Deep Neural Network for Dielectric Properties Prediction of PVDF/BaTiO_3 Nanocomposites for Flexible Capacitors

摘要

In this work, PVDF/BaTiO_3 nanocomposites consisting of polyvinylidene fluoride (PVDF) as matrix and BaTiO_3 (BT) as fillers were prepared by ball milling and hot-pressing process. It is known that nanofillers content and frequency affect the effective dielectric permittivity of the nanocomposites materials. Therefore, a developed model based on deep neural network (DNN) was used to study the effect of the input parameters on the dielectric permittivity of the nanocomposites. The volume fraction (vol%) of BT and frequency of alternating current (AC) were selected as the input parameters and the effective dielectric permittivity as the output response. The results show that the developed DNN model was able to predict the effective dielectric permittivity of PVDF/BT nanocomposites with a correlation coefficient (R) of 0.997. Thus, our study confirmed the accuracy and efficiency of the developed DNN model for predicting the relative dielectric permittivity of PVDF/BT nanocomposites.