模糊认知图与证据理论融合的质量控制决策模型  

Fuzzy cognitive maps and evidence theory based research on quality control decision-making model

作　　者：周娟[1] 余忠华[1] 侯智[1] 

机构地区：[1]浙江大学现代制造工程研究所,杭州310027

出　　处：《系统工程理论与实践》2016年第5期1288-1296,共9页Systems Engineering-Theory & Practice

基　　金：国家自然科学基金(50835008;71071138);武器装备预研基金(9140A19010408ZW04)~~

摘　　要：针对制造过程的质量问题层出不穷而企业常常疲于救火式管理、难以从根本上解决质量问题的现状,结合模糊认知图理论和证据理论,提出一种包含反应式、预防式、预报式和先行式等四种典型的质量控制模式的决策模型.论文首先分析这四种典型的质量控制模式及其决策因素;接着,讨论基于模糊认知图的质量控制决策模型及实施步骤,该模型立足于综合考虑质量问题属性、成本、功能符合性以及组织支持因素等维度,构建基于模糊认知图的决策网络.针对决策网络中多个专家所建立模糊认知图的合成问题,结合证据理论,以专家知识作为证据,节点间的因果强度的可能取值空间构成识别框架,对多个专家知识进行合成.最后,以某轴承企业深沟球轴承沟形误差超标的质量问题为例验证了该方法的可行性.Enterprises are continuously falling into fire management because of quality problems emerg- ing endlessly. To improve the situation, this work proposes a quality control decision model based on fuzzy cognitive map and evidence theory. The model includes four typical quality control modes： reactive, preventive, predictive and proactive. Firstly, the four typical quality control modes and decision factors were analyzed; then, the quality control decision model and implementation steps were presented. The decision strategy was studied from four perspectives including quality properties dimension, cost dimen- sion, flmction accord dimension and organization support dimension. Aiming at solving quality problems scientifically, the decision network was constructed and the most appropriate quality control mode was chosen for actual requirements. In order to construct multi-expert fuzzy cognitive map （FCM）, we use multi-expert＇s knowledge as evidences, the possible value of weight as frame of discernment, and use evidence rule combing to give fusion basic probability assignment. Finally, the proposed method was applied to the bearing groove shape error out of specification limits problems in the bearing enterprises, which proved that the method was effective.

关 键 词：模糊认知图 证据理论 质量控制 决策 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] F273.1[自动化与计算机技术—计算机科学与技术]