Discrepancy-based Optimization for Distributed Supply Chain Operations Planning
download the pdf
Year of publication :
GAUDREAULT, J., FRAYRET, J.M., PESANT, G, (2007) Discrepancy-based Optimization for Distributed Supply Chain Operations Planning, Ninth International Workshop on Distributed Constraint Reasoning.
This paper studies the case of a supply chain made up of autonomous facilities. They need to coordinate their manufacturing operations in order to optimize customer satisfaction. Although the coordination space can be described as a tree, simple coordination mechanisms used by industry allow them to visit only the first leaf. We show how they can implement a distributed search in order to evaluate alternative solutions. While chronological backtracking can be easily implemented in a distributed framework (e.g. Synchronous Branch and Bound); it is not the same for other backtracking strategies such as Limited Discrepancy Search (LDS). We therefore propose MacDS, a novel mechanism that allows agents to implement a search strategy based on discrepancies (e.g. LDS or others), while allowing concurrent computation. Use of this mechanism improved the quality of solutions and computation time for both real industrial problems and generated problems.
Member(s) who (co)wrote the paper
| Gilles Pesant || Jonathan Gaudreault || Jean-Marc Frayret |
Project(s) linked to this publication
Files linked to the publication