| Summary: | 碩士 === 國立東華大學 === 電機工程學系 === 91 === The purpose of complex science is to analyze the principles of Complex Adaptive System (CAS) in natural ecosystem, economic, etc., and wishing to attain the accurate control and predict the behavior of CAS. Because of many factors, such as time, money and morals etc., it couldn’t experiment on the real-world CAS for proving the theories. Therefore, it is very important work for establishing CAS simulation model on computer to assist CAS research. In general, it depends on domain expertise to establish CAS simulation model. The jobs are tedious and time-consuming to create the exactitude and integrity model.
This paper proposed a new method to establish CAS model by using fuzzy system auto-construction modeling technique through learning to substitute for expert experiences, and perform an agent-based CAS simulation model. By the way, a CAS complicated system’s behavior originates from the system members’ evolutionary process, and the evolving members can be regarded as adaptive process for the CAS world. These features cause CAS present dynamic behavior of self-evolution. The incorporating of evolutionary strategy (ES) in CAS simulation model can simulate the evolutionary behavior of real member via corresponding agents, and be able to imitate the real member of CAS system to evolve the behavior. This paper implements an Artificial Complex Adaptive System (ACAS) for aiding the research of complex science, and generate a grass ecosystem to prove our proposed method. The experiment results prove that the method is really right.
|
|---|
