Research And Simulation Of Artificial Community Competition Model

The complex system doesn’t have central control, scientists can use this theory tostudy different disciplines ranging from natural phenomena to the biological,management, economic, military, society and so on, and it has been hailed as “21stcentury science”. The Complex Adaptive System(CAS) is the most representativesystem among complex system, as behavior agent create adaptive survival anddevelopment strategies through studying，which lead to creative evolution of complexadaptive system. The core idea of CAS is “adaptability creates complexity”. Becauseof internal complexity、 uncertainty and the whole behavior nonlinear, traditionaltop-down research methods are not suited to describe the characteristics of the entiresystem. On the contrary, the bottom-up system simulation approach can be a goodsolution to the problem. At present, agent-based modeling is known as the mostdynamic approach of all complex system research approaches.Competition issue is a typical complex adaptive system involving agent diversity,competition target diversity, competition rules diversity and competition factordiversity, so it belongs to the study area of complex system. The current study incompetition generally concrete analysis of concrete problems and model for a specificcompetition phenomenon, research literatures under macro sense are relatively less.At present, there is no competition classification that based on the equality of theinformation that competition agent can attain. However, difference does exist inreality, which expresses as following: there is no communication between competitionagents in the competition where agent can acquire equal reference information. Theyfight and compete with other agents depending on their own strength. However,communication and negotiation exist in the competition where agent can acquireinequality reference information. For example, all agents can acquire equalcompetition intelligence in bidding competition, so there is no communication process.However, the interest competition between supplier and consumer existscommunication because the supplier can get more intelligence than consumer.The agent’s behavior interaction in the complex system is essentially a process ofgame. Therefore, according to the equality of the information that artificialcommunity can attain, artificial community competition is classified into artificialcommunity free competition and artificial community negotiation competition in this article, this classification is based on CAS theory and game theory. We proposedcompetition enhancement learning and history beliefs learning algorithm in the lightof behavior features in free competition and negotiation competition. We utilizecompetition enhancement learning and history beliefs learning to guide agent’sbehavior choice in the competition system. And we also construct an artificialcommunity competition model framework under the guidance of Multiple AgentModeling and Simulation method. The model framework consists of resourceallocation and adding mechanism, competition mechanism, incentive mechanism andlearning mechanism. Artificial community free competition and negotiationcompetition framework are constructed under the basis of competition modelframework. Finally, two example models are constructed to validate the feasibility andcorrectness of the artificial community free competition and artificial communitynegotiation competition model framework. The simulation results from the simulationplatform—SWARM verify the feasibility of the model we construct and theeffectiveness of the learning algorithm. The constructed model frameworks are notonly suitable for the analysis of competition problem under generally sense and reflectthe competition state under common situation, but also can provide valuablereference for competitive agent’s behavior choice.