Akademik Veri Yönetim Sistemi
European Food Research and Technology, cilt.252, sa.2, 2026 (SCI-Expanded, Scopus)
This study presents a parallel plate capacitance system operating at 15 MHz for non-destructive prediction of carrot quality during storage. Dielectric parameters such as dielectric constant (ε’), loss factor (ε’’), loss tangent (tan δ), dissipation factor (DF), and electrical parameters such as complex impedance, resistance reactance, output capacitance, voltage, current, power, signal attenuation, and phase angle values were measured for carrot samples stored over different periods. Regression models were developed to predict the soluble solid content (SSC), pH, titratable acidity (TA), electrical conductivity (EC), and flesh puncture force (F) of carrots using their electrical responses. Regarding storage times, the results showed that ε’ and ε’’ values decreased significantly, while DF values increased. The highest R2 (determination of coefficient), RMSE (root mean square error), and H (mean absolute error) results were obtained using Random Forest (RF) models for SSC 0.87, 0.1125, and 0.0892, and for TA 0.88, 0.1077, and 0.0891, respectively. In the pH and EC estimation, the lowest RMSE (0.1048) was found in MLR (Multiple Linear Regression) with the R2 and H values of 0.85 and 0.0867, respectively. In addition, the lowest H (0.0853) value was obtained in RF with the R2 and RMSE values of 0.83 and 0.1068, respectively. RF and MLR yielded the most accurate estimation results for the internal quality parameters of stored carrots.