A Hybrid Machine Learning and Non-Linear Programming Approach for Sustainable Inventory Optimization under Dynamic Cap-and-Trade Policies
Modern Dynamics: Mathematical Progressions · 10.07.2026 · 3 (3) · 13-26
DOI: 10.64170/mdmp.v3.i3.64 ↗Kurz gefasst
Fachartikel in der Zeitschrift Modern Dynamics: Mathematical Progressions, online erschienen am 10.07.2026. Verfasst von Preety Rani und 1 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY-NC 4.0).
Abstract (Originalsprache)
As global supply chains face increasingly stringent environmental regulations, classical Operations Research (OR) models struggle to reconcile deterministic demand assumptions with the financial realities of dynamic carbon markets. To address this methodological gap, this study develops a novel hybrid non-linear programming (NLP) inventory model that systematically integrates Machine Learning (ML) predictive analytics with a dynamic cap-and-trade regulatory framework. The proposed continuous-review, partial-backordering model pioneers the mathematical translation of ML prediction error (Root Mean Square Error) directly into the physical safety stock constraint, linking algorithmic accuracy to physical carbon abatement. Furthermore, the model decouples economic lot-sizing from baseline emissions by introducing endogenously decaying, targeted green capital investments for both transportation logistics and warehousing. To solve the conditionally pseudo-convex objective function, a Block-Coordinate Descent algorithm is developed, guaranteeing stable convergence. Comprehensive numerical simulations and sensitivity analyses reveal profound operational insights. Most notably, the study establishes that algorithmic accuracy functions as a "virtual green technology"; transitioning to a deep learning forecasting architecture reduces physical safety stock requirements by 66.7%, cutting warehouse holding emissions by 50% without requiring additional physical infrastructure.
Zitieren
APA
Rani, P., & Singh, J. (2026). A Hybrid Machine Learning and Non-Linear Programming Approach for Sustainable Inventory Optimization under Dynamic Cap-and-Trade Policies. Modern Dynamics: Mathematical Progressions, 3(3), 13-26. https://doi.org/10.64170/mdmp.v3.i3.64
Vancouver
Rani P, Singh J. A Hybrid Machine Learning and Non-Linear Programming Approach for Sustainable Inventory Optimization under Dynamic Cap-and-Trade Policies. Modern Dynamics: Mathematical Progressions. 2026;3(3):13-26. doi:10.64170/mdmp.v3.i3.64
BibTeX
@article{Rani2026,
author = {Rani, Preety and Singh, Jagatveer},
title = {A Hybrid Machine Learning and Non-Linear Programming Approach for Sustainable Inventory Optimization under Dynamic Cap-and-Trade Policies},
journal = {Modern Dynamics: Mathematical Progressions},
year = {2026},
volume = {3},
number = {3},
pages = {13--26},
doi = {10.64170/mdmp.v3.i3.64}
}
EUVIMED