Parameter identification of jiles-atherton model for transformer based on hybrid artificial fish swarm and shuffled frog leaping algorithm

Accurate modeling and simulation of the magnetization characteristic of transformer core is critical for the research of transformer DC bias phenomenon. Jiles-Atherton (J-A) model was applied to simulate the hysteresis loop of the experiment transformer. As the J-A model character is determined by its five parameters, it is necessary to identify the J-A model parameters of the transformer core under DC bias accurately. A hybrid artificial fish swarm and frog leaping algorithm was proposed to identify J-A parameters in this paper, which combined the advantages of fast convergence from artificial fish swarm algorithm and high local search accuracy from frog leaping algorithm. A transformer DC bias experiment was made, using two real transformers. According to the experiment result, the proposed algorithm was applied to identify the J-A model parameter of numerical simulation and experiment measured hysteresis loop, together with conventional artificial fish swarm, frog leaping algorithm, particle swarm and genetic algorithm respectively. The result shows that the hysteresis curve generated from the proposed hybrid algorithm has a great consistency with the measured curve, and compared to other two algorithms, the proposed algorithm has higher identification accuracy and computation efficiency, which makes it effective in the operating character analysis of the transformer under DC bias condition. © 2015 Chin. Soc. for Elec. Eng.