Autonomous kinetic modeling of biomass pyrolysis using chemical reaction neural networks

Modeling the burning processes of biomass such as wood, grass, and crops is crucial for modeling prediction wildland urban fire behavior. Despite its importance, solid fuels remains poorly understood, which can be partly attributed to unknown chemical kinetics most fuels. Most available kinetic models were built upon expert knowledge, requires insights years experience. This work presents a framework autonomously discovering pyrolysis from thermogravimetric analyzer (TGA) experimental data using recently developed reaction neural networks (CRNN). The approach incorporated CRNN model into ordinary differential equations predict residual mass in TGA data. In addition flexibility neural-network-based models, learned interpretable, by incorporating fundamental physics laws, law action Arrhenius , network structure. then translated classical forms facilitates extraction integration large-scale simulations. We demonstrated effectiveness predicting oxidation cellulose. successful demonstration opens possibility rapid autonomous fuels, wildfire industrial polymers.