基于Detroit模型和深度学习的交通流调度方法应用分析

Analysis of Traffic Flow Scheduling Method Based on Detroit Model and In-depth Learning

当前交通流调度方法,未考虑车流量的实际输出结果,且算法过于简单,难以得出到最优的调度方案,交通流调度效果差。据此提出基于Detroit模型和深度学习的交通流调度方法,采用基于Detroit模型的交通需求分布预测方法,获得预测年的车辆流;采用以深度信念网络(DBN)和高斯—伯努利GBRBM为基础的深度学习模型,对输入的预测年车辆流进行训练,获取对应道路的实际车辆流输出结果,结合最大最小蚁群算法求出交通流调度最优解,实现交通流的有效调度。实验结果说明,所提方法调度交通流的平均时间为31. 5 s、调度效率最高为99. 8%、最低规划误差仅为5%,说明该方法具有较高的交通流调度精确度和效率,调度性能高。

The current traffic flow scheduling method does not consider the actual output of the traffic flow,and the algorithm is too simple,it is difficult to get the optimal scheduling scheme,and the traffic flow scheduling effect is poor.Based on this,a traffic flow scheduling method based on Detroit model and deep learning is proposed.The traffic demand distribution prediction method based on Detroit model is used to obtain the vehicle flow of the forecast year.Based on the deep belief network(DBN)and GaussBernoulli GBRBM.The deep learning model trains the input predicted vehicle flow,obtains the actual vehicle flow output result of the corresponding road,and combines the maximum and minimum ant colony algorithm to find the optimal solution of traffic flow scheduling to realize the effective scheduling of traffic flow.The experimental results show that the average time for scheduling traffic flow is 31.5 s,the scheduling efficiency is 99.8%,and the minimum planning error is only 5%,which indicates that the method has high traffic flow scheduling accuracy and efficiency,and high scheduling performance.