Prediction of lignocellulosic pretreatment performance using deep eutectic solvents based on IWOA-Stacking.

Abstract

To enable rapid and accurate prediction of lignocellulosic pretreatment performance using deep eutectic solvents (DESs), an improved Whale Optimization Algorithm (WOA) was developed to optimize the hyperparameters of a stacking ensemble model, termed IWOA-Stacking. The stacking framework incorporated Random Forest (RF), Support Vector Regression (SVR), and eXtreme Gradient Boosting (XGBoost) as base learners, with SVR serving as the meta-learner. By integrating an improved Tent chaotic mapping, adaptive weighting, Lévy flight, and Gaussian-Cauchy hybrid mutation, the enhanced WOA exhibited superior optimization performance. The proposed IWOA-Stacking model achieved high predictive accuracy on the test set, with R2 values of 0.951 and 0.953 for cellulose retention (R-Cellulose) and lignin removal (d-Lignin), respectively. An interpretability analysis based on SHapley Additive exPlanations (SHAP) was conducted to quantify feature contributions, and an intelligent prediction system was further developed. This study provides a reference for the accurate prediction and process optimization of DES pretreatment performance.