A discrete event model for multi-agent real-time process control is introduced in the context of United Airlines' Operations Control Center (OCC). The model represents the OCC agents as a network of queuing servers, and organizes the individual communication and decision tasks into discrete event processes flowing through this queuing network. The input to the agent network is the superposition of a deterministic periodic demand, a stochastic demand and an uncertain forecast of future demand. The identification of the dynamic characteristics of each agent in the network is carried out through a combination of on-site measurements and statistical analysis of transactional data. A computer simulation of this OCC model is presented. © 1998 AACC.
|Original language||English (US)|
|Title of host publication||Proceedings of the American Control Conference|
|Number of pages||4|
|State||Published - Dec 1 1998|