Date Published: May 23, 2019
Publisher: Public Library of Science
Author(s): Xin Fu, Hao Yang, Chenxi Liu, Jianwei Wang, Yinhai Wang, Jinjun Tang.
Accurate Origin-Destination (OD) prediction is significant for effective traffic monitor, which can support operation decision in traffic planning and management field. The enclosed expressway network system like toll gates system in China can collect mounts of trip records which can be gathered for OD prediction. The paper develops a novel neural network, which is named Expressway OD Prediction Neural Network (EODPNN) for toll data-based prediction. The network consists of the following three modules: The Feature Extension Module, the Memory Module, and the Prediction Module. In the process, the attributes data which can reflect the city attribute such as GDP, population, and the number of vehicles are considered to embeded into the notwork to increase the accuracy of the model. For the applicability improvment of the model, we categorize the cities in multiple classes based on their economy and population scales in this paper, which can provide a higher accurate prediction of OD by EODPNN. The results shows that, comparing to the traditional model like ARIMA and SVM, or typical neural networks like Bidirectional Long Short-term Memory, the EODPNN delivers a better prediction performance. The method proposed in this paper has been fully verified and has a potential to transplant to the other OD data-based management systems for a more accurate and flexible prediction.
The origin-destination (OD) data is the fundamental source of transportation planning and management research no matter in urban or rural road systems . Taking advantage of trip information based on OD data, many studies have significant achievements in traffic flow analysis and demand recognition . In practice, OD prediction can be applied to multiple traffic flow and infrastructure management in expressway such as performance evaluation, investment decision-making, traffic volume balancing, deployment of personnel and resources and congestion mitigating [2,3,4]. However, the attributes of the OD data like continuity and granularity have a high impact on the final result, which indicates a reliable and concise method to collect the OD data is significant.
In this study, we used the enclosed expressway toll data and corresponding attributes data to construct the EODPNN for implementing OD prediction with varied time granularity. From the perspective of application and implementation of the method, such prediction work will help to improve the operation level of the entire road network. In general, the toll amount of China’s expressway usage is determined by the vehicle type (or weight) and mileage, which also indicates that there is a close relationship between the traffic volume within toll gates and the intensity and financial income, so, the OD prediction results with high accuracy and variable time granularity will have irrefutable impacts on the expressway network’s sustainability of finance and maintenance, and have a positive impact on operational and management improvements. The method proposed in this paper can predict the OD flow data of the whole expressway network based on toll data within a certain precision and accuracy range, therefore, it will be more practical and effective than some methods and experimental analytics which may provide prediction results for single certain toll gate or a road segment. In most cases, the toll system of expressway network is based on provinces. In a province’s network, a relatively complete collection system for accumulating toll data has been formed, therefore, the method proposed in this paper can be transplanted into the toll data management system to provide multi-purpose predictions. It will help to develop financial policies for the entire network, including financing and maintenance inputs, and allocate resources within the network with greater efficiency. It can also achieve efficient segment traffic prediction based on OD prediction results, thereby reducing the cost of road segment flow observation and prediction. Other than that, we believe that another noteworthy issue of this work is to provide a large-scale network prediction solution based on distributed data. Although the data set formed by exclosed road networks is relatively rare, with the continuous improvement of survey and detection methods, the distribution data formed by vehicle trajectory data observation or large-scale travel surveys can be continuously obtained, and this method is also applicable for those scenarios. It will also facilitate the establishment of prediction networks based on other types of massive real-time data set and attributes data.
In this paper, we put forward an expressway origin and destination (OD) prediction neural network (EODPNN) based on Bi-LSTM. City attributes and many general features are combined to improve the prediction accuracy. After training and testing EODPNN by using a provincial scale dataset, the results show the EODPNN with a steady and encouraging accuracy and the NN has the ability to predict OD based on various time intervals (15mins, 30 mins, 45 mins and 1 hour) with relatively high accuracy. Also, EODPNN has a bright future of implementation and artificial intelligent technology shows the power in transportation area again. Researchers will carry on improving the accuracy of EODPNN and further do more researches on expressway flow prediction using other advanced neural network.