多个体模拟是自底向上的，研究复杂系统的模型化方法。多个体系统由多个拥有自主决策能力的个体组成，这些个体被称为agent。每个agent都可以根据周围的环境和自身的判断规则，独立自主的做出决策。这使得agent模型具有适于描述微观相互作用，适于描述系统适应性的特点，从而成为研究系统宏观集体行为的微观基础的有效工具。在科学研究中，以经典的自驱动粒子模型（SPP）为基础，多个体模型被应用到群体行为各个方面的研究，这其中就包括对群体抓捕行为的研究。由于在群体抓捕行为中个体行为的多样性和特殊性，对这一行为给出纯粹的理论描述是十分困难的。而多个体模拟则为这一困难提供了有力的解决方案，使得可以人为的构建一个群体抓捕行为的模拟环境，从而能够全面综合的观察和理解个体的微观机制对系统宏观结果造成的影响。本文正是从研究两个群体的抓捕和逃避行为入手，研究了群体的聚集性对抓捕效率和群体生存时间的影响。在现实环境中，被捕食者总是倾向于聚集在一起，从而降低觅食的成本，方便求偶和提升生存能力。在逃避追捕行为的时候，成群的被捕食者也可以混淆追捕者的目标，从而提高逃生的概率。从这个角度出发，本文对群体抓捕行为引入了三种聚集策略，即向所有个体坐标矢量平均靠近，向最近的个体靠近，和向所有个体靠近的聚集策略。对不同策略下被捕食群体生存时间的分布变化做了深入的研究。并通过改变捕食者和被捕食者的比例，逐步增加被捕食者的聚集倾向，对比分析了三种聚集策略对群体生存时间和抓捕效率的影响，研究了随着捕食者数量增加被捕食者的平均生存时间的相变行为的特征。还给出了在不同的聚集策略指导下，被捕食者群体行为在典型生存时间下的典型特征，对其宏观表现的具体成因做了具体的分析。研究发现，当被捕食者向所有的个体的坐标矢量平均靠近时，即依据全局的位置信息指导聚集行为的情况下，聚集策略在捕食者数量相对较大时也能明显的提升被捕食者的生存效率。当被捕食者采取向最近个体靠近的聚集策略，即根据局部的距离信息指导聚集行为时，在捕食者数量相对较大的情形下，整体生存效率提升效果不明显，而对于两者数量相当的情形却影响较大，会大大延长群体的最大生存时间。而在引入向所有个体靠近的聚集策略后，即根据全局的距离信息指导聚集行为时，则集合了之前两种策略的共同点，在捕食者数量较多的情况下，可以明显的提升生存效率，对于两者数量相当的情形，也大大延长了整体的最大生存时间。我们的结果说明，聚集策略之所以能够提高被捕食者的生存效率，是由于被捕食者之间的聚集行为为彼此提供了正确逃生信息。而且在被捕食者形成有效的聚集之后，捕食者互相之间的拥塞也是被捕食者生存时间延长的原因之一。我们相信，这些结论能够帮助人们进一步理解自然界动物群体的行为机制，为军事和社会科学的相关应用提供有力的指导。

Multi-agent simulation is a bottom-up modeling approach for the study of complex systems. Multi-agent system consists of multiple individuals with independent decision-making ability, and these individuals are defined as agent. Each agent is an independent decision-making individual based on the system environment and their own rules. The agent-based model is suitable to describe the microscopic interactions and the characteristics of the system, which makes it an effective tool for researching the microscopic basis of the macroscopic collective behavior in a complex system.Based on the model of self-driven particles (SPP), multi-agent model is applied to many aspects of science research, including the group chase behavior. Due to the diversity and particularity of individual behavior in groups, it is very difficult to give a purely theoretical description for group chase. Multi-agent simulation provides a powerful solution for building the simulation environment, which can artificially design a group chase behavior, helping us comprehensively observing and understanding the micro-mechanism and the individual impact to the macro-results.In reality, preys tend to come together to reduce the cost of foraging, facilitate courtship and increase survivability. Collective motion of prey can also confuse the pursuers, and thereby increasing the probability of escape.From this perspective, we introduce three aggregation strategies, getting close to the average of all individuals’ coordinate vectors, getting close to the nearest individual, and getting close to all individuals by distance. By changing the ratio of predators and prey, and gradually increasing the prey aggregation tendency, we analyze the impacts of the three aggregation strategies in survival time and efficiency, and the phase transition behavior of the mean survival time of the prey when the number of predators increases. We also analyze the typical characteristics of the prey’s behavior in the typical survival time under different aggregation strategies, and explain the reasons of the macro-performance.We find that when the prey get close to the average of all individuals’ coordinate vectors, i.e. based on the global position information, aggregation strategy can significantly enhance the Predator survival efficiency even when the predator is larger. When the prey choose the strategy of getting close to the nearest individual, the overall survival time will obviously increase only if the number of the predator is almost equal to that of the prey. According to the global distance information，the strategy of getting close to all individuals play the role both like the former two strategies, whether the predator is equal or more than the prey, the maximum survival time of the targets is significantly prolonged. The outcome indicates that aggregate strategy is able to increase the survival efficiency because the aggregate behavior among each prey offer correct survival information for each other. Besides, the blocking behavior among predators after they have formed effective aggregation is another reason for why their survival time is longer. We believe that these conclusions can help people understand behavioral mechanism of collective motion, and provide relevant guidance for military and social science application.