AUTOMATION OF SIMPLIFICATION IN DISCRETE EVENT MODELING AND SIMULATION

This article discusses simplification in the context of discrete event modelling and simulation. An attempt is made to automate simplification. To achieve this, the definition of homomorphism for discrete event models is weakened to be used as a formal basis. In this approach, in addition to the models themselves, the modeller specifies a set of functions which map the input, output and state sets of a model into their corresponding simplified images. Our method uses this information along with the histories of original models to identify simplified equivalents of the original models. The concepts are implemented in DEVS-Scheme, a general purpose theory-based event simulation environment. The approach uses observation data to generate lumped model code. For experimentations, the lumped model codes are substituted in place of their original models and simulations are conducted to compare the behaviors of lumped and original models. The results are reported. The article also contains a section on future directions of simplification research. © 1990, Taylor & Francis Group, LLC. All rights reserved.