The Model And Application Of Urban Traffic Emergency Management And Safety System

Aiming at the safety problem of urban traffic system, this thesis focuses on the following topics in an urban traffic emergency response and real-time traffic safety analysis system using information processing technologies and smart agent technologies: the framework of the system, automatic traffic incident detection and alarming, the real-time traffic safety analysis, the validity evaluation of traffic emergency response. At the aspect of emergency response, the safety of urban traffic and its related problems are explored briefly. The development of advance traffic management system(ATMS) and the traffic incident response are discussed. In addition, the research content of the system is analyzed. Following the 'real-time forecast, incident detection, emergency response' model, the shortcomings of the recent emergency response system are analyzed. The concept of urban traffic emergency management and safety is presented based on the system level. The constituent parts of the system and their working mechanisms are discussed detailedly. The parts including forecast, detection and response are combined organically. The key techniques of applying data warehouse technology and network technology into the traffic emergency management system are stated. Moreover, the reconstruction of traffic incident data based on network data warehouse method is presented. Main aspects of the model are analyzed thoroughly such as subject organization of the model, the design of the metadata, the method of data refinement and data integration and online analytical processing. The ways of realization are given. Conformed to the multi-incident multi-response principle, there are some fatal problems of ordinary programming model, so the expressions of opportunity cost is inducted. Based on these expressions a new traffic incident response resource assigning model is found and in addition the arithmetic is simplified in order to be applied in project. In this paper some methods are employed to enhance the function efficiency of Dijkstra arithmetic such as dwindling searching area and improving the store construction of map data. The dynamic weights are set according to the time variation. According to the practical requirements of traffic video surveillance engineering, firstly, on the basis of analysis several background estimation and updating algorithms and studying on the complexity of surveillance scenes, this paper throws light on the background prediction algorithm frame and real-time updating background strategy, these algorithms is tested and analyzes the validity of these algorithm. Secondly, on the basis of analysis popular motion detection algorithm, considering practical hardness of motion detection in the traffic surveillance scenes, several motion detection algorithms is integrated and modified, by these means the proper algorithm is combined. Thirdly, the edge pixels of motion object are not continuous and sparse, what's more, the edge of motion object may be broken, in view of these facts motion object extraction algorithm is present, elimination noise algorithm based on pixels spatial distribution and the road characteristic direct line extraction algorithm based on Hough transformation theory are set out in detail. The logical framework of real-time traffic safety analysis is presented. Based on the information get from the traffic surveillance system, the speed limit, Bayesian network traffic incident forecast model and driving characteristics evaluation model analyze the traffic safety, report the primary risks instances, thereby, assisting the road safety control decision A traffic incident forecast model based on the Bayesian network is made. The parameters set and the relevant range of values are analyzed and presented. The studying and inference arithmetic are put forward according to the features of the statistical data. Different drivers have different estimate of the range keeping. If we can describe and evaluate these differences concisely, we can supervise the driving behavior easily. And we also can help drivers find their driving behavior problems to decrease the traffic accident. In this paper, the driving behavior data are processed orderly through establishing statistic model. A estimate method based on triangle-fuzzy number is presented. The information characteristics of driving are described in mathematic way. The test data are analyzed. Finally Multi-criteria cost-benefits analysis is adopt to bring forward the economy and society benefit evaluation methods, which provide the theory basis for the application and popularization of traffic emergency management system...