Artificial neural network models to predict density, dynamic viscosity, and cetane number of biodiesel

Abstract

Biodiesel is considered as an alternative source of energy obtained from renewable materials. This paper presents models based on artificial neural networks (ANNs) to predict the density, dynamic viscosity, and cetane number of methyl esters and biodiesel. An experimental database was used for the developing of models, where the input variables in the network were the temperature, number of carbon atoms and hydrogen atoms, as well as the composition of methyl esters. The learning task was done through hyperbolic and linear functions, while the Levenberg–Marquardt algorithm was used for the optimization process. Correlation coefficients of 0.91946–0.99401 were obtained by comparing the experimental and calculated values, while a mean squared error (MSE) of 1.842×10−3 was obtained in the validation stage. All models met the slope-intercept test with a confidence level of 99%. The ANN models developed here can be attractive for their incorporation in simulators.