Construction of a magnetostrictive hysteresis operator using a tripod-like primitive hopfield neural network

A A Adly,S K Abd-El-Hafiz

doi:10.1063/1.5007009

Abstract

It is well known that accurate modeling of magnetostrictive hysteresis is crucial to different industrial applications. Although several magnetostrictive models have been developed in the past, the accuracy-efficiency balance has always been crucial. Recently, the possibility of constructing a primitive vector hysteresis operator using a tri-node Hopfield Neural Network (HNN) was demonstrated. Based upon the fact that mechanical stress along a certain direction results in dimensional deformation, this paper introduces a novel extension to the aforementioned recently developed approach. More specifically, a stress-driven evolution of a tri-node HNN hysteresis operator pair is proposed, thus yielding a tripod-like HNN pair having different input offset values. Model identification, sample simulation results and comparison with experimental measurements are given in the paper.

Highlights

Accurate modeling and simulation of magnetostrictive hysteresis is crucial to different applications such as fine positioning actuation and energy harvesting.[1,2,3,4,5,6] While several magnetostrictive hysteresis models have been developed in the past, the accuracy-efficiency balance of such models has always been critical when assessing their practicality
Inspired by the fact that mechanical stress (σ) results in dimensional deformation (δ), this paper introduces a novel extension to the aforementioned recently developed approaches
2π, Nt φt where, φt is the orientation of the line joining the ensemble center to the tth triangle centroid, Nt represents the total number of triangles, while φv is the angle subtended by each triangle vertex located at the ensemble center

Summary

INTRODUCTION

Accurate modeling and simulation of magnetostrictive hysteresis is crucial to different applications such as fine positioning actuation and energy harvesting.[1,2,3,4,5,6] While several magnetostrictive hysteresis models have been developed in the past, the accuracy-efficiency balance of such models has always been critical when assessing their practicality Among those models, the notion of simulating 1-D magnetostrictive hysteresis using 2-D models has been introduced.[7] the approach of efficiently constructing a primitive vector hysteresis operator using a tri-node Hopfield Neural Network (HNN) with positive feedback factors was demonstrated.[8,9]. More details are given of the paper

THE PROPOSED APPROACH

SIMULATIONS AND EXPERIMENTAL VERIFICATION

CONCLUSIONS