Progress and Prospects of Artificial Intelligence Applications in Emergency Management
摘要
为系统梳理人工智能在应急管理全生命周期的应用特征、研究局限与发展方向,基于中国知网(CNKI)与 Web of Science(WoS)数据库2015-2025 年相关文献,采用 CiteSpace 文献计量分析与内容分析法,构建灾前预防与准备、灾中响应与处置、灾后恢复与评估三阶段分析框架,归纳人工智能在自然灾害与社会灾害场景中的应用现状.结果表明:人工智能已覆盖应急管理全流程,灾前侧重风险预测预警,灾中聚焦灾情感知与应急调度,灾后多用于损失评估与恢复决策;自然灾害领域研究较为成熟,面向社会灾害的应用广度相对有限;中文文献侧重实践治理应用,英文文献更关注算法性能与模型泛化能力;当前还存在低频灾害样本匮乏、跨部门数据协同不足、长期恢复场景应用薄弱、数据开放与隐私保护难以平衡等现实问题.未来应加强全生命周期智能应急体系建设,推动人工智能与应急管理实践深度融合.
To systematically examine the application characteristics,research limitations,and future development directions of artificial intelligence across the entire lifecycle of emergency management,based on relevant literature from 2015 to 2025 in the China National Knowledge Infrastructure(CNKI)and Web of Science(WoS)databases,this study employs CiteSpace bibliometric analysis and content analysis to construct a three-stage analytical framework comprising pre-disaster prevention and preparedness,during-disaster response and handling,and post-disaster recovery and assessment,and summarizes the current applications of artificial intelligence in natural disaster and social disaster scenarios.The results indicate that artificial intelligence has been applied across the entire emergency management process,focusing on risk prediction and early warning before disasters,disaster perception and emergency dispatch during disasters,and loss assessment and recovery decision-making after disasters;The research in the field of natural disasters is relatively mature,but the application breadth for social disasters is relatively limited.Chinese literature focuses on practical governance applications,whereas English-language litera-ture places greater emphasis on algorithmic performance and model generalization capabilities;Current challenges include a scarcity of low-frequency disaster event samples,inadequate cross-departmental data collaboration,limited application of long-term recovery scenarios,and the difficulty of balancing data openness with privacy protection.In the future,efforts should be made to strengthen the development of a full-lifecycle intelligent emergency management system and promote the deep integration of artificial intelligence into emergency management practices.