The annual global food waste production is approximately 100 × 106 tons, and its landfill treatment contributes more than 30% of the carbon emissions of the agricultural food system. The greenhouse effect intensity of methane emissions is more than 80 times that of CO2 (carbon dioxide). To address this dilemma, this paper proposes an innovative intelligent decision-making system that integrates the Internet of Things, Artificial Intelligence, and Life Cycle Assessment, aiming to optimize the synergistic benefits of reducing pollution and carbon emissions in the food waste energy pathway. The system comprises three modules: real-time perception of multi-source data, intelligent path decision-making driven by DRL (deep reinforcement learning), and dual-objective optimization of resource and climate. It has a 15-min resolution and second-level decision response capabilities, can dynamically match waste characteristics with the optimal treatment path, and integrate a real-time feedback loop of carbon footprint. By deploying a distributed sensor network to collect waste composition, humidity, and generation data in real time, and combining image recognition technology to achieve accurate classification monitoring (accuracy ≥ 95%). The system uses an intelligent path decision engine based on DRL. In terms of resource-climate benefit optimization, it constructs a multi-objective planning model, where carbon emission minimization and energy output maximization serve as dual constraints, and outputs a Pareto optimal solution set. In the empirical study, 12 Chinese cities covering different climate zones are selected for empirical deployment for 18 months. The system demonstrates excellent regional adaptability, adjusting the optimal strategy based on the composition of waste, climate, and policy. In the high-humidity area of the Yangtze River Delta, the biogas yield increases by 18%, and the Chengdu–Chongqing region alleviates the pressure on arable land through feed technology. The system significantly improves processing efficiency in the pilot cities, increasing the proportion of anaerobic digestion technology to 78%, reducing carbon emissions by 1.2 tons of CO2 equivalent (tCO2e) per ton compared to traditional landfills, and achieving an energy density of 18.5 MJ/m3 in the generated biogas. By converting 30% of food waste into feed technology, the demand for corn-soybean feed can be reduced by 4.8 × 106 hectares of arable land, equivalent to a 9.3% reduction in import dependence. The system provides an expandable tool chain for constructing “zero waste cities” around the world, which is expected to reduce carbon emissions from the global agricultural food system.
Jiangwei Li (2026) studied this question.