TY - JOUR
TI - Artificial neural network approach to predicting engine-out emissions and performance parameters of a turbo charged diesel engine
AU - Özener Orkun
AU - Yüksek Levent
AU - Özkan Muammer
JN - Thermal Science
PY - 2013
VL - 17
IS - 1
SP - 153
EP - 166
PT - Article
AB - This study details the artificial neural network (ANN) modelling of a diesel  engine to predict the torque, power, brake-specific fuel consumption and  pollutant emissions, including carbon dioxide, carbon monoxide, nitrogen  oxides, total hydrocarbons and filter smoke number. To collect data for  training and testing the neural network, experiments were performed on a four  cylinder, four stroke compression ignition engine. A total of 108 test points  were run on a dynamometer. For the first part of this work, a parameter  packet was used as the inputs for the neural network, and satisfactory  regression was found with the outputs (over ~95%), excluding total  hydrocarbons. The second stage of this work addressed developing new networks  with additional inputs for predicting the total hydrocarbons, and the  regression was raised from 75 % to 90 %. This study shows that the ANN  approach can be used for accurately predicting characteristic values of an  internal combustion engine and that the neural network performance can be  increased using additional related input data.