الفهرس 
AbstractOnly 14 pages are availabe for public view
 |  | from | 150 |  |  |
| | | from | 150 | | |
Abstract
Power transformers are important electrical equipments that need fast protection, because of their essential role in power system operation and their expensive cost. The most common technique used to protect a transformer is the differential relay, but it doesn’t provide discrimination between internal faults and inrush currents.
This dissertation is a systematic study of artificial intelligence (AI) applications for the diagnosis of power transformer incipient fault. The AI techniques used in this thesis include artificial neural networks (ANN), support vector machine (SVM) for monitoring the three phase two windings transformer. Distinguishing between the magnetizing inrush and internal fault currents is always a consideration for power transformer protection. Existing methods, mostly doesn’t based on harmonic restraint, are not very reliable for modern transformer protection.
In this work, a comparison among the performances of three neural networks based classifiers is presented. These networks are: FFBPNN (feed forward back propagation), cascade-forward back propagation network (CFBPNN), and recurrent network (RNN). The test results prove that the RNN is stable and give good behaviors for different fault conditions. It is good reliable for recognition of transformer inrush and internal fault currents.
The SVM has two output states; The SVMs fault detection designed as a four cascade SVM to discriminate between different conditions. The first SVM1 is trained to identify the normal state. The second SVM2 is trained to identify the internal fault. The third SVM3 is trained to identify the saturation state. The fourth SVM4 is trained to identify external fault.
The results of SVM are compared with corresponding results obtained by applying RNN. The comparison proves that the SVM algorithm is more accurate and faster in detecting and classifying different transformer conditions.