【摘要】 【目的】随着信息技术和人工智能的快速发展，基于知识图谱和计算机场景仿真模拟对应急预案知识的表达研究方兴未艾。针对当前城市内涝应急预案无法与内涝全周期场景模拟动态匹配且难以满足不同用户动态应急需求的问题。【方法】本文提出一种基于预案知识映射的城市内涝应急场景模拟方法。通过构建涵盖灾害各阶段的城市内涝应急预案知识图谱，探究应急知识与场景要素的交互映射方法，在此基础上进行城市内涝应急预案场景的多类型用户按需设计与模拟。【结果】以北京市丰台区“23·7”特大暴雨内涝事件为案例，利用LDA模型从16万条社交媒体数据中挖掘出6类高频舆情主题，构建了包含12类基础地理场景与39类应急管理场景的知识-要素映射表。基于该方法研发了原型系统，并通过115份有效问卷进行量化验证。结果显示，系统在交互便捷性、运行效率及功能完备性3个维度的评分均值分别为4.06、4.09和4.07（满分5分），均达到良好水平。【结论】本文研究完成了从应急预案知识建模与评估、场景要素解析与实体映射到内涝全周期场景的多用户按需模拟的实验验证，定量评估结果表明该方法能有效满足多类型用户在城市内涝全周期阶段的动态应急决策需求。更多还原

【Abstract】 [Objectives] With the rapid development of information technology and artificial intelligence, research on the expression of emergency plan knowledge based on knowledge graphs and computer scenario simulation is gaining momentum. Addressing the issues that current urban waterlogging emergency plans cannot dynamically match full-cycle scenario simulations and fail to meet the dynamic emergency needs of diverse users, [Methods] this paper proposes a scenario simulation method for urban waterlogging emergencies based on plan knowledge mapping. By constructing a knowledge graph of urban waterlogging emergency plans covering all disaster stages, the interactive mapping method between emergency knowledge and scenario elements is explored. On this basis, the on-demand design and simulation of urban waterlogging emergency scenarios for multiple types of users are carried out. [Results] Taking the “23·7” extreme rainstorm waterlogging incident in Fengtai District, Beijing as a case study, the LDA model was utilized to mine 6 categories of high-frequency public opinion topics from over 160 000 social media data entries, and a knowledge-element mapping table containing 12 types of fundamental geographic scenarios and 39 types of emergency management scenarios was constructed. Based on this method, a prototype system was developed and quantitatively validated through 115 valid questionnaires. The results show that the average scores of the system in the three dimensions of interaction convenience, operational efficiency, and functional completeness are 4.06, 4.09, and 4.07(out of 5), all reaching a satisfactory level. [Conclusions] This study completes the full-process experimental verification, ranging from knowledge modeling and evaluation of emergency plans, scenario element analysis and entity mapping, to the on-demand simulation of full-cycle waterlogging scenarios for multiple types of users. The quantitative evaluation demonstrates that this method can effectively satisfy the dynamic emergency decision-making requirements of multiple types of users throughout the full cycle of urban waterlogging.更多还原