Exhaust emissions prognostication for DI diesel group-hole injectors using a supervised artificial neural network approach

Abstract Broad information on exhaust emissions facilitates the design of modern machinery and processing equipment with modified quality specifications. This paper is aimed at investigating soot and NOx emissions as affected by crank-angle, liquid mass evaporated, mean diesel mass fraction and heat release rate of group-hole injectors utilizing computational fluid dynamics (CFD) while the objective parameters are prognosticated by a supervised artificial neural network (ANN). A feed-forward ANN with standard back propagation (BP) learning algorithm was adopted for problem modeling with varying number of neurons in the hidden layer. A 4-17-2 topology with Levenberg–Marquardt training algorithm (trainlm) denoted mean squared error (MSE) and mean relative error (MRE) of 0.8051 and 0.0818, respectively. The supervised ANN also represented coefficient of determination, R 2 of 0.9716 and 0.9678 for NOx and soot emissions, respectively. The obtained results have shed light on promising ability of ANN as a powerful modeling tool for prognostication of soot and NOx emissions due to some spray specifications.