A Prediction Method Based On Bayesian Network Of Road Congestion,

Traffic congestion, especially with regard to those occur in urban area, is a worldwide problem. The serious contradiction between supply and demand, namely the capacity of our roads which cannot meet or not always meet the heavy traffic needs in those areas is the root of the traffic congestion. What's worse, the improper management always aggravates this situation. Therefore, it's really very urgent to upgrade the urban traffic management. Those short time solutions, the unprofessional suggestions and the incomplete management have to be improved. Consequently, traffic flow will ultimately to be predicted if the dynamic, sophisticated and intelligent multi-spectral collaboration on traffic management is realized.It's usually not that difficult to find out the shortest way with the help of the navigator. But unfortunately, the best shortcut is not always the shortest way for us. It would be so hard to find the best short cut because no one can predict the congestion situation on a certain period of time accurately.As for these urgent practical problems, this paper proposes a kind of method to predict the traffic congestion situation in some certain region over any certain period of time effectively. This system will identify the various possible congestions in all sub-regional in a pointed area by making those data in actual road traffic management into application as well as the route planning. In this method, the road will be divided into several sub-regions firstly. Then the historical statistics on each sub-region will be calculated. Those data, including the topology model of the actual road and the capacity of those driving directions will be combined together to build up the Bayesian Networks. After that, the current traffic flow data of the pointed region will be regarded as the presupposition of the established Bayesian network. Then the traffic flow conditions in the following period of time can be speculated in every sub-region. When combined the speculation with the fixed design of the road capacity in every sub-regional, it will be easy to find out whether there will be traffic congestion in the following time or not. Finally with the help of DBSCAN Algorithm all data on traffic congestions will be put together and illustrate us the whole image of the traffic situation in the following period of time.This method is effectively according to the results of the experiment in a certain area, Panlong Region, Kunming.