Performance prediction of bended radio-frequency capacitors and inductors on plastic substrates using artificial neural network
Abstract
Flexible radio-frequency (RF) capacitors and inductors on the plastic substrates have been fabricated and characterized under mechanical bending conditions. A novel method to predict the RF performance for them on different bending states is demonstrated. Artificial neural network (ANN) shows good modeling accuracy for the flexible RF passive components with bending strains from dc to resonant frequency (∼13/2∼13/2 GHz for the capacitor/inductor). More importantly, the automatically generated ANN model, with no need of repeatedly tuning the model parameters, has demonstrated the ability to predict the RF responses for the flexible capacitors and inductors under arbitrary bending conditions with only a few sets of experimental data. Once established, this model can automatically learn the structure of the input date and predict the actual results on specific bending state which can provide an original method to measure the performance for flexible electronics on even extreme bent radius. The ANN model indicates good potential for accurate design, characterization and optimization of the high-performance flexible electronics.
