Multiobjective Decision Making for Risk Management in Large-Scale Systems

随着规模和复杂性的增加，与风险管理相关的问题解决与群体之间涉及个人价值观的冲突有关。为了达到一个公平合理的目标，我们需要考虑多目标群体决策中的冲突问题，在系统方法（GDSS）的帮助下，我们考虑了群体决策支持系统。首先，假设（1）每个群体对决策的影响是相等的，（2）每个目标都是数学函数。运用层次分析法（AHP）揭示了各群体的价值体系，表明该过程的主要框架可以归结为多目标优化问题的交互式求解。提出了群体间冲突解决的关键是共同拥有的广场信息和妥协的精神，并通过放宽上述限制，对群体决策支持系统进行了扩展场所我们已经应用遗传算法（GA）来解决由此产生的离散优化问题，并表明它是有效的，只有轻微的负荷，在人机交互过程中，通过应用于放射性废物管理系统的数值实验，我们已经表明每个GDSS是非常有前途的处理各自的问题解决这里关注。

According to increase both in scale and complexity, problem-solvings associated with risk management are related to conflicts referring to individual sense of value among the groups. To attain at a square and rational goal, therefore, we need concern with the conflicts in terms of group decision making under multiobjectives.In the aid of systems (GDSS) approach, we have considered a group decision support system. First the GDSS has been developed assuming (1) each group has equal effect on decision and (2) every objective is given as mathematical function. Using the analytic hierarchy process (AHP) to reveal value system of each group, we have shown the major frame of the procedure can be reduced as an interactive solution of the multi-objective optimization problem.As known in the paradigm of risk communication, jointly owned square information and the spirit of compromise are cited essential for the conflict resolution among the groups.Then the GDSS has been extended by relaxing the above premises. We have applied the generic algorithm (GA) to solve the resulting discrete optimization problems, and shown it is effective and imposes only mild loads in the man-machine interaction process.Through numerical experiments applied to a rad-waste management system, we have shown each GDSS is very promising to deal with respective problem-solvings concerned here.