Artificial neural network applications high voltage technique

11 ARTIFICIAL NEURAL NETWORK APPLICATIONS IN HIGH VOLTAGE TECHNIQUE (M.Sc. Thesis) Süleyman Sungur TEZCAN GAZI UNIVERSITY INSTITUTE OF SCIENCE AND TECHNOLOGY July 2004 ABSTRACT Sulphur-hexafluoride (SFö) is an electronegative gas used frequently for circuit breakers in power systems because of its high dielectric strength and good thermal conductivity. Argon (Ar) is also used for pulsed power switching applications because of its low arc inductance and its ability to conduct high currents. The major disadvantage of argon is its low dielectric strength. This disadvantage can be overcome by mixing argon with SFö. In this study, the breakdown process in gases are theoretically investigated and employing the artificial neural networks, SFö-Ar gas mixture breakdown voltages estimated with a program written in Matlab. After the calculation of the effective ionization coefficient for SFg-Ar gas mixtures using ionization coefficient values of SFg and argon from the literature, the breakdown voltages for the gas mixtures are calculated theoretically. The results obtained are in good agreement with the measurements ones given in the literature. Science Code : 6080207 Key Words : Artificial neural network, high voltage, discharge in gases, SFö and Argon Page Number : 85 Advisor : Prof. Dr. M. Sezai DİNÇER