A Multiple Time Series Approach to Modeling the Manufacturing Job-Shop as a Network of Queues

Abstract

The behavior of a uniformly sampled network of job-shop queues is characterized and interpreted using multiple time series. An analysis of actual queue data for a network of five functionally related machine centers shows that queue behavior is adequately described as a first order vector valued autoregressive model. This model describes the stochastic nature of each queue in the shop and its relationship to the other queues comprising the network. A network of queues obtained from simulating this job-shop under a composite dispatching rule is also modeled and interpreted as a first order, multivariate autoregressive process.