Research on the Application of Purchasing Management in W Company Based on Deep Learning

Polyester Pre-Oriented Yarn (POY) is a fundamental and essential raw material in the polyester fiber industry and serves as one of the primary inputs for China’s textile sector. Due to the oligopolistic nature of upstream suppliers, downstream enterprises have limited bargaining power, making POY price fluctuations a significant factor affecting the procurement costs of Company W. To manage procurement strategically, Company W must aggregate and analyze market supply and demand information to optimize its raw material purchasing plans.In response to this challenge, this study employs a hybrid CEEMDAN-BO-LSTM model to perform time series forecasting. Based on the predicted POY price trends for the first quarter of 2025, and considering Company W's demand, a Deep Q-Network (DQN) from reinforcement learning is utilized to develop an optimized procurement strategy aimed at minimizing total procurement cost. The strategy includes both scheduling of procurement time points and allocation of purchase quantities. A strategic procurement plan for the first quarter of 2025 is developed and compared against traditional procurement approaches. The results demonstrate that the proposed method significantly contributes to risk mitigation and cost control in procurement, offering tangible cost-saving benefits. This strategy can serve as a practical reference for other companies seeking to enhance their strategic procurement processes.