A reinforcement learning approach for a lot sizing and production scheduling problem with energy consideration

Abstract

with climate change, many companies are looking to reduce their carbon footprint and ensure a sustainable manufacturing. To meet this challenge, one of the alternatives is to replace carbon intensive processes with low-carbon processes involving electrical and/or renewable energies. Within this scope, a novel scheduling approach is proposed to take into account the introduction of onsite renewable energy. In particular, a lot sizing and production-scheduling problem in flexible flow line with renewable energy integration is formulated as a versatile optimization model. With regard to associated complexity issues, a multi-agent reinforcement learning approach is advocated to solve the lot sizing and scheduling problem. Finally, the approach is evaluated with a benchmark case and other numerical experiments.