Avaliação de técnicas de aprendizado de máquina para detecção de eventos em sistemas elétricos de grande porte a partir de dados de medição fasorial sincronizada

Abstract

The growing need to ensure the efficiency and reliability of the National Interconnected System (SIN), which represents the backbone of the country’s energy distribution, is what drives this work. Faced with the challenges posed by the territorial vastness and complexity of the Brazilian electrical grid, the use of Phasor Measurement Units (PMUs) emerges as a promising solution for real-time network monitoring. However, the effectiveness of this monitoring is linked to the ability to detect and respond to anomalies quickly and accurately, minimizing the risks of failures and supply interruptions. The study addresses the challenge of managing and evaluating the country’s interconnected electrical grid, where accurate information is pertinent for preventive and corrective actions in a continental-scale distribution system. The core of this research lies in the innovative exploration of advanced data compression techniques combined with unsupervised machine learning algorithms, aiming to optimize the interpretation and analysis of the large volumes of data generated by the PMUs. This approach points to a significant improvement in the quality and precision of the information extracted and offers a scalable solution to the challenge of processing and analyzing the data from a large-scale distribution system. The effectiveness is assessed through the detection of significant events and anomalies in the network at points geographically distant from the event’s origin. The results of this study validate the efficacy of the proposed algorithms, highlighting their practical relevance and substantial impact on improving the quality and reliability standards in the electricity supply. They demonstrate the feasibility of implementing these techniques in real scenarios, showcasing the potential for transformation in failure prevention and critical event management, contributing to a more stable, effective, and secure energy distribution network.