ANN Modeling of Performance Exhaust Emissions of CRDI Engine Under Diesel-Butanol Blends

The current work searches the feasibility and capacity of artificial intelligence-based modeling study for forecasting the intake and outlet relationship of an existing, CRDI engine fueled with 5 and 10% (by vol.) Butanol blended Diesel operated with six loading conditions from 5 to 30 N·m with 5 N·m step sized at 1500 (±10) rpm. The model performances are measured in view of few statistical calculations, namely, correlation coefficient (R), Nash–Sutcliffe coefficient of efficiency (NSE), Kling–Gupta efficiency (KGE), absolute fraction of variance (R2), mean square error (MSE), mean squared relative error (MSRE), mean absolute percentage error (MAPE) and Theil uncertainty (Theil U2) for mapping the relationship between input viz. load and Butanol share, and output viz. brake thermal efficiency (BTE), oxides of nitrogen (NOX), unburned hydrocarbon (UBHC) and carbon monoxide (CO). Utilizing Levenberg–Marquardt feed-forward backpropagation (trainlm) training algorithm with single hidden layer and logistic sigmoid (logsig) activation function, a topology (2–7–4) is noticed to be the preeminent with an overall R value of 0.99931. The noteworthy statistical performances beckoned the superiority of ANN model forecasting quality under Diesel-Butanol platforms.