Partial disharge recognition using artificial neural network
Partial discharge (PD) seriously affects the reliability of the distribution system due to electrical stress and the duration of the installation. Recent technology advance brings the analysis of the PD act as the guideline and maintenance strategy can be carried out when a parameter exceeding the p...
Saved in:
| Format: | Thesis |
|---|---|
| Language: | English |
| Subjects: | |
| Online Access: | http://dspace.unimap.edu.my:80/xmlui/bitstream/123456789/72711/1/Page%201-24.pdf http://dspace.unimap.edu.my:80/xmlui/bitstream/123456789/72711/2/Full%20text.pdf http://dspace.unimap.edu.my:80/xmlui/bitstream/123456789/72711/4/Nur%20Afifah.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Partial discharge (PD) seriously affects the reliability of the distribution system due to electrical stress and the duration of the installation. Recent technology advance brings the analysis of the PD act as the guideline and maintenance strategy can be carried out when a parameter exceeding the predefined level. This thesis presents an artificial neural network (ANN) modeling in recognizing the PD signal. PD signals are generated from experimental measurement and simulation by using electromagnetic transient program-alternative transient program (EMTP-ATP). There are two analyses are carried out; classification and de-noising of PD signal. The first analysis is aim to discriminate between PD and noise signals. Multilayer perceptron with back propagation algorithm is used to perform this task. The result shows that the number of nodes in hidden layer affects the accuracy of classification. Second analysis presents the de-noising performance of PD signal using three different techniques; ANN, fast Fourier transforms (FFT) and discrete wavelet transform (DWT). The objective of this analysis is to yield the PD signal from the measured signal which is the combination of PD and noise signals. Only PD signals generated from EMTP-ATP simulation environment is considered. The de-noising algorithm is implemented to discover a clean PD signal from disrupted signal. The performance of the de-nosing techniques was evaluated by comparing the signal to noise ratio (SNR). In order to de-noise the disturbed PD signal, the knowledge of interference peak needs to take into account. The result of this analysis shows ANN is the best de-noising technique as all the other techniques produce a peak higher than PD signal peak. |
|---|