Modelling the performance and emissions of ethanol–gasoline blend on a gasoline engine using ANFIS

Abstract

ABSTRACT This research investigates the importance of the adaptive neuro-fuzzy inference method (ANFIS) to model the gasoline engine's performance and exhaust emissions using various fuels. To collect data for the creation of the planned ANFIS model, an experiment was performed on a SI engine using gasoline-ethanol blends at varying engine loads. The ANFIS model was designed to provide an association between all parameters using specific gasoline-ethanol blends and different engine loads. Then, using experimental data, engine performance parameters, and emissions were forecasted by the ANFIS model. The model findings were then contrasted with experimental values to determine the accuracy of the ANFIS predictions. The maximum correlation coefficient (R) of 0.9900–0.9999 and 95.3594% accuracy for both performance and exhaust emissions values were generated by the ANFIS model. The Mean relative errors (MRE) ranged from 0.055 to 8.396% whereas the root mean square errors (RMSE) values were small. The findings confirmed that the ANFIS is competent enough to forecast the performance and emissions of gasoline engines very efficiently.