Energy Performance Contracting (EPC) is a market-driven collaboration mechanism between enterprises and energy service companies, offering substantial potential to improve energy efficiency and reduce carbon emissions. However, integrating EPC into large-scale sustainable supply chain networks presents complex multi-criteria decision-making challenges, particularly in balancing economic performance, environmental sustainability, and social responsibility under dynamic operational conditions. To address these challenges, this study proposes a holistic EPC-integrated mathematical model and a hybrid solution framework based on a Dual Attention Graph Neural Network (DAGNN). The model extends the traditional triple-bottom-line framework by incorporating two additional dimensions—operational efficiency and product quality, and explicitly captures temporal dynamics, such as seasonal fluctuations in cost, profit, and demand, to more accurately assess EPC’s impact on supply chain sustainability. The dual attention architecture adopts two separate neural networks that learn context-aware importance of strategic and operational attributes by leveraging historical expert decisions. These learned weights enable adaptive prioritization of participant attributes under varying supply chain contexts and enhance the efficiency and interpretability of decision logic. The mathematical model is then transformed into an attention-enhanced bipartite graph representation and solved through a graph neural network, enabling efficient and accurate decision-making in large-scale settings. Experimental results on multi-period, multi-product instances demonstrate that the proposed approach achieves 92.88% solution accuracy relative to commercial solvers while reducing computational time by 99.96%. These results highlight the framework’s potential to provide real-time, scalable, and transparent decision support for EPC-integrated sustainable supply chains, thereby advancing the alignment of energy efficiency initiatives with holistic supply chain performance optimization.
扫码关注我们
求助内容:
应助结果提醒方式:
