Optimizing Adaptive Supply Chain Strategies Uncertain Demand Planning through Deep Learning-Based Reinforcement Models
JOIV : International Journal on Informatics Visualization · 31.05.2026 · 10 (3) · 1388
DOI: 10.62527/joiv.10.3.4548 ↗Kurz gefasst
Fachartikel in der Zeitschrift JOIV : International Journal on Informatics Visualization, online erschienen am 31.05.2026. Verfasst von Lena Magdalena und 5 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY-SA 4.0).
Abstract (Originalsprache)
The integration of artificial intelligence into agricultural supply chain management offers new opportunities to enhance decision-making through data-driven optimization. This study applies Deep Reinforcement Learning (DRL) to optimize distribution and sustainability strategies within a corn supply chain using real-world production, demand, logistics, and environmental data collected from West Java, Indonesia, over a defined observation period. Two DRL algorithms, Proximal Policy Optimization (PPO) and Deep Q-Network (DQN), are developed, implemented, and systematically compared to evaluate their performance in terms of reward, revenue, distribution costs, waste, and emissions. The empirical results indicate that PPO achieves an average reward of 201.03 with the highest revenue of approximately 54.1 million, albeit at the expense of higher distribution costs (≈5.04 million) and emissions (2,520). In contrast, DQN attains a slightly higher reward of 202.36 with lower revenue (≈51.5 million), while demonstrating superior efficiency in reducing costs (≈1.2 million) and emissions (600). Notably, both algorithms completely eliminate waste, highlighting the potential of DRL to support sustainable agricultural supply chain operations. Further analysis of distribution patterns reveals that PPO provides a more balanced allocation across major demand regions (Jakarta, Bandung, and Cirebon), ensuring equitable regional coverage, whereas DQN adopts a more concentrated but cost-efficient allocation strategy. These findings suggest that PPO is particularly effective in maximizing profitability and maintaining distribution fairness, while DQN offers advantages in minimizing operational costs and environmental impacts. Overall, this study confirms DRL's capability to manage complex trade-offs between economic and sustainability objectives in real-world agricultural supply chains and provides a foundation for future research on advanced or hybrid DRL approaches.
Zitieren
APA
Magdalena, L., Hatta, M., Suwandi, S., Ardiningrat, R. M. R. P., Kuraesin, S., & Dewi, A. S. N. (2026). Optimizing Adaptive Supply Chain Strategies Uncertain Demand Planning through Deep Learning-Based Reinforcement Models. JOIV : International Journal on Informatics Visualization, 10(3), 1388. https://doi.org/10.62527/joiv.10.3.4548
Vancouver
Magdalena L, Hatta M, Suwandi S, Ardiningrat RMRP, Kuraesin S, Dewi ASN. Optimizing Adaptive Supply Chain Strategies Uncertain Demand Planning through Deep Learning-Based Reinforcement Models. JOIV : International Journal on Informatics Visualization. 2026;10(3):1388. doi:10.62527/joiv.10.3.4548
BibTeX
@article{Magdalena2026,
author = {Magdalena, Lena and Hatta, Muhammad and Suwandi, Suwandi and Ardiningrat, Raden Muhamad Rama Poetra and Kuraesin, Sinsin and Dewi, Ade Silva Nurfitria},
title = {Optimizing Adaptive Supply Chain Strategies Uncertain Demand Planning through Deep Learning-Based Reinforcement Models},
journal = {JOIV : International Journal on Informatics Visualization},
year = {2026},
volume = {10},
number = {3},
pages = {1388},
doi = {10.62527/joiv.10.3.4548}
}
EUVIMED