Increasing global demand for sustainable energy and carbon materials and pressing environmental challenges require innovative approaches to resource management. Biomass, a versatile renewable resource, offers great potential for conversion into an alternative fuel and valuable carbon material, biochar. However, efficiently converting diverse types of biomass into high-quality biochar remains a challenge. In a recent survey, Yuan Zhilong, Wang Yeh, Zhu Lingfeng, Chan Kong Kongand Sun Yifei from Beijing Hangzhou University and hainan universitywe address this issue by developing a sophisticated machine learning framework to optimize. biochar production from aquatic biomass. This study fills a critical gap, as previous modeling efforts have largely overlooked aquatic resources.
Data-driven transformation for cleaner energy
This comprehensive study compiles an extensive dataset of 586 data points from existing literature to detail the properties of hydrocarbon carbon (hydrothermal carbonization, HTC) and pyrolytic carbonization (pyrolytic carbonization, PLC) derived from aquatic biomass such as macroalgae, microalgae, and duckweed. The researchers trained and evaluated five tree-based machine learning algorithms to predict biochar yield and properties such as nitrogen recovery, energy density, energy recovery, and residual sulfur. The selected input parameters spanned 10 feedstock and process variables, including elemental composition, industrial components, and reaction conditions. This robust approach allows Random Forest Regression (RFR) and Extreme Gradient Boosting (XGB) This is the model that exhibits the best performance in terms of prediction accuracy.
Accuracy of hydrocarbon and dry coal properties
The RFR model showed excellent prediction accuracy for hydrocarbons, achieving R2 values of 0.89 to 0.98 for hydrocarbon yield, nitrogen recovery, energy recovery, and residual sulfur. Analysis of the importance of features shows that beyond process parameters such as temperature and time, Raw material element compositionespecially the nitrogen and sulfur contents have a significant impact on the properties of biochar. Similarly, the XGB model showed strong performance for charcoal, with R2 values for hydrocarbon energy density, charcoal yield, and charcoal nitrogen recovery ranging from 0.84 to 0.94. The main factors influencing the properties of charcoal include pyrolysis temperature, ash content, and carbon content of aquatic biomass. Understanding these relationships is the basis for tailoring biochar for specific applications such as solid fuels and catalysts.
Innovative iterative learning method
Distinctive aspects of this study include; Iterative learning application method. Initially, the model showed moderate generalization ability when directly predicting new data. However, by incorporating even a small amount of new experimental data (just 6 samples) into the original dataset and retraining the model, the prediction accuracy for subsequent new data was significantly improved. For example, the retrained XGB model achieved R2 0.97 for new charcoal yield data, and the RFR model reached R2 0.98 for new charcoal yield data. With this significant enhancement, generalization ability It provides a highly efficient and cost-effective strategy for researchers, greatly reducing the need for trial and error for large-scale experiments.
Promoting sustainable use of biological resources
This study not only fills a critical knowledge gap in modeling biochar production from aquatic biomass, but also provides a powerful data-driven framework to optimize these processes. Although the current study focused on specific biochar properties due to data availability, future studies could be expanded to include other important parameters such as surface area and pore volume as more data become available. The iterative learning approach developed here is very promising and offers versatility that can be deployed in a variety of machine learning applications beyond biochar research.
“Our work represents a significant advance in utilization. machine learning model transform aquatic biomass and turn it into a high-value biochar product,” says the corresponding author. Sun Yifei. “Proven iterative learning techniques provide a practical and efficient path for researchers to quickly adapt and enhance predictive models with minimal new data, thereby accelerating sustainable research and development. biochar production Improve processes and reduce the associated labor and financial burden in pursuit of cleaner energy solutions. ”
Corresponding author: Sun Yifei
Original source: https://doi.org/10.1007/s44246-024-00169-2
contribution: All authors contributed to the conception and design of the study. Material preparation, data collection, and analysis were performed by Zhilong Yuan, Ye Wang, Lingfeng Zhu, Congcong Zhang, and Yifei Sun. The first draft of the manuscript was written by Zhilong Yuan, and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.
Disclaimer: AAAS and EurekAlert! We are not responsible for the accuracy of news releases posted on EurekAlert! Use of Information by Contributing Institutions or via the EurekAlert System.
