Equivalent Circuit And Neural Network Modeling Research Of Flexible Radio-frequency Capacitor And Inductor

Over the past decades,the flexible electronic devices employing both inorganic and organic semiconductor materials have made a great development.However,the current flexible electronics are mainly used in low-speed flexible devices.With the progress in device structures and materials,flexible electronic devices have run much faster.The newly developed transfer technology and nanometer thin film devices provides a good direction for the flexible radio-frequency(RF)electronics with a lot of flexible radio-frequency passive and active components created.The RF characteristic research of flexible electronic components and the establishment of the models under different bending state are the basic need of flexible electronics research and flexible integrated circuit design and manufacturing.This paper mainly studies the equivalent circuit and neural network modeling of flexible RF capacitor and inductor.Firstly the manufacturing of flexible RF capacitor and inductor are introduced,and the RF characteristics of different mechanical bending working on these components are tested.Next,the equivalent circuit models of flexible RF capacitor and inductor under different bending radius were carried out.Equivalent circuit models showed high accuracy.And through this traditional modeling method,the internal effects of RF characteristics caused by mechanical bending are studied for these flexible RF components.Then the artificial neural network theory is introduced into the flexible component modeling.The neural models of flexible RF capacitor and inductor are set up with the curvature added into the input variables.These neural models demonstrate better accuracy and higher speed than the equivalent circuit models under bending state.The modeling errors of Flexible RF capacitor and inductor are less than 1% and total time are less than 10 minutes.The neural models also can accurately predict the RF characteristics in mechanical bending state,and the final prediction error of flexible RF capacitor and inductor neural network models are all below 2%.This will take a significant impact on the study of RF characteristics of flexible capacitor and inductor under bending state,which will also greatly influence the flexible IC design and manufacturing.