Evaluation of the Performance of Computational Models for the Analysis of the Protection of Mirror Coatings with Data Obtained Using Electrochemical Impedance Spectroscopy

Abstract

This research predicts the impedance function of mirror coatings using three curve fitting models implemented in MATLAB, such as: Polynomial fit (Polyfit), Cubic Spline fitting (CSAPS) and Local Regression Smoothing (LOESS) to carry out the analysis of the protection of mirror coatings in which the Electrochemical Impedance Spectroscopy (EIS) technique has been used. The electrochemical impedance parameters and their equivalent circuits have been measured from the data recorded at 5 and 24 hours of exposure of the sample under study to the aggressive agent sodium chloride at weight 3%. The methodology is structured into training (parameter calibration) and testing (performance evaluation) stages. The performance of the models in the training stage is carried out by cross-validation and in the testing stage it is quantified using the Root Mean Square Error (RMSE) and the determination coefficient (R²).

The main results obtained show that the CSAPS method at 5 and 24 hours after exposing the system to the electrolyte is the most robust and consistent in cross-validation and the most accurate and reliable in performance when evaluating the complete data set with an  3.40 and   3.78, and    in both cases .