Abstract This study is about soot emission prediction of a turbo-charged DI diesel engine in different opening ranges of waste-gate using artificial neural network. For training and testing the ANN model, different opening ranges of waste-gate were supplied using an adjustable spring to load the actuating rod of the waste-gate in which, increasing the opening range of the waste-gate decreases the inlet manifold pressure. The maximum inlet manifold pressures in test were 0.1 bar, 0.15 bar, 0.2 bar, 0.23 bar, 0.26 bar, 0.35 bar and 0.52 bar over atmosphere and experiments were conducted under the ECE-R49, 13 mode standard test. Using six ranges of the experimental data for training, an ANN model based on standard back-propagation algorithm for the engine was developed. Inputs for the ANN are inlet manifold pressure, inlet manifold temperature, mass flow rate of inlet air, fuel consumption, torque and engine speed. Output is density of soot in the exhaust manifold. The accuracy of the ANN was tested by comparing the predictions with seventh range of experimental results. Root mean squared-error (RMSE), fraction of variance ( ) and mean absolute percentage error (MAPE) were found to be 3.4 , 0.998 and 8.1% respectively.

Keywords