Design Of Enterprise Intelligent Agent Based On The Norms

With the development of enterprise information project, the economic environment is more uncertain, and the competition between enterprises are increasingly fierce. The enterprise which can be looked as complicated system must learn and evolve continuously in order to survive and develop. The evolution of the enterprise becomes a focused topic in management region. In order to study the evolution characteristic of enterprises, a reasonable enterprise model is usually established for the enterprise system. Limited to their structure and operation mechanism, the current enterprise models that are widely used are rigid to some degree. Agent-Based modeling and simulation (ABMS) is considered as an efficient approach to research for complex adaptive system, which is regarded as the most active methodology for researching complex systems as well.This paper build the agent model under the direction of norm which comes from Organizational Semiotics. It helps to describe subjective initiative of agent as well as enhance its intelligence. The evolution of enterprise mainly manifest in the dynamic decision process of enterprise agent. And the implementations of dynamic decision depends on how the agent finding and using knowledge.Based on analysis of agent, norm and so on, this paper combine norm and agent-based modeling method while analyzing the basic structure of enterprise agent model. And an example of application is presented at last. The main research results and the innovations in this paper as follows:First, this paper designed the structure of enterprise intelligent agent model combined with norm and agent-based modeling method by analyzing norm which comes from Organizational Semiotics.At the same time, this paper proposed a kind of design pattern of norm module, named RHN which is short for Requst Handler Norm.Second, the implementation of each module is described. In particular, this paper studied and designed the agent intelligent decision mechanism. A kind of intelligent agent decision model based on reinforcement learning was proposed. At last, the structure of agent model which are proposed before was verified feasible by the simulation experiment based on Swarm.