Methodologies and algorithms for fault locators in modern power systems.
Following world wide deregulation of the electrical power industry, the requirements
for high quality and highly reliable power supplies are increasing. Comprehensive
and systematic computer based methodologies and algorithms for fault locators in
power systems are needed to support reliable, independent and comprehensive fault
diagnosis or accurate fault location, fast fault detection and correct fault
classification. This thesis describes work to develop novel fault diagnosis techniques
and accurate fault location algorithms for fault locators using system-wide
information of modern power systems. Fast and efficient modern communication
techniques make it possible to apply the novel fault locator system in practice.
The fault diagnosis techniques use fault messages obtained from current and voltage
sensors installed in radial and meshed networks respectively. A faulty section in the
meshed network can be detected by analysing the network's topologic structure and
the measurements of the fundamental frequency voltages obtained from voltage
sensors; while for the radial network, current sensors are used to trace the path of a
fault. The fault diagnosis techniques can work independently from protection
devices, and circuit breakers, and can be implemented economically in the low- and
medium-voltage power systems, because current and voltage sensors can be made
In order to achieve very accurate fault location for the EHV transmission systems
with long transmission lines and complex network topology, algorithms based on the
wavelet transform, travelling wave concept and GPS synchronisation have been
developed. When a fault occurs, fault generated transient waves propagate from fault
point to line terminals and then to the other nodes in the whole network. Fault
transient detectors are installed at nodes in the meshed network, at substations and
customer terminals in the radial system to capture the time of fault transient signal
arrival. From the time recorded and topologic network structure, very accurate fault
location can be achieved.
The thesis also describes and compares applications of digital Fourier transform,
least squares method and Kalman filters for fast measurements in current and voltage
sensors, for fault classification and fault detection, and the wavelet transform used in
fault transient detectors. Methodologies and algorithms developed have been
validated by ATP/EMTP simulation on different networks, such as, traditional twosource
system and IEEE 14-bus system.
The maIn advantage of the developed methodologies and algorithms for fault
locators in power systems is the use of system-wide fault messages in the electrical
networks. The fault locators will efficiently support the operators in the control
centres and relevant substations with clear fault information allowing them to take
suitable actions for fault emergency and restoration of power supplies. This will be
beneficial in reducing the outage time required for inspection, damage repair and
restoration of the power supply. Reliability, security and power quality will be
improved for the customers.