Inverse discrete event modeling for facility parameter estimation

Abstract

Particular applications require analysts to estimate plant throughput from external observables via inverse modeling techniques. For example, auditors, law enforcement personnel, and financial planners might need to perform these types of analyses. Researchers at the SimCenter at The University of Tennessee Chattanooga have elected to model several simple basic production models as well as a fictional bicycle factory to do a preliminary investigation into the viability of implementing an inverse model using a discrete- event simulation software package. The fictional bicycle model will eventually include several simulation features such as a discrete event component, a flow portion, an agent based part, equation based power portion, and optimization. The results indicate that the approach is viable and that inverse modeling can be used to estimate internal activities. Future work will involve more detailed models with larger parameter sets.