Agent Based Process Simulation: Modelling Complex Systems Using an Hybrid Methodology

摘要

Process based enterprise simulations can be efficiently applied to what-if analysis and business process re-engineering, while agent based paradigms are effective when used to model complex social systems. Though, neither of the two can be successfully used to create enterprise simulations in which both the human factor and the structure are relevant and dynamically linked. A hybrid approach is thus presented in this paper: it uses prepositional logic to model the deterministic structures acting as a background, while it employs evolutionary methods derived from the research on artificial intelligence to create agents capable of learning and self organizing. With this technique it becomes possible to find the optimal organization for a given enterprise and realistically model various situations in which deterministic structures, e.g. machineries, are operated by human beings and affected by exogenous factors. Some operational examples are discussed, and a building methodology is presented.