A suboptimal class of decentralized hedging production policies in single part-type, stochastic M-machine flow shop

摘要

Production optimization is considered for stochastic single-part M-machine flow shops over the class of so-called decentralized hedging production policies. A decomposition methodology previously developed for tandem two-machine systems isgeneralized to M-machine flow shops. Using a machines decoupling approximation and a so-called demand averaging principle, the considered M-machine flow shop is decomposed into a set of M decoupled single machine systems with known analytical productioncosts. The overall production cost obtained by adding individual costs is parameterized by M critical levels of parts to be determined. Subsequently, the best decentralized hedging production policy is found by minimizing the overall analytical cost overthe hedging levels space. The predicted performance of the proposed class of decentralized policies is tested and validated on a number of three-machine flow shops with the help of Monte Carlo simulation techniques. Diverse results obtained indicate thatthe new hierarchical optimization scheme proposed runs very quickly, has a good performance, and appears to be very competitive for production scheduling in unreliable tandem manufacturing systems.