Prediction of Performance and Emissions of Diesel Engine Fueled with Undi-Diesel Blends: An ANN Approach

Main Article Content

Pravin Ashok Madane
Jitender Singh Rawat
Viresh Payak
Rajsekhar Panua

Abstract

The present study highlights the significance and applicability of the artificial intelligence based artificial neural network model in mapping the performance and exhaust fumes of an existing diesel engine under Undi-diesel strategies. The diesel engine load and Undi share were preferred as input parameters of the ANN model for mapping the diesel engine output parameters, namely brake thermal efficiency, oxides of nitrogen, unburned hydrocarbon and carbon monoxide emissions. A single hidden layer with feed-forward back-propagation, logistic sigmoid (logsig) transfer function and Levenberg-Marquardt training algorithm for developing an optimal mode is developed for current optimisation. The optimal model was found to be a (2-8-4) topology with an overall correlation coefficient (R) of 0.99744, a mean square error of 0.00023 and a mean absolute percentage error of 0.74%.

Article Details

Section
Articles