Research On Signal Timing Optimization Of Intersections Based On Traffic Flow Forecast

As a node of the urban road network,the intersections of urban roads directly affects the traffic efficiency of the entire urban road network.The signal timing scheme of intersections plays a decisive role in its traffic efficiency and it also as the core part of urban traffic signal control.Based on the short-term traffic flow prediction data of intersections,this paper studies the signal timing optimization of intersections.This paper explains the types of traffic signal control according to different control methods and control ranges,analyzes and selects the evaluation indicators of traffic signal control.In this paper,the research work is divided into two parts: the study of short-term traffic flow prediction methods at intersections and the study of timing optimization at intersections.For the prediction part,based on analyzing the traffic flow characteristics of intersections,this paper plans the data format as 5min traffic flow data of twelve traffic lows at intersections,and selects LSTM as the prediction algorithm for short-term traffic flow data.The PCA-LSTM prediction model is established and the validity of the model is verified.For the timing part,this paper improves the particle swarm optimization algorithm by analyzing and summarizing the research scenarios of timing optimization and the principle of multi-objective timing optimization.Taking the intersection traffic capacity,average vehicle delay and average vehicle parking times as optimization goals.Considering the deviation between the predicted data and the real data,the MNMX-MOPSO timing algorithm is designed.Finally,combining the research results of the prediction part and the timing part,a multi-objective timing optimization model based on traffic flow prediction is constructed.In the simulation verification,taking the intersection of Huangshan Road and Science Avenue in Hefei as an example,the simulation model was constructed using VISSIM traffic simulation software.The Webster method and the MNMX-MOPSO timing algorithm proposed in this paper were carried out during peak and peak periods respectively.By comparison,the results show that the MNMX-MOPSO timing algorithm has better performance at the peak period,and it is similar to the Webster method at the peak period and relatively more stable,which verifies the effectiveness of the multi-objective timing optimization model proposed in this paper.