Innovative model-optimized machine learning for high-accuracy predicting and exploring nitrogen transformation in...

Area of Science:

• Biomass thermochemical conversion
• Environmental engineering
• Computational chemistry

Background:

• Producing zero-carbon fuels from biomass pyrolysis requires minimizing nitrogen oxides (NOx) emissions.
• Controlling nitrogen behavior during pyrolysis is key to source reduction.
• Traditional methods struggle to analyze complex nitrogen migration pathways.

Purpose of the Study:

• To develop a predictive model for nitrogen migration during biomass pyrolysis.
• To optimize pyrolysis conditions for reduced NOx formation.
• To offer a novel approach for understanding and controlling elemental behavior in biomass conversion.

Main Methods:

• Development of the Pyro-SPIN (Source Parameter-based Integrated Nitrogen Migration) model.
• Synergistic analysis of model predictions with experimental data.
• Machine learning techniques including chain modeling and multi-output joint training.

Main Results:

• Temperature, nitrogen content, and oxygen content are primary factors influencing nitrogen migration.
• Optimized conditions (particle size <200 μm, temp <500°C, duration <60 min, slow heating) favor nitrogen enrichment in solid char.
• The model achieved mass conservation and improved prediction consistency.

Conclusions:

• The Pyro-SPIN model accurately predicts and optimizes nitrogen migration pathways in biomass pyrolysis.
• This approach offers a new method for understanding and controlling NOx formation at the source.
• The developed strategies can be applied to other elements in biomass thermochemical conversion.