Experimental and statistical evaluation of biohythane fuelled thermal barrier coated engine using machine learning algorithms

Machine learning (ML) algorithms, linear regression (LR), random forest regression (RFR), decision tree regression (DTR), and support vector regression (SVR), were utilised to predict the performance, combustion, and emission of thermal barrier coating (TBC) compression ignition engine fuelled with various blends biohythane and diesel. The mixture of supplied gaseous fuel was blended with 85%-95% biomethane and 5%-15% biohydrogen. Results indicated that up to 15% of biohydrogen enrichment with TBC improved the engine BTE by 6% compared to the diesel-only mode. Combustion parameters such as in-cylinder pressure and heat release rate were improved up to 16.5%-20% with TBC. Further, HC, CO, and smoke emissions were reduced up to 16.2%, 29.1%, and 62.6%, respectively, with TBC and biohythane. Evaluating the ML models, DTR produced the best prediction, with an R2 value range between 0.9-0.99 and an RMSE range between 0.1-0.9. It is closely followed by RFR, SVR, and LR, respectively.