Gradient/sensitivity estimation in discrete-event simulation
WSC '93 Proceedings of the 25th conference on Winter simulation
Likelilood ratio gradient estimation: an overview
WSC '87 Proceedings of the 19th conference on Winter simulation
Single run optimization of a SIMAN model for closed loop flexible assembly systems
WSC '87 Proceedings of the 19th conference on Winter simulation
Simulation optimization: methods and applications
Proceedings of the 29th conference on Winter simulation
Simulation-optimization using a reinforcement learning approach
Proceedings of the 40th Conference on Winter Simulation
ISNN 2009 Proceedings of the 6th International Symposium on Neural Networks: Advances in Neural Networks - Part II
Hi-index | 0.00 |
This paper deals with stochastic optimization of a discrete-event simulation model for the solution of a manufacturing system operation problem. Gradient estimates are obtained by the application of the infinitesimal perturbation analysis (IPA) technique. We begin with background material on stochastic approximation (SA) and the IPA technique, their potential value in finding optimal solutions to manufacturing system operation problems, and limitations concerning their applicability. Next we present our attempt to solve a real problem (the design of a partially-automated assembly line in an electronics manufacturing facility) using this approach. A sequence of models is described moving from one which embodies some restrictive assumptions through to models which more closely approximate the real system. All of the models are implemented in the SIMAN IV simulation language incorporating user-written code (written in C++) implementing the SA and IPA algorithms. We report and interpret the results obtained with the different models and close with concluding remarks on the current value of this technique in solving this kind of system design problem.