点击次数：

论文类型：会议论文

发表时间：2017-05-01

收录刊物：EI、CPCI-S、Scopus

页面范围：1010-1011

摘要：This paper presents a deep learning based time series model to predict the traffic flow of transportation systems, DeepTFP, which exploits the effectiveness of time series function in analyzing sequence data and deep learning in extracting traffic flow features. Accurate and timely prediction on the future traffic flow is strongly needed by individual travelers, public transport, and transport planning. Over the last few years, with the exploding of traffic data, various big data analytics based methods have been proposed to predict the traffic flow. However, it is hard to provide timely prediction by processing real-time traffic data. This paper proposes DeepTFP, which conducts the prediction with a time series function which considers the spatial and temporal correlations of traffic data to track the changes of traffic flow, and DeepTFP uses deep learning to extract the feature of traffic data as the basis of the time series function. Contrast experiments are used to demonstrate the performance of the proposed model.