Study And Implement On Parallel Simulation Algorithm Of Dynamic Route Solution For Traffic Network

With the quick development of society economy and continuous accelerationof urbanization process, urban vehicle population continues to increase and urbantraffic demands increase sharply. Many problems about traffic that includes trafficcongestion, traffic accidents, environmental pollution and energy shortage havebecome a universal problem. Because the construction of urban roadway can't meettraffic travel demand at present, traffic problems will deteriorate without moreeffective traffic management measures. Intelligent transportation system that takesinto account traffic participant, vehicle, roadway and environment appliesadvanced information technology, communication technology and traffic flowtheory to urban traffic management and control, and then guides travelers choosingrational route. It is one of effective approach of resolving urban traffic congestion.Dynamic traffic assignment is an important aspect of implementing intelligenttransportation system item. Especially, Advanced Traffic Management Systems,Advanced Traveler Information Systems and Vehicle Route Guidance Systems areall nearly correlated with dynamic traffic assignment. Dynamic route solution iscore and pivotal technology of dynamic traffic assignment. Due to characteristic ofcomplexity, dynamic time-dependence and randomicity in traffic system, driverswhich are unfamiliar with city street network can produce many noneffectivetravels. This will increase travel time and tend to bring traffic jam. Dynamic routesolution algorithm researches how time-dependent traffic demand is reasonablyassigned to time-space traffic network by estimating and forecastingtime-dependent traffic network condition, and then providing real-time trafficinformation to travelers through ATMS, ATIS and VRGS. It can guide travelerschoosing rational route and reduce individual travel cost or general system cost,improve traffic network performance and alleviate traffic congestion. Hence,dynamic route solution model and algorithm is primary essential technology ofimplementing traffic guidance.Dynamic route solution model and algorithm using microscopic trafficnetwork simulation can evaluate feasibility of pre-implement traffic infrastructureconstruction and traffic management and control measure, guide traffic traveldemand, and provide decision-making for route choice of drivers. With the increaseof simulation network scale, a great deal of real-time data obtained from trafficnetworkneed to be rapidly handled. However, computing capacity of singleprocessor can't cater for need of large-scale network simulation. Distributed andparallel computing technology can greatly improve speed and efficiency of trafficnetwork simulation and cater for real-time dynamic route guidance.Firstly, domestic and international development of dynamic route solution viaanalytical approach and simulation approach for traffic network is introduced. Theadvantages and disadvantages of two models are also compared. The paper putsforward to adopt parallel simulation technology to improve speed and efficiency ofdynamic route solution, briefly reviews application of parallel simulationtechnology in traffic field, and points out development prospects of parallelsimulation of dynamic route solution.The framework of dynamic route solution is designed, based microscopictraffic network simulation. The time-dependent OD travel demend matrixs areestimated using average link flow ratio method. Link travel time estimating modeltakes into account intersection queue delay, signal control delay and intersectionturning delay. Forecast model of link travel time based on gray equal-dimensionand new-info model is established. The proposed model is proved to have betterforecast precision with several examples. The time-dependent optimal routesbetween OD are searched by using Dijkstra mending algorithm and are used todetermine optional route set. The multi-user route choice models and algorithmsare built up according to different obtaining information manner under ATIS,describing multi-user stochastic dynamic route choice behavior in detail frommicroscopic level. Because route solution model integrates traffic control measures,it can describe actually dynamic characteristics of traffic network congestion.The paper constructs object-oriented parallel data structure by analyingparallel characteristic of dynamic route solution for traffic network simulation.Distributed memory parallel computer system and master-slave pattern developingstrategy of parallel program are adopted to build up framework of parallelsimulation system of dynamic route solution. Parallel algorithm of dynamic routesolution simulation is designed, based on network-partition method and processmatching strategy. Network-partition method based on number of vehicles loadingis put forward, which assigns subnetwork to corresponding slave process accordingto the average number of vehicles that each process is responsible for and numberof vehicles that each intersection contains. At the same time, communicationmechanism and information exchange between processes are analyzed. The paperdescribes detailedly parallized strategy of module, flow control and communicationpattern between processes in parallel simulation implement of dynamic routesolution.According to research outcome of microscopic traffic network simulation anddynamic route solution model and algorithm, the paper develops integrated trafficparallel simulation system, designs distributed parallel flatform scheme and datatransmission technology to support system, and builds up parallel computingplatform based on message passing interface. The proposed system can evaluateimplement effects of ITS items and guide drivers choosing optimal route.The proposed dynamic route solution model and algorithm are evaluated byexample application that traffic network operation condition is assessed by fourmeasures including different user proportion, assignment interval of dynamic routesolution, iterative number and average link flow ratio. At the same time, parallelalgorithm of dynamic route solution is implemented in distributed and parallelcomputing platform. Parallel benefits are analyzed by four measures includingprogram run time, speedup, parallel efficiency and expandability.Study on distributed and parallel simulation is a rapidly developing researcharea of international simulation field. Dynamic route solution is core of urbantraffic guidance. Their amalgamations are frontiers of traffic simulation field. Thedynamic route solution model and parallel implement algorithm proposed in thispaper can evaluate traffic network performance and meet real-time route guidance.It is very significative in both theory and practice.