121 / 2025-05-15 22:33:55

Spare Parts Demand Forecasting and Reserve Optimization: A Gradient Descent Approach with Real-Data Validation

Aviation spare parts, demand analysis, reserve layout optimization, deep reinforcement learning

Aiming at the problem that aviation spare parts inventory management is difficult to dynamically respond to demand fluctuations and lack of multi-airport reserve layout optimization, this paper proposes a data-driven three-stage framework: Firstly, we establish the consumption rule of spare parts. Secondly, the Deep Q-Network (DQN) was used to construct a multi-warehouse dynamic inventory optimization. Finally, the aviation spare parts demand under the condition of multiple airports was solved, and the spare parts reserve amount of each airport was reasonably allocated. The results show that the proposed framework has high prediction accuracy, supports adaptive spare parts inventory optimization among multiple airports, and lays a foundation for the application of dynamic optimization and predictive maintenance fusion technology for aviation equipment.