Self-Organizing Neural Network architecture to detect HI faults in power distribution lines
Conference Paper
Overview
Additional Document Info
View All
Overview
abstract
A common problem associated with distribution feeders is power the High Impedance Fault (HIF). A number of algorithms and some neural networks have been proposed to detect these faults, but to date, none of these techniques can conclusively claim that an HIF will always be detected. This paper describes a Self-Organizing Neural Network (SONN) based approach to develop a system to detect HIF more discriminately. Kohonen's feature mapping was used to create a two-dimensional topology preserving map of input data. Once features were mapped onto the processing elements, we generated a pattern joining the selected processing elements and generated different patterns for different types of events represented by the input data.