Inventory control of multiple perishable goods using deep reinforcement learning for sustainable environment

Abstract

Perishable goods like fresh produce are growing part of today’s e-commerce projects. Due to their perishable nature, traditionally they have accounted for a high share in the inventory cost management. In current pandemic times, when all the business are suffering, inventory control becomes all the more important for a sustainable environment. There is a need of calculated buying and selling of products as the supply chain has effected in a major way. Aiming at the cost control problem for these enterprises, this paper proposes use of Deep Reinforcement Learning (RL) techniques for the inventory management of perishable goods. The paper models the retailers inventory limitation factors and also models real-world parameters like overdue cost, shortage cost, lead time and corruption costs across multiple products. The simulation experiments show that the proposed RL models are able to reduce inventory cost and spoilage rate of the goods when the products lead times and life cycles are known along with the distribution of the demand of the products.