Main Article Content
In this study, waste fish oil biodiesel and pure diesel fuel were mixed containing Fe2O3 nanoparticles as a catalyst. The Fe2O3 nanoparticles were evaluated at 50 and 100 ppm added to B10 at different engine speed under full load condition. Results show that adding nanoparticles could improve the combustion and efficiency of diesel engine as well as emissions. Then,multi-layer networks with feed-forward back diffusion neural network model, the algorithm of levenberg-marquardt (trainlm) as the algorithm of training, and thetan-sig, log-sig and purelintransmission function as an activation function were employed in the present research. The input or independent parameters included fuel mixture, engine speed, and fuel consumption, while, the target parameters individually included engine power, UHC, CO2, CO, NOX and , torque. The result shows that in the optimal network, there are two hidden layer with 15-15 neurons and transmission function of logsig- logsig, for hidden layer one and two, respectively. This study implies that ANNcan bea robust toolfor predicting efficiency, and emission of diesel engines with high correlation between experimental data and predicted model.