Artificial Neural Network Models for Prediction of Density and Kinematic Viscosity of Different Systems of Biofuels and Their Blends with Diesel Fuel. Comparative Analysis

Abstract

In the present article, two models based on the artificial neural network methodology (ANN) have been optimised to predict the density ( ρ ) and kinematic viscosity ( μ ) of different systems of biofuels and their blends with diesel fuel. An experimental database of 1025 points, including 34 systems (15 pure systems, 14 binary systems, and 5 ternary systems) was used for the development of these models. These models use six inputs, which are temperature ( T ) in the range of −10 – 200 °C, volume fractions ( X 1, X 2, X 3) in the range of 0–1, and to distinguish these systems, we used kinematic viscosity at 20 °C in the range of 0.67–74.19 mm2 s–1 and density at 20 °C in the range of 0.7560–0.9188 g cm–3. The best results were obtained with the architecture of {6-26-2: 6 neurons in the input layer – 26 neurons in the hidden layer – 2 neurons in the output layer}. Results of comparison between experimental and simulated values in terms of the correlation coefficients were: R 2 = 0.9965 for density, and R 2 = 0.9938 for kinematic viscosity. A 238 new database experimental of 4 systems (2 pure systems, 1 binary system, and 1 ternary system) was used to check the accuracy of the two ANN models previously developed. Results of prediction performances in terms of the correlation coefficients were: R 2 = 0.9980 for density, and R 2 = 0.9653 for kinematic viscosity. Comparison of validation results with those of the other studies shows that the neural network models gave far better results. U ovom članku dva modela zasnovana na metodologiji umjetne neuronske mreže (ANN) optimizirana su za predviđanje gustoće ( ρ ) i kinematičke viskoznosti ( μ ) različitih sustava biogoriva i njihovih mješavina s dizelskim gorivom. Za razvoj tih modela upotrijebljena je eksperimentalna baza podataka od 1025 točaka, uključujući 34 sustava (15 čistih sustava, 14 binarnih sustava i 5 ternarnih sustava). Ti modeli koriste šest ulaza: temperatura ( T ) u rasponu od −10 do 200 °C, volumni udjeli ( X 1, X 2, X 3) u rasponu 0 – 1, a za razlikovanje tih sustava korištena je kinematička viskoznost pri 20 °C u rasponu 0,67 – 74,19 mm2 s–1 i gustoća pri 20 °C u rasponu 0,7560 – 0,9188 g cm–3. Najbolji rezultati dobiveni su arhitekturom {6-26-2: 6 neurona u ulaznom sloju – 26 neurona u skrivenom sloju – 2 neurona u izlaznom sloju}. Rezultati usporedbe eksperimentalnih i simuliranih vrijednosti u smislu korelacijskih koeficijenata bili su: R 2 = 0,9965 za gustoću i R 2 = 0,9938 za kinematičku viskoznost. Za provjeru točnosti dva prethodno razvijena modela ANN upotrijebljeno je 238 novih eksperimentalnih baza podataka s 4 sustava (2 čista sustava, 1 binarni sustav i 1 ternarni sustav). Rezultati performansi predviđanja s obzirom na korelacijske koeficijente bili su: R 2 = 0,9980 za gustoću i R 2 = 0,9653 za kinematičku viskoznost. Usporedba rezultata validacije s rezultatima drugih studija pokazuje da su modeli neuronske mreže dali znatno bolje rezultate.