Research On Traffic Prediction And Multipath Optimization Of Intelligent Transportation Systems

Intelligent Transportation System(ITS)has gradually become one of the effective means to solve various traffic problems.This thesis mainly focuses on the short-term traffic flow prediction and path planning.The research results of this thesis are mainly displayed in the form of forecasting research programs and planning research programs.The research programs designed in this thesis are mainly used to solve the problem without real-time data.How can traffic application software provide traffic flow prediction results quickly and how to provide customized road planning solutionsFor short-term traffic volume prediction,how to predict short-term traffic flow in real time and accurately is the main problem.In this thesis,a combination model of extreme random tree and Kalman filter is proposed to realize real-time traffic volume prediction.This prediction model combines the advantages of Kalman filter recursive algorithm and extreme random tree.Specifically,Kalman filtering recursive algorithm can improve the operational efficiency and has high real-time signal processing ability.Extreme random tree can extract parameters according to the importance of the characteristic parameters,which can be used to solve Kalman filtering divergence problem.Therefore,the short-term traffic flow forecasting proposed in this study is advanced and achievable.On the other hand,for the traveler’s road path planning,the solution of the problem not only needs to consider the traffic situation of the current path,but also should consider the traveler’s preference for the way of travel.Personal preferences can be calculated based on historical travel data and related data mining methods.In this thesis,Naive Bayesian algorithm is used to calculate and mine personal preferences.Then GS algorithm and ant colony optimization are used to solve the path planning problem respectively.Naive Bayesian algorithm is a classical prediction algorithm,which has the advantages of low computational requirements and low data dependence.GS algorithm is a resource allocation mechanism,which can better solve the problem of stable matching.Ant colony optimization(ACO)is a modern heuristic algorithm based on swarm intelligence,which can be better used in multi-choice decision-making under multiconstraints.According to the specific model proposed in this thesis,the confidence of path selection is calculated by Naive Bayesian algorithm,and then different path planning schemes are designed by using GS algorithm’s stable matching and ant colony optimization’s pheromone definition and volatilization rules.By comparing the simulation results,we can verify the efficiency and practicability of the planning scheme proposed in this thesis.A brief summary of the specific work of this thesis is as follows:(1)The combination model of extreme random tree and Kalman filter is used to realize the design of real-time short-term traffic volume forecasting scheme.Firstly,the extreme random tree is used to select the characteristic parameters which have great influence on the traffic volume from the historical traffic data,and then Kalman filter is used to predict the short-term traffic volume through the selected parameters.(2)Based on the traveler’s personal history travel data,the naive Bayesian algorithm is used to calculate the traveler’s preference for travel mode,which is an important parameter in the design of route planning scheme.(3)Based on the results of traffic volume prediction and personal preference calculation,GS algorithm and improved ant colony optimization algorithm are used to realize the design of personalized path planning.The criterion of optimization is to ensure the shortest travel time,avoid congestion,and fully consider the user’s personal preferences.(4)Through simulation and comparison with several common schemes,the superiority of the proposed scheme is proved,and the shortcomings in the scheme are pointed out.Through the work of this thesis,a new combination model based on traditional method is proposed,and the design of real-time short-term traffic volume forecasting scheme is completed.A simple improvement of the mature algorithm is proposed to adapt to the new scenario,and the design of personalized path planning scheme is completed.