Arturo S. Bretas, Karen C.O. Salim, and Rodrigo H. Salim
Fault diagnosis, wavelet transforms, artificial neural networks, underground distribution feeders, power systems protection
This paper presents further formulation details of an implemented hybrid fault diagnosis method for unbalanced underground distribu- tion systems (UDS). The proposed formulation is hybrid approach based, using both artificial neural networks (ANNs) and wavelet transforms (WTs) for the fault diagnosis process. Traditional fault location and detection approaches are digital Fourier transforms based, which have an inverse time–frequency resolution relationship that provides a low level of robustness to the fault diagnosis process. UDS are characterized by having a significant shunt capacitance component, non-symmetrical and non-transposed lines, time-varying loads and single or double phase laterals. These characteristics make UDS feeders operation highly unbalanced, which compromises the traditional digital Fourier transforms based fault diagnosis methods performance. This paper describes further formulation details of an implemented hybrid fault diagnosis method and discusses the obtained test results. The test results demonstrate the techniques capability and robustness and its potential for on-line applications.
Important Links:
Go Back
