Developing a Hybrid Algorithm Integrating Robust Optimization and Reinforcement Learning for Solving Large-Scale Dynamic Optimization Problems

Large-Scale Dynamic Optimization Problems (LSDOPs) have become increasingly important in Operations Research due to their wide range of applications in manufacturing, logistics, transportation, energy systems, and intelligent decision-making. These problems are characterized by continuously changing environments and uncertain parameters, making conventional optimization techniques less effective in producing reliable and adaptive solutions.This study proposes a novel hybrid optimization algorithm that integrates Robust Optimization (RO) and Reinforcement Learning (RL) to address large-scale dynamic optimization problems under uncertainty. The proposed framework combines the ability of Robust Optimization to generate reliable solutions that remain feasible under uncertain conditions with the adaptive learning capability of Reinforcement Learning, which continuously improves decision-making through interaction with the optimization environment.The research presents the mathematical formulation of the hybrid model, followed by a detailed mathematical implementation describing the interaction between the optimization process and the learning mechanism. The proposed algorithm is designed to enhance solution quality, improve adaptability to environmental changes, and maintain computational efficiency while preserving robustness against uncertainty.