Classification of Polluted Silicone Rubber Micro Nanocomposites Based on ESDD Using ANN

Abstract

Silicone rubber micro nanocomposites are coated with various types of pollutant with variation in concentration and the laser-induced breakdown spectroscopy (LIBS) technique is used to understand the pollution performance of test specimens. The chemical composition of the contamination present on the sample was effectively established via elemental analysis of LIBS spectra. The equivalent salt deposition density (ESDD) and the normalized intensity ratio of LIBS spectral data have a direct relationship. In order to correlate the normalized LIBS spectral data intensity ratio and ESDD, the regression coefficient ($R^{2}$) is employed to determine its performance. The LIBS spectral data is used to implement an artificial neural network (ANN) approach to the categorization of contaminated silicone rubber micro nanocomposite samples based on ESDD and pollutant type. In this work, the total hidden layer neurons are selected based on classification accuracy and number of epochs needed for convergence. The developed ANN model is successful in classifying contamination level and type of contamination on silicone rubber specimens with a classification accuracy of 100%.