Research On The Mining Of Abnormal Behavior Pattern Based On Traffic Spatio-temporal Bigdata

In intelligent transportation system,the technology for Internet of things is used to monitor and manage the flow of traffic participants,with massive traffic spatio-temporal data generated.These data provide valuable base for our study of the complex and changeable traffic network.However,the increasing amount of data makes the traditional data analysis methods have been unable to meet the actual application requirements.So,how to use the big data processing method,and to carry out the relevant research on the specific characteristics of the traffic spatio-temporal data is very important to provide technical support for promoting the development of transportation network.In this paper,we focus on the research of abnormal behavior pattern mining in the traffic spatio-temporal data,and propose the models of velocity outlier and floating car out-of-bounds outlier and the detection algorithms.For the velocity outlier detection,The key lies in how to get the velocity of the vehicles quickly from the massive vehicle monitoring data.we realize the corresponding outlier detection algorithms based on the specific application by using big data processing methods and data stream technology.For the floating car out-of-bounds outlier detection,the essence of this problem lies in the judgment of the position relation between the the spatial point and area.So,we propose the multilayered grid model base on the geographic grid,and realize the storage and query of this model by using HBase database.Besides,the performance of the two spatio-temporal outlier detection algorithms is tested.The experimental results show that our algorithms have high feasibility and efficiency.Finally,we realize the application of velocity outlier monitoring and floating car traveling area monitoring in the road real-time monitoring system.