Short-Term Traffic Flow Uncertainty Prediction Based on Deep Kernel Adaptive Interval Grey Model

To better capture the uncertainty in the random fluctuations of traffic flow and improve the accuracy of short-term traffic flow predictions, this study proposes a Deep Kernel Adaptive Grey Interval Model for uncertainty prediction of short-term traffic flow. First, an adaptive grey model is constructed to predict the mean of short-term traffic flow, and the optimal parameters of the model are obtained in real time using the Hogon Swine Optimization Algorithm. Then, a deep kernel learning module is employed to predict the residual values. Finally, the mean prediction results are combined with the residual results to generate a prediction interval, thereby quantifying the uncertainty of short-term traffic flow. The proposed model is evaluated using traffic flow data from UK highways, and the results demonstrate that the model can generate feasible traffic flow prediction intervals with improved prediction accuracy.