الفهرس 
INTRODUCTIONOnly 14 pages are availabe for public view
 |  | from | 177 |  |  |
| | | from | 177 | | |
Abstract
The insulating oil performance can be enhanced in high voltage apparatus using barriers. The importance of the barrier in increasing the dielectric strength of the insulating oils in order to reduce the oil failure stresses has not been sufficiently studied. The Influence of barriers on the breakdown voltage under Inhomogeneous field conditions (point-plane electrods), and moderately homogenous field conditions (plane-plane, hemisphere -hemisphere, and sphere- sphere), with and without barrier, under AC voltage was investigated.
An experimental setup was constructed to study the effect of the barrier parameters on the dielectric strength of the insulating oil. These parameters are, the gap space, the barrier location relative to the high-voltage electrode, barrier diameter, barrier thickness,electrode configurations, the presence of contaminating particles, the weight of the contaminating particles, the temperature of the insulating oil, a hole within the barrier, the number of the barriers in the gap space, and the inclined angle of the barrier with the horizontal axis.
Two mathematical models were built to predict the magnitude of the breakdown voltage with the variation of the aforementioned parameters. The first model was the Artificial Neural Network model which was built using 784 experimental results with another 25 samples for testing the model. There were 8 parameters as the inputs such as, the presence of contaminating particles, and breakdown voltage as an output. The second model was the Box Behnken Design model which needed only 15 experimental results to build it. In this model, a prediction equation was constructed to develop the relation between the gap space, the barrier location relative to the high-voltage electrode, barrier diameter, and breakdown voltage. Some statistical analysis was performed to indicate the significance of the studied parameters on the magnitude of the breakdown voltage such as a cumulative distribution function, probability distribution function, and t-test.
The results indicated that the prediction Artificial Neural Network model had a high ability to expect the breakdown voltage for other different experiment cases. The results obtained from Box Behnken Design model showed that minimum prediction errors can be obtained.