EUVIMEDEUVIMED entdecken
Crossref

Adaptive Multi Agent Intelligence for Coordinated Resource Management in Sustainable Farming Systems

Muhammad Farooq

Journal of Smart and Sustainable Farming · 03.10.2026 · 225

DOI: 10.64026/jssf/2026019 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Smart and Sustainable Farming, online erschienen am 03.10.2026. Verfasst von Muhammad Farooq. Die Arbeit nennt 15 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Farms must share scarce resources, groundwater, pumping capacity, solar energy and irrigation labour, among fields that differ in crop, soil and economic value, under weather and market conditions that change from season to season. Centralised optimisation requires complete information and a single decision maker, whereas uncoordinated, field-by-field decisions can waste capacity at peak demand. This paper develops an adaptive multi-agent system for farm resource management in which parcel agents learn irrigation policies by Q-learning, resource agents (water, energy, crew) publish adaptive prices that reflect congestion and the pace of seasonal water use, and requests are allocated within pump, crew and seasonal caps. The system was evaluated on a simulated 16-parcel farm with three crops, a seasonal groundwater allocation of 70 % of unconstrained demand and a limited pump and crew capacity, in six scenarios (normal, dry, wet, pump failure, price shock and a combined dry–failure–shock case) with three independent training seeds. The proposed system achieved 95 % of the net value of an omniscient central planner in a normal year and 84 % in a dry year, with a yield Gini of 0.025 against 0.085 for the planner and a lowest parcel yield of 0.82 against 0.66, while independent Q-learners reached only 80 % with large variation between training runs. Ablation showed that the adaptive price signal was the essential component (net value fell to 75 % without it) whereas the bid-based priority ranking added nothing. Simple heuristics were competitive when capacity was not scarce, and the central planner remained clearly better in dry years. After a severe combined shock, continued learning with drift-triggered exploration did not outperform the frozen policy, so most of the robustness came from the daily price adaptation.

Zitieren

APA

Farooq, M. (2026). Adaptive Multi Agent Intelligence for Coordinated Resource Management in Sustainable Farming Systems. Journal of Smart and Sustainable Farming, 225. https://doi.org/10.64026/jssf/2026019

Vancouver

Farooq M. Adaptive Multi Agent Intelligence for Coordinated Resource Management in Sustainable Farming Systems. Journal of Smart and Sustainable Farming. 2026:225. doi:10.64026/jssf/2026019

BibTeX
@article{Farooq2026,
  author = {Farooq, Muhammad},
  title = {Adaptive Multi Agent Intelligence for Coordinated Resource Management in Sustainable Farming Systems},
  journal = {Journal of Smart and Sustainable Farming},
  year = {2026},
  pages = {225},
  doi = {10.64026/jssf/2026019}
}
Kostenlos starten →