Research on Optimization of Intersection Signal Control Based on Traffic Flow Forecasting

To solve the problem of congestion at intersections, an optimization method of intersection signal timing based on traffic flow forecasting is put forward. Firstly, a Principal Component Analysis - Long-Short Term Memory (PCA-LSTM) model is established to predict the traffic flow at the intersection based on intersection data from Huangshan Road-Science Avenue in Hefei. The predicted traffic flow is then taken as the input of Minimum Maximum - Multi-objective Particle Swarm Optimization (MNMX-MOPSO). Finally, the signal timing is optimized by considering the intersection capacity, average delay, and average number of stops. The results show that the REMS and MAE of the proposed forecasting model are smaller than those of the traditional forecasting model, and the traffic capacity after timing optimization is also better than the typical timing scheme. This study can improve the traffic efficiency of intersections and alleviate the problem of intersection congestion to a certain extent.