| With the rapid changes in the global economy and social life,various emergencies occur frequently.Among them,the incidence of emergencies such as violent terrorism and the spread of epidemics has shown a rapid upward trend worldwide,which has had a serious impact on the country’s economic development,social harmony and stability,and people’s lives and health.In response to these problems,this thesis proposes two crowd behavior simulation models,namely antagonistic crowd behavior simulation model and crowd behavior control simulation model,which solve two types of similar emergencies.In-depth analysis and study of group behaviors under emergencies through crowd simulation and reasonable management and control can provide theoretical support and decision-making basis for formulating solutions to similar incidents,thereby quelling incidents as soon as possible and maintaining social harmony.Aiming at the research on antagonistic crowd behavior,this thesis takes the sudden group riot event as an example,combines emotional contagion and deep reinforcement learning to build an antagonistic crowd behavior simulation model.First of all,this thesis fully considers the influence of emotions on group behavior.We build an emotional infection model based on the improved SIS model combined with specific combat situations,and use this model to calculate the emotions of combat individuals.Then,we combine deep reinforcement learning and mean field theory to build a network model,and the agent learns combat behavior efficiently and autonomously through neural networks.Finally,we analyze whether the learned behavior is reasonable according to the emotion of the agent,and determine the final combat behavior of the agent according to the behavior rules under different emotions.Experiments have proved that this method is helpful to study the antagonistic behavior of both sides in the riots,can formulate a real and reasonable combat plan for the square group,and improve the combat victory rate.Aiming at the research on crowd behavior control,this thesis takes the intra-school epidemic incidents that may be caused by infectious viruses after students return to school during the outbreak of the epidemic as an example,and proposes a simulation of crowd behavior management and control that is oriented to a closed environment,frequent crowd contact,and strong mobility.model.Against the background of the new crown pneumonia epidemic that occurred at the end of 2019: First,the thesis model the scenes and individuals,and simulate the daily movement of college groups based on social forces.Then,we build an Agent-based virus infection model,taking into account the heterogeneity between individuals and the characteristics of virus transmission in colleges,and simulate the situation of virus infection in college groups from a micro perspective.Finally,we build a crowd behavior control model based on the simulation results,formulate control measures such as batch trips,staggered trips,and mandatory isolation,and plan the group’s movement behavior during the epidemic prevention and control period.Experiments have proved that this method is helpful for studying the spread of viruses in semi-enclosed areas and high-density populations,and the thesis proposed crowd behavior control measures can reduce the virus infection rate in college.The method proposed in this thesis fully considers the key factors that affect the development of emergencies and group sports behaviors,such as group emotions,infectious viruses,etc.The experimental results show that analyzing the transmission mechanism of these factors between groups and establishing an accurate and applicable infection model are helpful to study group behaviors in emergencies and formulate practical and reasonable response plans,thereby protecting personnel safety,reducing resource consumption,and maintaining social stability. |
