On one hand, electrochemical impedance spectroscopy (EIS) is a powerful technique to characterize a wide variety of electrochemical systems, which allow us to evaluate the behavior at the electrode - electrolyte interface and helps in the interpretation of the corrosion mechanisms. This paper aims to develop an artificial neural network with the ability to learn from experimental values to predict EIS-Bode diagrams of 304 stainless steel in biodiesel. In this work, a feed-forward is proposed with three layers; composition of biodiesel divided in saturated, monounsaturated, and polyunsaturated methyl esters, blends biodiesel / diesel, Total acid number (TAN), exposure time, angular frequency, square root and square of angular frequency were considered as input variables on the model and the output layer contains two variables, impedance modulus and phase angle. The best fitting training data was acquired with 9-16-2 considering a Levenberg –Marquardt learning algorithm, a logarithmic sigmoid transfer function in the hidden and output layer. Experimental and simulated data were compared satisfactorily through a linear regression model with a correlation coefficient, R>0.98, mean square error, MSE < 9.13x10-4 in the validation stage and met the slope-intercept test with a confidence level of 99%. ANN model developed can be attractive to understand the synergy of the parameters and variables on corrosion behavior of metal in biodiesel besides to extract interest information related to phenomena.
