Machine learning assisted predicting and engineering specific surface area and total pore volume of biochar

Area of Science:

• Materials Science
• Environmental Science
• Chemical Engineering

Background:

• Biochar, a porous carbon material from biomass pyrolysis, is crucial for applications like hydrogen storage and pollutant removal.
• Key properties such as specific surface area (SSA) and total pore volume (TPV) dictate biochar's performance.
• Traditional methods for biochar engineering are inefficient and resource-intensive.

Purpose of the Study:

• To explore the application of machine learning (ML) for predicting and optimizing biochar properties.
• To identify key factors influencing biochar yield, SSA, and TPV.
• To validate ML-driven optimization strategies through experimental verification.

Main Methods:

• Utilized machine learning algorithms, specifically Random Forest (RF) and Gradient Boosting Regression (GBR), for predictive modeling.
• Employed ML model interpretation to identify significant features affecting biochar characteristics.
• Optimized pyrolysis parameters and biomass mixing ratios using a three-target GBR model.

Main Results:

• Achieved high prediction accuracy for biochar yield, SSA, and TPV, with test R-squared values ranging from 0.89 to 0.94.
• Identified pyrolysis temperature, biomass ash content, and volatile matter as critical factors influencing biochar properties.
• Experimentally validated optimized production schemes, confirming the effectiveness of ML in achieving high SSA and TPV.

Conclusions:

• Machine learning offers a powerful and efficient tool for the prediction and engineering of biochar properties.
• ML-driven optimization significantly reduces the time and labor associated with traditional experimental approaches.
• This study highlights the potential of ML to advance biochar technology for diverse environmental and energy applications.