Probability-Box Informed Hysteresis Modelling through Metaheuristic Search Algorithms

Jone Ugarte Valdivielso, Manex Barrenetxea Iñarra, Jose I. Aizpurua

Keywords: Hysteresis, Jiles-Atherton, partial differential equation, probability-box, metaheuristic-search algorithm, power transformer

TL;DR: We present a parameter initialization strategy based on probability-boxes for metaheuristic-search algorithms and applied to hysteresis modelling

Abstract:

Hysteresis modelling is crucial for many industrial applications ranging from mate-
rial science to power and electrical energy systems. A frequently-used approach
in the magnetic materials used for power and energy sector is the Jiles-Atherton
(JA) model, which approximates the hysteresis curve through a partial differential
equation (PDE). However, the parameter-estimation for the PDE is challenging.
The present study evaluates the JA parameter estimation through the integration
of probability-box (p-box) parameter initialization with metaheuristic search algo-
rithms. The proposed p-box informed parameter initialization approach is tested for
two different iron core materials integrated with three different metaheuristic-search
algorithms, including Genetic Algorithms (GA), Differential Evolution (DE), and
Particle Swarm Optimization (PSO). Then, the p-box approach is compared against
the classical uniform and normal distribution based parameter initialization strate-
gies. The results show that p-box parameter initialization can be used to estimate
JA parameters accurately when there is little knowledge about the transformer
data.

Submission Number: 20