ANN modeling for justification of thermodynamic analysis of experimental applications on combustion parameters of a diesel engine using diesel and safflower biodiesel fuels

Experimental works were carried out and results were modeled with an artificial neural network (ANN) for robust outputs. In this context, combustion indicators such as cylinder pressure, mass fraction burned (MFC), average gas temperature (AG), rate of heat release (HRR), cumulative heat release (CHR) and velocity of heat transfer (VHT) variations for each fuel under different loads have been measured, calculated and robust estimation were done by application of ANN. The combustion parameters mentioned were analyzed in an ANN models for comparison of testing parameters. The selected ANN model is the Feed-Forward Back Propagation Levenberg- Marquardt algorithm with performance function MSE with 3 numbers of layers and 10 neurons. Comparisons were made for load and fuel variations and deeply analyzed and presented. Results showed that the calculated parameters and the estimated values of the selected ANN were well matched. The production level of the ANN model is meaning that the testing installation and works are done well. The measured values on combustion indicators are well matched with the values estimated by the ANN showing that the robust estimations were performed.