TS-HDBSCAN Based Load Profile Characteristics of Residential Consumers for Demand Response Applications

Demand response plays a vital role in the electric grid balancing mechanism. The deployment of smart meters motivates residential consumers to participate in demand response programs. Residential consumers have an inherent variability in consumption patterns and have variable shapes, sizes, and densities. The consistency of users' consumption habits over time acts as a vital part of the demand response program. One must analyze the consumers’ load profile characteristics to shift the demand from peak to off-peak hours. Consumers’ consumption behaviors are commonly described using clustering techniques due to the presence of a large number of datasets. However, traditional clustering methods, such as K-means, hierarchical clustering, and Density-Based Spatial Clustering of Applications with Noise (DBSCAN), are less effective with datasets of varying density. To address this, the time series Hierarchical Density-Based Spatial Clustering of Applications with Noise (TS-HDBSCAN) method is proposed for load profile analysis. This approach supports clustering across different shapes, sizes, and densities by creating a condensed hierarchy and selecting clusters based on stability. In this study, the TS-HDBSCAN method is applied to consumption data from 651 residential consumers, demonstrating its suitability for profile selection, classification, and operational stability.