Dynamic evolutionary pathway analysis of urban rail transit flood risks and intelligent decision support based on knowledge graphs

With the intensification of global climate change, rainstorm disasters have become increasingly frequent and catastrophic. Urban rail transit (URT) systems, which are primarily constructed underground, possess structural features that make them particularly vulnerable to severe impacts during heavy rainfall events. Such disasters can result in significant casualties and substantial losses. Meanwhile, extensive domain-specific knowledge has been accumulated from historical disaster events. Effectively extracting and utilizing such knowledge is essential for improving disaster risk identification and enhancing emergency management practice. To address these challenges, this study proposes a method for analyzing risk evolution mechanisms by integrating Knowledge Graph and Natural Language Processing (NLP) technologies. The knowledge graph enables structured knowledge representation and facilitates effective knowledge reuse. Building on this, a knowledge-driven decision support model is established by combining the language understanding capability of NLP with the inferential capacity of knowledge graphs. Case studies of representative examples are conducted to validate the effectiveness of the proposed method in this study. The findings show that structuring knowledge in the form of a graph network offers significant advantages for the intelligent analysis of disaster risk evolution. On one hand, a large amount of multi-source, heterogeneous knowledge related to URT flood risks is systematically structured and represented, thereby enhancing the efficiency of knowledge utilization by decision-makers. On the other hand, integrating NLP with knowledge graph–based risk network analysis enables the accurate identification of potential risk paths, providing valuable insights and a foundation for disaster prevention and mitigation decision-making.