Bayesian Network Model for the Prediction of Traffic Incident Duration

Traffic incident is one of the main factors that lead to traffic congestions.Through controlling methods such as real-time traffic guidance,its impacts on traffic operation can be reduced.Accurately prediction of traffic conges-tion duration is a prerequisite for effective traffic control.Based on MIT scoring functions,an S-ACOB algorithm as the core of the Bayesian network model is developed.The networks are generated from top to bottom with an ant colony algo-rithm searching for the optimal network structure.To increase the robustness of the proposed Bayesian network,a ran-dom selection mechanism for the nodes and a partial probabilistic selection model for the local structure are introduced. Through an empirical study and comparative analyses,the average precision is up to 87.82%,which is superior to the al-ternatives reported in the previous research.regarding those nods with the complete and incomplete node properties,the accuracy of the network prediction model is up to 76.97% and 93.23%.The results show that this model can effectively predict the duration of traffic congestions.