摘要
准确的短时交通流量预测有利于主动交通控制和出行者的出行规划,文章提出了一种改进的长短期记忆模型(ILSTM)来对短时道路交通流量进行预测。LSTM是RNN的变体形式,在处理时间序列数据上具有优势,所以适合来预测短时交通流量,并通过仿真实验来验证所提方法的有效性,构建了以LSTM为基础的深度学习模型,与其它传统模型支持向量机(SVR),长短记忆模型(LSTM)进行了比较分析,并调整了模型的超参数以分析对模型性能的影响。
Accurate short-term traffic flow prediction is conducive to active traffic control and travel planning of travelers.The paper proposes an improved long short-term memory model(ILSTM)to predict short-term road traffic flow.LSTM is a variant form of RNN,with advantages in processing time series data,which is suitable to predict short-term traffic flow,and can verify effectiveness of proposed method with simulation experiments.The paper constructs deep learning model based on LSTM,and compares it with other traditional models,including support vector machine(SVR)and long short memory model(LSTM),and adjusts super parameters of the model to analyze effect on model performance.
作者
郑友康
王红蕾
ZHENG You-kang;WANG Hong-lei(Electrical Engineering College,Guizhou University,Guiyang,Guizhou 550025,China)
出处
《软件》
2020年第5期72-74,共3页
Software
基金
贵阳市交通运输智能化建设规划资助项目《贵阳市智能公交体系专项规划》(项目编号:2018-1114)。
