Reinforcement-learning-based optimization method for ecological driving behavior at an urban road intersection

The invention provides a reinforcement-learning-based optimization method for an ecological driving behavior at an urban road intersection. The method comprises the following steps: establishing a simulation platform by using a cellular automata model; placing a vehicle on the simulation platform for reinforcement learning; after reinforcement learning on the vehicle, obtaining a Q matrix; and optimizing the driving behavior of the vehicle by using the obtained Q matrix to realize ecological driving. With the evaluation method based on reinforcement learning and the microscopic simulation platform, the influences on vehicle exhaust emission and traffic characteristics by the ecological driving behavior are studied to provide useful information for improvement of the strategy of the vehicleecological driving. 本发明提供种基于强化学习的城市道路交叉口生态驾驶行为优化方法。该方法包括以下步骤：利用元胞自动机模型建立仿真平台；将所述车辆置于所述仿真平台进行强化学习；所述车辆经过强化学习后，得到Q矩阵；利用得到的Q矩阵对车辆的驾驶行为进行优化，以实现生态驾驶。通过该基于强化学习的评价方法和微观的仿真平台，来研究生态驾驶行为对车辆尾气排放和交通特性的影响，从而为车辆生态驾驶的策略改进提供有用的信息。