Pemprosesan penggera dalam perlindungan sistem kuasa melalui neuro-kabur
This research develops alarm processing in a methodology combination of Probability Neural Network (RNK) and Fuzzy Relation (HK) is named as Neuro- Fuzzy (NK) in power system protection. In a power system protection, alarm processing is illustrated as an input of alarm pattern. While fault occurs, o...
Main Author: | |
---|---|
Format: | Thesis |
Published: |
2007-11.
|
Subjects: | |
Online Access: | Get fulltext |
Summary: | This research develops alarm processing in a methodology combination of Probability Neural Network (RNK) and Fuzzy Relation (HK) is named as Neuro- Fuzzy (NK) in power system protection. In a power system protection, alarm processing is illustrated as an input of alarm pattern. While fault occurs, operated protective devices will send fault information in the form of alarm pattern. The RNK has alarm pattern as an input for every neuron output. RNK is also responsible to calculate the membership degrees of component system at faulted component class. On the other hand, HK is expended based on rule base of Sagittal Diagram. The Sagittal Diagram has been developed using HK equation known as parametric operators to find the membership degree. These membership degrees mean to detect fault and non fault conditions in alarm processing. NK method has been used to calculate the membership degrees for the existence of fault. The developed method has been conducted on several IEEE systems and also Perusahaan Listrik Negara (PLN) Sumbar-Riau system. The simulation results give the membership degrees of NK method close to ideal value as compared to HK and RNK methods. Ideal of membership degrees indicates the best application in power system protection. The percentage of working system of NK method is also found higher than that produced by HK and RNK methods. This revealed that the NK method gives the best working system to fault detection. |
---|