Research And Optimization Of Pedestrian Evacuation Algorithm Based On Complex Network And Game Theory

With the acceleration of urbanization,a large number of people begin to gather in large and medium-sized cities.In the case of emergency,such as fire,terrorist attacks and so on,it will threaten the personal safety of pedestrians and may cause serious social impact,so it is necessary to evacuate a large number of people in a short time.In these cases,panic crowd may lead to stampede,causing serious or even fatal consequences.Therefore,it is of important practical significance in emergency safety management for the panic pedestrian evacuation.Game theory is widely used in the modeling of pedestrian emergency evacuation decisionmaking process.However,there is little research on pedestrian cooperation during evacuation.This thesis establishes a pedestrian relationship network,and proposes a new spatial evacuation model of pedestrians based on the coordination game.Game theory is used to guide the decisionmaking process of pedestrians,and game theory is used to analysis the decision-making process of pedestrians.Moreover,the pedestrian evacuation algorithm in the case of terrorist attacks is designed,and the evolutionary dynamics characteristics of the game in different social dilemmas are analyzed.The main contributions of this thesis are summarized as follows.(1)In order to explore the evolution process of pedestrian strategy and maximize the efficiency of evacuation,a pedestrian space evacuation model is established based on the complex network and game theory.The pedestrian relationship network is established through complex dynamic network,and the pedestrian decision is described by game theory.Considering the strategy evolution of each pedestrian in the evacuation process,the replication dynamics mechanism is used to update the game strategy in the evacuation process.(2)We discuss the impacts of the game scope and the maximum average payoff on evacuation time,which have been missed in the previous work.To maximize the evacuation efficiency,the factors influencing the evacuation time are further investigated based on the dynamic network of pedestrians' game relationships.The impacts of all different parameters,including the panic level,the proportion of initial cooperators,the game scope and the maximum average payoff,are presented.Moreover,pedestrians' cooperative behaviors are investigated along with the coevolution of coordination games and evacuee interactive network.(3)Considering the terrorist attacks,an emergency evacuation optimization algorithm is designed to optimize the evacuation efficiency and reduce casualties based on the game theory and SIS model.The influence of perceived radius on average time steps and casualties is analyzed in four types of social dilemmas.The strategy evolution of panic crowd under the attack of assailants is simulated in four types of social dilemmas.Compared with classical agent-based model,both evacuation time and casualty in our model are less than those in the classical agent-based model when the perceived radius is small.