Optimization of a launcher integration process: a Markov decision process approach

TL;DR

This paper models and optimizes an industrial launcher integration process using specialized Markov Decision Processes, resulting in an effective policy that outperforms existing approaches.

Contribution

It introduces a novel application of MDPs to a complex inventory-production system with unique transition and cost features, along with tailored simulation algorithms.

Findings

Optimal policy significantly outperforms reference policies

Simulation-based algorithms effectively handle complex transition laws

Model provides practical solutions for launcher integration optimization

Abstract

This paper is dedicated to the numerical study of the optimization of an industrial launcher integration process. It is an original case of inventory-production system where a calendar plays a crucial role. The process is modeled using the Markov Decision Processes (MDPs) framework. Classical optimization procedures for MDPs cannot be used because of specificities of the transition law and cost function. Two simulation-based algorithms are tuned to fit this special case. We obtain a non trivial optimal policy that can be applied in practice and significantly outperforms reference policies.