A stochastic simulation calibration framework for real-time system control

A stochastic simulation model is often used to guide decision making for a complex real system, such as scheduling decisions for semiconductor production. To provide a reliable guidance, we propose a simulation calibration framework. We first develop a spatial-temporal metamodel to estimate the system dynamic behaviors at different settings of calibration parameters. Then, assisted by the metamodel, we introduce a calibration model so that the dynamic behaviors of the calibrated simulation model match with those of the real system. Thus, for any feasible decisions, the calibrated simulation model can predict the future outputs for the real system and deliver prediction intervals.