变压器分层绕组模型变焦遗传算法的参数辨识

电力变压器分段绕组的传递函数模型可用来实现局部放电的精确定位，受制于测量条件，其模型参数辨识较为困难，文章应用遗传算法来解决电力变压器分段绕组模型参数辨识的问题，在绕组套管上施加一脉冲信号，通过在接地端测量其响应，在分段绕组的RLC数学模型基础上，假定分段绕组各单元的绕组之间绝缘损耗RS，对地电阻Rg，线圈之间电容Cs，线圈对地电容Cg，绕组损耗R和漏抗L相等，分段绕组之间的互感M通过一个电流控制的电压源模拟，对变压器分段绕组数学模型的参数进行辨识，同时进一步研究了参数变化范围对辨识结果的影响，引入变焦算子大搜索范围与辨识精度之间的矛盾，辨识结果表明，该方法是可行的。

Parameter Identification for Transformer's Sectional Winding Model Based on Zooming Genetic Algorithms

Transfer function of transformer's sectional winding can be used for the evaluation and localization of partial discharges(PD),but the estimation of the winding parameters is difficult when not knowing the sectional winding voltages.This paper applies genetic algorithms(GA) to estimate the parameters of sectional winding.The RLC models of transformer's sectional wind,which are based on transmission line theory,are improved by assumption that the six main circuit elements R_S,R_g,C_S,C_g,R and L are assumed identical in all sections while the mutual inductances Mi keep different.The transformer's frequency response to an impulse signal is used to estimate the parameters.A zoom operator is introduced to accelerate the GA's convergence.The results are compared with the results of least mean square error method and maximum likelihood method under the same condition.