Hybrid Response Surface Methodology–Extreme Gradient Boosting Optimization of a Hemp Biodiesel–Diesel Fueled Compression Ignition Engine

The current study develops a multifunctional optimization and forecasting model to determine the combustion, performance, and emission characteristics of hemp biodiesel-diesel mixtures in a compression engine by using a combined response surface analysis and extreme gradient boosting. Hemp biodiesel blends and various engine parameters were evaluated on a single-cylinder, four-stroke, direct-injection diesel engine under controlled-load conditions. Response surface methodology was used to develop statistically significant correlations between operating variables and key responses, including brake thermal efficiency, brake-specific fuel consumption, and emissions of carbon monoxide, hydrocarbons, and nitrogen oxides. An extreme gradient boosting model was developed, which showed high correlations between measured and estimated outputs, especially for brake thermal efficiency and brake specific fuel consumption, with test coefficients of determination of 0.992 and 0.959, respectively. The predictions of hydrocarbons and nitrogen oxides also exhibited high generalization, with test R^2 values of 0.947 and 0.974, respectively, and the carbon monoxide prediction was also acceptable, albeit with relatively greater variability. The combined approach established that the use of hemp biodiesel as a renewable alternative to diesel is technically feasible when the data-driven optimization is in place. The integrated statistical and machine learning platform offers a scalable path to multi-response engine calibration and contributes to the realization of a sustainable, low-emission biodiesel combustion system in line with clean energy and decarbonization demands.