Research on Omni-Channel Inventory Management Based on Deep Reinforcement Learning

Abstract

This study aims to address the complexity of inventory management in an omni-channel retail environment by proposing an intelligent inventory management model based on Deep Reinforcement Learning (DRL). The research objective is to develop a comprehensive framework that integrates various stages of the retail supply chain to enhance operational efficiency and customer service. By introducing Deep Q Networks (DQN) and a multi-agent collaboration framework, the model achieves the coordinated optimization of various stages, including inventory allocation, demand forecasting, order fulfillment, and return processing. The research innovatively integrates the return management mechanism with inventory replenishment strategies, proposing dynamic inventory allocation and real-time decision-making mechanisms that significantly enhance the efficiency and flexibility of omni-channel inventory management. Experimental results show that this model can significantly improve service levels (from 85% to 96.8%) in dynamic environments, optimize inventory costs, and enhance the operational capabilities of enterprises under complex market conditions. The study not only enriches the theoretical framework for omni-channel retail inventory management but also provides enterprises with a practical intelligent decision-making tool, which is of great significance for promoting the digital transformation of retail businesses.