基于深度学习短时交通流预测.PDF

优先出版 计 算 机 应 用 研 究 第34 卷 基于深度学习的短时交通流预测* 罗向龙，焦琴琴，牛力瑶，孙壮文 (长安大学 信息工程学院，西安 710064) 摘 要：针对现有预测方法未能充分揭示交通流内部的本质规律,提出了一种基于深度学习的短时交通流预测方法。该 方法结合深度信念网路模型 （DBN ）与支持向量回归分类器 （SVR）作为预测模型，利用差分去除交通流数据的趋势 向，用深度信念网络模型进行交通流特征学习，在网络顶层连接支持向量回归模型进行流量预测。实际交通流数据测 试结果表明：文中提出的预测模型与传统预测模型相比，具有更高的预测精度，预测性能提高了 18.01%，是一种有效 的交通流预测方法。 关键词：交通流预测；深度学习；短时交通流；支持向量回归 中图分类号：U491.1 Short-term traffic flow prediction based on deep learning Luo Xianglong, Jiao Qinqin, Niu Liyao, Sun Zhuangwen (School of Information Engineering, Changan University, Xian 710064, China) Abstract: In view of the existing prediction methods fail to fully reveal the nature of the traffic flow, this paper proposed a short-term traffic flow prediction model based on deep learning. The method combined the deep belief network model (DBN) and support vector regression (SVR) classifier as predictive model .The trend of the traffic flow was removed by using data difference, traffic flow features were extracted by deep belief network model, and the traffic flow prediction was carried out with support vector regression in the top level of the network. Experiment results with actual traffic flow data show that the proposed method has a higher accuracy compared with others, prediction performance increased by 18.01%, and it is a effective traffic flow prediction method. Key Words: traffic flow prediction; deep learning; short-term traffic flow; support vector regression [4] 0 引 言 Wang 等人 提出了一种新的贝叶斯组合预测方法，通过基 交通