An LSTM network-based genetic algorithm for integrated procurement and scheduling optimisation

• Published in: International journal of production research Vol. 63; no. 11; pp. 4036 - 4065
• Main Authors: Bubak, Alexander, Rolf, Benjamin, Reggelin, Tobias, Lang, Sebastian, Stuckenschmidt, Heiner
• Format: Journal Article
• Language: English
• Published: Taylor & Francis 03.06.2025
• Subjects: GA: Genetic algorithm; genetic algorithm; hybrid flow shop scheduling; integrated procurement production problem; LSTM: Long short-term memory; MILP:Mixed-integer linear program; OAP: Order allocation problem; OR: Operations research; PCB: Printed circuit board; RNN: Recurrent neural network; rolling horizon planning; supervised learning; Supply chain management; TS: Tabu search; VNS:Variable neighbourhood search
• ISSN: 0020-7543; 1366-588X; 1366-588X
• DOI: 10.1080/00207543.2024.2434948

Modern supply chains are characterised by high complexity, requiring effective management through coordinated activities across interrelated functions. This study aims to move from isolated optimisation to integrated decision-making, which offers new potential for efficiency. We investigate an integrated procurement-production problem based on a real case study from a German company specialising in printed circuit board assembly. We propose a novel solution approach that combines a genetic algorithm with a neural network to increase computational efficiency. Our comprehensive evaluation scheme demonstrates the viability of the approach in generating integrated decisions within a limited time frame. Specifically, we quantify the benefits of integrated over separated decision-making at the operational level, extending previous research focussed on the tactical level. The results indicate considerable benefits of integrated decision-making across a wide range of cost factors, although the exact savings depend on specific cost parameters. In addition, we evaluate our model on a rolling horizon planning basis, which is crucial for modelling realistic supply chain behaviour and remains underrepresented in the literature.