基于Q学习的Agent在交叉口航班排序中的应用

Application of Q-learning Agent in Aircraft Sequencing at En-route Intersection

为了解决航路交叉口的交通压力问题,针对交叉口位置的空中交通排序展开研究,把Q-learning与Agent联系起来作为解决该问题的技术手段。首先对Q-learning算法的学习原理及Agent结构模型的实现进行说明,然后根据对汇聚的空中交通流进行控制的过程提出了奖惩函数,仿真结果证明,这种方法的控制效果要比FCFS算法更好。

Great traffic pressure has been posed on air route intersections due to development of civil aviation industry. To address this issue,research on air traffic sequencing at intersections has been carried out,in which the Q- learning algorithm and Agent technology are employed. The Q- learning mechanism along with the realization of Agent model are introduced in the first place,then a reward function is proposed based on the control process over the converging air traffic. The simulation results show that the effect of this control method is better than that of the FCFS approach.