基于知识元的突发事件系统结构模型及演化研究

发布时间：2018-06-04 10:41

本文选题：知识元 + 突发事件系统　；参考：《大连理工大学》2013年博士论文

【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。
[Abstract]:With the economic growth and social development, the environment for human survival is deteriorating, natural disasters, accident disasters and all kinds of sudden events caused by complex disasters are frequently occurring, which seriously threaten the harmonious development of human society. There are various types of sudden events, and there are great differences in personality characteristics displayed by different types of emergencies. There are different situations in the occurrence and development of emergencies, the relationship between emergencies is different, and the same sudden event will have different evolutionary paths. In the process of emergency response, the emergency manager faces a large number of, multi source, heterogeneous information, and the development trend and performance of the sudden events need to be identified and the overall level recognizes the development situation and performance of the unexpected events. In order to make the emergency managers understand the hierarchical structure and evolution path of the emergency system from the common point of view, the emergency managers can understand the evolution process of the emergencies and provide the scientific and effective coping strategies. Theoretical guidance.
From the different aspects of natural disasters, accident disasters, public health events, and social security events, scholars at home and abroad study the evolution law of sudden events in specific fields. There are few unexpected events and situations as a complex system. From the perspective of the system, the common structural features and common evolution laws of sudden event systems are studied. From the perspective of knowledge, the knowledge element is independent of the specific domain knowledge. It makes it possible to use knowledge element to describe the common attribute characteristics of the system elements of different types of emergencies, and then construct the general structure model of the emergency system. The knowledge management model is used to explore the common structure characteristics and evolution rules of all kinds of emergency system, and a unified description of the scientific symbols is the key problem in emergency management.
Under the above background, by analyzing the composition of the emergency system, this paper constructs the event knowledge element and the knowledge element model in the emergency system, analyzes the common structure of the emergency system on the basis of the knowledge meta model, defines the concept of the system entropy of the emergency, reflects the degree of disorder within the system, and the risk entropy and the decision making. Based on the knowledge metamodel, in the framework of the common structure of the emergency system, the entropy change process of the sudden event system is studied and the evolution law of the system is identified. The risk entropy is the basis of the entropy evolution of the emergency system, without the risk entropy, the emergency managers need not inject the decision entropy into the system, and put forward the unexpected events based on the event knowledge element. The prediction method of risk entropy is studied as follows:
(1) the sudden event and the situation are regarded as a complex system, and the composition of the emergency system is analyzed. The event knowledge element and the disaster body knowledge element, which describe the sudden events and the objective things in the emergency system, are constructed. The general structure of the emergency system is analyzed, the various levels are formally described, and the sudden events based on the knowledge element are established. The common structure model of part system. The time domain and the space domain of the burst event evolution are divided. The common structure features of the emergency system are analyzed, the entropy phenomenon in the emergency system is expounded, and the essence, the essence of the evolution and the objective basis of the evolution are revealed.
(2) the concept of system entropy, which is used to characterize the degree of disorder within the system, is proposed. It is composed of the entropy of risk and the decision entropy provided by the set of environmental input attributes of the entity object of the event knowledge element in the system. The entropy evolution process of the system is analyzed, and the entropy change process of the system is summed up as the product of entropy. The abrupt period of entropy and the decline period of entropy. By using the catastrophe theory, based on the knowledge element, the catastrophe model of the system entropy of the sudden event is established, the catastrophe condition of the system entropy is obtained and the parameter region of the entropy stable state of the system is determined. The evolution of the system entropy of the sudden event has two kinds of evolution form, including the gradual change and the sudden change, including the lag mutation and the advance mutation. Four evolutionary paths for synchronization and asynchronous instability.
(3) aiming at the problem that there are many influential factors in the process of sudden event evolution, the problem that the event risk entropy can not be described quantificationally in real-time observation sample data and high dimension, a method of risk entropy prediction based on knowledge element is proposed. The projection pursuit method is used to reduce the multi factor observation data of the unexpected events and obtain the event knowledge. The one dimension projection characteristic value of the high dimensional input attribute set of the meta entity object reflects the data characteristics of the multiple factors by single factor, and the probability of the occurrence probability of different risk grade emergencies is obtained by using the information diffusion theory to spread the single factor projection characteristic value of the input attribute set to the control point of the risk index domain of the output attribute. Entropy theory is used to calculate the risk entropy of contingencies and predict the possibility of unexpected events.
(4) taking the rain and snow freezing event system in southern China in 2008 as an example, taking Changsha as the research area, this paper makes an example test on the composition method of the freezing rain and snow event system and the prediction process of the risk entropy of the snow and snow events. The example analysis results show that the general structure model of the sudden event system based on the knowledge element proposed by this study is easy to describe. The law of the occurrence of sudden events and the evolution law in different levels. The risk entropy prediction method based on the knowledge element can predict the risk entropy according to the time series of the observed samples. The change trend of the risk entropy is consistent with the actual situation and conforms to the entropy change law.
Based on the knowledge element, this paper systematically analyzes the common structure of the emergency system, discusses the evolution law of the system entropy, and puts forward the prediction method of the risk entropy of the emergency, and provides the reference and reference for the theoretical research and practice of emergency emergency management.
【学位授予单位】：大连理工大学
【学位级别】：博士
【学位授予年份】：2013
【分类号】：N941.4;X913

【参考文献】