Research on Market Demand Forecasting and Dynamic Adjustment Model for Manufacturing Enterprises Based on Long Short-Term Memory Networks

As manufacturing companies increasingly face the challenges of dynamic changes in market demand and resource optimization, traditional forecasting methods can no longer meet practical needs due to their reliance on linear assumptions and their inability to effectively model complex nonlinear relationships. In this study, we propose a market demand forecasting model based on Long-Short-Term Memory Network (LSTM) and design a dynamic adjustment strategy framework with reinforcement learning.The LSTM model captures the long-term dependence and short-term fluctuation characteristics of market demand through time series modeling, while the reinforcement learning module achieves the dynamic optimization of resource allocation in real time by defining the state, action, and reward functions. The experimental results show that the proposed LSTM model outperforms traditional methods (e.g., ARIMA and SVM) in terms of prediction accuracy, with a root mean square error (RMSE) of 12.39 and a mean absolute error (MAE) of 12.14. In addition, the dynamic adjustment strategy effectively improves the resource utilization rate (by 10%-15%) and inventory turnover rate (optimized by 5%-8%), which validates the feasibility and practicality of the proposed method.