Institute for Aero Engine, Tsinghua University, China

Speech Title:
AI-Enabled Research on Rotating Detonation Combustion
 
Abstract:
This talk presents a review and discussion of the application of artificial intelligence (AI) to research on Rotating Detonation Combustors (RDCs). Rotating detonation combustion offers several advantages, including quasi-constant-volume combustion, high specific impulse, and wide operational adaptability, making it a promising technology for next-generation aerospace propulsion systems. However, the strongly unsteady and multi-scale coupled flow structures within RDCs, together with extreme high-temperature and high-pressure environments and complex reactive flow processes, pose significant challenges to conventional approaches in modeling, measurement, and numerical simulation.
To address these challenges, this paper highlights representative applications of AI in RDC research through several case studies. In the modeling of RDC, models like CAE-LSTM, DeepONet, and self-attention-based temporal prediction networks are employed for real-time combustion mode identification and multi-step forecasting, providing new data-driven approaches for investigating mode bifurcation, wave competition, and instability transition mechanisms. In the measurement of RDC, DDPM and  GAN are utilized to reconstruct spatiotemporal flow fields within the combustor from sparse sensor measurements and limited optical observations, enabling super-resolution flow field reconstruction. In the numerical simulation of RDC, a GPU-accelerated differentiable reactive flow solver is developed and integrated with an adjoint optimization framework to improve the efficiency of inverse parameter identification and optimization. n addition, a wide-operating-condition detonation characteristic database is established by integrating experimental and simulation data.
Finally, considering recent advances in computational power and large-scale AI models, future prospects are discussed, including foundation models for fluid dynamics and combustion, intelligent CFD simulation platforms, and data-driven design methodologies. The rapid development of AI provides new technical pathways and opportunities for overcoming limitations in rotating detonation research.