关键词: 智能交通, 短时交通流预测, 变分模态分解, 长短时记忆神经网络, 深度学习

Abstract: Traffic flow has the characteristics of nonlinearity, volatility and randomness. In order to further improve the prediction accuracy of short-term traffic flow, this paper proposes a short-term traffic flow prediction method based on variational modal decomposition (VMD) and long and short-term memory (LSTM) neural network. Firstly, VMD is used to decompose the original traffic flow data into k stable Intrinsic Mode Functions (IMF). Then, LSTM models are input into each modal component for prediction, and all of the predicted values are summarized and superimposed to obtain the final traffic flow prediction results. Based on the induction coil data of Shanghai North-South Elevated Expressway, the results show that the prediction result after VMD decomposition is more accurate. Compared with the prediction results of BPNN, LSTM, EMD-LSTM and EEMD-LSTM, the mean absolute error (MAE) is optimized by 35.5%, 28.25%, 21.1% and 13% respectively, showing a high prediction accuracy.