Research On Multi-vehicle Path Planning Based On Traffic Information Prediction And Dynamic Pricing Mechanism

In recent years,the development of society has been speeding up the process of urbanization and the expansion of urban transportation network.Problems such as urban traffic congestion have seriously impact on people’s daily lives,which not only increases people’s time for commuting,but also brings problems such as environmental pollution and economic losses.The emergence of Intelligent Transportation Systems(ITS)makes it possible to solve this problem.The Intelligent Transportation System combines computer,communication,electronics,operations research,traffic big data and artificial intelligence technique,monitor and analyze the real-time traffic information,and plays an important role in guaranteeing the road network’s safety and efficiency.As the key link of intelligent transportation system,vehicle path planning plays an important role in urban vehicle travel navigation.Vehicle path planning is mainly divided into single vehicle path planning and multi-vehicle path planning.Compared with the single-vehicle path planning,multi-vehicle path planning can co-ordinate most of the road network factors,which is conducive to alleviating traffic congestion.Traffic information prediction and vehicle path planning are important components of Intelligent Transportation Systems,which provides effective basis for urban traffic managers or travelers to make decision.This paper mainly researches on the effect of multi-vehicle dynamic path planning based on traffic information prediction and dynamic pricing mechanism on reducing traffic congestion.Firstly,this paper discusses some basic problems of multi-vehicle path planning,defines the traffic road network,and transforms the multi-vehicle path planning problem into a mixed integer linear programming problem so that the problem can be discussed with the optimization methods.Then the common optimization goals of the path planning algorithm are discussed later,and the simulation platform and tools used in this paper are also introduced.Secondly,this paper analyzes some static and dynamic path planning algorithms in multivehicle path planning(including Dijkstra algorithm,A* algorithm,D*Lite algorithm etc),defines the generalized price of roads for path planning,and proposes a heuristic dynamic pricing mechanism for updating iterative formula of generalized price.In this mechanism,the generalized price of roads will increase when traffic jams occur,and remain in other cases.The generalized price is monotonically increasing,which considers the cumulative impact of road congestion on future time.The simulation experiment is carried out to verify the effect of this method on reducing traffic congestion.Thirdly,this paper discusses short-term traffic information prediction,and analyzes several short-term traffic information prediction methods,including Autoregressive Integrated Moving Average model(ARIMA),BP neural network model,Long Short-Term Memory model(LSTM)and Stacked AutoEncoders model(SAEs).Then the experiments for comparing their performance are made.The results reveals that the artificial intelligence methods are better than the traditional time series methods in general,and the SAEs model works best.Finally,the paper analyzes the dynamic price mechanism and make some improvements.One is to replace the real-time average travel speed with the predicted average travel speed.The other is to add a momentum term in the iterative formula of the dynamic price mechanism.The simulation results show that the improved multi-vehicle path planning method based on traffic information prediction and dynamic price mechanism is better than the preimprovement method in reducing traffic congestion and reducing travel time.