Localização e classificação de faltas em linhas de transmissão utilizando o Método dos Mínimos Quadrados e Inteligência Computacional

• Ano de defesa: 2013
• Autor(a) principal: Carlos Alberto Severiano Junior
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Minas Gerais; UFMG
• Palavras-chave em Português: Localização de faltas; Estimativa de impedância; Linhas de transmissão; Classificação de faltas; Engenharia elétrica
• Link de acesso: http://hdl.handle.net/1843/BUOS-9ACGZT

This work analyses fauts in power transmission lines and presents techniques for fault location and classification of the causative events (lightning or fire). The presented fault location technique was developed from the application of the Kirchhoffs Voltage Law to an equivalent circuit. The method, which defines the distance between the fault and one of the line terminals, also estimates fault impedance. Its formulation followsthe model of an optimization problem whose solution is obtained by means of the Least Squares Estimate. It can work with a window of voltages and currents, making the algorithm less sensitive to noisy signals. The model presented for classification of faults based on the causative event uses some information originally applied to problems offault location or fault classification under its electrical nature (such as line-to-ground and line-to-line), since the kind of fault studied in this work is still underexplored. The technique evaluates some features extracted from current and voltage signals such as fault duration, impedance, harmonic energy and decompositions under the timefrequency domain. The features are preliminarily ranked by a feature selection method.According to the results of the selection, the most relevant features are used to develop a classifier model based on Support Vector Machines (SVMs). The presented techniques are evaluated by simulated data and real fault records provided by a power supplier company.