Real-time Traffic State Estimation Method For Urban Expressway

Real-time traffic state estimation of urban expressway is the premise andfoundation of effective traffic control. The development of intelligent transportationsystem, especially the progress of the AVI (Advanced Vehicle Information)technology, makes the urban road traffic information to obtain more accurate andcomprehensive. Based on mobile sensing technology (such as floating check-out car,smart phones), the real-time estimation of road traffic is one of the significantresearch direction. Compared with the traditional fixed traffic detection technology,mobile and portable traffic detection method has obvious advantages in the real-timeand accuracy of the information retrieval with a wide range of urban road.This paper studies the real-time road traffic state estimation problem of urbanexpressway based on the portable speed detector network. Vehicle speed informationis used to estimate the urban expressway road traffic density, flow and other trafficparameters. A single simple vehicle speed detection device, such as the traffic policeuse portable radar gun, combined with mobile communication technology, form thenetwork used in a wide range of average velocity testing of wireless sensor network.Compared with the usual traffic state estimation research, this paper have two features:first, the average speed of vehicles information can be collected in any space and timeposition of urban expressway, at the same time in different sections for trafficestimates. Detector network has the characteristics of self-organization to keep thenormal operation of the network traffic. Second, the observability of system isdifferent when the urban expressway traffic flow is in different traffic patterns, suchas freedom or congestion. This can cause traffic detection information is no effective,after the road traffic patterns change. In order to deal with large data traffic effectively,play the advantages of mobile traffic detection technology, gathering traffic flowinformation in the key location for state estimation is also a discussing question in thispaper.In fact, as a kind of space-time distributed parameter systems, system evolutiontrend of urban expressway traffic flow is determined by the current state of roadtraffic and road network boundary (upstream and downstream, the entrance ramp). Inorder to realize the synchronization estimation o the traffic state and boundary flow,this paper is mainly the following three aspects:First, in order to study the observability of urban expressway traffic flow under different traffic patterns, based on the traffic transmission relationship betweensections, a state switching system model is established, and the observability analysesin the paper. Through the analysis of the model, we found when urban expresswaytraffic flow in different traffic state, the effectiveness of traffic detection informationchanges. According to the effectiveness of traffic detection information, we canreasonable use of resources, reduce detector data transmission time, and effectivelycope with the traffic data problem.Second, the traditional estimation method is mainly focused on the study of mainroad with the known boundary conditions. In this way, the traffic estimation problemis restricted to detector space and time position. To solve above problems, in thisarticle, a kind of to the traffic flow model is proposed, and this model obtains not onlythe state variable (density), but also the unknown input (boundary flow). In order toachieve the traffic condition and boundary flow synchronization estimation, arecursive filter is designed in this paper, and the filter can estimate traffic state andtraffic flow at the same time. Because the system equation is not limited by theboundary flow, we select speed as observed variables, through the speed detectionequipment, combined with mobile communication technology, and construct thewireless sensor network of speed detection. According to information, traffic state isestimated with no limit, and the traffic model and filtering method has higherapplicability.Finally, we study the large scale of urban expressway traffic state estimationproblem. The order of large scale traffic network is high and the algorithm is complex.In this paper, we divide a long road into several sections, and then estimate trafficstate and boundary flow in each section. According to make a fusion of the boundary(or overlapping cell density), we can get the traffic state of a long road. Dividing thelong road into several sections greatly reduce the order of the model, and improve theefficiency of the algorithm.