Quantitative Research On Emergent Behavior In Swarm Emergent Computation Model

Emergent behavior exists in many natural, social and other complex systems. In the complex science field, research on the emergent behavior in complex adaptive systems has become a hot topic. Many researchers have been aware of the research of emergent properties will be conducive to the development of complex systems. However, research carried out in this area has just started, and most researches are concentrated on the summary and the concluding in theory, lacking of quantitative conclusions and knowledge interpretations. Therefore, how to build a systematic research method to explain the inherent laws of the emergent behavior in complex systems and make out a reasonable scientific quantitative analysis method is a problem calling for immediate solution in emergent computation field.On the basis of summarizing demestic and abroad research status of emergent computation, the thesis analyzes the current difficulties and focus existent in the area, and focuses on the studies of the scientific explanation to the inherent variation in emergent behavior.Currently, since there are no perfect quantitative assessment metrics to evaluate the inherent information of emergent behavior, firstly, a series of metrics that can be used to evaluate the inherent characteristics of emergent behavior is choosed and analyzed. With the emergent behaviors in swarm emergent computation model as its research object, according to the characteristics in the model, the quantitative problems of emergent behavior in Swarm model are analyzed qualitatively. Then, combining quantitative assessment metrics in related traditional disciplines with metrics analysis in Swarm model, the thesis gives out the selection basis and calculation method of the metrics, and uses them for assessment and analysis in swarm model. At the same time, the thesis also analyzes the general application of part metrics.Secondly, according to the complexion of the inherent information of emergent behavior and the difficulties in finding the potential rules, a series of systematic quantitative analysis methods for emergent behavior in swarm model is proposed. With data mining based on rough set, the thesis studies the kinetic parameters and quantitative assessment metrics of the model to obtain rules from different granularity and fuse them. Finally effectively finds out the key factors that play a decisive role in different emergent environment and gets the key rules used to determine critical state of emergence and different emergence behaviors. The method puts a new thought into the emergent behavior research of complex systems.Lastly, a swarm-based prototype system for quantitative analysis of the emergent behavior which contains above contents is designed and established.