Prediction of emission characteristics of a diesel engine using experimental and artificial neural networks

The focus of the study is to investigate effects of corn blends on exhaust emissions using Artificial Neural Network (ANN) approach. A series of experiments were conducted on the water-cooled multi-cylinder engine to calibrate the emissions of CO, THC, and NOx. The biodiesel was prepared using the transesterification process. Furthermore, the MgO nanoparticles of 10, 15, 20 and 30 ppm was added to the corn blends through ultrasonication. The ANN is developed to anticipate the emission characteristics of the compression ignition engine. As engine load increases, the emission of carbon monoxide and total hydrocarbons decreases significantly. On the contrary, the emission of NOx gases spiked at higher load. The ANN back propagation algorithm is developed with four input network and one output network to predict the results. The blends C10, C15, C20, and C30 were studied with the developed ANN by varying the engine load. Besides, the highest and lowest value of mean square errors and correlation coefficient were found for CO, THC, and NOx. Meanwhile, the optimized regression coefficients for the emission parameters ranged between 0.8875 and 0.9858. The predicted correlation coefficients for CO, THC, and NOx were 0.9985, 0.9978 and 0.9986, respectively.