Low-Cost Electronics for Automatic Classification and Permittivity Estimation of Glycerin Solutions Using a...

Area of Science:

• Applied Physics
• Chemical Sensing
• Machine Learning Applications

Background:

• Glycerin is a key component in various industries and biodiesel production.
• Accurate classification of glycerin solutions is crucial for quality control and process optimization.
• Existing methods may lack portability or cost-effectiveness for widespread application.

Purpose of the Study:

• To develop and evaluate a dielectric resonator (DR) sensor for classifying glycerin solutions.
• To compare the performance of a commercial Vector Network Analyzer (VNA) with a novel low-cost portable electronic reader.
• To assess the efficacy of machine learning algorithms in analyzing sensor data for glycerin concentration.

Main Methods:

• Utilized a small-cavity dielectric resonator sensor to measure glycerin solutions across a relative permittivity range of 1 to 78.3.
• Collected data for air and nine distinct glycerin concentrations using both a commercial VNA and a low-cost electronic reader.
• Applied Principal Component Analysis (PCA) and Support Vector Machine (SVM) for classification, and Support Vector Regressor (SVR) for permittivity estimation.

Main Results:

• Both the VNA and the low-cost reader achieved excellent classification accuracy (98-100%) using PCA and SVM.
• Permittivity estimation using SVR yielded low Root Mean Square Error (RMSE) values (around 0.6 for VNA, 1.2 for the electronic reader).
• The study confirmed the viability of low-cost electronic readers for glycerin sensing.

Conclusions:

• The developed DR sensor effectively classifies glycerin solutions.
• Machine learning techniques enable low-cost electronic readers to achieve performance comparable to commercial instrumentation.
• This research paves the way for affordable and portable glycerin sensing solutions.