Application of Time Series-Based Demand Forecasting in Inventory Management for Energy Enterprises

This paper proposes a material requisition demand forecasting method based on TimeGPT by comparing the results of various models. The method improves the accuracy of material demand forecasting through optimizations such as adjusting the training and testing datasets split ratios, weighted forecasting of critical dates, and parameter tuning. This approach reduces inventory uncertainty for energy enterprises, enhances planning and execution efficiency, and plays a crucial role in inventory management. It not only lowers operational costs but also improves supply chain agility, enabling enterprises to gain a competitive advantage.