基于知识元的非常规突发洪水事件演化风险研究

为解决非常规突发洪水事件风险分析中所存在的多因素高维数据与小样本数据信息不完备问题,将非常规突发洪水事件系统进行粒度分解,在知识元层建立基于投影寻踪及信息扩散的非常规突发洪水事件演化风险分析模型.利用知识元理论,将事件间的演化风险转变为事件所处情境内承灾载体状态属性变化引发输出属性变化的风险.根据承灾载体知识元内部状态属性阀值建立后续事件演化规则,利用投影寻踪方法与信息扩散理论,将监测样本所包含风险信息扩散到输出属性的风险指标论域控制点,计算事件演化概率分布,确定非常规突发洪水事件演化风险.以桃曲坡水库为例,在上游发生暴雨的情景下,计算水库水位上升向漫坝事件演化风险.实例分析结果表明,该演化风险分析模型能够根据风险等级标准和少量监测数据,对非常规突发事件风险概率分布进行动态定量分析与评估,帮助应急决策者实时掌握事件发展状况,及时做出应急响应,减少次生事件所造成的损失.

Research on unconventional flood emergency evolution risk analysis based on knowledge element

In order to solve the problem about multi factors, high dimensional data, small samples and incomplete information when analyzing unconventional flood emergency evolution risk, in the knowledge element layer by granularity decomposition, model of unconventional flood emergency evolution risk analysis is built based on projection pursuit and information diffusion theory. The evolution risk between emergency events is transferred to the risk that the change of status attributes of hazard bearing body in current situation result in the mutation of output attributes based on knowledge element theory. The rules of emergency events evolution are established based on status attributes threshold of hazard bearing body, risk information in the observed data is diffused into the control point of the risk index universe on the output attributes by projection pursuit and information diffusion theory, then the emergency events evolution probability, and the unconventional flood emergency evolution risk is calculated. Taking Taoqupo reservoir as an example, the evolution risk is analyzed between water level rising and dam overtopping at the scenario of serious rainstorm in the upstream. The results show that the model of events evolution risk analysis could sequential and quantitative analyze and evaluate risk probability distribution according to the risk grade level and few monitoring data, and help the emergency management decision makers grasp the development of the event and emergency response in real time, reduce the losses caused by secondary events.