A Combined Approach Using Lorentzian Fitting and ANNs for Microwave Resonator Modeling

This work aims to investigate a novel modeling approach for microwave sensors. As a case study, a microwave relative humidity sensor is considered. It consists of a microstrip resonator combined with an Ag@$\alpha$-Fe
2
O
3
-based sensing material. The frequency-dependent response of the device at different values of relative humidity has been described through a mathematical model. The proposed modeling approach is based on a combination of the Lorentzian fitting method, usually employed for the accurate determination of the quality factor and the resonant frequency, with artificial neural networks (ANNs). Both methods have been exploited for microwave device modeling and they exhibited pros and cons, depending on the selected application. A combination of the two approaches is proposed in this paper with the aim of taking the advantages of both and minimizing their disadvantages.