基于岭估计的变压器绕组参数辨识

变压器绕组参数辨识中普遍采用的方法是最小二乘法。这个方法仅能在变压器过励磁、空载合闸或外部故障等暂态过程中，辨识出变压器绕组的电阻和漏感，且电阻辨识的误差较大，而在变压器正常运行工况下，由于辨识模型的系数矩阵呈现病态，此时仅能辨识漏感，电阻值无法辨识，最小二乘法失效。为了解决此病态问题，本文将岭估计法应用于变压器绕组参数辨识中，并将所得估值与最小二乘估值进行对比，大量EMTP仿真及MATLAB验证结果表明，该方法能有效改善系数矩阵的病态性，使其在稳态过程中能有效辨识出变压器两侧绕组参数。

The Winding Parameter Identification of Transformer Based on Ridge Estimation

Least- squares method is the commonly adopted method for the winding parameter identification of transformer. This method can only recognize the transformer winding resistances and leakage inductances during transient state process, such as over excitation, no -load switching on or external fault, and the resistances encounter larger identification errors. Also, when the transformer is working on the normal state, it can only recognize leakage inductances not resistances, due to the coefficient matrix of the identification model are morbid. The Least - squares method will be invalid at this moment. To solve this morbid problem, ridge estimation is applied to the winding parameter identification of transformer in this paper. Through comparison to the Least - squares identification result by using EMTP simulations and MATLAB program, it concludes that ridge estimation can improve the morbid of the coefficient matrix effectively, and it can identify the transformer winding parameters at the normal state availably.