USING ARTIFICIAL NEURAL NETWORKS TO APPROXIMATE A DISCRETE EVENT STOCHASTIC SIMULATION MODEL

A computer simulation model may be regarded as a stochastic function that maps a set of inputs to a set of outputs; in many cases computer simulation models are quite computationally expensive. It would be beneficial to have fast, accurate approximations of computer simulation models to perform such tasks as quick turnaround studies, sensitivity analyses, model aggregation/reduction, and simulation optimization. This paper examines the use of two methods, artificial neural networks (ANN) and multiple linear regression, for approximating a lot size reorder point inventory system simulation.