AI-Enabled Forecasting and Performance Optimization in Sustainable Supply Chains[Prévision basée sur l’intelligence artificielle et optimisation des performances dans les chaînes d’approvisionnement durables : une approche par réseaux de neurones utilisant le modèle numérique SCOR®]

This chapter presents a computational framework for forecasting supply chain performance using augmented SCOR®4.0 metrics and Multilayer Perceptron (MLP) neural networks. The authors operationalize an eight-module SCOR® architecture, incorporating digital, cost, working capital, cash cycle, responsiveness, reliability, and risk dimensions. Each MLP model is optimized through systematic topology selection, normalized input processing, and rigorous cross-validation. Empirical results demonstrate high predictive accuracy, with correlation coefficients exceeding 0.997 and statistical tests confirming model reliability. Comparative analysis indicates that MLP-based forecasting significantly outperforms traditional linear methods, capturing non-linear interactions among operational, financial, and sustainability indicators. The chapter highlights the practical utility of AI-driven models for decision support in complex, cross-border supply chains.