AI-DRIVEN OPTIMIZATION MODELS FOR E-COMMERCE SUPPLY CHAINS: INTEGRATING HMO, MAXENT RL, TS, AND MDVRP FOR ENHANCED DEMAND FORECASTING, INVENTORY MANAGEMENT, AND DELIVERY EFFICIENCY

Abstract

Background: E-commerce expansion requires supply chain optimization with state-of-the-art 
techniques. In this paper, an AI-based optimization framework for improving demand forecasting, 
inventory management, and delivery efficiency through Hypervolume-Based Multi-Objective 
Optimization, Reinforcement Learning, Tabu Search, and Multi-Depot Vehicle Routing is 
proposed. 
Methods: The paper combines Hypervolume-Based Multi-Objective Optimization (HMO), 
Maximum Entropy Reinforcement Learning (MaxEnt RL), Tabu Search (TS), and Multi-Depot 
Vehicle Routing Problem (MDVRP) for supply chain optimization. 
Objectives: Improve demand forecasting accuracy, reduce the holding costs of inventories, 
improve delivery efficiency, and scale up the AI framework for cost-effectiveness and 
sustainability in modern e-commerce supply chains.Empirical Results: The model enhances forecasting accuracy by 12%, cuts down on the inventory 
costs by 15%, and improves delivery efficiency by 20% in comparison to the standard supply chain 
model. 
Conclusion: The proposed AI-driven framework highly optimizes performance in the supply chain 
by offering real-time deployment, edge computing, and federated learning to enhance agility and 
sustainability in modern e-commerce systems.