【作者】 李磊；

【导师】 王富章；

【作者基本信息】 中国铁道科学研究院 ， 交通信息工程及控制， 2013， 博士

【摘要】 当前对于铁路突发事件应急决策，主要依据决策者的自身经验，然而由于决策者的知识局限性，做出的决策可能会存在一定的不足，另外，决策者的自身压力可能较大，正常的决策能力会受到影响。因此，如何高效、及时地对铁路突发事件进行有效决策，已成为迫切需要解决的问题。本文综合运用模糊数学、评估理论、多属性群决策理论及计算机技术等多学科知识，采用分析与综合相结合、理论研究与数值试验相结合的方法，针对铁路突发事件应急决策的若干问题进行了研究。本研究的具体内容如下：（1）研究了铁路突发事件及铁路应急决策相关理论，包括铁路突发事件定义、分类分级、特点、基本应急属性，以及铁路应急决策定义、特点、流程和方法。（2）提出案例推理（CBR）应用到铁路应急决策中，①针对铁路突发事件案例表示，提出采用四元组（主题域、问题域、解决方案域、及效果域）描述案例，并提出该四元组中包含的属性；②提出铁路突发事件案例以及铁路应急预案框架法表示方法；③案例属性值上考虑了数值型、有序枚举型、无序枚举型、区间数型及模糊数型五种形式属性值，提出五种形式属性值的案例相似度计算方法；④结合案例属性的特点，采用层次分析法确定铁路突发事件案例属性权重，研究了案例属性局部相似度计算方法、案例总体相似度计算方法，以及基于结构相似度算法，研究案例属性缺失条件下的案例总体相似度计算方法；⑤提出基于案例推理的铁路突发事件应急决策流程。（3）在铁路应急预案评估方面，考虑评估值的不确定性和模糊性以及评估者的偏好和知识不完备性，本文提出四种铁路应急预案评估方法：①改进VIKOR方法评估铁路应急预案，针对的评估指标值为语言值，指标权重的求解运用基于熵的主客观赋权法，另外对铁路预案评估指标提出改进；②针对评估指标值为混合值，提出基于二元语义的混合多属性群决策评估方法，提出模型及求解算法；③针对指标评估值为区间语言值，定义区间二元语义混合平均(IVTHA)算子及区间二元语义混合几何(IVTHG)新算子，提出基于区间二元语义混合平均(IVTHA)算子的决策评估方法，给出算法步骤；④针对评估指标具有不同优先级，评估值为语言值，提出二元语义优先加权平均（2TLPWA）算子、二元语义优先几何平均（2TLPWG）算子、二元语义优先有序加权平均（2TLPOWA）算子、二元语义优先有序加权几何（2TLPOWG）算子，提出基于二元语义优先有序平均算子（2TPOWA）算子的与预案评估方法，并给出算法步骤。（4）在决策支持系统的理论基础上，给出铁路突发事件应急决策系统的定义、目标，构建了铁路突发事件应急决策系统的框架，提出铁路突发事件应急决策系统的总体设计，并对功能模块做了较详细设计，另外对铁路突发事件应急决策系统“多库”系统中的知识库、模型库、方法库、基础信息库系统做出设计。本文的研究对改进铁路突发事件应急决策方法，提高铁路突发事件应急决策水平具有一定的现实意义，同时对案例推理人工智能技术、多属性群决策评估理论以及决策支持系统的结合研究提供一定的理论基础。更多还原

【Abstract】 Railway emergency decision-making is mainly based on their own experience ofdecision-makers. However, due to the limitations of knowledge of the decision-makers andtheir own pressure, the decision may exist some deficiencies. Therefore, how to makeeffective decision efficiently and timely has become an urgent.In this thesis, some issues around the railway emergency decision-making are researchedby the method of combining analysis and synthesis, theoretical research and numericalexperiments, and using multi-disciplinary knowledges, such as fuzzy mathematics,assessment theory, multi-attribute group decision theory and computer technology. The maincontents of this reasearch are as follows:(1) The theories of railway emergencies and railway emergency decision are discussed.The theories include definition, classification, grading, characteristics, and basic emergencyproperty of railway emergency, as well as the definitions, characteristics, processes, methodsof railway emergency decision.(2) Case-based reasoning (CBR) is applied to the railway emergency decision. Fourtuples of subject domain, problem domain, solution domain, and the effect domain are used todescribe railway emergency case. Attributes of the four tuples are presented. Framework ofrailway emergency is represented. Different attribute values including numeric, orderlyenumeration type, disorderly enumeration type, type of interval numbers,and fuzzy numbersare studied. A method of five attributes value type is proposed to calculate their similarity.Attribute weights of railway emergency are determined combing with attribute characteristicsby the analytic hierarchy process (AHP) method. Calculation method of local similarity andglobal similarity of the current case and the source case is studied. Calculation method ofglobal similarity is studied for values of attributes missing conditions, based on structuresimilarity algorithm.(3) Due to uncertainty and ambiguity of assessment value and knowledge incompletenessand preferences of evaluators in the aspects of railway emergency plans evaluation, four kindsof evaluation methods for railway emergency plan evaluation are presented. Aiming atlinguistic evaluation value of attribute, VIKOR method is improved for evaluting railwayemergency plan. Weights are solved by the method of subjective and objective weight basedon entropy. In addition, evaluation indexes of railway emergency plan is improved. In view of the evaluation indexes value is hybrid, a method based on2-tuple linguistic is presented forevaluating railway emergency plan with weight information known, and algorithm are given.Aiming at evaluation value of interval linguistic form, interval-valued2-tuple linguistichybrid average (IVTHA) operator and interval-valued2-tuple linguistic hybridgeometric(IVTHG) operator are defined. The method of multi-attribute group decisionmaking based on IVTHA operator is proposed for evaluating railway emergency plan, inaddition algorithm and steps are suggested. According to the evaluation indexes with differentpriorities and evaluation value of interval linguistic form,2-tuple prioritized weightedaverage(2TLPWA) operator,2-tuple prioritized weighted geometric(2TLPWG) operator,2-tuple prioritized ordered weighted average (2TLPOWA) operator, and2-tuple prioritizedordered weighted average(2TLPOWG) operator are suggested. The method of multi-attributegroup decision making based on2TLPOWA operator is presented, and algorithm is proposed.(4) Based on the theory of decision support system, the definition of the railwayemergency decision-making system, and the goal of the system are proposed. The frameworkof railway emergency decision system is built. The overall design of the railway emergencydecision system is presented. The detailed design of functional modules is made. In additionto knowledge base, model base, method base, and basic information database system aredesigned for the "library" system of the railway emergency decision system.In short, improving railway emergency decision-making methods and the level of railwayemergency decision-making has a certain practical significance. Combination study of CBRof artificial intelligence technology, the theory of multi-attribute group decision makingassessment and the technique of decision support system provides certain theoretical basis forrailway emergency decision-making.更多还原