Prediction of engine performance and exhaust emissions using artificial neural network in a compression ignition engine fueled with butanol-diesel fuel

Abstract

In this study, engine performance and exhaust emission of a Diesel engine (CI) at full load and various speeds conditions using only Diesel fuel and five different blends with butanol (3, 6, 9, 12 and 15 v/v %) were modeled by using Artificial Neural Network (ANN). A single-cylinder diesel engine was used in the experimental studies. Single layer, logistic sigmoid transfer function Scaled Conjugate Gradient and Levenberg-Marquardt algorithms were used in the presented ANN model. Input layer includes engine speed and blending ratio. Output layer includes parameters of brake specific fuel consumption, effective efficiency, NOx emission and CO emission. Mean absolute percentage error (MAPE) data and mean square error (MSE) and were calculated for performance of the networks. It was obtained that there was a consistency among the presented ANN model and the data obtained from experiments.