Simulation modeling decision support through belief networks

This paper presents an automated approach aimed at optimizing simulation model performance by means of intelligent agents and belief networks. The method described embeds intelligent agents in simulation models to conduct ‘simulation real-time’ model evaluation. These agents can be considered ‘intelligent observers’ placed within a model that inspect and make real-time decisions regarding the overall model performance. The knowledge encapsulation for the agents is provided via belief networks, which allow the agents to record their observations and make inferences. To exemplify this approach, a general-purpose resource allocation model describing an earthmoving operation is provided.