GIS Partial Discharge Pattern Recognition Based on Dimension Reduction Based on Rough Set Theory
Abstract
It is an important precondition to judge the relationship between partial discharge (PD) type and PD envelope signal so as to evaluate the insulation condition of gas insulated switchgear (GIS) and formulate reasonable maintenance strategies. Utilizing the dimension reduction based on rough set theory, this article performed discretization and differential matrix reduction on the feature matrix composed of 37 feature vectors that were used to characterize UHF PD envelop signals. Tightly followed, reduced feature vectors, together with a BP neural network classifier, were used to perform the pattern recognition on the four different types of UHF PD envelope signals. The results revealed this method had relatively higher recognition rate.
Keywords
UHF PD envelop signal, PD type, Pattern recognition, Dimension reduction based on rough set theory, BP neural network
DOI
10.12783/dteees/peem2016/5028
10.12783/dteees/peem2016/5028
Refbacks
- There are currently no refbacks.