基于改进人工鱼群算法的互感器Jiles-Atherton模型参数辨识

Parameter identification of Jiles-Atherton model based on an improved artificial fish swarm algorithm

为开展仿真平台下的电磁式互感器特性研究,需要对试验互感器建立精确可靠的磁滞模型。Jiles-Atherton(J-A)模型广泛应用于互感器铁芯磁滞模型与仿真实验,其5个关键参数的辨识准确度直接影响J-A模型与试验互感器的拟合程度。提出一种基于人工鱼群和模拟退火算法的混合智能寻优算法对J-A模型进行参数辨识。改进算法初期使用变步长人工鱼群算法将搜索域快速锁定在全局最优解的附近。当J-A模型拟合达到一定精度后,转而使用并行模拟退火算法继续进行局部的精确搜索。通过Matlab仿真证实:改进混合算法同时解决了鱼群算法后期寻优效率较低以及退火算法难以大范围搜索的问题,且算法稳定性较高,能有效提高电磁式互感器J-A模型参数辨识的时效性与精确度。

In order to research on the characteristics of electromagnetic transformer in the simulation platform,it is necessary to establish a precise and reliable magnetic hysteresis model for test transformer.Jiles-Atherton(J-A)model is widely used in the field of hysteresis modeling and simulation experiments of ferromagnetic materials.The accuracy of the five key parameters of the J-A model directly affects the fitting degree of the J-A model and tested sensor.This paper proposes an improved hybrid intelligent optimization algorithm,combining the artificial fish swarm algorithm(AFSA)and simulated annealing algorithm(SAA),which can identify the parameters of J-A model.In the improved hybrid algorithm,the variable-step AFSA is employed to quickly locate the search domain in the vicinity of the global optimal result.When the fitting degree reaches a certain accuracy,the parallel SAA is used to search the global optimal result precisely in a small range.Simulation results show that the improved hybrid algorithm is able to solve the problem that the AFSA is not efficient enough,and it can effectively improve the time-validity and the accuracy of J-A model parameter identification with higher stability.