Proposta de uma metodologia para o tratamento de alarmes e diagnóstico de falta em centros de operação e controle de sistemas de potência

Abstract

This dissertation proposes a methodology for alarm processing aiming to solve the fault section estimation in electrical power system. The main motivation for this study is the fact that operators of control centers being subject to information overload during great contingencies. Accordingly, this work aims to support the operator decisions in order to enhance the service reliability and reduce the power restoration time. The approach integrates a legacy SCADA interpretation system working together with a new network topology processor to determine the protection alarms (circuit breakers, switches, protective relays and protection logical schemes), and the set of disconnected equipment after occurrence of fault. The fault diagnosis has been treated as an optimization problem, solved through two stages: event classification at equipment level, based on Bayes s Theorem; and the fault section estimation, which is formulated as a mixed integer programming problem, using the commercial software CPLEX to solve instances. The developed approach also identifies the malfunctioned protective devices as well the missing and false alarms. Possible fault scenarios were considered in part of a real Brazilian power system to validate the methodology. The results show that the proposed approach can find the optimal solution even in case of multiple faults or in case of failure of protection devices.