Clasificación de perfiles de carga en consumidores comerciales mediante análisis de conglomerados

Abstract

This work presents methods for segmenting electric load profiles of commercial users, allowing to choose the most relevant one according to the types of data. The identification and understanding of consumption profiles are valuable for analyzing energy efficiency, tariff adjustments, planning, and decision-making. Therefore, this study introduces a method to assess the performance of K-means clustering, Hierarchical Clustering, Fuzzy C-means, Self-Organizing Map, Gaussian Mixture Model, and Decision Tree techniques applied to customer load shape factors. The results indicate that the Self-Organizing Map and Fuzzy C-means algorithms show homogeneity in their categorization. In all three case studies, the results are evident due to their low percentage of outliers.