Research on Traffic Signal Control Optimization Based on Time Period Division and Traffic Flow Prediction

Abstract

With the acceleration of urbanization, the surge in traffic flow makes traffic signal optimization a key issue, and the traditional fixed-time control method is difficult to cope with the dynamically changing traffic flow, which urgently needs to be optimized and adjusted based on actual data. Specifically, for traffic flow analysis, this study first preprocesses the dataset, extracts the records of the longitudinal middle road - latitudinal middle road, and divides the data from 1 April to 6 May 2024 into weekdays, double holidays, and holidays. By the equidistant binning method, the day is divided into four time periods, and the steering rate is introduced for traffic flow estimation. Next, the traffic signal period, maximum number of passing vehicles and delay time are calculated, and a nonlinear planning model is established to minimize the delay. In contrast, the optimization of the traffic signal configuration by applying the SLSQP algorithm yields an average speed of 60km/h for the low peak, 50km/h for the flat peak, and 40km/h for the peak, which improves the average speed.