Based Multi-agent Traffic Information

Several information technology issues related to traffic are studiedin this thesis.This include:intersection-centric route guidancealgorithm and architecture design for traffic simulation based onmulti-agents,intersection-centric priority algorithm and architecturefor traffic simulation based on multi-agents,multi-agent network flowarchitecture for traffic application performance evaluation,and vehicleconvert coefficient research.At the part of intersection-centric route guidance algorithm andarchitecture design for traffic simulation based on multi-agents,algorithm which could be used in large scale vehicles and person' sevacuation when they have own destination addresses is proposed.Basedon the analysis and summarize of existing research of traffic simulationrouting,combined with data flow control methods and protocols of computernetworks,this paper creatively proposed network protocols could be usedin traffic simulation to simulate the relationship of roads and vehicles.After the function development practice proved that this architecture andalgorithm are practical,algorithm is studied in further.Road map relatedrandom map and IBF* algorithm proved that design' s complexityexpectation is much lower than the worst case.Multi-dimensional scalingis introduced into intelligent routing algorithm proposed in trafficrouting issue.Performance simulation result in simulation test provedthe correctness of algorithm.A mathematic proof is also given to proveit.At the part of intersection-centric priority algorithm for trafficsimulation based on multi-agents,evacuate algorithm to special vehicleis proposed.Based on last chapter' s algorithm,a priority guaranteeframework and a priority algorithm run on intersection referenced withcomputer networks' QoS algorithm is proposed.For urgent or specialvehicles,controllable priority is new requirement.With the simulationverification of algorithm,the feasibility of it is proved.To large scale traffic flow evacuating,algorithm is useful to guarantee special objectsrunning.New algorithms are based on the intelligent agents ofintersections,so these algorithms framework systemly solves the correctguidance issue and the priority guarantee issues.At vehicle conversionfactor research part,using multi-view Expectation Maximizationalgorithm based on Gaussian mixture model to calculate vehicle convertcoefficient is proposed first time.With the real roads' vehicles datacolleted by video device,algorithm was trained by real data.This provedthat algorithm could reduce the value of entropy and improve view' sagreement rate,so the vehicle convert coefficient is believable.At a multi-agents' network flow architecture for applicationperformance evaluation part,traffic related application perforce wasconsidered.With the new software architecture appearing,the applicationperformance evaluation and bottle neck finding becomes more difficultbecause of the complexity in software components and networks'interaction.This part is about how to use multi-agents on networks tocapture analysis and replay network data flow to evaluate the applicationperformance.Both active and passive agents cooperate to simulate andtest,they communicate and interact to find the performance result andthe bottle neck of the system.Through real traffic application systems'test practice,the advantage of this method is proved.This method cangive out multi-data at same time with high performance and have someautomatic functions.In conclusion,the main innovation is to use internet protocols andmulti-agents as research platform to study traffic related routing area.Related algorithm' s new application and algorithm' s improvement andproof proposed include:Dijkstra algorithm,Intersect-Bellman-Ford *algorithm,n regular random graph,multi-dimensional scaling,FastmapAlgorithm,QoS algorithm,multi-view Expectation Maximization algorithmbased on Gaussian distribution hybrid model and so on.Simulation,mathematic proof and real system practice are used to prove the correctionof these architectures,methods and algorithms.