Machine learning-aided combustion reaction fluid model

and computing method



We focus on combining combustion science, computational fluid dynamics, computational chemistry with artificial intelligence and high-performance computing, develops original machine learning-assisted combustion computing methods, builds open-source platform, database and standardized workflow, and promotes the application of combustion scientific computing in aerospace, energy and chemical industry.


Focus on the application of the next generation of artificial intelligence methods to solve core scientific problems,including multi-scale modeling, first-principles molecular dynamics, engine turbulent combustion simulation.


[Deep Flame] Open source computing platform of combustion reaction in “AI for Science” era

DeepFlame is AI-assisted numerical simulation software of combustion reaction initiated by AISI combustion research team, which is based on open-source platforms such as OpenFOAM, Cantera, pyTorch, etc., and combined with heterogeneous parallel and AI accelerator. Cooperate with DeepModeling community to build open-source computing platform and database for the combustion simulation researchers.


Deep learning-based First-principles molecular dynamics simulation for methane combustion

Deep learning-based DeePMR method for simplifying N-heptane reaction mechanism


Multi-scale sampling and ODE solving of reaction kinetics

About Us



Our Team



Address:No.150 Chengfu Road, Haidian District, Beijing

Email :