Deep Reinforcement Learning Based Traffic Signal Control in Multi-Intersection Environment: A Comparative Study of DQN Variants

This dataset contains simulation data and trained models for the study "Deep Reinforcement Learning Based Traffic Signal Control in Multi-Intersection Environment: A Comparative Study of DQN Variants". The dataset includes:- SUMO simulation environment files for two scenarios (regulated left-turn and free left-turn)- Training and evaluation results for DQN, DDQN, Dueling DQN, and Dueling DDQN- Performance metrics: reward, average waiting time, and average speed- Scripts used for training and testing The experiments are conducted in a multi-intersection environment with heterogeneous vehicle types (ambulance, fire truck, police, bus, truck, car, and motorcycle) with priority weights. This dataset supports reproducibility of the results reported in the associated publication.本数据集为研究论文《多交叉口环境下基于深度强化学习的交通信号控制：DQN变体对比研究》（"Deep Reinforcement Learning Based Traffic Signal Control in Multi-Intersection Environment: A Comparative Study of DQN Variants"）提供支撑。 本数据集涵盖以下内容： - 两种场景的SUMO仿真环境文件：受控左转场景与自由左转场景 - 深度Q网络（Deep Q-Network, DQN）、双深度Q网络（Double Deep Q-Network, DDQN）、对决式深度Q网络（Dueling Deep Q-Network, Dueling DQN）以及对决式双深度Q网络（Dueling Double Deep Q-Network, Dueling DDQN）的训练与评估结果 - 性能指标：奖励值、平均等待时长与平均行驶速度 - 用于模型训练与测试的脚本文件 本次实验在多交叉口环境中开展，实验纳入带有优先级权重的异质车型，涵盖救护车、消防车、警车、公交车、货车、轿车及摩托车。 本数据集可支撑关联发表论文中所报告结果的可复现性。