A Machine Learning approach for the validation and optimization of permittivity mixing rules for binary liquids

Abstract

This paper presents the application of Support Vector Regressor models trained with glycerin-water mixture signals from a Dielectric Resonator sensor. Each signal is labeled with a concentration considered. The performance of these models indicates which mixing rule fits the most with experimental permittivity values. Some modifications of these formulas are validated to acquire better estimations.

This paper presents the application of Support Vector Regressor models trained with glycerin-water mixture signals from a Dielectric Resonator sensor. Each signal is labeled with a concentration considered. The performance of these models indicates which mixing rule fits the most with experimental permittivity values. Some modifications of these formulas are validated to acquire better estimations.