Prediction Model Of The Load Of Urban Traffic Sites During Snow Disaster

Urban transport system is facing new pressures with the accelerating process of urbanization and theincreasing traffic, especially in recent years, unconventional incidents which give a more serious impactto the transportation system have occurred frequently, such as in the case of snow disatster, transportcapacity of urban traffic sites become weaker dramatically, which results that passengers get stranded,and the load of traffic sites increases significantly. In emergency situations, how to evaluate the pressuredistribution of the transport system, especially the load of traffic sites, has become a new hotspot anddifficult problem.Based on the principle of Maximum Entropy and Bayesian network theory, the load of urban trafficsites during snow disaster is studied deeply in this paper.Firstly, classification and characteristics of snow is described, and the components of urban transportsystem is analyed, then the impact on site traffic analysis by snow disaster is studied, including roadtransport, railway transport, air transport and sea transport.Secondly, a new prediction model of load of urban traffic sites during snow disaster is presented. Themodel overcomes the shortcomings of traditional Markov Chain method. Considering the dynamicchanges of passengers’ preferences and the interaction of traffic sites, the author establishes themaximum entropy model of distribution of passengers during snow disaster, and on this basis a methodof calculating the load rate of traffic sites is presented, with which the dynamic changes of the load oftraffic sites can be deduced. This provides a reference for the relevant departments to make responsemeasures.Thirdly, based on the Maximum Entropy prediction Model of the load of urban traffic sites duringsnow disaster, considering the impact of government measures and other uncertainties, using theknowledge of Bayesian network, the author established a amendment prediction model of the load ofurban traffic sites basedon the Bayesian network. The model can amend the prediction results accordingto the update changes of some uncertain factors.Finally, a case of south snow disaster occurred in2008is studied, the models which are provided inthis paper are used to predict the load rate of Guangzhou Railway Station during the snow disaster, theprediction result is more ln line with actual data, which indicated that the models have good predictionaccuracy.