Short-Term Traffic Flow Prediction Based on Hybrid Model

Abstract

AbstractAccurate and reliable short-term traffic flow prediction can provide effective help for people’s travel and road planning. In order to improve the accuracy of short-term traffic flow prediction, this paper proposes a hybrid model of improve long-term short-term memory (LSTM) and radial basis function neural network (RBFNN). Firstly, according to the temporal and spatial characteristics of traffic flow, LSTM and RBFNN models are constructed. Then, by adding the percentage error term to balance the loss function of the LSTM, and an improved LSTM (ILSTM) is proposed. Finally, the prediction results of these models are weighted by the Entropy method to obtain the final result. The experimental results show that the ILSTM-RBFNN model can achieve higher prediction accuracy compared with traditional models.KeywordsTraffic flow predictionLSTMRBFNNEntropy method