Property Analysis And Judgment Method Of Emergence In Swarm Model

Emergence, as one of the most important properties of complex system, makes a way for complex system to produce its spatial or temporal function. At present, many researchers try to find the generating mechanism of emergence through varies methods. But there are no systematic methods for emergence analysis. It is a compact way to analyze the cause and condition of emergence by establishing appropriate computer model and observing the evolution process in emergent behavior. Swarm model is established on the basis of nature cluster phenomenon in animals. With simple rules and various emergent behaviors, it is provided a simple way for emergence analysis in complex systems.In the thesis, with kinetics modeling method, systemic definition of Swarm model is given, and the kinetic property and chaotic property during emergence process are analyzed. On the one hand, mathematic method is used to analyze the kinetic property in swarm model. It is proved that under certain initialization, the mean square deviations of each kinetic parameter would fall into certain value space. Experimental results show that under certain conditions, by setting the initial state and parameters of the control equations, the behavior of Swarm can be effectively bounded. Then based on the kinetic property of Swarm model, a kinetic method is elicited to judge emergence. It is validated by experiment that the method can determine whether it is emergence or not, and calculate the critical point of emergence. On the other hand, the thesis calculated the Kolmogorov entropy and correlation dimension of each period during the evolution of Swarm system. It is found in the process of Swarm emergence, the system is getting orderly.According to characters of emergent system, the thesis collects all foregoing works. Then all metrics are assorted into three kinds: complex network metrics, informatics metrics and kinetic metrics. Experiment is designed to measure efficiencies of the metrics, to find that different metrics perform vary in different kinds of systems.In the end, Swarm behavior and analysis system is designed, including behavior exhibition module and data analysis module. Testing show that the system can not only exhibit all kinds of swarm behavior, but also analyses data list produced in Swarm model and other models.