Design And Implementation Of Highway Management And Early Warning Platform Based On SPARK

In recent years,with the rapid development of social economy and the rapid growth of transportation demand in all provinces,the highway management departments are faced with great challenges in the road transportation,which puts forward higher requirements for the highway management departments in the road network operation management and the ability to deal with emergencies.In road traffic management,for the purpose of this paper,combining the domestic and foreign advanced experience in the field of traffic emergency management,integrated application of distributed technology,data processing technology,video monitoring technology,deep learning and Java EE technology highway management and early warning platform is designed and developed,on the premise of meet the actual demand to achieve convenient traffic management.The platform is divided into four subsystem modules: platform information management,emergency management and disposal,road network operation monitoring and highway information service.Among them,platform information management provides basic data and personnel information management;Emergency management and disposal to provide risk hidden danger information,emergency resource information management and emergency statistical analysis;Road network operation monitoring provides intuitive and convenient real-time traffic flow data and statistical analysis results of traffic flow data display in the way of data visualization,and designs and realizes the function of traffic flow prediction.The highway information service provides the function of news release and highway incident management.According to the timeliness of traffic flow and the characteristics of Streaming data,this paper proposes to apply the Spark Streaming component based on the Spark platform to process data,so as to ensure the reliability of data and reduce the coupling degree between systems.The data is transmitted via the Kafka messaging system to the Spark cluster for processing,and the real-time data is classified into different categories to calculate the traffic in a given time interval.In view of the periodicity and trend of short-time traffic flow,a traffic prediction method based on Hyper Networks was proposed.The learning experience of the past time was obtained by taking the time series of days as the input of the supercoding layer,and then the prediction results were obtained by building a model with the traffic series of half an hour as the input of the inference coding layer.As an engineering project,the system has successfully passed the testing stage,and each function module runs stably with good reliability and stability.The application of the system will certainly improve the ability of road network management and emergency response of highway authorities,and will also provide assistance to ensure the smooth road.