Power Transformer Abnormal State Recognition Model Based on Improved K-Means Clustering

Abstract

At present, most of the data of the running transformers are normal state data. In order to utilize the historical normal state data to identify efficiently whether the new data are abnormal, a power transformer abnormal state quick recognition model based on improved K-means clustering, is proposed in this paper. To solve the problems of traditional K-means clustering, an improvement about choosing K value and initial cluster center based on data density and distance is proposed. The improvement can make K-means clustering get stable cluster centers and K value to greatly decrease the times of iterations and make the process of clustering quicker, more stable and efficient. Most of the power transformer data are normal state data. And the normal state data gradually change according to a certain trend while the abnormal state data change rapidly. Based on the historical normal data and improved K-means clustering, a quick recognition model for power transformer is established. According to the clustering results of normal data, thresholds can be calculated to identify new data. The example analysis shows that the improved algorithm can effectively identify the abnormal state of the power transformer quickly and accurately.