Data-Driven Based Digital Twin Leak Detection for Hydrogen Networks for the Gas Turbines’ Applications

Abstract The transition to a hydrogen economy requires the development of robust and reliable monitoring systems for hydrogen distribution networks, particularly to ensure safety and operational efficiency. This study explores the application of deep learning models for detecting leaks in hydrogen transport systems, using a digital twin approach. A simulated hydrogen network, including various leakage scenarios, is used to train a neural network model. The network leverages pressure sensor data, enhanced with noise augmentation, to predict leak occurrences and locations with high accuracy. Results demonstrate that the model achieves more than 96% accuracy under typical noise conditions, with its performance improving when multiple leaks occur. The model also shows resilience to noise, providing a reliable solution for real-world applications in hydrogen infrastructure. This study highlights the effectiveness of deep learning in improving leak detection and safety in hydrogen networks.