Proposta de metodologia baseada em sistema de suporte à decisão espacial e aprendizado não supervisionado para análise e mitigação de faltas em redes de distribuição modernas

Abstract

The availability and feasibility of deploying new technologies (smart meters, distributed generation and storage resources, monitoring structures, sensors, etc.), as well as the expansion of the topological structures of power systems (network extension and equipment) and the increasing pressure from the regulatory agency and society for a higher quality service resulted in a signifcant increasing in the complexity of the planning and operation tasks of electricity distribution networks. To allow an operation with quality, reliability and security in this environment, it is a consensus of the need to integrate the physical layer of the “physical network” network with the communication and information layer, “cyber network”. The way to perform this integration is through the so-called Smart Energy Networks or Smart Grids. This work proposes the development of a methodology that will allow to evaluate the performance of the technologies, as well as to mitigate faults in electric power distribution systems. The proposed method includes the evaluation of the main factors that influence the quality od the services. Network topology, availability of protective equipment, maneuvering, telecontrol and trafc in the access routes are evaluated together with the reliability of the power supply, in order to combine the results of these criteria with quality indicators, which will allow mitigating the impacts of the faults and propose more efcient short and medium term investments. The paper will also approach an analysis methodology based on Spacial Decision Support System, Database Knowledge Discovery Technique and Unsupervised Machine Learning, providing in a practical and easy way the comparison between feeders, substations or even regions that make up a distribution system. Keywords: smart grid, reliability optimization, fault mitigation, electrical distribution systems, database knowledge discovery, unsupervised machine learning, simulation of trafc on the access routes.