Summary
The oscillating voltage test method is widely used in partial discharge detection for its non-destructive characteristics, and the pattern recognition of partial discharge signal of different defects is the key to the detection of partial discharge. According to the common cable defect type, four kinds of artificial defect models of cable joints such as outer semi-conductive layer cutting-edge, outer semi-conductive layer air gap, high potential metal tip and the stress cone lap too long in 10 kV cross linked polyethylene cable are made, oscillating voltage is applied to them and the characteristics of partial discharge is tested. A recognition method on back propagation artificial neural network is introduced to apply to four kinds of artificial defect models in this paper, which using statistics operators of positive and negative half wave of partial discharge signal as input of back propagation neutral network. The result shows that partial discharge signal of different defects cable has significant difference under oscillating voltage. Using statistics operators of positive and negative half wave as input of back propagation neutral network can reflect the characteristics of partial discharge information well. The pattern recognition method of back propagation neutral network can identify partial discharge pattern of various defects and has considerable value for practical application.
Additional informations
| Publication type | ISH Collection |
|---|---|
| Reference | ISH2015_53 |
| Publication year | 2015 |
| Publisher | ISH |
| File size | 506 KB |
| Price for non member | Free |
| Price for member | Free |
Authors
SAVIC, HADŽIC, Hrólfsdóttir
