Therefore,the improved algorithm will reconcile the calculation time and the optimization quality,and significantly enhance the construction efficiency of optimal experimental design.