Materials classification by partial least squares using S-parameters

This paper presents a simple and usable algorithm, which is called Partial Least Squares, to classify the samples in different thicknesses and frequency range. This model employed to reconstruct the samples using S-parameters which are collected by Vector Network Analyzer in Free Space Measurement method. The relationship of between S-parameters is showed to classify the materials using proposed model. The Partial Least Squares algorithm has a good potential to classify the different samples with non-contactless and non-destructive measurement method. The classification process will be easy by using the proposed model, due to its practical and simple usage. In addition, the extraction techniques, which are Nicolson Ross Weir, Newton–Raphson, and Genetic Algorithm, are used in order to extract the dielectric constant of samples, in this study. The Newton–Raphson algorithm is selected to obtain the complex permittivity of samples since it gives best results than other extracting techniques and the results are used for classification.