CarbonKG: Industrial Carbon Emission Knowledge Graph-Based Modeling and Application for Carbon Traceability of Complex Manufacturing Process
Tao Wu, Jie Li, Jinsong Bao, Qiang Liu, Zhengkai Jin, Jun Gao
Abstract
Abstract In the context of the 2030 carbon dioxide emissions peak target, achieving carbon neutrality in manufacturing is essential. However, the complex and extensive supply chain, ranging from raw materials to the final product, presents significant challenges in assessing the carbon footprint throughout the lifecycle. This paper concentrates on the manufacturing process and introduces a modeling approach for quantifying and predicting carbon emissions, utilizing industrial intelligent technologies like process mining and knowledge graphs. First, it is crucial to establish a comprehensive carbon emission quantification model for each manufacturing stage, encompassing the energy, material, personnel, and carbon flow. Subsequently, this paper proposes an industrial carbon emission knowledge graph-based model (CarbonKG) to record and compute emissions at each production stage. Furthermore, process mining technology aids in analyzing the global distribution and movement of carbon emissions within the manufacturing process. Finally, this research presents a two-stage predictive approach for manufacturing process carbon emissions based on CarbonKG. The first stage involves initializing a local order carbon graph for prediction and developing a model to find similar cases. The second stage uses a graph-matching model to identify the Top-K similar order cases, using their carbon emission data for comprehensive prediction. Moreover, the feasibility of the proposed method was verified using actual production process data from industrial companies.