ABC method for hysteresis model parameters identification

Abstract

The paper deals with the application of the artificial bee colony (ABC) method for hysteresis parameters identification. For the first time, the ABC method will be applied on hysteresis model optimization. For this purpose, two hysteresis models are tested: the first is based on a physical magnetic material behavior, which is Jiles-Atherton and the second is simpler, Fröhlich hysteresis model built on mathematical considerations. This method's robustness will be assessed, by comparing the experimental signals to model results. I. INTRODUCTION Electrical engines and electromagnetic machines are designed according to the customer requests and safety norms. Furthermore, the advanced technologies have to target the optimized operating conditions. For this purpose, it is necessary to build robust numerical modeling based on the magnetic material behavior. In literature, several models have been proposed and studied in large material application. It can be splitted into families: analytical ones: such as Rayleigh [1], Potter [2] or Fröhlich [3][4][5], which offer comfortable implementation in finite element code, but still limited to low or high magnetization behavior description. The second family is based on physical considerations such as the Preisach model [6] and the Jiles-Atherton model [7]. These are considered as the most robust and reliable models can be applied on soft and hard magnetic materials by parameter identification process. This task remains complex due to the inter dependency of each parameter to the other. The first investigation realized in this topic is based on iterative procedure [8]. It leads to good approximation of parameter's values but often presents convergence problems and can engender numerical instability. Other authors have proposed deterministic optimization methods [9][10], where the results were successful but request a great time calculation. Since two decades, stochastic optimization methods like generic algorithms and neural network [11][12][13] have been investigated. Other works combined between generic algorithms and simulated annealing [14]. It results accurate solution in a very short time. Recently, swarm intelligence method is used in electromagnetic applications. The most famous one is the particle swarm optimization (PSO) [15], inspired by the collective behavior of birds and fishes. Many experts have investigated this method

Integral Methodology for the Multiphysics Design of an Automotive Eddy Current Damper

“…Szewczyk et al [33] then refined the numerical integration strategy to attain the 5 unknown parameters. Currently, more efficient algorithms, such as the Genetic Algorithm [34], Artificial Bee Colony [35] and improved Shuffled Frog-Leaping Algorithm [36], are available to estimate the J-A model parameters.…”