Prediction of engine emissions and performance with artificial neural networks in a single cylinder diesel engine using diethyl ether

In the present study, the performance and exhaust emissions of a single-cylinder, direct-injection and air-cooled diesel engine using diethyl ether (DEE)-diesel fuel mixtures were estimated by artificial neural networks (ANN). The test engine was run with pure diesel and diesel-DEE blends at different engine speeds and loads to obtain the test and training data required to build the ANN model. In the designed ANN model, brake specific fuel consumption (BSFC), exhaust gas temperature (EGT), brake thermal efficiency (BTE), nitrogen oxides (NOx), hydrocarbons (HC), carbon monoxides (CO) and smoke were selected as the output layer while engine load, engine speed and fuel blending ratio were selected as input layer. An ANN model was developed using 75% of the experimental results for training. The performance of the ANN model was measured by comparing the test data generated from the unused part of the training. According to the obtained data, ANN model predicts exhaust emissions and engine performance with a regression coefficient (R2) at 0.964-0.9878 interval. At the same time, mean relative error (MRE) values ranged from 0.51% to 4.8%. These results show that the ANN model is able to use for estimating low-power diesel engine emissions and performance. (C) 2018 Karabuk University. Publishing services by Elsevier B.V.