Cumulative Flow Diagram-Based Fixed-Time Signal Timing Optimization at Isolated Intersections Using Connected Vehicle Trajectory Data

Abstract

Time-dependent fixed-time control is a cost-effective control method that is widely employed at signalized intersections in numerous countries. Existing optimization models rely on traditional delay models with specific assumptions regarding vehicle arrivals. Recent advancements in intelligent mobility have led to development of high-resolution trajectory data of connected vehicles (CVs), thereby providing opportunities for improving fixed-time signal control. Taking advantage of CV trajectories, this study proposes a cumulative flow diagram (CFD)-based signal timing optimization method for fixed-time signal control at isolated intersections, which includes a CFD model and a multi-objective optimization model. The CFD model is formulated to profile the time-dependent vehicle arrival and departure processes under varying signal timing plans, where the intersection demand is estimated based on a weighted maximum likelihood estimation method. Then, a CFD-based multi-objective optimization model is proposed for both undersaturated and oversaturated traffic conditions. The primary objective is to minimize the exceeded queue dissipation time, whereas the secondary objective is to minimize the average delay at the intersection. Considering the data-driven property of the CFD model, a bi-level particle swarm optimization-based algorithm is then specially designed to solve the optimal cycle length (and the reference point if it is considered) and green ratios separately. The proposed method is evaluated based on simulation data and compared with Synchro. The results indicate that the proposed method outperforms Synchro under various traffic conditions in terms of average delay and queue since the time-dependent vehicle arrivals during the cycle are considered in a CV trajectory data-driven way.