Page 505 - Decision Making Applications in Modern Power Systems

Pattern-recognition methods for decision-making Chapter | 17 465

straight relationship between the input and output of a protection scheme.
These impose a considerable difficulty when employing conventional
protection schemes. Some of these uncertainties and complexities are as
follows: different power system operating conditions such as changes in
load demand and generation type (renewable energy source, distributed
generation, conventional energy resource, DC storages, etc.); different
power system topology and configuration such as single/double line,
untransposed/transposed line, uncompensated/compensated line, two-
terminal/multiterminal line, and overhead/combined line; different pre-
fault and fault conditions such as fault type, fault inception angle, fault
resistance, fault location, loading level, short-circuit level, and references
values of compensator device; different phenomena in transmission lines
such as fault, power swing, fault during power swing, and transients of
lightning strike; different access point to relaying data such as only local
data and synchronized/unsynchronized local and remote data; error
caused by CT saturation or CCVT transients; noisy conditions caused by
various interferences. In addition, the random nature of some parameters
and conditions increase the complexity of the problem. When some con-
ditions of the power system have been changed, a pattern-recognition-
based function can adapt itself to the new condition through considering
a new training data set in the learning phase of classifier or estimator
model(s).
Since different training patterns can be employed in the training phase,
the functions obtain high generalization ability.
The hardware implementation of pattern-recognition-based functions con-
firms the robust performance of the smart function from practical
viewpoint.
The pattern-recognition-based functions are efficient in terms of com-
plexity and speed. Consequently, they can stand as a candidate for a pro-
tection function.

17.4.3 Disadvantages
There are some challenges to apply pattern recognition in numerical distance
relay and generally protective relays. There are many papers proposed the
smart protection functions based on pattern recognition theoretically, but
there are a few papers that consider the smart protection functions from prac-
tical viewpoint. In this regard a basic question may be raised: “what are the
challenging issues to employ pattern-recognition-based protection function in
commercial relays?”. The response to this question can be considered as the
most disadvantages and shortcomings of smart functions [16,84]:
Before commercial application the smart functions should be investigated
in more details from a perspective of hardware implementation, whereas