An Integrated Demand Forecasting and Inventory Optimization Model Using EOQ/EPQ and Hybrid Metaheuristic Algorithms Under Uncertainty

Abstract

This study presents an integrated system for inventory optimization under uncertainty with a combination of AI-forecasting techniques, a stochastic EOQ/EPQ model, a deterioration model, adaptive safety stock, and a hybrid metaheuristic optimizer. The proposed model considers the demand uncertainty by using a probabilistic demand with a forecasting-error component, while it is deterministic in the canonical EOQ/EPQ models, and accounts for the deterioration explicitly. The proposed model does not apply to the constant-cost EPQ since it does not consider deterioration or shortage. A novel stochastic derivation for EOQ/EPQ is developed to capture the uncertainty in forecasts and production dynamics, and an adaptive rule of safety stock is added to the EOQ/EPQ decision variables. The three production characteristics: order quantity, reorder point, and safety stock are simultaneously optimized by a hybrid genetic algorithm – particle swarm optimization (GA–PSO) under conditions of uncertainty to minimize total inventory cost. The proposed integrated approach is numerically tested and compared with the classic EOQ and EPQ policies through sensitivity analysis to show the advantages of the integrated approach in terms of the lowest total cost and more robust operation. The findings show that the integration of stochastic forecasting with hybrid metaheuristics can be beneficial for improving the stability and cost-effectiveness of inventory management systems, providing a practical approach for intelligent and sustainable supply chain management.