The incompleteness and scarcity of life cycle inventory (LCI) data constitute a critical challenge for life cycle assessment (LCA), limiting database coverage for a wide range of current activities and emerging technologies. This paper proposes a novel computational framework that leverages statistical relational learning for knowledge graphs to extrapolate LCI data for machining activities when data are incomplete or don't exist within LCI databases. First, an ontology-based knowledge graph for LCI data (LCIKG) of machining activities in the Ecoinvent database is developed, which provides an explicit semantic representation of LCI data concepts and interrelations. Second, the LCIKG is embedded into a real-valued vector space using a tensor factorization-based relational learning model, which captures the latent semantic similarity of entities and relations in LCIKG. Missing data are modeled as incomplete triples, and a score vector was computed for each to predict the missing flow type or value. Finally, the model's efficacy was demonstrated through two validation pathways: accurately estimating intentionally omitted data within the Ecoinvent database and successfully extrapolating data for a new machining activity from an external database. Quantitative evaluation yields a high predictive accuracy, with a mean squared error (MSE) of 5.30 and a mean absolute percentage error (MAPE) of 2.36 %. This research establishes a new, knowledge-driven paradigm for bridging LCI data gaps, offering a robust and scalable solution to enhance data completeness and reliability in LCA.
扫码关注我们
求助内容:
应助结果提醒方式:
