Bernoulli memristor modeling

Application of the signal splitting method for behavioral modeling of electrical nonlinear element – a memristor is considered. Behavioral models based on a harmonic signal class are built for the transfer characteristic, which describes the input-output relationship of the Bernoulli memristor. A splitter is implemented as a delay line. A piecewise neural network and a piecewise polynomial are specified as the mathematical forms of models. The piecewise representation of models is driven by the severely nonlinear transfer characteristic of the Bernoulli memristor. The modeling error is estimated on the basis of the uniform and root-mean-square norms. Comparative analysis shows that the piecewise neural model of the transfer characteristic for the Bernoulli memristor gives higher approximation accuracy in comparison with the piecewise multidimensional polynomial model.

Authors: E. B. Solovyeva, A. P. Serdyuk

Direction: Electrical Engineering

Keywords: memristor, modelling, mathematical model, polynomial, neural network

---

View full article