基于数字孪生的城市交通流智能预测与导引策略

物联网和人工智能等信息技术的快速发展极大地推动了交通系统的变革,同样也带来了机遇与挑战。针对现有导航系统忽略交通流时空特征而产生的策略重复性拥堵问题,对宏观交通流和微观车辆驾驶分别建模,并挖掘其耦合关系,进而提出一种基于数字孪生的城市智能交通流分层预测与导引策略,为减缓交通拥堵提供新思路。在该策略中,虚拟空间中的上层道路孪生通过扩散卷积递归神经网络预测时空交通流量,并显式作用于车辆路径规划决策。在此基础上,提出一种时空协同深度强化学习方法,用于实现车辆面向未来的协作式路径规划,指导虚拟空间中的下层车辆孪生选出最优策略反馈于真实世界。基于SUMO仿真平台进行了仿真验证。实验结果表明,本文所提方法在提高出行达成率、缓解拥堵等方面显著优于现有算法,能够有效提升城市交通出行效率。

Digital twin based intelligent urban traffic forecasting and guidance strategy

As the technology of ubiquitous Internet of things and artificial intelligence improves by leaps and bounds, the transportation system revolution is flourishing and bringing new opportunities and challenges.Considering the defect in the existing navigation system, and the neglect of the temporal and spatial characteristics of traffic flow, the macro traffic network and micro vehicle network were modeled and their coupling relationship was mined.Then, a digital twin based urban traffic forecasting and guidance method was proposed to alleviate the problem of traffic congestion.The spatial-temporal traffic flow information was predicted through the diffusion convolution recurrent neural network, which was explicitly applied to the vehicle path planning decision.On this basis, a spatial-temporal collaborative deep reinforcement learning method was proposed to implement the future-oriented collaborative path planning of vehicles.It also guided the underlying vehicle twins to select the optimal strategy for the real world.With SUMO for simulation verification, the experimental results show that the proposed method is significantly better than the existing algorithms in improving the travel completion ratio and congestion relief, and can improve the efficiency of urban traffic travel.