Machine learning-driven dual-objective optimization of biomass-derived carbon quantum dots toward high conductivity
Heng Jiang1,2, Haoyu Zhang1,2, Ninghui Xu3
1State Key Laboratory of Heavy Oil Processing Beijing 102249 China jianping.su@cup.edu.cn liyeqingcup@126.com.
RSC Advances
|July 13, 2026
Summary
Machine learning optimized carbon quantum dots (CQDs) from straw for enhanced anaerobic digestion. This approach achieved high CQD yield and conductivity, improving methane production and process stability.
Area of Science:
- Biomass conversion
- Materials science
- Environmental engineering
Background:
- Biomass-derived carbon quantum dots (CQDs) enhance anaerobic digestion efficiency.
- Simultaneously optimizing CQD yield and conductivity is challenging.
Purpose of the Study:
- To optimize straw-based CQD preparation using machine learning for high yield and conductivity.
- To evaluate the performance of optimized CQDs in anaerobic digestion.
Main Methods:
- Developed a predictive model using AutoGluon with 273 experimental data points.
- Employed interpretable machine learning and NSGA-II algorithm for multi-objective optimization.
- Tested optimized CQDs in practical anaerobic digestion scenarios.
Main Results:
- Achieved maximum conductivity of 0.713 mS cm⁻¹ with 32.02% CQD yield under optimized parameters (C/N=1:2, T=187.7°C, t=323.18 min).
- Optimized CQDs significantly improved methane productivity, process stability, and acid buffering capacity in anaerobic digestion.
- Demonstrated the effectiveness of machine learning in guiding CQD preparation for dual objectives.
Conclusions:
- Machine learning effectively optimizes CQD preparation for high conductivity and yield.
- The prepared CQDs offer practical benefits for anaerobic digestion processes.
- This study provides a pathway for developing advanced CQDs for bioenergy applications.
