Enhancing gas concentration prediction and ventilation efficiency in deep coal mines: a hybrid DL-Koopman and Fuzzy-PID framework.

With the increasing depth of coal mining operations, traditional ventilation systems are becoming insufficient to address the growing safety and operational challenges, particularly in dynamic underground environments. To enhance the sustainability and environmental performance of the coal mining industry, this study proposes an innovative framework that integrates deep learning with the DL-Koopman operator theory for accurate gas concentration prediction and a fuzzy adaptive PID (Fuzzy-PID) control strategy for optimized airflow regulation. The DL-Koopman-based model significantly improves prediction accuracy under fluctuating ventilation conditions, effectively addressing the challenges posed by variable wind speeds and other dynamic factors. By analyzing historical data on gas concentrations and wind speeds, the model identifies underlying patterns to develop a robust predictive framework. Furthermore, the Fuzzy-PID control strategy dynamically adjusts PID parameters in real-time, incorporating a dead zone mechanism to mitigate disturbances and enhance system stability. This dual approach not only ensures rapid adaptation to changing underground conditions but also significantly improves energy efficiency and safety. The proposed method demonstrates a practical pathway toward intelligent ventilation systems, contributing to cleaner and more sustainable mining practices. This research aligns with the global energy transition goals by reducing the environmental footprint of coal mining operations while maintaining high safety standards.