中国安全科学学报 ›› 2021, Vol. 31 ›› Issue (9): 1-7.doi: 10.16265/j.cnki.issn1003-3033.2021.09.001

• 安全科学理论与安全系统科学 •    下一篇

基于深度长短记忆模型的民航安保事件分析*

冯文刚1,2 副教授   

  1. 1 中国人民公安大学 国家安全学院,北京 100038;
    2 中国人民公安大学 民航安保研究中心,北京 100038
  • 收稿日期:2021-06-24 修回日期:2021-08-15 出版日期:2021-09-28 发布日期:2022-03-28
  • 作者简介:冯文刚 (1982—),男,浙江杭州人,博士,副教授,博士生导师,主要从事民航安保、公安情报分析、模式识别等方面的研究。E-mail:wengang.feng@qq.com。
  • 基金资助:
    国家重点研发计划项目(2017YFC0822502);中央高校基本科研业务费资助项目(2021JKF220);中国人民公安大学公共安全行为科学实验室开放课题资助(2020SYS09)。

Research on civil aviation security event analysis based on deep LSTM model

FENG Wengang1,2   

  1. 1 School of State Security, People's Public Security University of China, Beijing 100038, China;
    2 Research Centre for Civil Aviation Security, People's Public Security University of China, Beijing 100038, China
  • Received:2021-06-24 Revised:2021-08-15 Online:2021-09-28 Published:2022-03-28

摘要: 为辅助公安民警分析民航安保事件,采用深度长短记忆(LSTM)模型,研究民航安保事件行为主体识别问题。通过搭建民航安保事件数据库,对民航安保事件概念信息进行多模态信息表示,提取安保事件时序特征,构建深度LSTM模型,进而实现安保事件行为主体的学习与预测。结果表明:该模型可基于事件时序特征分析事件行为主体,预测精度更优,且在有噪声情况下也可得出良好结果,相关研究成果已在SZX机场成功应用。

关键词: 民航安保事件, 深度长短记忆(LSTM )模型, 行为主体, 多模态, 时序特征

Abstract: In order to assist police officers to analyze civil aviation security events, deep LSTM model was used to study subject identification of these events. Firstly, multi-modal information of them was presented by establishing a database, and their time series characteristics were extracted. Then, a deep LSTM model was developed to study and predict subjects of security events. The results show that the proposed model can predict subjects more accurately based on time series characteristics of the events, and even in the presence of noise, it can obtain better prediction results. Moreover, related research results have been successfully applied in SZX international airport.

Key words: civil aviation security event, deep long short-term memory(LSTM) model, event subject, multi-modal, temporal feature

中图分类号: