Dynamic Path Planning Research In Intelligent Transportation Of Cities

With rapid development of urbanization in China since 21 st century,the number of vehicles in urban areas is growing exponentially,which exerts great influence on people's daily life in terms of traffic congestion,traffic safety and so on.Intelligent transportation can effectively solve current traffic problems by introducing advanced technology of computer science and electrocommunication.As an important part of intelligent transportation system,dynamic vehicle path planning mainly seeks out the optimal path and maximizes the use of the whole traffic network according to the needs of travelers and the real-time variable of traffic network.This thesis mainly studies the optimal planning path about urban traffic.First,the paper discusses the traditional shortest path algorithm,and then summarizes the defect of this algorithm in the real traffic situation.Paper improves the classic A* algorithm by importing Angle factor to the algorithm of the heuristic function,shielding the nodes with large angle between routes and destination and reducing the evaluation function as to the nodes with small angle.Improved A* algorithm takes the road network features into account,so it can ensure search precision and at the same time improve search efficiency.In addition,dividing road network into two layers can increase the ratio of upper layer road,which guarantees the shortest driving time although the path is not the shortest.Second,the paper analyzes the dynamic characteristic of road network,and proposes that using BP neural network to predict future traffic flow information.Take a typical intersection as an example,the simulation results show that the model can predict future road traffic flow accurately.When road traffic flow data is transformed into dynamic time weight,since BPR road impedance function can not reflect the fact that road condition changes from unblocked to crowded,speed-density relation is used to calculate the real driving time when traffic is clear or in jam,which provides forecast information of dynamic path planning.Finally,the paper designs the dynamic path planning framework predicted by traffic flow,which is mainly composed of vehicle-mounted navigation system and traffic control center system.The concept of dynamic path planning predicted by traffic flow refers to whether the vehicles re-plans the routes at the crossroad,calculated by discriminant ? which is improved from A* algorithm.Simulation results show that the dynamic calculation with forecast plans a route more effective and changeable to the real-time than that without forecast,resulting in different travel times under each departure time than other ways.