Eco-Intelligent Planning: A Multi-Objective Hybrid Reinforced Optimization Framework for Low-Carbon Resilient WEEE Reverse Logistics

Abstract

To address dual challenges of WEEE growth and sustainability, this paper proposes Green-LAtt, a multi-objective hybrid reinforced optimization framework. By integrating lifecycle assessment into a bi-objective model, Green-LAtt tackles facility location problems under non-stationary spatiotemporal volatility. The framework features a “perceive-decide-refine” architecture employing a spatiotemporal encoder (MHA+LSTM) to capture demand dynamics and a collaborative DRL-MOPSO strategy for efficient solution generation and refinement.