Fuzzy Cognitive Map for Health Assessment of IEC 61850 ... - aceps
Fuzzy Cognitive Map for Health Assessment of IEC 61850 ... - aceps
Fuzzy Cognitive Map for Health Assessment of IEC 61850 ... - aceps
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
<strong>Fuzzy</strong> <strong>Cognitive</strong> <strong>Map</strong> <strong>for</strong> <strong>Health</strong> <strong>Assessment</strong> <strong>of</strong> <strong>IEC</strong><br />
<strong>61850</strong> Based Devices and Functions<br />
Salman Mohagheghi<br />
ABB Inc., US Corporate Research Center, Raleigh, NC, USA<br />
© ABB Group<br />
March 3, 2011 | Slide 1
Contents<br />
Introduction<br />
Communication within the Substation<br />
Overview <strong>of</strong> <strong>IEC</strong> <strong>61850</strong><br />
Communication Services<br />
Data Model<br />
<strong>Health</strong> State <strong>of</strong> Devices and Functions<br />
<strong>Cognitive</strong> <strong>Map</strong>s<br />
Historical Approach<br />
<strong>Fuzzy</strong> Logic<br />
<strong>Fuzzy</strong> <strong>Cognitive</strong> <strong>Map</strong>s<br />
Case Studies<br />
Concluding Remarks<br />
© ABB Group<br />
March 3, 2011 | Slide 2
Introduction<br />
Smart Grid<br />
Communication and metering infrastructure<br />
Self-healing mechanisms<br />
Desired power quality level<br />
Advanced monitoring systems<br />
Efficient demand responsive programs<br />
Future pro<strong>of</strong> design<br />
Interoperable infrastructure<br />
© ABB Group<br />
March 3, 2011 | Slide 3
Communication within the Substations<br />
<strong>IEC</strong> <strong>61850</strong> Overview<br />
<strong>IEC</strong> <strong>61850</strong> is a standard specifically designed <strong>for</strong> implementing<br />
substation automation (SA) systems.<br />
The main features <strong>of</strong> the standard are:<br />
Data Modeling- Object oriented plat<strong>for</strong>m-independent<br />
modeling <strong>of</strong> all the functions.<br />
Fast Transfer <strong>of</strong> Events and Messages- GOOSE and<br />
GSSE <strong>for</strong> the peer-to-peer communication mode.<br />
Setting Groups- Control blocks allow the user to switch<br />
from any active group <strong>of</strong> setting values to another.<br />
Sampled Values- Transfer <strong>of</strong> sampled values from modern<br />
switchgear and nonconventional CT/VTs.<br />
Data Storage and Representation- Configured data <strong>of</strong> the<br />
substation is stored in XML <strong>for</strong>mat using the Substation<br />
Configuration Language (SCL).<br />
© ABB Group<br />
March 3, 2011 | Slide 4
Communication within the Substations<br />
<strong>IEC</strong> <strong>61850</strong> Substation Automation Topology<br />
© ABB Group<br />
March 3, 2011 | Slide 5
Communication within the Substations<br />
Services<br />
© ABB Group<br />
March 3, 2011 | Slide 6
Communication within the Substations<br />
Logical Node<br />
<br />
<br />
<br />
<br />
Virtual Units: objects defined in<br />
the object oriented context <strong>of</strong><br />
the standard.<br />
LN is a collection <strong>of</strong> data<br />
objects, data-set objects,<br />
descriptive attributes, report<br />
control objects, log-control<br />
objects and a list <strong>of</strong> sampled<br />
values.<br />
The type <strong>of</strong> data contained in a<br />
LN can be operational (e.g.<br />
measurement values, position<br />
status) or configuration data<br />
(e.g. IED self descriptive).<br />
The substation automation<br />
functions are broken down into<br />
LNs.<br />
© ABB Group<br />
March 3, 2011 | Slide 7
Communication within the Substations<br />
<strong>IEC</strong> <strong>61850</strong> Data Models<br />
© ABB Group<br />
March 3, 2011 | Slide 8
Communication within the Substations<br />
<strong>Health</strong> State in <strong>IEC</strong> <strong>61850</strong><br />
<br />
<strong>IEC</strong> <strong>61850</strong> defines health attributes <strong>for</strong> the logical nodes and<br />
the external devices connected to them. These include:<br />
<br />
<br />
<br />
<strong>Health</strong>: state <strong>of</strong> the logical node related to hardware or<br />
s<strong>of</strong>tware. Mandatory data with three states:<br />
<br />
<br />
<br />
Ok (green): no problems, normal operation<br />
Warning (yellow): minor problems, but in safe operation<br />
mode<br />
Alarm (red): severe problem, no operation possible<br />
EE<strong>Health</strong>: state <strong>of</strong> the external equipment, <strong>for</strong> example a<br />
circuit breaker controlled by the logical node XCBR.<br />
Optional data with the above three states.<br />
Phy<strong>Health</strong>: Mandatory attribute that refers to the health<br />
state <strong>of</strong> the physical device.<br />
© ABB Group<br />
March 3, 2011 | Slide 9
<strong>Cognitive</strong> <strong>Map</strong>s<br />
Historical Approach<br />
Introduced in 1970s <strong>for</strong> representing social scientific<br />
knowledge.<br />
Signed digraphs where a positive link from vertex A to vertex<br />
B indicates that A causally increases B, whereas a negative<br />
link from A to B means that should A happen, the chances <strong>of</strong><br />
B happening are reduced.<br />
Indirect Effect <strong>of</strong> A on B over the path pi is negative if the<br />
number <strong>of</strong> negative signs (<strong>of</strong> the edges) is odd, and<br />
positive otherwise.<br />
Total Effect <strong>of</strong> A on B would then be derived from the<br />
indirect effects over all possible paths: the total effect <strong>of</strong><br />
A on B is negative if all the indirect effects are negative,<br />
is positive if all the indirect effects are positive, and<br />
indeterminate otherwise.<br />
© ABB Group<br />
March 3, 2011 | Slide 10
<strong>Cognitive</strong> <strong>Map</strong>s<br />
Weighted Approach<br />
Alternative Solution: creating a weighted digraph<br />
removes the uncertainty related to the sign calculations<br />
requires higher level <strong>of</strong> granularity in defining the<br />
dependencies <strong>of</strong> causes on one another<br />
w<br />
AB p<br />
= wA<br />
, j<br />
× w<br />
j , j<br />
× L w<br />
j , B<br />
,<br />
×<br />
i<br />
1 1 2<br />
w<br />
AB<br />
wAB,<br />
p<br />
= ∑<br />
i<br />
i<br />
n<br />
© ABB Group<br />
March 3, 2011 | Slide 11
<strong>Cognitive</strong> <strong>Map</strong>s<br />
<strong>Fuzzy</strong> Logic<br />
Causality is fuzzy<br />
<strong>Fuzzy</strong> logic can deal with imprecise and vague in<strong>for</strong>mation. It<br />
enables representation <strong>of</strong> human knowledge with<br />
approximate terms and values<br />
© ABB Group<br />
March 3, 2011 | Slide 12
<strong>Fuzzy</strong> <strong>Cognitive</strong> <strong>Map</strong>s<br />
Qualitative Approach<br />
FCM provides a qualitative output (not a crisp number)<br />
w<br />
w<br />
AB<br />
AB<br />
= min { wA,<br />
j<br />
, w<br />
j , j<br />
, L,<br />
j ,<br />
, p<br />
w<br />
i 1 1 2<br />
n B<br />
m<br />
= max { wAB<br />
p<br />
, L,<br />
wAB,<br />
, 1 pm<br />
}<br />
}<br />
In other words:<br />
Indirect Effect: weakest causal link in a path<br />
Total Effect: Strongest <strong>of</strong> the weakest links<br />
© ABB Group<br />
March 3, 2011 | Slide 13
Case Study<br />
Trans<strong>for</strong>mer Bay<br />
Trans<strong>for</strong>mer bay with protection/control functions:<br />
line distance protection, tap changing functions with related protection<br />
functions, interlocking and reclosing functions, general monitoring<br />
functions, and a generic control function <strong>for</strong> regulating the trans<strong>for</strong>mer<br />
tap positions<br />
© ABB Group<br />
March 3, 2011 | Slide 14
Case Study<br />
<strong>Fuzzy</strong> <strong>Cognitive</strong> <strong>Map</strong><br />
© ABB Group<br />
March 3, 2011 | Slide 15
Case Study<br />
Voltage Trans<strong>for</strong>mer<br />
© ABB Group<br />
March 3, 2011 | Slide 16<br />
TVTR1: “Alarm” state <strong>for</strong> health.<br />
The impact <strong>of</strong> this health status on the IHMI (at the operator<br />
desk):<br />
TVTR1 → MMXU → IHMI<br />
TVTR1 → PTOV → AVCO → GAPC → IHMI<br />
TVTR1 → CPOW → GAPC → IHMI<br />
TVTR1 → MMXN → GAPC → IHMI<br />
TVTR1 → PDIS → RREC → IHMI<br />
The total effect <strong>of</strong> the Alarm health status <strong>of</strong> the TVTR1 on the<br />
IHMI is determined as:<br />
max{ min(V.H, V.H) + min(H, M, H, M) + min(L, H, M)<br />
+ min(H, M, M) + min(V.H, V.H, V.H) }<br />
= Very High
Case Study<br />
Circuit Breaker<br />
CB controlled by XCBR: “Warning” state <strong>for</strong> health (i.e. Warning<br />
state <strong>for</strong> the EE<strong>Health</strong> data <strong>of</strong> XCBR)<br />
The impact <strong>of</strong> this health status on the per<strong>for</strong>mance <strong>of</strong> the<br />
generic GAPC control function:<br />
XCBR → GAPC<br />
XCBR → CILO (Bay) → CILO (Station) → GAPC<br />
XCBR → CSWI → CILO (Bay) → CILO (St.) → GAPC<br />
The total effect <strong>of</strong> the Warning health status <strong>of</strong> the CB on the<br />
GAPC is determined as:<br />
max{ min(M) + min(V.H, M, M) + min(V.H, V.H, M, M) }<br />
= Medium<br />
© ABB Group<br />
March 3, 2011 | Slide 17
Case Study<br />
Tap Changer<br />
YLTC: “Alarm” state <strong>for</strong> health<br />
The impact <strong>of</strong> this health status on the per<strong>for</strong>mance <strong>of</strong> the<br />
ATCC logical node:<br />
YLTC → ATCC<br />
YLTC → AVCO → GAPC → ATCC<br />
The total effect <strong>of</strong> the Alarm health status <strong>of</strong> YLTC on the<br />
ATCC is determined as:<br />
max{ min(V.H) + min(H, H, V.H) } = Very High<br />
© ABB Group<br />
March 3, 2011 | Slide 18
Concluding Remarks<br />
<br />
<br />
<br />
<br />
<br />
<strong>IEC</strong> <strong>61850</strong> standard defines a data attribute indicating the health<br />
status <strong>of</strong> logical nodes (i.e., basic functions) and physical devices.<br />
This attribute tags the data points traveling across the system.<br />
While this identifier attribute can be easily interpreted and utilized<br />
in the device immediately connected to the source <strong>of</strong> the data, it is<br />
<strong>of</strong>ten not easy to find the impact <strong>of</strong> a non-ideal health status in a<br />
web <strong>of</strong> multiple control and protection functions.<br />
A fuzzy cognitive map (FCM) is proposed <strong>for</strong> evaluating the impact<br />
<strong>of</strong> non-ideal health status <strong>of</strong> logical nodes and physical devices in<br />
an interconnected network on the per<strong>for</strong>mance <strong>of</strong> the rest <strong>of</strong> the<br />
system functions.<br />
FCM can monitor the devices and functions with a predefined<br />
frequency to raise flags upon detecting non-ideal health.<br />
Expert knowledge can be used to determine the weights <strong>of</strong> the<br />
interdependences <strong>of</strong> logical nodes on one another.<br />
© ABB Group<br />
March 3, 2011 | Slide 19
© ABB Group<br />
March 3, 2011 | Slide 20