Machine learning–assisted multi‐objective optimization of hydrogen‐integrated nanoparticle biodiesel blends for sustainable diesel engine decarbonization

Abstract In recent years, biodiesel fuel has been developed as a crucial alternative for fossil fuels, since it is seen as a sustainable form of energy with lower effluent content. Despite its superior benefits over conventional oils, its consumption in engines may often lead to certain difficulties such as fuel filter clogging and injector deposits. Hydrogen gas coupled with nanoparticles in biodiesel blends enhances fuel performance while simultaneously lowering greenhouse emissions effectively. This further aids in obtaining improved fuel economy, stable combustion, and enhanced fuel atomization in diesel engine operation. The study explores the biodiesel developed from Pontederia crassipes with different nanoparticles in combination with hydrogen gas for diesel engine performance and emission characteristics. The combined adaptive neural network (ANN)‐k means analysis is applied to transform nonlinear multi‐fuel interactions into a structured decision framework. This identifies Titanium Dioxide as the best nanoparticle in the “Best” cluster with a centroidal distance of 5.22, Zinc Oxide in the “Average” cluster at 5.23, and Iron Oxide in the “Worst” cluster with a distance of 7.91. Titanium nanoparticle‐based blends reported highest BTE (33%) and the lowest BSFC (240 g/kWh), along with reduced emissions of UBHC (20 ppm) and NOx (8 g/kWh) compared to other nanoparticles. Obtaining an optimal nanoparticle, as well as hydrogen mixing, are critical in improving the characteristics of biodiesel, enabling a cleaner and sustainable option for future diesel engines.