Hybrid Deep-Learning Frameworks for Prediction of Industrial Material Demands

This research introduces an advanced hybrid forecasting framework designed to enhance the accuracy of industrial material demand prediction. The proposed system integrates deep-learning architectures with the principles of chaos theory to effectively model complex temporal and nonlinear dependencies in industrial datasets. By combining Chaotic Long Short-Term Memory (LSTM) and Chaotic N-BEATS networks, the framework captures intricate seasonal and dynamic demand variations. The dataset utilized includes both historical sales and environmental attributes such as temperature, humidity, precipitation, and activity metrics to provide a comprehensive understanding of consumption trends. Rigorous preprocessing and feature engineering techniques were applied to ensure high data quality. Performance was evaluated using metrics such as MAE, MSE, RMSE, MAPE, accuracy, and training time. Experimental results reveal that the hybrid chaotic models deliver superior predictive performance and improved generalization capability, making the proposed approach a valuable tool for intelligent supply chain planning and inventory management.