EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

A Practical Roadmap for Deep Reinforcement Learning in Inventory Control

Ala H. Bani-Irshid, Mohammad M. Hamasha

Journal of Engineering · 2026-01 · 2026 (1)

DOI: 10.1155/je/9751734 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Engineering, online erschienen am 13.11.2026. Verfasst von Ala H. Bani-Irshid und 1 weiteren Autor:innen. Die Arbeit nennt 49 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Deep reinforcement learning (DRL) can support replenishment decisions under uncertain demand, stochastic lead times, and operational constraints, but reliable use requires more than selecting an algorithm. This review presents a practical roadmap linking inventory formulation, forecasting‐informed state design, algorithm selection, benchmarking, robustness, safety, interpretability, and deployment. It clarifies how external time‐series models can provide forecasts, uncertainty summaries, latent demand features, and drift indicators to the DRL state without conflating forecasting with control. The roadmap is evaluated through a real‐data simulation using 739 days of UCI Online Retail II demand data for one retail SKU. Under a matched 50,000‐transition budget and a common forecast‐policy warm start, forecasting‐informed PPO achieved the lowest nominal mean cost (230.28 units/day) while maintaining a 97.45% fill rate; DQN and Dueling DDQN missed the 95% service gate. Under lead‐time stress, PPO retained a 95.35% fill rate and a mean cost of 245.38, whereas the tuned (s, S) policy fell below the gate. These results show that algorithm choice, strong classical comparators, paired uncertainty analysis, and explicit deployment gates materially change the conclusion about DRL readiness.

Zitieren

APA

Bani-Irshid, A. H., & Hamasha, M. M. (2026). A Practical Roadmap for Deep Reinforcement Learning in Inventory Control. Journal of Engineering, 2026(1). https://doi.org/10.1155/je/9751734

Vancouver

Bani-Irshid AH, Hamasha MM. A Practical Roadmap for Deep Reinforcement Learning in Inventory Control. Journal of Engineering. 2026;2026(1). doi:10.1155/je/9751734

BibTeX
@article{BaniIrshid2026,
  author = {Bani-Irshid, Ala H. and Hamasha, Mohammad M.},
  title = {A Practical Roadmap for Deep Reinforcement Learning in Inventory Control},
  journal = {Journal of Engineering},
  year = {2026},
  volume = {2026},
  number = {1},
  doi = {10.1155/je/9751734}
}
Kostenlos starten →