具有复杂约束混料试验的渐近D-最优设计

摘要

It is difficult to get an accurate optimum design when the experimental design area is very irregular under complex constraints.This paper constructs a random search algorithm for mixture experiments designed (MDRS).Firstly,generating an initial points set in areas with complex constraints by the Monte-Carlo method,then use MDRS algorithm iterative to approximate optimum set.By way of example verification,this method is effective.It can be used as a standard measure of other designs,that is the only effective when given superior to other designs approximate optimal solution.%复杂约束条件下试验设计区域极不规则,通常难以得到精确的最优设计.本文构造一种针对混料试验设计的随机搜索算法(MDRS),在具有复杂约束的区域内由Monte-Carlo方法产生一组初始点集,并通过MDRS算法迭代至逼近最优点集.通过实例验证,这种方法是有效的.它可以作为衡量其他设计的一个标准,即只有当给出的其他设计优于近似的最优解时才是有效.