Abstract: For frequently congested sections of an expressway, this paper proposes a short-term traffic congestion prediction model combining a fuzzy C-means (FCM) clustering algorithm and a radial basis function (RBF) neural network. In the model, the FCM clustering algorithm is used to obtain the congestion status labels of historical traffic flows and clustering centers of different traffic states, and the RBF neural network algorithm is used to predict short-time traffic flow parameters. The short-term traffic flow parameters predicted by the RBF neural network are brought into the result of the FCM clustering algorithm to obtain the short-term traffic congestion status label. In this way, the basic computational architecture of the fusion model is built through the implicit relationship between the traffic flow parameters and the traffic state. The results show that the classification results are more stable and effective after being trained by the FCM clustering algorithm. The RBF neural network has higher prediction accuracy than the control methods, and the relative error of the prediction results is basically less than 1.2%. The classification accuracy of the short-term traffic congestion state predicted by the FCM-RBF model reaches 95%, demonstrating the accuracy and reliability of the proposed model.