An energy minimization approach to vector hysteresis modeling in objects having arbitrary shapes

It is known that proper modeling and simulation of vector hysteresis is crucial to the precise design and/or performance estimation of electric power devices and magnetic recording processes. Within this context, only computationally efficient vector hysteresis models may be practically utilized in numerical field computation methodologies which are essential for handling complicated device geometries and excitation schemes. Recently, substantial efforts, focusing on the efficiency enhancement of vector hysteresis models, have been reported [1, 2]. Moreover, discrete Hopfield neural networks (DHNN) have been successfully configured to construct vector hysteresis models [3]. This paper presents a novel DHNN approach to model vector hysteresis in triangular regions, which are the most commonly used discretization sub-domain components in field computation engines.