Scheduling Shutdowns for Manufacturing Systems with an Application to Automotive Production Lines:Optimization Models and Computation.

Dynamic Programming;End-state;Optimization;Manufacturing;Fictitious Play;Industrial and Operations Engineering;Engineering;Industrial & Operations Engineering

Nicholson, Blake E.Reaume, Daniel J. ;

University of Michigan

关键词: Dynamic Programming;    End-state;    Optimization;    Manufacturing;    Fictitious Play;    Industrial and Operations Engineering;    Engineering;    Industrial & Operations Engineering;   

Others  :  https://deepblue.lib.umich.edu/bitstream/handle/2027.42/61772/blaken_1.pdf?sequence=1&isAllowed=y

瑞士|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

Manufacturing systems are complicated networks of many potentially unreliable parts.The analytical study of such complex systems can prove difficult, if not impossible, and computational models are frequently used to analyze such systems.Through construction of both analytical and computational models, we analyze and optimize the shutdown process of a case study of an automotive production line.We first start with a simple serial line, and analyze the trade-off between goals we would like to achieve to facilitate work during the downtime and the costs associated with lost production or overtime if we do not finish on time.This yields a technique for developing a shutdown policy for a serial line that performs better in a stochastic environment than a rule of thumb modeled after approaches used in practice.Next we consider a hierarchical decomposition of the line which enables the analysis of non-serial lines.We can now consider splits and merges where components of the same job are sent on separate paths for processing, to be merged together later.This approach expands the applicability of the model, can handle a wider variety of goals, and also has superior computational characteristics to the earlier algorithm when applied to a serial line.Finally, we return to the analysis of a serial line, but now incorporate stochasticity into the development of a shutdown policy.We combine the deterministic dynamic program of the earlier chapters with a revised sampled fictitious play algorithm to develop a policy that performs well in a stochastic environment with respect to a representative plant manager;;s objective.