Towards Intelligent Combustion: A Bibliometric and Machine Learning Analysis of Hydrogen-Fueled Gas Turbines

Abstract

Thermo-acoustic instability poses significant challenges to the safe and efficient operation of hydrogen-fueled gas turbines, particularly in industrial applications. This study conducts a comprehensive bibliometric analysis to explore research trends, identify gaps, and evaluate predictive modeling opportunities for thermo-acoustic instability in hydrogen combustion systems. Using Scopus and Web of Science data analyzed with Bibliometrix (R-package), the study maps the thematic evolution of key research areas. The findings reveal extensive work on performance optimization, flow dynamics, and combustion propagation, yet limited attention to hydrogen-specific thermo-acoustic instability. Additionally, while numerical simulations and active control mechanisms are well-developed, real-time predictive modeling using machine learning (ML) remains underexplored. To bridge these gaps, this study proposes a hybrid AI-CFD framework incorporating neural networks for enhanced instability prediction and control. The insights gained contribute to advancing hydrogen combustion technologies, enabling safer and more efficient gas turbine operations.

Introduction

• The global energy industry is shifting towards decarbonization, with hydrogen emerging as a viable clean energy alternative due to its high energy content and zero-carbon combustion.

• Hydrogen’s use in gas turbines and combustion engines faces a major technical barrier: thermoacoustic instability (TAI)—the coupling of heat release with pressure oscillations, leading to unstable combustion.

2. Traditional vs. Emerging Solutions

• Conventional CFD methods (LES, DNS) are accurate but computationally expensive and unsuitable for real-time instability prediction.

• Machine Learning (ML) techniques—especially neural networks (NNs)—offer faster, scalable alternatives for:

  • TAI prediction

  • Performance optimization

  • Active combustion control

3. Role of Bibliometric Analysis

• Bibliometric analysis was used to map trends, collaborations, and research gaps in hydrogen combustion and ML-based TAI prediction.

• Data from Scopus and Web of Science (209 documents) were analyzed using Bibliometrix (R package) and visualization tools (bubble charts, tree maps, word clouds, network graphs).

4. Key Research Trends Identified

• Increasing use of ML methods such as:

  • ANNs, CNNs, DNNs, LSTMs, GANs, PINNs

  • Hybrid models combining ML with CFD

  • Reinforcement Learning (RL) for adaptive control

• Integration of sensor data and multi-modal signals for real-time detection of combustion instabilities.

5. Highlights of Recent ML Applications

• Wang et al.: ANN + CFD for MILD combustion prediction.

• Mondal et al.: DNNs for TAI prediction and real-time monitoring.

• Zhang et al.: ANNs to accelerate LES in turbulent flames.

• An et al.: U-Net-based CNN for hydrogen combustion.

• Cellier et al.: CNN + RNN for early instability detection.

• Mariappan, Son & Lee: Applied PINNs to embed physics into ML models.

• Grenga & Xu: Used GANs for instability detection and simulation enhancement.

• Lyu, Sengupta, Bhattacharya: Applied hybrid and probabilistic ML models to control and detect TAI in various systems.

6. Identified Research Gaps

7. Future Research Directions

• Develop hydrogen-specific ML models tailored to unique combustion dynamics.

• Conduct comparative analyses of NN architectures for TAI detection.

• Build hybrid CFD-ML frameworks for accurate, real-time modeling.

• Leverage PINNs, GANs, Transformers, and GNNs for next-gen modeling.

• Integrate sensor networks + ML for dynamic, adaptive control.

• Apply reinforcement learning for real-time fuel injection optimization.

8. Bibliometric Visualization Insights

• Keyword Analysis: Dominant terms include “combustion,” “hydrogen,” “instability,” “performance,” and “propagation.”

• Tree Map & Word Cloud: Reveal focus areas in modeling, emissions, flame dynamics, and increasing relevance of AI and hydrogen-specific studies.

Conclusion

The comparative analysis of keyword trends in hydrogen combustion research, conducted through a variety of bibliometric visualization tools, has provided a comprehensive overview of the current research landscape. This study has identified dominant themes, emerging trends, and the interconnectedness of key research topics, offering valuable insights into the evolving priorities and challenges in the field. The findings underscore the critical role of hydrogen as a clean energy carrier and highlight the need for continued innovation and interdisciplinary collaboration to address the technical challenges associated with hydrogen combustion. Additionally, the bibliometric analysis in Section III serves as a quantitative validation of the literature review in Section II, particularly regarding the role of machine learning in thermoacoustic instability prediction. The keyword frequency trends and network visualizations confirm the increasing research focus on deep learning, hybrid AI-CFD models, and predictive analytics, aligning with the discussion in Section II. Furthermore, the research gap analysis highlights unresolved challenges, such as the need for specialized neural network architectures and the integration of high-fidelity CFD data, reinforcing the importance of the literature review’s findings. By bridging qualitative insights from the literature review with quantitative trends from bibliometric analysis, this study provides a comprehensive, data-driven roadmap for future research in hydrogen combustion. Moving forward, continued exploration of machine learning techniques, hybrid modeling approaches, and advanced instability control strategies will be essential for accelerating the development of hydrogen-based energy solutions. Based on the dominant themes, emerging trends, and research gaps identified by the literature view and the bibliometrics analysis , the following recommendations are proposed to guide researchers, policymakers, and industry stakeholders in advancing hydrogen combustion technologies and addressing the associated challenges: Enhance Computational Modeling Capabilities The continued development and application of advanced computational tools, such as LES and DNS, are essential for addressing the complex dynamics of hydrogen combustion. Future research should focus on improving the accuracy and scalability of these models, particularly for large-scale industrial applications. Foster Interdisciplinary Collaboration: The multidisciplinary nature of hydrogen combustion research necessitates collaboration across fields such as mechanical engineering, chemical engineering, computational fluid dynamics, and machine learning. Interdisciplinary research efforts can lead to innovative solutions for mitigating combustion instabilities and optimizing system performance. Explore Hybrid Energy Systems: The integration of hydrogen fuel cells with combustion-based systems represents a promising avenue for achieving sustainable energy solutions. Future research should investigate the design and optimization of hybrid systems that leverage the strengths of both technologies. The recommendations outlined above provide a strategic framework for advancing hydrogen combustion research and addressing the critical challenges associated with its implementation. By prioritizing combustion stability, leveraging advanced computational tools, fostering interdisciplinary collaboration, and exploring innovative combustion techniques, researchers can contribute to the development of sustainable and efficient hydrogen-based energy systems. These efforts will not only support the global transition to clean energy but also pave the way for the widespread adoption of hydrogen as a key component of the future energy landscape.

References

Q. Hassan et al., \"The renewable energy role in the global energy Transformations,\" Renewable Energy Focus, vol. 48, p. 100545, 2024. [2] F. Belaïd, A. Al-Sarihi, and R. Al-Mestneer, \"Balancing climate mitigation and energy security goals amid converging global energy crises: The role of green investments,\" Renewable Energy, vol. 205, pp. 534-542, 2023. [3] A. Islam et al., \"Accelerating the green hydrogen revolution: A comprehensive analysis of technological advancements and policy interventions,\" International Journal of Hydrogen Energy, vol. 67, pp. 458-486, 2024. [4] J. Hwang, K. Maharjan, and H. Cho, \"A review of hydrogen utilization in power generation and transportation sectors: Achievements and future challenges,\" International journal of hydrogen energy, vol. 48, no. 74, pp. 28629-28648, 2023. [5] S. Qiu et al., \"Multi-sensor information fusion based on machine learning for real applications in human activity recognition: State-of-the-art and research challenges,\" Information Fusion, vol. 80, pp. 241-265, 2022. [6] J. H. Jeong, H. S. Park, Y. K. Park, and T. S. Kim, \"Analysis of the influence of hydrogen co-firing on the operation and performance of the gas turbine and combined cycle,\" Case Studies in Thermal Engineering, vol. 54, p. 104061, 2024. [7] M. SaberiKamarposhti et al., \"A comprehensive review of AI-enhanced smart grid integration for hydrogen energy: Advances, challenges, and future prospects,\" International Journal of Hydrogen Energy, vol. 67, pp. 1009-1025, 2024. [8] Y. Wang, S. Liang, Z.-Q. J. Xu, T. Zhang, and L. Ji, \"Artificial neural network aided unstable combustion state prediction and dominant chemical kinetic analysis,\" Chemical Engineering Science, vol. 300, p. 120567, 2024. [9] S. Mondal, A. Chattopadhyay, A. Mukhopadhyay, and A. Ray, \"Transfer learning of deep neural networks for predicting thermoacoustic instabilities in combustion systems,\" Energy and AI, vol. 5, p. 100085, 2021. [10] R. Gaudron and A. S. Morgans, \"Thermoacoustic stability prediction using Deep Learning,\" in INTER-NOISE and NOISE-CON Congress and Conference Proceedings, 2023, vol. 265, no. 7: Institute of Noise Control Engineering, pp. 582-591. [11] M. Zhang, R. Mao, H. Li, Z. An, and Z. X. Chen, \"Graphics processing unit/artificial neural network-accelerated large-eddy simulation of swirling premixed flames,\" Physics of Fluids, vol. 36, no. 5, 2024.

Copyright

Copyright © 2025 Agonga Oyinbonogha Fred, Norazila Othman, Mohd Fairus Mohd Yasin . This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.