Research On Traffic Organization Optimization Of Plane Signalized Intersections Based On Flow Prediction By LSTM

The problem of urban traffic is a difficult problem in today’s world.Because of the characteristics of mixed traffic in our country,it is more difficult to manage.The urban traffic management is a huge open system,in which each subsystem affects the normal operation of the huge system.The social economy and the level of urban motorization in Lanzhou rapidly improved,the contradiction between the original urban planning,road infrastructure,road network structure and the construction of lanzhou into a livable city of mountains and rivers has been increasingly prominent.At the same time,the contradiction between traffic supply and demand is more and more prominent.Urban traffic congestion has spread from point to line,from line to surface,from local area to a larger range of the city,and also extends from peak to flat peak.Traffic congestion has become the normal problem.Further strengthening the management of city’s traffic safety,maintaining good traffic order,effectively improving the traffic environment,ensuring the safety of road traffic,making the traffic smooth and orderly is impending.This paper aimed at optimizing traffic organization of planar signal intersection,introduced background of the paper’s topic and the current research of domestic and foreign firstly.Today,traffic congestion increasingly serious.It is realistic to forecast the traffic flow through the artificial intelligence method,thus optimizing traffic organization from space and time.Secondly,the concept and ideas of traffic organization optimization are briefly introduced,and the methods of optimizing the traffic organization for planar signal intersections by channelization of intersections,signal control of intersections and management through relevant rules and policies are proposed.The specific methods include design of lane,design of diversion island,increasing the number of lanes at the entrance,setting up waiting areas for vehicles to turn left or go straight,setting up variable lanes in the same direction,setting up reverse variable lanes,and reasonably determining the phase sequence,cycle length and the allocation of green time.Finally,the RNN and LSTM models are introduced.The traffic flow at the intersection of tianshui north road and yanning road is predicted based on the cyclic neural network model.The comparison of relevant parameters between pre-optimization intersection and optimized intersection shows that through the traffic organization optimization,the traffic congestion and the traffic efficiency at the intersection of tianshui north road and yanning road is significantly improved,which verifies the feasibility of the intersection traffic organization optimization method based on the artificial intelligence method to predict traffic flow.