A Simple Method for Fault Classification Based on Two Stages of Self Organizing Map

Recently, the use of algorithms based on artificial intelligence for fault analysis in power systems has had a great increase mainly due to the ability of these techniques to model systems with non-linear behavior such as an electrical system. This paper presents a method for fault classification in...

Full description

Saved in:
Bibliographic Details
Main Authors: Guama, Edison G., Claros, Ivan D.
Format: Conference Proceeding
Language:English
Subjects:
Analytical models
Fault Analysis
Fault diagnosis
Power transmission lines
R-DFT
Self Organizing Maps
Self-organizing feature maps
Symmetrical components
Training
Transmission line measurements
Unsupervised Learning
Voltage measurement
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Recently, the use of algorithms based on artificial intelligence for fault analysis in power systems has had a great increase mainly due to the ability of these techniques to model systems with non-linear behavior such as an electrical system. This paper presents a method for fault classification in transmission lines based on 2 stages of Self Organizing Maps: the first one determines if the fault is to ground using the voltage and current of zero sequence, and the second one identifies the failed phase through the phase current. The measures of current and voltage signals, available in the protection devices through the COMTRADE standard format, were used as model inputs. Data processing is realized using symmetrical components and the R-DFT algorithm, model settings and training are also reviewed in detail. Finally, the proposed method was evaluated through a case study under several real failure conditions, the results confirm the effectiveness of the prooosed method.
ISSN:2643-8798
DOI:10.1109/ISGTLatinAmerica52371.2021.9543044