Comparative evaluation of machine learning models for predicting <scp>IC</scp> engine performance and emission behavior of <scp>ZOME20</scp> hybrid Nano‐additive diesel blends

Abstract This article presents the synergistic effect between the ZOME20 biodiesel blend (i.e., 20% ZOME +80% diesel) and metal oxide nanoadditives (CeO 2 , Al 2 O 3 , TiO 2 , and GO) in the amount of 50–200 ppm in the performance and emission characteristics of a single‐cylinder CI engine and its prediction models by machine learning (ML). Out of all operable combinations in a diesel engine, the strongest enhancement of brake thermal efficiency (BTE) was found by the experimental addition of 150 ppm CeO 2 and TiO 2 , rising from 28.9% (diesel) to 31.5%, and the drop in Brake Specific Fuel Consumption (BSFC) is from 0.29 to 0.27 kg/kWh. Emission analysis showed a 21% reduction in HC, a 16% reduction in CO, and a 24% reduction in smoke opacity, while CO 2 and NOₓ showed reasonable increases of 9% and 7%, respectively (over baseline diesel). A series of machine learning models (Random Forest, Gradient Boosting, AdaBoost, and Ensemble Regression) were developed to predict the performance as well as emission results. In the present work obtained the highest total accuracy using Random Forest ( R 2 = 0.82; RMSE = 7.91 and MAE = 4.10), followed by Gradient Boosting ( R 2 = 0.78). Mean absolute percentage error (MAPE) of significant parameters was below 15%, indicating good model generalization. The combined experimental and ML efforts validated that nanoadditive‐assisted biodiesel combustion improves thermal efficiency and emission control, a sustainable and model‐based line of advance for optimizing future diesel engine technologies.