Machine learning and response surface optimization to enhance diesel engine performance using milk scum biodiesel with alumina nanoparticles

) emissions were minimized at 850 ppm for DSMEB10 + 50 ppm due to accelerated combustion and reduced high-temperature residence time, facilitated by the oxygen-rich biodiesel composition. Additionally, machine learning models were employed to predict and analyze the relationship between engine load, thermal efficiency, and emissions. Among the models tested, linear regression (LR) yielded a marginally lower mean squared error (MSE = 3.04) compared to Huber regression (MSE = 3.13), while the mean absolute error (MAE) for LR was 1.33 against 1.37 for Huber regression, suggesting LR's slightly better predictive accuracy. These findings highlight the potential of nano-enhanced DSME biodiesel blends as a sustainable and cleaner alternative to conventional diesel, with favorable engine performance and emission profiles.