Time Scale Analysis Techniques for Flexible Manufacturing Systems

摘要

This thesis uses results on the aggregation of singularly perturbed Markov chains to analyze manufacturing systems. The basis for this analysis is the presence in the system of events and processes that occur at markedly different rates - operations on machines, set-ups, failures and repairs, etc. The result of the analysis is a set of models, each far simpler than the full model, describing system behavior over different time horizons. In addition, a new theoretical result is presented on the computation of asymptotic rates of particular events in perturbed Markov processes, where an event may correspond to the occurrence of one of several transitions in the process. This result is used to compute effective production rates at different time scales, taking into account the occurrence of set-ups and failures.