الفهرس 
INTRODUCTIONOnly 14 pages are availabe for public view
 |  | from | 165 |  |  |
| | | from | 165 | | |
Abstract
Power transformers are one of the most important parts of the electrical
network. Maintaining the safety of transformers and the safety of their
performance is of utmost necessity to ensure the continuation of proper
electrical feeding for consumers and to ensure the safety of operators and
the safety of the environment. One of the most serious issues confronting the
insulating materials inside the transformer is partial discharge (PD). The
partial discharge leads to slow and continuous erosion of the insulating
materials inside the transformer. Un-detected PD activity will cause
deterioration in insulation and could lead to catastrophic damage to the HV
equipment.
Electro-magnetic (EM) waves radiated by PD activity could also be
captured in the ultra-high frequency (UHF) range. The UHF detection
technique’s main advantages are its high immunity to environmental
background noise. Background noise in power plants is usually in the range
of a few kHz to MHz, which is below the desired operating range. Therefore,
in this thesis, a design, manufacturing, and validating process for the UHF
wideband 4th order Hilbert Fractal antenna to be used for capturing the
electromagnetic waves radiated from the PD source are presented. Then,
different artificial PD activity sources are created, simulating different
common defects that could be found in oil-immersed transformers. The
artificial PD sources are simulating typical faults such as corona discharge,
surface discharge, sharp edges in oil discharge, as well as internal oil void
discharge. Then, two alternative techniques, i.e., discrete Fourier Transform
(DFT) and discrete Wavelet Transform (DWT), are proposed to process the
captured signals and extract descriptive features. An investigated artificial
neural network (ANN) classifier will be capable of successfully differentiating among the study cases with recognition rates above 84%. The
DWT has shown higher recognition rates as compared to the DFT for the
same number of samples and test conditions.
Further, a numerical investigation of PD current pulses produced due
to applying a high voltage on the needle to plane geometry at atmospheric
air pressure using COMSOL Multiphysics is achieved. The investigation
aims to simulate the negative corona current pulses with the charged species
distribution in the gap during different current pulse stages. Three charged
species are taken into account; electrons, positive ions, and negative ions.
Two-needle tip radii are selected at 35µm and 250µm. The VI
characteristics, the pulse repetition frequency, and the temporal current pulse
characteristics are obtained. Then, the results are compared with previously
published research data.
Finally, an experimental study of the negative corona current pulses in
atmospheric air pressure is conducted. The impact of changing the negative
applied voltage on the average DC current, the pulse repetition frequency,
and the temporal current pulse characteristics is shown. Then, a comparison
between the results obtained from experiments and those obtained from the
numerical model is made. Good agreement is shown regarding the average
DC current values and the pulse shape. Then, it is confirmed through the
numerical model and experimental observations that the pulse temporal
characteristics, i.e., rise time, fall time, pulse width, and pulse peak to peak,
do not depend on the negative applied voltage.