MGCP: A Multi-View Diffusing Graphs Based Traffic Congestion Prediction for Roads Around Factory

Traffic congestion easily occurs on the roads around factories due to limited road space, which will be worsen especially when many trucks stay at the roadside of the same road. Actually the truck’s movement or staying at the road depends on the phase of its implementing cargo-loading task, e.g., the truck may strand in the road outside the factory due to waiting for loading cargoes, or move toward the road near the gate of factory when receiving loading notification. Thus, to predict traffic jams of the roads around the factory, it is necessary to consider the influence of the truck’s cargo-loading task phase on road traffic situation. However, the influences of different task phases that the trucks are on traffic situation of the roads are not the same, and such influences may change with the transition of the trucks’ task phases, which brings severe challenges for precise prediction. In this paper, we put forward a multi-view task diffusing graphs based traffic congestion prediction method for Roads around Factory, called MGCP. To capture discrepant impacts of task phases on traffic situations of the roads, we present a task phase based multi-view diffusing graphs generating method. In addition, we leverage a Markov process to denote the transition of each task phase, and build a transition matrix of task phases to extract the transited task phases in the future. Experimental results on real steel logistics data sets demonstrate that our proposed method outperforms the existing prediction approaches in terms of prediction accuracy.