An iterative optimization method for sustainable environmental improvement under mixed CAV-HDV traffic

This paper proposes an iterative optimization method for trajectory control of connected automated vehicles (CAVs) at an off-ramp bottleneck, aiming to improve energy consumption and pollutant emissions in the traffic system. The methodology integrates real-world data from the Next Generation Simulation (NGSIM) dataset and generated data from each iteration. A comprehensive cost function is developed to evaluate safety, efficiency, comfort, equilibrium, lane-changing, energy, and emissions. The lane-changing process is divided into two stages: lane-changing decision-making (LCD) and lane-changing execution (LCE), modeled using advanced artificial intelligence algorithms. Specifically, gcForest is applied to model LCD, while long short-term memory (LSTM) is used for LCE, allowing for more precise control of lane-changing behavior. The analysis considers fuel consumption and key vehicular emissions, including carbon dioxide (CO2), nitrogen oxides (NOx), volatile organic compounds (VOC), and particulate matter (PM). The results indicate that energy consumption and pollutant emissions are reduced by 20 % after three iterations of optimization. Furthermore, the iterative method demonstrates significant environmental improvements, particularly when the CAV market penetration rate (MPR) reaches approximately 50 %. Higher MPR levels further enhance the sustainability benefits of CAVs, making them more advantageous for promoting sustainable traffic development.