Application of ANN-ICA Hybrid Algorithm toward Prediction of Engine Power and Exhaust Emissions

Artificial neural network was considered in previous studies for prediction of engine performance and emissions. ICA methodology was inspired in order to optimize the weights of multilayer perceptron (MLP) of artificial neural network so that closer estimation of output results can be achieved. Current paper aimed at prediction of engine power, soot, NOx, CO2, O2, and temperature with the aid of feed forward ANN optimized by imperialist competitive algorithm. Excess air percent, engine revolution, torque, and fuel mass were taken into account as elements of input layer in initial neural network. According to obtained results, the ANN-ICA hybrid approach was well-disposed in prediction of results. NOx revealed the best prediction performance with the least amount of MSE and the highest correlation coefficient(R) of 0.9902. Experiments were carried out at 13 mode for four cases, each comprised of amount of plastic waste (0, 2.5, 5, 7.5g) dissolved in base fuel as 95% diesel and 5% biodiesel. ANN-ICA method has proved to be selfsufficient, reliable and accurate medium of engine characteristics prediction optimization in terms of both engine efficiency and emission.