Research of Traffic Flow Forecasting Based on the Information Fusion of BP Network Sequence

• Published in: Intelligence Science and Big Data Engineering. Big Data and Machine Learning Techniques Vol. 9243; pp. 548 - 558
• Main Authors: Zhang, Wei, Xiao, Ridong, Deng, Jing
• Format: Book Chapter
• Language: English
• Published: Switzerland Springer International Publishing AG 2015; Springer International Publishing
• Series: Lecture Notes in Computer Science
• Subjects: BP neural network; Graphics programming; Image processing; Information fusion; Pattern recognition; Traffic volume forecasting
• ISBN: 3319238612; 9783319238616
• ISSN: 0302-9743; 1611-3349
• DOI: 10.1007/978-3-319-23862-3_54

Traffic flow forecasting is an important aspect of the ITS as accurate traffic predication can alleviate congestion, save traveling time and reduce economical loses. The forecasting process may rely on historical data, current data, or both, to forecast the traffic volume in the future. In this paper, we compare three different approaches in traffic forecasting, study the input data and output data for these approaches, as well as some general insights, and also propose BP neural network to estimate accurate traffic flow for a roadway section. By means of three layers-BP neutral network model, in which mechanism algorithm are used to preprocess the multi-source data, error data is eliminated, multi-source data fusion is realized and accurate traffic forecasting is achieved.