基于改进云模型的语言偏好信息多属性大群体决策方法

摘要

针对具有语言评价标度、属性权重未知、专家权重未知的多属性大群体决策问题,提出了基于改进云模型的语言偏好多属性大群体决策方法.该方法首先提出了一种改进的云生成方法,将语言值转化成正态云模型;然后在偏好信息已知的情况下,用熵权法确定属性权重;在此基础上,提出了一种基于泛概念树的云模型聚类的方法以确定成员权重;最后通过一个算例验证了方法的合理性和可行性.%This study considers that the evaluation values of experts are given in the form of linguistic values which are closer to human perception.The scale of decision group tends to be large and complicated.A multi-attribute large group decision-making method with linguistic preference information based on improved cloud model is proposed to solve multi-attribute large group decision-making problems.To solve the problem,linguistic evaluation values are given,but the attribute weights and decision-makers' weights are unknown.Firstly,an effective cloud generation method needs to be discovered in order to solve decision problems with linguistic assessment information based on cloud methods.In the first part,this paper analyzes the limitations of cloud generation methods in the existing literature,and proposes a new method for transformation between linguistic variables and clouds based on golden section.The new method can overcome limitations that the expectations of cloud models generated by traditional methods probably exceed the scope of the universe or different expectations of clouds.The principle and steps are introduced in detail.The calculation of the numerical characteristics is shown by taking the generation method of seven clouds as an example in this paper.The second part discusses the clustering of the group preference as one of the premises of large group decision making on the basis of the climbing of pan-concept-tree.Large group preference clustering based on pan-concept-tree is proposed.The number of clusters is determined by decision makers before clustering.Two preference cloud models with minimal expectation difference at the same level are constantly promoted by Soft-Or until an ideal number of cluttering satisfying a high level is reached.In contrast,Soft-Or should be conducted only once at each level to promote two cloud models.After linguistic information in the decision matrix is transformed into normal cloud model by using the proposed improved cloud generation method,attributes' weights are determined based on entropy weight.Experts' weights are determined based on clustering results.The preference cloud model of each altemative is aggregated to develop a synthetic cloud model,followed by ordering alternatives and choosing feasible options according to the statistical results of cloud drops.In the last part,a numerical example is presented to verify the feasibility and rationality of the method.The clustering and decision results are analyzed and discussed,including the relation between attributes' weights and decision result,the cluttering number K and decision result,and the difference between new method and traditional method.These results validate the feasibility and reliability of the new method proposed in this paper.In this paper,the improved cloud model is applied to the process of large group decision,and a decision method with linguistic preference based on improved cloud model.The improved model introduces examples and solves multi-attribute large group decision method with linguistic values.The application of cloud model in large group decision can better quantify the uncertainty of human perception.Besides,the cloud model preference clustering method proposed is more flexible,and realizes the soft partition of cloud model.