lecture 15 stateestimation
TRANSCRIPT
-
8/17/2019 Lecture 15 StateEstimation
1/45
Lecture 15
Power system state estimation
Kun Zhu2013-05-02
-
8/17/2019 Lecture 15 StateEstimation
2/45
Outline
•
State estimation
- What
- Why
• Weighted Leaset Square (WLS) algorithms
- Mathmatics
- Concepts
• Operation challenges
-
8/17/2019 Lecture 15 StateEstimation
3/45
Course road map
-
8/17/2019 Lecture 15 StateEstimation
4/45
What is state estimation?
-
8/17/2019 Lecture 15 StateEstimation
5/45
How could the operator know thesystem?
62 MW
6 MW
-
8/17/2019 Lecture 15 StateEstimation
6/45
The truth is out here!
Courtesy of ABB Ventryx
-
8/17/2019 Lecture 15 StateEstimation
7/45
Power system states: x
• The power system states are those parameters that can beused to determine all other parameters of the power system
•
Node voltage phasor
- Voltage magnitude V k-
Phase angle !k• Transformer turn ratios
- Turn ratio magnitude t kn- Phase shift angle "kn
•
Complex power flow
- Active power flow P kn , Pn
k- Reactive power flow Qkn , Qnk
-
8/17/2019 Lecture 15 StateEstimation
8/45
Analog measurements
•
Voltage magnitude
• Current flow magnitude & injection
• Active & reactive power
- Branches & groups of branches
- Injection at buses
-
In switches
- In zero impedance branches
- In branches of unknown impedance
• Transformers
- Magnitude of turns ratio
-
Phase shift angle of transformer• Synchronized phasors from
Phasor Measurement Unit
-
8/17/2019 Lecture 15 StateEstimation
9/45
Network topology processing
Bus breaker model Bus branch model
-
8/17/2019 Lecture 15 StateEstimation
10/45
Power system measurements: z
z=ztrue+e
z=ztrue+e
z=ztrue+e
ztrue: power system truthztrue=h(x)
e: measurement errore=esystematic+erandom
-
8/17/2019 Lecture 15 StateEstimation
11/45
Measurement model
• How to determine the states (x) given a set ofmeasurements (z)?
z j = h j ( x ) + e j
known unknown unknown
where
- x is the true state vector [V 1 ,V 2 ,…V k , !1 , !2 ,… !k ]
- z j is the jth measurement- h j relates the jth measurement to states
- e j is the measurement error
-
8/17/2019 Lecture 15 StateEstimation
12/45
State estimation process
-
8/17/2019 Lecture 15 StateEstimation
13/45
Why do we need state estimation?
-
8/17/2019 Lecture 15 StateEstimation
14/45
Measurements correctness
• Imperfections in- Current & Voltage transformer
-
Transducers
• A/D conversions
•
Tuning-
RTU/IED Data storage
- Rounding in calculations
- Communication links
• Result in uncertainties in the measurements
-
8/17/2019 Lecture 15 StateEstimation
15/45
Measurement timeliness
• Due to imperfections in SCADA system themeasurements will be collected at different pointsin time, time skew.
• If several measurements are missing how long towait for them?
• Fortunately, not a problem during quansi-steadystate.
•
State estimation is used for off-line applications
t
Pab
Pbc
Pcb Pdc
-
8/17/2019 Lecture 15 StateEstimation
16/45
How can the states be estimated?
-
8/17/2019 Lecture 15 StateEstimation
17/45
Approaches
•
Minimum variance method- Minimize the sum of the squares of the weighted
deviations of the state calculated based onmeasurements from the true state
• Maximum likelihood method
-
Maximizing the probability that the estimate equalsto the true state vector x
• Weighted least square method (WLS)
- Minimize the sum of the weighted squares of theestimated measurements from the true state
2
1
( ( ))( )
ii
m
i i
i
z h x J x
R=
!="
-
8/17/2019 Lecture 15 StateEstimation
18/45
Least square (Wiki)
• "Least squares" means that the overall solutionminimizes the sum of the squares of the errors madein the results of every single equation.
• The method of least squares is a standard approach
to the approximate solution of over determinedsystem, i.e., sets of equations in which there are moreequations than unknowns.
• The most important application is in data fitting.
• Carl Friedrich Gauss is credited with developing the
fundamentals of the basis for least-squares analysis in1795.
-
8/17/2019 Lecture 15 StateEstimation
19/45
WLS state estimation
• Fred Schweppe introduced state estimation to powersystems in 1968.
• He defined the state estimator as “a data processingalgorithm for converting redundant meter readings
and other available information into an estimate of thestate of an electric power system”.
• Today, state estimation is an essential part in almostevery energy management system throughout theworld.
Felix F. Wu, “Power system state estimation: a survey”,International Journal of Electrical Power & EnergySystems, Volume 12, Issue 2, April 1990, Pages 8
-
8/17/2019 Lecture 15 StateEstimation
20/45
WLS state estimation model
z j = h j ( x ) + e jknown unknown unknown
-
8/17/2019 Lecture 15 StateEstimation
21/45
Error characteristics
• The errors in the measurements is the sum ofseveral stochastic variables
- CT/VT, Transducer, RTU, Communication…
• The errors is assumed as a Gaussian Distribution
with known deviations
- Expected value E[e j ] =0
- Known deviation # j • The errors are also assumed to be independent
- E [eie j ] =0
-
8/17/2019 Lecture 15 StateEstimation
22/45
WLS objective function
)]([)]([))((
)( 1
12
2
xh z R xh z xh z
x J T
m
i i
i !!=!
=
!
=
"#
wherei=1,2,…m
Solution to above is iterative using newton methods
R = diag{! 12,! 2
2,! 3
2,...! m
2} =Cov(e) = E [e !e
T ]
-
8/17/2019 Lecture 15 StateEstimation
23/45
Newton iteration
• At the minimum, the first-order optimality conditionswill have to be satisfied
0)]([)()()( 1 =!="
"=
! xh z R x H x x J x g T
])(
[)( x
xh x H
T
!
!= is the measurement Jacobian matrix
-
8/17/2019 Lecture 15 StateEstimation
24/45
0.....))(()()( =+!+= k k k x x xG x g x g
Newton iteration cont’d
• Expanding the g(x) into its Taylor series around statevector
k x
)()(
)(
)(
1 k k T k
k
x H R x H x
x g
xG
!=
"
"=
where
-
8/17/2019 Lecture 15 StateEstimation
25/45
Newton iteration cont’d
• Neglecting the higher order terms leads to an iterativesolutions scheme know as the Gauss-Newton methodas :
)()]([ 11 k k k k
x g xG x x !"= "+
k is the iteration index,is the solution vector at iteration k k x
-
8/17/2019 Lecture 15 StateEstimation
26/45
Newton iteration IV
• Convergence
• If not, update
Go back to the previous step
! "# )max( k x
1
1
+=
!+=+
k k
x x x k k k
-
8/17/2019 Lecture 15 StateEstimation
27/45
Why the measurements areweighted?
),,,(
),,,(
ji jiij
ji jiij
vv f Q
vv f P
! !
! !
=
=
voltage mag
complex power flow
-
8/17/2019 Lecture 15 StateEstimation
28/45
Weight
• Weight is introduced to emphasize the trustedmeasurement while de-emphasize the lesstrusted ones.
•
WLS
2
1
i
iW
!
=
-
8/17/2019 Lecture 15 StateEstimation
29/45
Observability
• Based on system topologyand location ofmeasurements parts of the
power system may beunobservable.
• Unobservable parts of the
system can be madeobservable via data
exchange (CIM), pseudo
measurements, etc…
-
8/17/2019 Lecture 15 StateEstimation
30/45
Observability example
Pij = f (vi,v j ,! i,! j )
Qij = f (vi,v j ,! i,! j )
voltage mag
complex power flow
-
8/17/2019 Lecture 15 StateEstimation
31/45
Observability criterion
Necessary but not sufficient condition
m: Number of measurements
n: Number of states
Is the system’s oservability guranteed in this
case?
nm !
-
8/17/2019 Lecture 15 StateEstimation
32/45
Observability example II
),,,(),,,(
ji jiij
ji jiij
vv f Qvv f P
! !
! !
=
=
voltage mag
complex power flow
-
8/17/2019 Lecture 15 StateEstimation
33/45
n of m measurements has to be independent
),,,(
),,,(
ji jiij
ji jiij
vv f Q
vv f P
! !
! !
=
=
voltage mag
complex power flow
-
8/17/2019 Lecture 15 StateEstimation
34/45
Summary of assumptions
• Quansi-steady system
- No large variations of states over time
- State estimator will be suspended when largedisturbance happens.
• Errors in measurements are
- Gaussian in nature with known deviation
- Independent
•
Strong assumption- Power system topology model is correct
-
8/17/2019 Lecture 15 StateEstimation
35/45
Refinements
• Refinements of method for State Estimation hasbeen the objective of much research.
• Reduce the numerical calculation complexity in
order to speed up the execution.- 3000 bus node in 1-2 seconds
• Improve estimation robustness
- less affected by erroneous input
-
8/17/2019 Lecture 15 StateEstimation
36/45
Bad Data Detection
-
8/17/2019 Lecture 15 StateEstimation
37/45
Data quality
• Analog measurement error
• Parameter error
• Topological error
- Discrete measurement error
- Model error
-
8/17/2019 Lecture 15 StateEstimation
38/45
Bad Data Detection (Analog)
• Putting the measurements up to a set oflogical test (Kirchhoff's laws) before they areinput into the State Estimator
• Calculating J(x) and comparing with a pre-determined limit. If the value exceeds the limit,we can assume there is “something” wrong in themeasurements (Chi-square)
-
8/17/2019 Lecture 15 StateEstimation
39/45
Bad Data Detection (Analog)
• Once the system state has been identified, i.e. wehave an estimate of x
• We can use the estimate to calculate
• If there is any residual in this calculation that “stands-out” this is an indication that particularmeasurement is incorrect
)ˆ( xh z r iii !
=
-
8/17/2019 Lecture 15 StateEstimation
40/45
Largest normalized residual
• The normalized residual follows the standardnormal distribution
i
i N
i
xh z r
!
)ˆ("
=
-
8/17/2019 Lecture 15 StateEstimation
41/45
Limitations
• The critical measurements have zero residuals,hereby they can not be detected
!
!!!!!
"
#
$
$$$$$
%
&
=
t
t
t t t t t
S
0000
....
.0...
00000
Sz z K z z r
Kz z R H H H H z T T
ii
=!=!=
==
!
)1(ˆ
)(ˆ 1
-
8/17/2019 Lecture 15 StateEstimation
42/45
Critical measurement example
),,,(),,,(
ji jiij
ji jiij
vv f Qvv f P
! !
! !
=
=
voltage mag
complex power flow
-
8/17/2019 Lecture 15 StateEstimation
43/45
Parameter and structural processing
•
Similar theory can be applied
iiiiii e p f z e xh z +=!+= )()(
-
8/17/2019 Lecture 15 StateEstimation
44/45
State estimation functions
1. Bad data processing and elimination givenredundant measurements
2. Topology processing:
create bus/branch model (similar to Y matrix)
3. Observebility analysis:
all the states in the observable islands haveunique solutions
4. Parameter and structural processing
-
8/17/2019 Lecture 15 StateEstimation
45/45
Challenges
• Perception of the process is from themeasurements whose quality is, to largeextent, out of our control
•
The quality of estimates relies on the inputwhose uncertainty is highly dependent on theICT infrastructure
• Power system model can contain errors