Machine Learning Aided Supply Chain Analysis of Waste Management Systems: System Optimization for Sustainable Production

This study employs a bespoke Machine Learning infrastructure to unravel the complex dynamics of e-waste supply chain networks (SCNs) by optimizing a Euler-Lagrange cost function. The framework analyzes the interconnections of the concerned variables (and parameters) with the three pillars of sustainability, eventually ranking their relative contributions towards the optimization of the SCNs. To analyze the resilience of the economic SCN and its ability to converge to optimality when perturbed by external factors, using data from an Indian e-waste plant, we use Monte Carlo simulation to generate four stochastically perturbed modules around the optimal point of the original dataset. To identify the emerging nonlinear patterns, a Neural Network (NN) model has been combined with a Random Forest Model (RFM) for ensemble-based feature selection. Our results consistently point to NN as a better predictor of arbitrage conditions compared to RFM. The juxtaposition of these two models reveals a trade-off between predictive power and interpretability. While the NN serves as a potent tool for uncovering hidden complexity and emergent behaviors, the RFN proves invaluable for feature engineering, scenario simplification, and system pruning. Together, they outline an optimized toolbox that ensures sustainable production through smart engineering solutions.