Shop floor production community design: artificial agents based approach

We model automated shop floor production as a multi-agent community. An artificial agent based approach has been developed to improve the design and control of such a community. The resulting community has two types of agents. The first is a cost simulation agent, and the second is an optimization agent whose leaving mechanism is based on generic algorithms. The output from the cost simulation agent is used as input for the optimization agent. We compare the performance of this agent-based approach with infinitesimal perturbation analysis, and find out that the results using the former approach are promising. Finally, we discuss the convergence behavior of the optimization agent. The results shed light on modeling an enterprise as a community of agents.