基于故障树的电网灾害应对失效问题识别

发布时间：2018-08-16 17:57

【摘要】：自然灾害、人为灾害以及电网体系内的偶然故障都可能造成电网设施损坏，进而扩散至其他基础设施部门，负面后果被成倍放大，爆发关键基础设施相继故障的“连锁效应”，给社会经济体系带来巨大的损失。电网灾害应对任务体系应支持掌握足够的灾情数据信息，以支持应急决策分析，从而识别应对风险，在有限的应对能力下控制灾情避免其扩大。灾害应对时可能面对不确定的、模糊的、随机的复杂情境，如何发现关键应对失效是一项难度较大的课题。本文就电网灾害应对体系中的失效事件作为研究对象，以故障树方法为逻辑主线提出一套对应的理论与方法框架。首先，界定了电网灾害应对情景中的致灾因子、承灾体、孕灾环境与抗灾体等基本要素，特别是提出情景应对失效驱动因子，用来体现应对失效导致的灾情演变。给出灾害应对失效的广义概念使其与传统失效理论狭义定义区别开来，并从应对失效的来源角度，分灾害情景意外、应对准备不适与应对响应不当三个方面探究其溯源。基于以上基础分析，刻画了应对系统能力丧失与能力衰退的表征，分析这两类失效发生在初始与中间阶段对电网灾情救援的影响。最后，基于三阶段任务的失效子系统概念模型得以建立。其次，构建了电网应急的基本故障树模型，分为应对情景认知、应对资源管理、临时供电任务与抢修执行任务四类模型，，用以寻求应对响应体系中的非期望顶事件。面向特定情景，给出了基于Delphi法的情景故障树生成流程。为便于数据共享与传输，使用XML文档存储案例中的情景故障树。然后，在定性方面，利用二元决策图转换方法结合相邻节点优先法得到最小割集；在定量方面，为解决不确定环境下的概率评估问题，结合D-S证据理论与梯形模糊数的语义评价法生成失效事件概率，根据获得的专家评估概率计算顶事件的发生率以及各底事件的敏感度与重要度度量，最后通过灰色关联度识别出关键失效事件集合。最后，以深圳电网为例，验证了本文所提出的模型与方法。
[Abstract]:Natural disasters, man-made disasters and accidental failures within the grid system can all cause damage to power grid facilities, which can spread to other infrastructure sectors, with negative consequences multiplied and "cascading effects" of successive failures of critical infrastructure, Bring huge losses to the social and economic system. The grid disaster response task system should support to grasp enough disaster data information to support emergency decision analysis, so as to identify the response risk, and to control the disaster situation to avoid its expansion under the limited response ability. Disaster response may be faced with uncertain, fuzzy, random complex situation, how to find the key response failure is a difficult task. In this paper, the failure events in the disaster response system of power grid are studied, and a set of corresponding theory and method framework is put forward, which takes the fault tree method as the logical main line. First of all, the basic factors of power grid disaster response scenarios, such as disaster response factors, disaster tolerance, disaster environment and disaster resistance, are defined. In particular, a scenario response failure driving factor is proposed to reflect the disaster situation evolution caused by response failure. The generalized concept of disaster response failure is given to distinguish it from the narrow definition of traditional failure theory. From the point of view of the source of response failure, this paper probes into its source from three aspects: accident, malaise and improper response. Based on the above basic analysis, this paper describes the characteristics of capacity loss and capability decline in response system, and analyzes the effects of these two kinds of failures in the initial and intermediate stages on power grid disaster rescue. Finally, the concept model of failure subsystem based on three-stage task is established. Secondly, the basic fault tree model of power grid emergency response is constructed, which is divided into four types: response scenario cognition, response to resource management, temporary power supply task and emergency repair execution task, in order to seek to deal with unexpected top events in response system. A scenario fault tree generation process based on Delphi method is presented for specific scenarios. To facilitate data sharing and transmission, XML documents are used to store situational fault trees in cases. Then, in the qualitative aspect, the minimum cut set is obtained by using the binary decision graph transformation method combined with the adjacent node priority method, and in the quantitative aspect, in order to solve the problem of probability evaluation in uncertain environment, Combined with D-S evidence theory and the semantic evaluation method of trapezoid fuzzy number, the probability of failure event is generated. According to the expert evaluation probability obtained, the incidence rate of top event and the sensitivity and importance of each bottom event are calculated. Finally, the key failure event set is identified by grey correlation degree. Finally, taking Shenzhen Power Grid as an example, the model and method proposed in this paper are verified.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TM73

【参考文献】