A Neural Network Approach for Solving Traffic-Flow Forecasting Based on the Historical Voyage Datasets: A Case Study on Hai Phong Roads

Traffic congestion is one of the most common issues in big cities in the world. Therefore, traffic warning and forecasting always play a vital role for traffic participants. Currently, in Vietnam, drivers only know the information and level of congestion through experience and some traffic information channels. Meanwhile, the journey data of vehicles participating in traffic has been stored and transmitted to a management center continuously. These data play an integral part but they are not fully exploited to provide useful information for road users such as average velocity information at each time interval or at each road. In this paper, the authors propose an approach to address this problem. There are two main steps: the first one is to convert raw data to a time series dataset that can provide moving status on each road section. The next step is to use neural networks to make a forecast of average velocity on each road at different times. Experimental results with data on Le Hong Phong and Nguyen Binh Khiem streets (Ngo Quyen District, Hai Phong City, Vietnam) show that the proposed approach gives feasible results that can be applied to many different areas in Vietnam. © 2021, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.