基于加权模糊C均值聚类算法的变压器故障诊断

Fault Diagnosis on Transformer Based on Weighted Fuzzy C-means Clustering Algorithm

模糊C均值聚类(FCM)算法具有将数据集合进行相等划分的趋势,每一个样本对数据集分类的影响相同。在变压器绝缘故障中,不同故障类型产生的主要特征气体及气体组分含量存在很大差异。因此,为了区别各类数据对故障划分的影响程度,可考虑对各类数据施加一个权。文中提出了一种加权模糊C均值聚类(WFCM)算法,该算法可实现故障聚类。与FCM算法相比,WFCM算法明显提高了故障划分的正确性和鲁棒性。

The fuzzy c-means clustering （FCM）algorithm has the trend of dividing data sets equally,the effect of each sample on classification of the data sets is the same. In the insulation fault of transformer, different types of fault produce different main characteristic gasses and the components. Therefore,in order to distinguish the influence degree of all kinds of data on fault partitioning ,this paper uses the weights to express the relative degree of the importance of various data in fault partitioning, and presents a weighted fuzzy c-means clustering algorithm which can accomplish fault clustering. Compared with FCM algorithm, the proposed algorithm obviously improved the accuracy and robustness of the fault partitioning.