Slope-Based Shape Cluster Method for Smart Metering Load Profiles
Yue Xiang, Juhua Hong, Zhiyu Yang, Yang Wang, Yuan Huang, Xin Zhang, Yanxin Chai, Haotian Yao
Abstract
Cluster analysis is used to study the group of load profiles from smart meters to improve the operability in distribution network. The traditional K-means clustering analysis method employs Euclidean distance as similarity measurement, which is insufficient in reflecting the shape similarities of load profiles. In this letter, we propose a novel shape cluster method based on the segmented slope of load profiles. Compared with traditional K-means and two improved algorithms, the proposed method can improve the clustering accuracy and efficiency by capturing the shape features of smart metering load profiles.
Topics & Concepts
Cluster analysisMetering modeOperabilityEuclidean distanceSimilarity (geometry)Cluster (spacecraft)Computer scienceData miningMetric (unit)Pattern recognition (psychology)EngineeringArtificial intelligenceComputer networkOperations managementMechanical engineeringImage (mathematics)Software engineeringElectricity Theft Detection TechniquesSmart Grid Energy ManagementWater Systems and Optimization