Research on Intelligent Supply Chain Optimization for Dongguan Enterprises Based on Long Short-Term Memory Networks Driven by Digital Economy

In the context of the digital economy, this study focuses on constructing a set of intelligent supply chain optimization models for Dongguan manufacturing enterprises by using LSTM neural networks to deal with key problems such as demand fluctuation, improper inventory control, and response delay. We collected the actual operation data of the enterprise, including historical demand, inventory level, and transportation efficiency, and constructed an LSTM model for demand forecasting and inventory optimization based on these data. The experimental results show that the LSTM model outperforms traditional methods such as ARIMA and linear regression in terms of accuracy and efficiency. In demand forecasting, the mean square error (MSE) of the model is only 0.012, which reduces the forecasting error by about 65%; in inventory management, the inventory cost is reduced by 20.8%, which significantly optimizes the operating cost. This study not only theoretically expands the application boundary of LSTM in supply chain management, but also provides a scalable and intelligent supply chain optimization solution for enterprises to improve responsiveness and management efficiency.