考虑多阶段决策信息集结的新算法及其应用

针对多阶段区间信息集结与决策问题，提出一种考虑最小化集结矩阵与阶段区间矩阵之间距离的新方法，寻求更趋近帕累托最优的集结结果，使最终评价值更符合多阶段评价的目标。 首先，根据多阶段专家评价值将区间信息转化为二维坐标点，并将其映射到二维坐标系中。然后，构建区间信息离差最小化集结模型，并基于植物模拟生长算法(PGSA）进行群体判断信息的集结，再通过合成各方案的属性评价值，给出各决策方案的综合评价值并进行排序，进而给出最优决策方案。最后，以物流服务商的多阶段绩效评价为例，验证了该方法的合理性和有效性。

A new algorithm for aggregating interval information of multistage decision making problems and its application

A new method is proposed to multistage interval information aggregation and decision making. This method considers minimizing the distance between the aggregation matrix and multistage interval matrices. The method aims to seek the aggregation result which is closer to Pareto optimum so as to make the final evaluation value more in line with the goal of multistage evaluation. Firstly, the interval information is converted into two dimensional coordinate points according to the multistage expert evaluation values， and then these points are mapped to planar reference frame. Next, the dispersion minimization aggregation model is constructed and solved by plant simulation growth algorithm. Later, the ranking of each decision plan is given by synthesizing the attribute evaluation values of each plan and then the optimal decision plan is given. Finally, the comprehensive evaluation values and the rationality and effectiveness of the method are verified by the example of the multistage performance evaluation of the logistics service providers.