Machine Learning-Based Prediction Techniques for the Evaluation of Diesel Engine Performance
The current study investigates the performance parameters of the DI diesel engine when it is powered with a diesel and mixed with various blends (CB10 to CB50) of Calophyllum Inophyllum biodiesel. The characteristics of diesel fuel and C. Inophyllum biodiesel are very similar and its properties also meet the ASTM requirements. C. Inophyllum is a feedstock that is economically attractive due to its high yield, when it is transesterified. A 4- stroke, single-cylinder diesel engine was adopted to test the performance using diesel and biodiesel blended fuel samples. From the results, The BTE for the biodiesel blended fuel samples were reduced maximum up to 12.68 % compared to diesel fuel. Lowest BSFC for the CB10 fuel (i.e. 0.24 kg/kW hr) were observed compared to remaining blended fuel samples and CB50 fuel exhibits highest EGT compared all the fuel samples. The performance of diesel engine running on mixtures of diesel and biodiesel blends can be predicted by using this study and also examines the potential Machine Learning (ML) techniques, such as Linear Regression (LR), Random Forest (RF) and Support Vector Regression (SVR). In the prediction of performance parameters, the random forest model shows better than other methods for attaining the highest accuracy. Major Findings: The CB10 blend showed the highest BTE among all biodiesel blends, though diesel outperformed all in both BTE and BSFC. Machine learning techniques proved effective for forecasting diesel engine performance, aiding biodiesel blend optimization.