Data-driven modeling required energy of fuel production process through biomass hydrothermal liquefaction

• ML predicts energy demand for biomass solvent extraction. • Four optimization algorithms tuned a Gradient Boosting tree. • The GBDT-GPO model achieved superior test R 2 of 0.9248. • SHAP reveals sulfur and temperature as main energy drivers. The precise estimation of energy requirements for biomass Hydrothermal Liquefaction (HTL) is critical for designing efficient reactors and ensuring the economic viability of sustainable fuel production. However, traditional thermodynamic models often fail to capture the complex, non-linear dependencies between heterogeneous feedstock compositions and operational energy demands. This study aims to bridge this gap by developing a robust, data-driven machine learning framework to accurately predict the required energy for this integrated process. A comprehensive dataset comprising 633 experimental data points was compiled, incorporating input variables such as elemental analysis, biochemical composition, and process conditions. A Gradient Boosting Decision Tree (GBDT) was selected as the primary learner, and its performance was maximized through four distinct hyperparameter optimization algorithms. Model reliability was ensured through 5-fold cross-validation and rigorous statistical assessment. The results demonstrated that the GBDT-GPO model achieved superior generalization capability, yielding a coefficient of determination (R 2 ) of 0.9248 on unseen test data, significantly outperforming the ES-optimized model which exhibited signs of overfitting. Furthermore, SHAP analysis identified sulfur content and process temperature as the most influential factors governing energy demand. In conclusion, the proposed GBDT-GPO framework provides a highly accurate tool for process optimization, enabling reduced energy consumption and improved operational stability in biomass-to-fuel conversion systems.