Frequency and temperature-dependence ZnO based fractional order capacitor using machine learning

This paper investigates the fractional order behavior of ZnO ceramics at different frequencies. ZnO ceramic was prepared by high energy ball milling technique (HEBM) sintered at 1300 degrees C to study the frequency response properties. The frequency response properties (impedance and phase angles) were examined by analyzing through impedance analyzer (100 Hz - 1 MHz). Constant phase angles (84 degrees -88 degrees) were obtained at low temperature ranges (25 degrees C-125 degrees C). The structural and morphological composition of the ZnO ceramic was investigated using X-ray diffraction techniques and FESEM. Raman spectrum was studied to understand the different modes of ZnO ceramics. Machine learning (polynomial regression) models were trained on a dataset of 1280 experimental values to accurately predict the relationship between frequency and temperature with respect to impedance and phase values of the ZnO ceramic FOC. The predicted impedance values were found to be in good agreement (R2 - 0.98, MSE - 0.0711) with the experimental results. Impedance values were also predicted beyond the experimental frequency range (at 50 Hz and 2 MHz) for different temperatures (25 degrees C - 500 degrees C) and for low temperatures (10 degrees C, 15 degrees C and 20 degrees C) within the frequency range (100 Hz - 1 MHz).