Rough K-Means Incremental Clustering Algorithm And Its Application In State Evaluation Of Electrical Equipment

Electrical equipment is an important part of the power grid system,and its operating status directly affects the reliability of the power distribution system.With the rapid advancement of ubiquitous sensing technology in power distribution system,the amount of monitoring data for electrical equipment has increased dramatically.At the same time,with the continuous development of power system technology,the proportion of abnormal data relative to normal operation data continues to decrease,which intensifies the degree of imbalanced scale of monitoring data clusters,making it more difficult to extract abnormal features of electrical equipment operating status assessment.This study gives full consideration to the impact of the imbalanced cluster size,and a dynamic incremental clustering algorithm considering the imbalanced cluster size is proposed to improve the effect of incremental clustering on the data processing of electrical equipment.A state evaluation method of electrical equipment based on incremental clustering is also proposed to improve the accuracy of electrical equipment state evaluation.The main research contents of this paper are as follows:(1)Dynamic Incremental Clustering Algorithm Considering Imbalanced Cluster SizeWith the continuous development of the power grid and the increasing monitoring data,the monitoring data of electrical equipment shows that the proportion of abnormal data is small and the data scale is imbalanced.Aiming at the problem that the imbalance of the data set scale is not considered by the traditional incremental clustering algorithm when dealing with incremental data,a dynamic incremental clustering algorithm considering the characteristics of the imbalanced cluster scale is proposed.The algorithm fully considers the influence of the imbalanced degree on the update iteration of the cluster center and dynamically adjusts the membership degree.The experimental results show that the algorithm can correctly divide the newly added data points.(2)Dynamic Incremental Clustering Algorithm Based on Mixed Metrics of Neighborhood InformationThe key to improve the quality of clustering is the processing of cluster-cross boundary data samples.In order to reduce the adverse effect of the uncertain data information in the boundary region on the cluster analysis,a dynamic incremental clustering algorithm based on mixed metrics of neighborhood information is proposed.This study fully considers the attribution,local density and other information of the neighborhood data,and adds the distance measure of the disequilibrium adjustment parameter to solve the problem of dividing newly added data points with uncertain information in the case of imbalanced cluster size.The cluster is also merged and split with reference to the idea of density peak to improve the accuracy of the incremental clustering algorithm for the division of newly added data points.(3)State Evaluation Method of Electrical Equipment Based on Rough K-means Incremental ClusteringAccording to the application requirements of electrical equipment status assessment,a state evaluation method of electrical equipment based on rough K-means incremental clustering is designed.By analyzing the monitoring data of the equipment and quickly processing the continuously generated data online,the method in this paper can detect the changes in the status of the electrical equipment and possible safety problems in time.It also facilitates maintenance personnel to find problems in advance to ensure the stable operation of electrical equipment,and provides decision support for maintenance personnel to flexibly adjust the maintenance plan.