الفهرس Only 14 pages are availabe for public view
 |  | from | 109 |  |  |
| | | from | 109 | | |
Abstract
Nowadays, the recent fault detection t~chniques are using high frequency components instead of the fundamental signal components. This leads to a new scheme named transient based protection. In such a protection system, it is essential that the fault signal has to be analysed accurately. Fourier Transform was used in the early stage as a mathematical tool to analyse the transient fault signal. However, it has been shown that the analysis based on Fourier Transform, sometimes, is not precise enough. In recent years, Wavelet Transform has been used extensively for signal and image processing. It has been found that the Wavelet Transform is capable of investigating the transient signals generated in a power system in a more accurate and explicit way.
In this thesis a wavelet-based protection scheme is proposed for the purpose of classifying the faults and identifying its locatiop. The proposed technique uses wavelet transform in order to extract the high frequency components present in the transient waveform.
The fault classification is based on detecting distinct frequency bands contained in the transient voltage signal using a sampled data window with a sampling rate 0 f 3 .2 kHz. Discrete wavelet transform. is used to capture the high frequencies in the transient voltage signal. The spectral energies of these frequencies are obtained and used to classify the fault lype according to predefined pattern recognition. An additional condition is added to distinguish between the three type of double phase to ground faults.
Fault location is done using discrete wavelet transform of sampled data at a rate of 204.8 kHz. The coefficients of a frequency band in the range of 102.4 - 51.2 kHz are obtained from the sampled data of the three phase voltages. The problem of far and close faults is treated with acceptable error in the results.
The basic principle of the protection scheme is described. Simulation studies are presented for various fault conditions, fault types, and fault locations and accurate results are 0 btained. T he scheme response to some critical cases such as faults occurring near zero voltage and faults occurring close to the measuring point is studied and satisfactory results are obtained.
Fault simulations are performed using PSCAD/EMTDC and the results are then interfaced to MA TLAB where the algorithm is implemented. It is found that the proposed method gives reliable results for fault classification and fault location.