Enhancing hydrogen combustion insights in dual-fuel engines: A study through explainable machine learning and feature quantification

Energy crises draw attention to the need for substitute fuels. These fuels contribute to achieving sustainability and energy security. Dual-fuel engines, which combine conventional and alternative fuels, offer improved efficiency and reduced emissions. This makes them indispensable in resolving the world energy issue. In the present study, third-generation biodiesel derived from algae was used in blended form with diesel as pilot fuel, while hydrogen was employed as the main fuel. The engine was tested at different operational settings by varying engine load, fuel blends, and gaseous fuel substitution rates. The experimental data was used to develop a prognostic model for engine performance. XGBoost outperformed others having the lowest test MSE, mean squared error (0.08), and highest test R 2 (0.9958) in brake thermal efficiency (BTE) prediction; followed by GBR. GBR demonstrated superior performance for CO predictions outperforming RF and DT. In HC and NOx forecasts with reduced test MAPE, GBR also performed better, highlighting its overall stability across parameters. The SHAP plot reveals that Engine Load has the most significant effect on NO x emissions.