Application of Data Fusion Based on Clustering-Neural Network for ETC Gantry Flow Capacity Correction

Electronic Toll Collection (ETC) gantries are beneficial on highways and feature non-stop charging, improve highway traffic efficiency, and alleviate congestion. ETC gantry data is highly accurate, however, abnormal data detection can occur in some adverse driving environments. To improve the accuracy of flow data, a multi-source data fusion model based on K-means RBF neural network is proposed. A multi-source data fusion is carried out using traffic survey and ETC gantry data. The fusion result is compared with the detection result of the ETC gantry and other fusion methods. Results show the model has adaptive learning characteristics and realizes the high-precision correction of the running state of traffic flow in abnormal environments by learning the historical law of multi-source data which overcomes the problem of data detection of a single detector due to a bad driving environment and demonstrates the potential of multi-source data fusion in the transportation field.