Machine Learning-Based Prediction of Biomass Pyrolysis Yield Under Diverse Conditions: A Comparative Analysis

Biomass pyrolysis offers a sustainable pathway for renewable energy generation and waste management by converting organic materials into valuable products such as bio-oil, bio-char, and gases. Since, the yield of pyrolysis products highly depend on different parameters like temperature, heating and the biomass composition, it becomes crucial to predict the ideal distribution of the product (bio-char, bio-oil, and gas) in different circumstances. For this, machine learning techniques offer a potent tool since they can effectively examine intricate, nonlinear correlations between these characteristics and the final product yields. In this study, we employed six advance ML algorithms: XGBoost, KNN, Random Forest Regressor, CatBoost Regressor, LGBM Regressor, and HistGradientBoosting Regressor and through our work we conclude that XGBoost Regressor outperforms in all phases of yield obtained through biomass pyrolysis.