摘要
针对多阶段群决策环境下的混合型偏好信息集结问题,引入证据理论对实数、区间数、语言信息及语言直觉模糊数信息进行统一度量,并提出一种新的、基于发展稳定和增长趋势的多阶段激励性集结方法.首先,利用基于改进灰色关联分析的模糊测度方法将混合偏好信息转化为基本信度分配函数形式(mass函数);其次,综合冲突系数及证据间的相似性程度确定证据冲突度,并以此计算专家权重,利用证据权值和Demsper组合规则对不同专家的mass函数进行修正融合;再次,利用多阶段方案的排序与平均序值的差值衡量备选方案的波动情况,并定义稳定性激励调节系数;最后,将该决策方法运用到商业银行对小微企业融资贷款的选择案例中,对比分析方法的合理性和可行性.
In order to solve the aggregation problem of hybrid preference information in multistage group decision-making environments,evidence theory is introduced to uniformly measure real number,interval number,linguistic information,and linguistic intuitionistic fuzzy number information,and a new multi stage incentive aggregation method based on development stability and growth trend is proposed.Firstly,a fuzzy measure method based on improved grey correlation analysis is used to transform the hybrid preference information into a basic reliability distribution function(mass function).Secondly,the degree of evidence conflict is determined by integrating the conflict coefficient and the degree of similarity between evidences,and expert weights are calculated based on this;after that,the mass functions of different experts are modified and fused using the evidence weights and Demper combination rules.Thirdly,the difference between the ranking values of multi-stage schemes and average ranking values is used to measure the volatility of alternative schemes,and define the stability incentive adjustment coefficient.Finally,this decision-making method is applied to the selection of financing loans for small and micro enterprises by commercial banks to compare and analyze its rationality and feasibility.
作者
张发明
张淋茜
韩江涛
ZHANG Faming;ZHANG Linqian;HAN Jiangtao(School of Business,Guilin University of Electronic Technology,Guilin 541004,China)
出处
《系统工程理论与实践》
EI
CSCD
北大核心
2023年第12期3619-3635,共17页
Systems Engineering-Theory & Practice
基金
国家自然科学基金(72161006)
教育部人文社会科学研究基金(21XJA630009)
广西自然科学基金重点项目(2023JJD110010)
广西自然科学基金面上项目(2021JJA180078)。
关键词
混合型偏好信息
证据理论
冲突系数
激励因子
多阶段决策
hybrid preference information
evidence theory
conflict coefficient
incentive factor
multi-stage decision making