Digital twin management platform integrating multi-agent system for resource utilisation of crop-livestock waste and its applications

To address the issues in traditional crop-livestock waste recycling, such as low resource utilisation efficiency and insufficient management effectiveness, this study proposes a digital twin-based management method incorporating multi-agent system. Firstly, the key factors influencing crop-livestock waste recycling were analysed, and a hierarchical five-tiered management architecture was proposed to achieve closed-loop coordination across the entire production process. Secondly, a dual data-model driven strategy was adopted to construct digital twin models of the production system. This approach deeply integrates the distributed decision-making advantages of multi-agent system with the simulation and predictive capabilities of digital twin, enabling multi-dimensional simulation, multi-objective optimisation, and prospective production capacity forecasting for all elements of the production process. Additionally, a virtual-real bidirectional interactive verification mechanism enhances prediction accuracy and resource allocation efficiency, forming an innovative application model of ‘digital pre-evaluation–parameter optimisation–physical validation’. Ultimately, case validation demonstrates that this platform reduces raw material transportation cost by about 15%, decreases unplanned equipment downtime by over 50%, and effectively improves resource utilisation efficiency. It offers practical value and broad application prospects for advancing green and low-carbon agricultural development.