【作者】 于海峰；

【导师】 王延章；

【作者基本信息】 大连理工大学 ， 管理科学与工程， 2013， 博士

【摘要】 随着经济增长和社会发展,人类赖以生存的环境不断恶化,自然灾害、事故灾难以及复合灾害引发的各类突发事件频繁发生,严重地威胁着人类社会和谐发展。突发事件类别多样,不同类型突发事件表现出来的个性特征存在着很大差异。突发事件发生发展的情境不同,突发事件之间的关联关系不同,同一突发事件将会出现不同的演化路径。在突发事件的应对过程中,应急管理者面对大量的、多源的、异构的信息,需要站在综合、总体的层面识别出突发事件的发展态势和演化方向,并在支持信息不完备的情况下快速、科学地进行决策。通过提炼突发事件系统的一般性规律,使应急管理者从共性角度了解突发事件系统的层次结构和演化路径,认知突发事件发生演化的过程,为制定科学有效的应对策略提供理论指导。国内外学者从自然灾害、事故灾难、公共卫生事件、社会安全事件等不同方面入手,研究特定领域内突发事件的演化规律。鲜有将突发事件与所处情境视为一个复杂系统,从系统视角研究突发事件系统共性结构特征和共性演化规律。在知识的视角下,知识元与特定领域知识无关,使得利用知识元描述不同类型突发事件系统要素的共性属性特征,进而构建突发事件系统共性结构模型成为可能。如何对各种不同属性特征的突发事件系统进行分析,如何利用人们已有知识和知识管理模型,发掘各类突发事件系统的共性结构特征和演化规律,并以科学的符号进行统一描述成为突发事件应急管理中的关键问题。本研究在上述背景下,通过分析突发事件系统的构成,构建突发事件系统中事件知识元和承灾体知识元模型,在知识元模型的基础上剖析突发事件系统的共性结构。界定突发事件系统熵的概念,反映系统内部无序程度,由风险熵和决策熵组成。以知识元模型为基础,在突发事件系统共性结构的框架内,研究突发事件系统熵变过程,辨识系统的演化规律。风险熵是突发事件系统熵演化基础,没有风险熵,应急管理者无需向系统注入决策熵,提出基于事件知识元的突发事件风险熵的预测方法。具体研究如下：(1)将突发事件与所处情境视为一个复杂系统,分析突发事件系统的组成,构建了突发事件系统中描述突发事件和客观事物的事件知识元和承灾体知识元。解析突发事件系统的共性结构,形式化描述各个层次,建立了基于知识元的突发事件系统共性结构模型。切分突发事件演化的时间域和情境的空间域,分析突发事件系统的共性结构特征,阐述突发事件系统内部的熵现象,揭示突发事件发生本质、演化实质以及演化的客观基础。(2)提出了用以表征系统内部无序程度的突发事件系统熵的概念,由系统中事件知识元实体对象的环境输入属性集提供的风险熵和应急输入属性集提供的决策熵构成。剖析了突发事件系统熵的演化过程,将系统熵变历程归纳为熵的积累期、熵的突变期和熵的衰退期。运用突变论,基于知识元建立了突发事件系统熵的突变模型,获得系统熵的突变条件,确定系统熵稳定态的参数区域。揭示了突发事件系统熵的演化具有渐变和突变两种演化形态,包含滞后突变、超前突变、同步渐稳和异步失稳四种演化路径。(3)针对突发事件演化过程中影响因素多、实时观测样本数据少且维数高导致的无法定量描述事件风险熵的难题,提出了基于知识元的突发事件风险熵预测方法。采用投影寻踪方法对突发事件的多因素观测数据进行降维,获得事件知识元实体对象高维输入属性集的一维投影特征值,以此单因素反映多因素的数据特征；利用信息扩散理论将输入属性集的单因素投影特征值扩散到输出属性的风险指标论域的控制点上,获得不同风险等级突发事件的发生概率。借鉴信息熵理论,给出突发事件风险熵的计算方法,预测突发事件发生的可能性。(4)以2008年我国南方雨雪冰冻事件系统为例,以长沙市为研究区域,对雨雪冰冻事件系统的构成方法和雨雪冰冻事件风险熵的预测过程进行实例检验。实例分析结果表明本研究提出的基于知识元的突发事件系统共性结构模型,便于描述突发事件发生规律和不同层次内的演化规律。基于知识元的突发事件风险熵预测方法能够根据观测样本的时间序列对风险熵进行预测,风险熵的变化趋势与实际情况相吻合且符合熵变规律。本研究以知识元为基础系统地剖析了突发事件系统所具有的共性结构,论述了系统熵的演化规律,提出了突发事件风险熵的预测方法,为突发事件应急管理的理论研究与实践工作提供参考和借鉴。更多还原

【Abstract】 Witn economic growth and social development, human beings encounter increasing degradation in the environment for survival. Various emergencies derived from natural and accidental disasters, along with those combined, have had severe threats on human society and its harmonious development. Given the variety of the emergencies, they present great differences in their respective features. With different situation and association in between, the same event may present different evolving paths. In the course of answering the emergency, the managers may be faced with plenty of multi-sourced heterogeneous information, and have to identify the development trend and evolving direction in a comprehensive and integral perspective, followed by a rapid and scientific decision-making based on incomplete supporting information. It is helpful for the emergency managers to understand the hierarchical constitution and evolving laws of emergencies by summarizing their general rules, thus is a better cognition of their evolving process and a response in time.Researchers at home and abroad have studied the evolving laws of emergencies in certain categories, ranging from natural disasters to accidents, public health incidents, social security events, to name a few, but few regard the emergency and its circumstances as a complex system to look into their systematically constitutional features and evolving laws in generality. In the perspective of knowledge, the knowledge source has nothing to do with the knowledge in specific field, which makes it possible to constitute the general structure model of the emergency system by means of describing the general features of various elements in the system. How to analyze emergency systems with different features? How to make use of the existing knowledge and its management models to find out the general features and evolving laws of various emergencies, and make them expressed in a scientific and unified way? It is essential for such questions to be answered in the management of emergencies.Under the above background, this thesis is to establish the modal of knowledge elements in events and hazard bearing body of the emergency system by the conceptualization of emergency and the analysis of its system. Based on the modal of knowledge element, the changing processes of the general structure and the system entropy of the emergency system are to be analyzed, followed by the identification of the general law of the system evolution. According to the knowledge element of the event, the prediction method for risk entropy of the emergency is to be offered. The details are as follows:(1) The concept of the emergency is defined from the source of its occurrence by regarding emergencies and their circumstances as a complicated system, followed by the analysis of the system’s composition and the establishment of its event knowledge element and hazard bearing body knowledge element. The general structure of the emergency system is analyzed, with format description of various levels and the setting of the general structure modal of the emergency system based on knowledge element. The temporal and spatial domains of the emergency evolution is segmented, thus is the analysis of its characters in dissipative structure and the disclosure of its essence in occurrence, evolution and objective foundation.(2) The entropy phenomenon inside the emergency system is explained in details by putting forward the concept of emergency system entropy to reveal the disorder degree inside the system, which is composed of the positive and negative entropies provided respectively by environmental input feature set and emergency response output feature set of the event knowledge element entity in the system. Following the analysis of the evolving process of the system entropy is the induction of the entropy’s accumulative, mutation and declining periods from the system entropy evolution. With the application of catastrophe theory, the mutation modal of the emergency system entropy is established based on knowledge element. After getting the changing conditions’of the system entropy, the thesis confirms the parameter domain of the system entropy’s stable status. It is revealed that emergency system entropy evolves in two forms, namely, gradual and sudden changes, with four paths of lagging mutation, leading mutation, synchronous gradual stabilization and asynchronous instability.(3) Given such challenges as many influencing factors in the course of emergency evolution, and the difficulty in quantitative description of the risk entropy derived from few on-site observation data and multiple dimensions, this thesis puts forward the emergency risk entropy prediction methodology based on knowledge element. By means of projection pursuit, the multi-element observation and measurement data is de-dimensioned, thus is the acquirement of the projected characteristic value from the multi-dimension input feature set of the subject in the event knowledge element, based on which data characteristics of the multi-elements are reflected by single elements. By means of information diffusion theory, the single element’s projected characteristic value of the input feature set is spread to the control point of output feature’s risk index discourse domain, thus is the acquirement of emergency occurrence probability at different risk levels. By means of information entropy theory, emergency risk entropy calculating method is provided, thus is the prediction of emergency occurrence possibility.(4) With reference to the sudden freezing rain and snow event system (SURSES) of China in2008, Changsha City in Hunan Province of China is chosen as the case study area to test the formation method of the SURSES and the risk entropy prediction process of sudden freezing rain and snow events. The case analysis results present that the general structure model of emergency system based on knowledge element put forward by this study is more convenient to be used in describing emergency occurrence law and the evolving rules at different levels. The risk entropy prediction method based on knowledge element can predict the risk entropy according to the time series of the sample. The changing trend of the risk entropy coincides with the rules of the entropy change.Based on knowledge element, this study systematically analyzes the general structure of emergency system, discusses the evolving rules of the system entropy, puts forward the prediction methodology of emergency risk entropy, and provides theoretical and practical reference for emergency response management.更多还原