Abstract: The traditional Multi-Criteria Group Decision Making (MCDM) approaches usually determined evaluation criteria weights subjectively,and did not consider the influence relationships among criteria.Data Envelopment Analysis (DEA) did not need criteria weights,but the optimized criteria weights obtained by DEA could not reflect the objective importance of criteria.Aiming at the problems of DEA,Decision Making Trial and Evaluation Laboratory (DEMATEL) was combined with DEA.DEMATEL was used to analyze the influence relationships among criteria to determine the importance degrees of criteria which were used to adjust the criteria weights optimized by DEA.The 2-tuple linguistic model was used to deal with the linguistic evaluation information and the information of influence relationships among criteria provided by experts.The dynamic weights of experts were determined by analyzing the closeness degree between the evaluation value of expert given and the value of others.The 2-tuple linguistic DEA model combined with DEMATEL was used to calculate the comprehensive evaluation values.A case study of sensor medical equipment selection in a hospital was presented to illustrate the effectiveness of the proposed approach.

Key words: multi-criteria group decision making, 2-tuple linguistic, decision making trial and evaluation laboratory, data envelopment analysis, expert dynamic weight

摘要： 传统多属性决策方法通常主观地确定评估指标权重,没有考虑指标间的影响关系;数据包络分析法无需指标权重,但其优化得到的指标权重不能反映客观的指标重要性。针对上述问题,将决策实验与评价试验法和数据包络分析方法相结合,采用决策实验与评价试验法分析指标间的影响关系,确定指标的重要度,从而修正数据包络分析优化得到的指标权重。采用二元语义表达模型处理专家给出的语义评估信息和指标影响关系信息,通过分析专家评估值与其他专家评估值的贴近度确定专家动态权重,并采用结合二元语义决策实验与评估试验法的数据包络分析模型计算候选方案的综合评估值。以某医院传感器医疗设备选择为例,验证了所提方法的有效性。

关键词: 多属性群决策, 二元语义, 决策实验与评价试验法, 数据包络分析, 专家动态权重