Abstract:
This thesis is a study of Partial Discharge(PD) Pattern Recognition in High Voltage Equipment program to automatically diagnosed. PD Pattern namely Corona, Internal and External discharge were measured from defective modeling. The calculated compulsory characteristics are stored in database by utilizing statistical and fractal method. The result of distribution of average value and repeated existing of PD corresponding voltage phase angle, provided the statistical value. In addition, the latter method was obtained from the distribution value consisting the correlation between the amplitude of PD, the voltage phase angle and the repeated existing of PD. Using these methods, statistical and, fractal method. The 13 parameters, inputted to Neural Network to training system, were obtained. Moreover the system recognition is utilized Genetics Algorithm to determine the appropriate weight for Neural Network as well. The tested result shown more than 80 percent correct diagnosis.