(IEEE, 2023) Quemada Mayoral, Carlos; Ederra Urzainqui, Íñigo; Beruete Díaz, Miguel; Gonzalo García, Ramón; Iriarte Galarregui, Juan Carlos; Estadística, Informática y Matemáticas; Ingeniería Eléctrica, Electrónica y de Comunicación; Institute of Smart Cities - ISC; Estatistika, Informatika eta Matematika; Ingeniaritza Elektrikoa, Elektronikoaren eta Telekomunikazio Ingeniaritzaren
This work addresses the lack of moisture content estimation models for food products in the millimeter-wave frequency range and showcases the potential of this range for designing compact, cost-effective, and in-line food moisture sensors. The moisture content estimation models developed in this study are intended for flour-based mixtures in the 67-110 GHz frequency range and are derived by means of a nondestructive and contactless monitoring system. To this aim, data obtained by continuous-wave (CW) vector network analyzer (VNA) spectroscopy is used to create two different models, both with a coefficient of determination ( R2 ) of 0.97. One model is based on the theoretical response obtained by means of the Looyenga effective medium theory (EMT) model, while the other is based on measured data. Both models have been experimentally validated with root mean square error (RMSE) values of 0.4% and 0.35%, respectively. These small estimation errors show the potential of this frequency range to design compact, cost-effective, and in-line food moisture sensors. This research contributes to improving quality control and monitoring of moisture levels in flour-based mixtures.
