Uma nova abordagem para a estimação da seção em falta em sistemas elétricos de potência através da geração de padrões de causa efeito em tempo real

Abstract

This work proposes a dynamic tool for solving the problem of fault section estimation in power systems. A Binary Integer Problem (BIP) is used to identify the faulty section trough analyses of circuit breakers states and power system protection devices signalizations. For the knowledge base construction it is proposed an innovative tool able to automatically generate the pattern of events and alarms for each topology configuration change. The outputs obtained from the pattern generator are used as base of cause end effect by the BIP model. Received alarms on SCADA are used as parameters in model the restrictions. The BIP model, based on the parsimonious set covering problem, is solved by CPLEX commercial software and it is able to deal with protection devices failure, data acquisition problems and occurrence of multiple events. To validate the approach it was used part of a brazilian power system, through the proposed technique the patterns were automatically generated and several faults situations were simulated. The proposed methodology achieved optimum results to all tests applied, being capable to deal with topology changes on power systems and with the problems inherent to the fault section estimation.