基于变压器最优漏磁信息的绕组变形分类方法

为了判别电力变压器绕组变形状况,提出一种基于最优漏磁信息及多层聚类编码和最优权重译码的ECOC分类器对变压器绕组变形分类的方法,首先利用多层聚类算法构建最优ECOC编码矩阵,然后通过最优权重译码算法对分类器输出进行译码得出变压器绕组具体变形类别,建立变压器绕组二维漏磁场有限元模型,计算得出绕组可能出现的变形形式和绕组区域变形前后的磁感应强度值数据,从而得出磁感应强度测量点;最后利用所建立模型得出的绕组变形数据进行仿真判别。结果表明,所建立的分类器在绕组变形判别时具有较高的准确性,可用于变压器绕组变形类型的检测。

Classification Method of Winding Deformation Based on Transformer Leakage Field Information

In order to determine the deformation status of power transformer windings, a classification method of the transformer windings deformation was proposed based on the magnetic field measurement and the improved ECOC classifier. The ECOC classifier based multi-level clustering coding and optimal weighting was proposed to classify the transformer winding deformation. Firstly, a multi-layer clustering algorithm was used to construct the optimal ECOC coding matrix. And then the optimal weight coding algorithm was adopted to classify the output of the classifier decoding transformers derived from the specific types of deformation. The finite element model of two-dimensional leakage magnetic field of transformer winding was established. The possible deformation forms of the windings and the values of the magnetic induction intensity before and after the deformation of the winding area were calculated to obtain the magnetic induction intensity measurement points. The simulation results show that the established classifier has high accuracy in the determination of winding deformation and can be used to detect the type of transformer winding deformation.