突发事件灾害后果推演模型研究

发布时间：2018-10-05 07:13

【摘要】：近年来各类突发事件频繁发生,造成了严重的灾害后果。突发事件是随时间不断向前演变的复杂系统,这种演变使得应急决策主体所面对的突发事件状态不断变化,因此,在信息不全、时间紧迫的情况下,如果能事先知道突发事件随时间发生发展的趋势,也就是对突发事件进行推演,并根据推演结果制定应急行动,以达到提高效率,降低损失,避免灾害扩大化的目的。现有对突发事件推演的研究多针对突发事件情景,但是对情景的定义却很少和决策者的需求联系起来,也没有建立关于突发事件与推演的关联。从应急需求考虑,应急处置的对象是灾害后果,决策者需要对灾害后果的严重程度及其发展趋势有所了解。因此,推演的内涵是灾害后果及其走向趋势。由于突发事件的发生是不可预测的,而且在不同的区域会产生不同的灾害后果。因此,本文将从灾害后果的角度出发,根据灾害后果的共性特征,构建突发事件灾害后果的推演模型。突发事件的灾害后果不仅包括单个突发事件的灾害后果,还包括该事件引发的连锁反应,由于两者发生机理的不同,应分别对其构建推演模型。对于单个突发事件灾害后果推演模型,本文依据灾害系统论,认为灾害后果是由承灾体体现的,单个突发事件灾害后果的推演也就是区域内大量的承灾体之间通过简单的规则构成的复杂动态系统的演化,该特性与元胞自动机模型相符。因此,提出基于元胞自动机模型的单个突发事件灾害后果推演模型。该模型用元胞表示承灾体,根据承灾体是否具有恢复能力以及恢复后是否还会受损提出三种不同类型承灾体的推演规则,并分别对这三种类型的承灾体进行仿真实验。对于突发事件连锁反应推演模型,本文在突发事件关联网络模型的基础上,针对从该网络中很难找出突发事件连锁反应路径的问题,提出了基于Hopfield (?)申经网络的突发事件连锁反应路径推演模型,该模型用神经元代表突发事件,通过建模将突发事件连锁反应路径推演过程映射为Hopfield (?)神经网络的演化过程,通过运行Hopfield网络,推演出初始事件引发的连锁反应事件。最后,推演了地震引发的连锁反应事件,验证了该模型的合理性。本文提出的突发事件灾害后果推演模型,便于决策主体提前了解突发事件各时刻的灾害后果情况,评估灾害损失,预测突发事件可能引发的次生突发事件,并制定行之有效的应急行动。
[Abstract]:In recent years, a variety of emergencies occurred frequently, resulting in serious disaster consequences. Emergency is a complex system that evolves continuously with time. This kind of evolution makes the emergency decision makers face constant changes in the state of emergencies. Therefore, in the case of incomplete information and tight time, If we can know in advance the trend of sudden events with time, that is to say, we can extrapolate the sudden events, and make emergency action according to the result of deduction, in order to improve efficiency, reduce losses and avoid the expansion of disasters. The current research on emergency inference mostly focuses on emergency scenarios, but the definition of scenarios is rarely related to the needs of decision makers, and there is no correlation between emergencies and extrapolation. Considering the emergency demand, the object of emergency treatment is the disaster consequence, and the decision-makers need to understand the severity of the disaster consequence and its development trend. Therefore, the implication of inference is disaster consequence and its trend. The occurrence of unexpected events is unpredictable and has different disaster consequences in different regions. Therefore, from the point of view of disaster consequences and according to the common characteristics of disaster consequences, this paper will construct the inference model of disaster consequences of unexpected events. The disaster consequences of a sudden event not only include the disaster consequences of a single emergency, but also include the chain reaction caused by the event. Due to the difference of their occurrence mechanism, the inference model should be constructed separately. According to the disaster system theory, this paper considers that the disaster consequence is reflected by the disaster bearing body. The evolution of complex dynamic systems formed by simple rules between a large number of disaster bearing bodies in a region is the deduction of the disaster consequences of a single unexpected event, which is consistent with the cellular automata model. Therefore, a model based on cellular automata (CA) is proposed to predict the disaster consequences of a single emergency. In this model, the disaster bearing body is represented by cells. According to whether the disaster bearing body has the ability to recover and whether the disaster bearing body will be damaged after recovery, the inference rules of three different types of disaster bearing bodies are proposed, and the simulation experiments are carried out on the three types of disaster bearing bodies. For the emergency chain reaction deduction model, based on the emergency correlation network model, aiming at the problem that it is difficult to find out the chain reaction path of the emergency from the network, this paper proposes a method based on Hopfield (?) In this model, neurons are used to represent the sudden events, and by modeling, the inference process of the chain reaction path of the sudden events is mapped to Hopfield (?) The evolution process of neural network, by running the Hopfield network, deduces the chain reaction event caused by the initial event. Finally, the chain reaction caused by earthquake is deduced and the rationality of the model is verified. The model proposed in this paper is convenient for the decision makers to understand the disaster consequences at each moment in advance, to evaluate the disaster losses, to predict the secondary emergencies that may be caused by the emergencies. And formulate effective emergency action.
【学位授予单位】：大连理工大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：X913

【参考文献】