الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 142 |  |  |
| | | from | 142 | | |
Abstract
When artificial intelligence AI was born in the mid-fifties, the primary concerns were centered on game playing, planning, and problem solving. In the last decades, the most important applications areas of AI centered on knowledge engineering.
Nowadays, modern utilities install high speed data acquisition equipment in the distribution substations. These equipment may be used in recording the transient waveforms of both voltage and currents due to any disturbances and transmit them to the control center through supervisory control and data acquisition SCADA system. In this way, all the recorded data will be stored and be available for analysis in a single information center. So it helps the dispatchers to do their work in a quicker and easier way.
The major objective of the present work is to develop algorithms for detecting and classifying the faults. Also, develop an algorithm for identifying the fault location using Artificial Neural Networks ANN`s, which consider one of the most important linds of the Artificial Intelligence techniques. These algorithms can be included in a digital fault recorder in the main control center to help the dispatcher to restore the network after any abnormal conditions.
This work contains the use of the statical function as a single processing for the phase currents. Many cases of processed signals for different fault types are used as input training and testing vectors of Artificial neural network ANNs. Current signals are obtained using several simulation studies developed by the use of Matlab 6.0 simulink-toolbox. The ANNs are trained off line using Matlab 6.0 neural network-toolbox, and gave excellent results.
Researchers in power systems have investigated fault locations techniques for a number of years, by accurately locating a fault, the service can be restored quickly. The work include fault locator algorithm based on calculating the magnitude of the ratio between the positive sequences for voltages measured at the two ends of the line, due to the fundamental frequency 50hz. This ratio is defined as distance factor. The proposed technique uses a simulated network to evaluate the fault voltages impostd at the two line ends. The distance factor and status of lines are used as training and testing vector of considered ANNs.
ANNs are trained off line with taking into account the outage of one or two line due to cleaning or testing of line`s equipment. The work includes includes also the mathematical background for the network under fault conditions.
Finally, both the fault detection with classification algorithm and fault location algorithm are connected together in one M-file algorithm under Matlab 6.0 media. This work was rested using 230 kv network. The network consists of 7-buses connected through8-line. The simulated process is applied using matlab6.0 power system blocksets-simulink toolbox.