الفهرس 
: SERIES COMPENSATED .LINE PROTECTIONOnly 14 pages are availabe for public view
 |  | from | 168 |  |  |
| | | from | 168 | | |
Abstract
Faults on senes compensated transmission lines need to be detected,
classified, and located accurately for being cleared as fast as possible.
This research presents an adaptive protection directional scheme by using
artificial neural networks for the series compensated transmission system.
The scheme is designed to detect the faults, classify the fault type and
identify the faulted phase for all types of faults.
The proposed topology of this scheme is composed of two levels of
neural networks. In level-I a neural network (ANNF) is used to detect the
fault, while in level-2, four networks (ANNA, ANNa, ANNe and ANNG)
are used to identify faulted phase(s). The output of ANNF activates
(ANNA. ANNa. ANNe and ANNG) if there is a fault. Therefore. the
proposed topology determines both the fault type and the faulted phase(s)
selection.
The proposed scheme is trained and tested using 10caJ measurements of
three phase voltage and current samples. These samples are generated
using EMTDC package. All ten possible fault .types are simulated. The
sampling rate is 16 samples per cycle of power frequency.
ANNs are trained by different methods until getting the proper number of
samples per input pattern and proper design of ANN.
The proper design for all ANNs used in this research consists of three
layers; an input layer having 24 input nodes (four recursive samples of
three phase voltages and currents), a hidden layer of 10 neurons, and an
output layer of one neuron.
The ANN approach is compared with a travelling wave-based relaying
technique for similar case studies. ANN shows higher resolution
regarding selecting faulted phases.Testing results proved that the proposed scheme is accurate, reliable and
more efficient.