Adaptive production strategy in vertical farm digital twins with Q-learning algorithms.

Area of Science:

• Agricultural Science
• Computer Science
• Sustainable Development

Background:

• Urban food production offers sustainability benefits like reduced land use and transport.
• Digital twin (DT) technology adoption in urban food systems is less explored than in manufacturing.
• Adaptive decision-making is crucial for optimizing urban food production under fluctuating demand.

Purpose of the Study:

• To investigate the impact of digital twin technology on adaptive decision-making in urban food production.
• To compare the effectiveness of different models (MILP and Q-learning) in enhancing production decisions.
• To assess the improvements in service level, resource utilization, and energy efficiency.

Main Methods:

• Utilized mixed integer linear programming (MILP) and Q-learning models.
• Employed digital twin data to inform production decisions.
• Focused on the 'Grow It York' case study for practical application.

Main Results:

• The Q-learning model achieved higher demand fulfillment ([Formula: see text]) compared to the MILP model ([Formula: see text]).
• Operational efficiency was significantly improved by the DT-enhanced Q-learning approach.
• Electricity usage per fulfilled demand decreased by approximately [Formula: see text].

Conclusions:

• Digital twin technology significantly enhances adaptive decision-making in urban food production.
• Q-learning models integrated with DT offer superior performance over traditional MILP methods.
• Broader application of DTs can foster economic resilience and environmental sustainability in urban food systems.