Modelling dielectric relaxation with neural networks

Nicola Barbalace

Abstract


We describe a software model for media with dielectric relaxation. In particular, a model identification is used to measure the phenomenological coefficients which occur in a mathematical description of dielectric relaxation phenomena in electromagnetic media. The identification is developed in order to substitute laboratory dielectric measurements on PMMA and PVC at different frequencies and at fixed temperature so as to obtain the phenomenological coefficients as a function of the frequency.

Full Text:

PDF


DOI: http://dx.doi.org/10.1478/C1A0902005