Short-term Vehicle Speed Prediction Based on Convolutional Bidirectional LSTM Networks

The optimal control based on the forecast of vehicle speed in the future is of great significance to vehicle safety system and energy management systems of hybrid electric vehicles. In this brief, a new vehicle speed prediction approach combining one-dimensional convolutional neural network with bidirectional Long Short-term Memory network (CB-LSTM), utilizing the information provided by V2V and V2I communication. Convolutional neural network (CNN) is used to receive input data and extract important features of the data, and bidirectional Long Short-term Memory network (Bi-LSTM) is used to receive the output of CNN layer, extract time series features, and produce final prediction results. The simulation results show that the prediction error increases with the increase of the prediction horizons, and the number of past values used in CB-LSTM has a certain impact on the prediction accuracy. Compared with the classical BP network, CB-LSTM has significantly improved the prediction accuracy for short-term vehicle speed prediction.