マルチソース特徴融合と戦略最適化に基づくサプライチェーン炭素排出意思決定システムの設計

Unofficial AI-generated summary based on the public title and abstract. Not an official translation.

🔬研究者:AI×ESGにおける炭素会計と最適化の統合手法として、特徴融合・残差補正・制約設計の貢献を検証できる。

🏢実務担当者:Scope 3排出算定と削減戦略の自動化・精度向上に活用でき、サプライヤー管理と開示対応を効率化できる。

🏛政策担当者:サプライチェーン排出削減の計算可能な枠組みとして、Scope 3政策・報告制度設計の参考になる。

A modular carbon carbon emission decision system is proposed based on multi-source data fusion, temporal residual correction, and constrained multi-objective optimization, enabling computable modeling and strategy generation within complex supply chain environments. At the data integration level, a constraint-aware fusion function is constructed to transform heterogeneous transactional sequences into unified carbon activity vectors under aligned accounting boundaries. These features are further embedded into a two-stage estimation framework that combines activity-factor-based baseline measurement with lightweight time-series residual correction, improving robustness and traceability in emission and cost estimation. The resulting outputs feed into an optimization module that formalizes carbon management decisions as constrained multi-objective problems, where explicit separation between objective functions and constraint rules enhances configurability and interpretability. Experimental evaluations on enterprise-level datasets containing over 21,000 business events demonstrate that the system achieves significant improvements in Mean Absolute Error (MAE) and Root Mean Square Error (RMSE) for emission prediction, while reducing carbon intensity by 13.4% under realistic delivery and compliance constraints. Comparative and ablation analyses validate the core contributions of fusion mechanisms, correction modules, and constraint frameworks, forming a closed-loop computational architecture applicable to low-carbon supply chain governance.

• semanticscholar https://doi.org/10.1145/3801438.3803155first seen 2026-09-11 05:26:25 · last seen 2026-09-21 04:54:00