Fault Diagnosis for Power System Transmission Line Based on PCA and SVMs - Intelligent Computing for Sustainable Energy and Environment - page 524

Fault Diagnosis for Power System Transmission
Line Based on PCA and SVMs
Yuanjun Guo, Kang Li, and Xueqin Liu
School of Electronics, Electrical Engineering and Computer Science,
Queen's University of Belfast, Belfast, BT9 5AH, UK
{ yguo01,k.li,xueqin.liu } @qub.ac.uk
Abstract. This paper presents the application of a fault detection
method based on the principal component analysis (PCA) and support
vector machine (SVM) for the detection and classification of faults in
power system transmission lines. Consider that the data may be huge
with a number of strongly correlated variables, method which incorpo-
rates both the principal component analysis (PCA) and support vector
machine (SVM) is proposed. This algorithm has two stages. The first
stage involves the use of the PCA to reduce the dimensionality as well as
to find violating point of the signals according to the confidential limit.
The features of each fault extracted from the data are used in the second
stage to construct SVM networks. The second stage is to use pattern
recognition method to distinguish the phase of the faulty situation. The
proposed scheme is able to solve the problems encountered in traditional
magnitude and frequency based methods. The benefits of this improve-
ment are demonstrated.
Keywords: Fault Diagnosis, Transmission Line Faults, Principal-
Component Analysis (PCA), support vector machine (SVM).
1 Introduction
In power systems, the transmission line is the vital link between the electricity
power production and usage. To find the accurate fault location in the trans-
mission line based on the measurement of the currents and voltages is of great
importance, since a more accurate location results in the minimization of the
amount of time spent by the line repair crews in searching for the fault [1].
Since the modern power system is well equipped with advanced measure-
ment and protection instruments, a huge amount of data has been collected
with greater dimensionality and quantity. Applications of statistical monitoring
techniques could be useful in extracting and interpreting process information
from massive data sets in order to discriminate between power system normal
or faulty states. At the same time, pattern recognition techniques could also be
used to distinguish which phase of the power system is faulty. Consequently, var-
ious multivariable statistical process control algorithms have been investigated
and implemented in power systems recently [2].