الفهرس 
AbstractOnly 14 pages are availabe for public view
 |  | from | 130 |  |  |
| | | from | 130 | | |
Abstract
Electric power systems are frequently subjected to different failures. The majority of these failures, i.e. faults, are related to transmission lines. It can cause undesirable damage of power system. So, to deal with such problem, it is important to detect fault, but just detect it is not enough to be quickly fix it. To solve the problem quickly, the fault must be classified, and located. Many methods have been processed to detect, classify, and locate it in transmission network. Artificial Intelligent (AI) techniques have been used to detect, classify and locate the faults in the transmission lines. Artificial neural network (ANN) and Support vector machine (SVM) are the most AI techniques used in pattern classification problems. Some of these techniques workout using the voltages and currents got from the local end, the other uses the voltages and currents from two ends. ANNs have been proposed in this thesis to detect, classify, and locate the fault for two different transmission networks. One system is homogenous single circuit transmission line (an overhead transmission line), and the second is a non-homogenous single circuit transmission line (the overhead lines combined with underground cables). The proposed ANNs have been designed, trained, and tested for these two transmission systems which are simulated by Matlab/Simulink. The data used to train and test the proposed ANNs are the three phase currents and voltages taking from one end of the Transmission line. It is used train and test three ANN modules one for detect, second for classify, and the third for locate the fault in transmission system. These modules have been tested for the fault present giving promising results. SVM has been proposed also for the two types of the transmission system using the same data collected from one end of the transmission line. SVM also has its module to detect, classify, and locate. It is shown that using SVM give an accurate results than using ANN.