Machine Learning-Based Prediction of Butanol–Diesel Dual Fuel Engine Performance and Emissions

Machine learning algorithms such as Random forest regression, Support vector regression, Polynomial linear regression, and Multiple linear regression are used in the present research article for modeling a common rail direct injection four-cylinder diesel engine. The engine utilizes butanol as a dual fuel, and the aim is to predict important parameters such as brake-specific fuel consumption, brake thermal efficiency, and exhaust emissions. To obtain the data needed for training and testing the recommended machine learning algorithms, the diesel engine was operated with varying concentrations of butanol (Bu10, Bu20, and Bu30) across different engine loads. The injection pressure was kept constant at 800 bar. The dataset has been split into two parts: 75% for the training set and 25% for the testing set. Three metrics were used in evaluating the models: those are, relative root mean squared error, root mean squared error, and coefficient of determination. These metrics were employed to assess the accuracy and predictive capabilities of the models. The results indicate that the developed machine learning models perform admirably in predicting the performance and emissions of the engine with R2 values approaching 1. Additionally, the relative root-mean-square error and root-mean-square error metrics for the dataset were found to be significantly low.