Travel Time Prediction: Based on Gated Recurrent Unit Method and Data Fusion

Abstract

Travel time prediction is the basis for the implementation of advanced traveler information systems and advanced transport management systems in intelligent transportation systems. Many studies have shown that the fusion of multi-source data can achieve higher precision prediction of travel time than the travel time prediction based on single source data. In recent years, with the continuous development of China's expressways, traffic detectors such as dedicated short-range communications (DSRC) and remote transportation microwave sensors (RTMS) have been installed on both sides of the road, which provides a basis for the prediction of travel time by fusing multi-source data. At the same times, the deep learning methods show good performance in prediction. So, this paper uses the deep learning algorithm to realize the travel time prediction based on DSRC data and the RTMS data. First, the travel times are, respectively, extracted based on the DSRC data and the RTMS data. Then, both travel time values are input into the gated recurrent unit (GRU) model to obtain travel time prediction results based on multi-source data. Finally, based on the data of the Jinggangao Highway, the accuracy of the algorithm is verified and compared with the traditional data fusion method. The results show that the GRU model can achieve better accuracy of travel time prediction with data fusion.