社区老年人空间行为轨迹异常分析方法  

Abnormal Analysis Method of Old People’s Spatial Behavior Trajectory in Community

作　　者：孟祥泽 胡啸峰[1,2] 沈兵 MENG Xiang-ze;HU Xiao-feng;SHEN Bing(School of Information Technology and Network Security,People s Public Security University of China,Beijing 100038,China;Public Security Behavioral Science Laboratory,People s Public Security University of China,Beijing 100038,China)

机构地区：[1]中国人民公安大学信息网络安全学院,北京100038 [2]中国人民公安大学公共安全行为科学实验室,北京100038

出　　处：《科学技术与工程》2021年第9期3676-3681,共6页Science Technology and Engineering

基　　金：国家重点研发计划(2018YFC0809702);公安部科技强警基础工作专项(2018GABJC01)。

摘　　要：为分析社区中老年人空间行为轨迹的规律并识别其异常空间行为轨迹,建立了基于动态时间规整(dynamic time warping, DTW)算法和基于密度的空间聚类算法(density-based spatial clustering of applications with noise, ST-DBSCAN)的社区老年人异常空间行为轨迹分析模型。首先,利用社区内老年人空间轨迹定位数据,采用ST-DBSCAN聚类算法对连续空间轨迹进行聚类分析,提取老年人的动态行为链。其次,针对两种常见的异常轨迹模式(出行轨迹偏离日常轨迹和停留时间过长),利用DTW算法对老年人的动态轨迹进行异常模式识别。最后,结合异常轨迹模式、老年人背景信息、气象信息,建立社区老年人异常行为风险分析模型,分析老年人在出现轨迹异常情况下的安全风险,并基于微软亚洲研究院的开源轨迹数据集GeoLife对建立的模型进行了验证。研究结果表明,模型可以识别老年人出行时的异常空间轨迹,分析其安全风险。研究成果可以为居委会、社区物业、养老机构等部门的老年人安全管理工作提供方法支持。In order to analyze the laws of the seniors spatial behavior trajectory in community and identify their abnormal trajectories,a model coupled dynamic time warping(DTW)algorithm with density-based spatial clustering of applications with noise(ST-DBSCAN)clustering algorithm was established.First,based on the spatial trajectory data made by the seniors in the community,ST-DBSCAN algorithm was used to cluster the trajectories and extract the dynamic behavior chain of them.Then,for two typical abnormal trajectory patterns(one was that the seniors travel trajectory deviated from usual trajectory and the other was that the seniors stayed at a specific place for a long duration),the DTW algorithm was used to identify them.Finally,comprehensively considering the abnormal trajectory pattern,the seniors background information as well as the weather data,the risk analysis model for analyzing the seniors abnormal behavior in the community was built.The proposed model was validated by the GeoLife trajectory data set from the Microsoft Asia Research Institute.The results show that the proposed model can identify the seniors abnormal trajectories when they move within the community,and evaluate their abnormal behavior risk.The results are expected to provide method support for neighborhood committee,property management company,nursing home and other departments safety management of the seniors.

关 键 词：社区 老年人 空间行为轨迹 异常分析 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]