Early-waring Model Of Traffic Risk Of Large-scale Activities Based On Dynamic Bayesian Network

With the development of China's social and economic level and the increasing richness of people's daily life,large-scale activities are held more and more frequently,which will cause a large-scale gathering of people and vehicles,pressure on the surrounding road traffic,and also face a huge traffic risk.Aiming at this phenomenon,this paper takes the main traffic risks of large-scale activities as the research object,constructs a dynamic bayesian network(DBN)traffic risk early-warning model,and realizes the function of traffic risk element identification and risk early warning.Based on the traffic characteristics of large-scale activities,this paper defines the large-scale activities and the affected areas;and analyses the hazard sources of traffic risks of large-scale activities,based on the research of traffic hazard sources,the traffic risk of large-scale activities is mainly divided into two categories: traffic congestion risk and crowd stampede risk,and the risk warning threshold is determined.The whole work flow of traffic risk early warning based on DBN model is determined.Combined with five hazard sources,the traffic risk elements are analyzed,and the fault tree(FT)models of traffic congestion risk and crowd stampede risk of large-scale activities are constructed respectively.Considering the polymorphism of some elements,the static Bayesian network model is constructed.Furthermore,this paper analyzes the influence of nodes' changes with time on traffic risk in Bayesian network,selects dynamic nodes,and then transforms them into DBN model.For the application scenario of 2022 Winter Olympic Games,the model is simplified and adjusted to establish a DBN early warning model of traffic congestion risk during the Winter Olympic Games in Beijing area.Based on the improved BWM method,the conditional probability allocation method is proposed to calculate the prior probability of the root node,the conditional probability table of the other node and the state transition matrix of the dynamic node,the probability distribution results of the DBN model are given;the model structure is constructed by using Ge NIE software,the sensitivity analysis of the static Bayesian network model is carried out,the main factors affecting the traffic congestion of large-scale activities are analyzed,and the impact of large-scale activities on the probability of traffic congestion is mainly analyzed;the early warning function of the traffic risk is realized by use the DBN model.The results show that the dynamic Bayesian network model can realize the realtime warning function of traffic risk,analyze the impact of various risk factors on traffic risk,and the model can reflect the specific impact of large-scale activities on the surrounding traffic risk,and provide the basis for the relevant management decisionmaking.