An Improved Subtractive Clustering Framework for Load Profile Analysis: Integrating Time-Frequency Feature Selection and Cluster Optimization

Electrical load profile (ELPs) analysis is an important foundation for data-driven decision-making, such as effective power system capacity planning and dynamic tariff development. However, one of the challenges in ELPs analysis is finding the right method to accurately and meaningfully group consumers into segments that represent similar consumption patterns. This study proposes a framework for consumer segmentation based on load profile patterns. The first stage uses a Discrete Wavelet Transform (DWT) approach to extract representative features from the daily electricity consumption data in the time-frequency domain. The second stage uses the extracted features as input for a segmentation process using improved Subtractive Clustering (SC), where the radius parameter is adaptively optimized using Particle Swarm Optimization (PSO) to obtain the optimal number of high-quality clusters. An empirical analysis and several evaluation metrics were used to assess the proposed framework. The results indicate that the proposed framework can improve the analysis efficiency and cluster quality in ELPs clustering and provides a systematic approach for addressing the specific characteristics of time-series data.