论文部分内容阅读
针对多指标群决策时专家评语中存在认知不确定性时的指标赋权问题,提出一种新的组合赋权方法。借鉴证据理论的表达方式和思想,采用DICE系数表征群决策信息之间的相似性,并对专家信息进行融合;采用区间数将评语等级定量化,根据融合信息构造指标重要程度的累积概率函数;采用Monte Carlo随机抽样方法得到指标权重。该方法解决了高冲突群决策信息难以有效融合、融合后信息损失较大的问题,整个处理过程较为客观的保留了专家的原始判断;累积概率函数构造和随机抽样的过程则无需任何人为干预和判断,简洁明了。实例仿真表明了本文方法客观、可靠、简单有效,便于在计算机上实现。
Aiming at the problem of indicator weighting when there exists cognitive uncertainty in expert reviews in multi-indicator group decision-making, a new combination weighting method is proposed. According to the expression theory and theory of evidence theory, DICE coefficients are used to characterize the similarity between group decision-making information and to fuse expert information. The interval number is used to quantify the rating scale, and the cumulative probability function is constructed based on the fusion information. Weights of indicators are obtained by Monte Carlo random sampling method. This method solves the problem of high conflict group decision-making information is difficult to effectively integrate, the information loss after fusion is large, the entire process more objectively retained the expert’s original judgment; cumulative probability function construction and random sampling process without any human intervention and Judgment, concise. The simulation results show that this method is objective, reliable, simple and effective, and easy to implement on the computer.