Optimal Clustering of Energy Consumers based on Entropy of the Correlation Matrix between Clusters

TL;DR

This paper presents a novel clustering method for energy consumption data using entropy-based metrics, genetic algorithms, and self-organizing maps to identify optimal customer groups efficiently.

Contribution

It introduces a new metric for determining the optimal number of clusters and combines genetic algorithms with self-organizing maps for improved clustering of smart meter data.

Findings

Effective identification of optimal clusters in energy data

Demonstrated method's success on datasets from Australia and Ireland

Improved energy demand management through better customer segmentation

Abstract

Increased deployment of residential smart meters has made it possible to record energy consumption data on short intervals. These data, if used efficiently, carry valuable information for managing power demand and increasing energy consumption efficiency. However, analyzing smart meter data of millions of customers in a timely manner is quite challenging. An efficient way to analyze these data is to first identify clusters of customers, and then focus on analyzing these clusters. Deciding on the optimal number of clusters is a challenging task. In this manuscript, we propose a metric to efficiently find the optimal number of clusters. A genetic algorithm based feature selection is used to reduce the number of features, which are then fed into self-organizing maps for clustering. We apply the proposed clustering technique on two electricity consumption datasets from Victoria, Australia and Ireland. The numerical simulations reveal effectiveness of the proposed method in finding the optimal clusters.