Determining representative load curves for consumer's classes by cluster analysis

Abstract

Performing new electronic meters installed at consumer substations allows obtaining a large set of daily load curves. This encourages the representation of load profiles for different types of consumers in South of Vietnam. The paper presents an approach to define the representative daily load curves based on fuzzy clustering algorithms. Fuzzy k-means algorithm and combination of 5 indexes on cluster validity will classify all load curves into different classes. From these classes the representative load curve for consumer, group of consumers are formed.