Reliability analysis for hydrogen-integrated composite renewable power systems through AI-augmented Monte Carlo simulations

Abstract

In this study, a new approach for improving the reliability assessment of hydrogen-integrated composite power systems is presented, based on AI-enhanced Sequential Monte Carlo Simulations (SMCSs). Although hydrogen technologies are essential for achieving carbon-neutral power systems, they introduce operational complexities that challenge traditional reliability analysis methods. To address these challenges, an AI-augmented framework is developed to enhance the computational efficiency of SMCSs by incorporating deep learning techniques for system state pre-classification. A 20-fold reduction in computation time is achieved compared to conventional methods, while maintaining a classification accuracy of 99%. The proposed framework is validated through case studies on the IEEE RTS-79 and RTS-96 systems under both constant and varying load conditions. The results indicate that the integration of hydrogen storage leads to a reduction in the Loss of Load Probability (LOLP) by up to 50% in scenarios with increasing renewable penetration. Furthermore, the economic and sensitivity analyses conducted in this study demonstrate that hydrogen-integrated systems achieve approximately 30%-40% lower total system costs over an average 10-year period compared to conventional configurations, and that hydrogen storage effectively mitigates reliability risks as renewable energy penetration rises. The developed AI-enhanced SMCS framework provides an efficient and scalable tool for supporting the reliable and economic integration of hydrogen technologies into sustainable power systems.