Multiple Simulation Response Surfaces for Robust Optimization In Inventory System

Computer simulation is a powerfil tool in evaluating complex systems. These evaluations are usually in the from of responses to "what if" questions. Simulation optimization provides a structured approach to system design and configurations. Simulation optimization provides a structured approach to system design and configuration when analytical expression for input/output relationships are unavailable. Nondeterministic inputs and their relation will results evaluation. in this paper we propose a robust optimization method based on multiple response Surfaces (MRS) to optimiza probabilistic inventory models with multiple performance objectives. MRS optimization as a branch of experimental design is useful procedure to find a series of regression models to the responses of the simulation ,odel evaluated at several points and trying to optimize the resulting regression functions. Then a numerical example on an inventory system has been developed and its results have been compared with evolutionary algorithms to show the accuracy of the proposed method.

Keywords:

Experimental Design. Robust Optimization , simulation , Inventory Modeling , Multiobjective Optimization