Вы находитесь на странице: 1из 28

4.2 State estimation 4.2.

Real-time security analysis


Network topology State vector
Measurements State
J(x), E{J(x)}
Variance of mea-
surement errors Estimation Location of
bad data
Network parameter

Model
Network
equivalent of
external system

Security constraints:
Contingency set • met
• lines
Security • not met
• transformers Exeeding limits:
• generation units analysis • Ith
• Vmin, Vmax
• Imax
4.2 State estimation 4.2.2

Definition of State Vector

In a network with i = 1…k nodes


the vector
θ 
V 
1
 1

θ 
 
2

V θi voltage angle
x=  
2

M  Vi voltage magnitude
 
M 
θ 
 
k

Vk 

is called the state vector of the network.


4.2 State estimation 4.2.3

State vector in 4-node network


V1,θ1 = 0 (Slack Node)
1
P14 P12

V4, θ4< 0 PG1


4 2 V2, θ2< 0
V3, θ3< 0

P43 P23
3

PL3

P12 =
1
( R12 + X12 )
2 2
[ ]
V12 R12 + V1V2 X12 sin(θ1 −θ 2 ) − V1V2 R12 cos(θ1 −θ 2 )

V1V2
P12 ≈ sin(θ1 −θ 2 )
X12
4.2 State estimation 4.2.4
x
Terms of state estimation true value of
state variables
Power System h(x)
functional relation
x ; h(x)
between measurable
quantities and the
state variables
v
measurement Measuring
errors Instrumentation

z
z = h(x) + v measured values

State estimation

x

estimated value of
x state variables
4.2 State estimation 4.2.5

Estimation by Weighted Least Squares:

Measuring equations:

z = h(x) + v
z vector of m measurements
x state vector with n variables (θi, Vi)
h(x) vector of m measurement functions
v vector of m measurement errors
m≥n
4.2 State estimation 4.2.6

Assumptions in respect to measurements

v1….vm random variables with Gaussian


probability function
E{v} = 0 mean value of measurement
errors = 0
E { v vt } = R measurement covariance
matrix

σ12 0
  no interdependence between
R= O  measurement errors
0 σm 
2

4.2 State estimation 4.2.7

Normal or Gaussian probability function


Probability density function Ф(y) of a random variable Y
1  y −µ 2
−  
1
Φ (y ) = ⋅e 2 σ 
σ 2 ⋅π

− ∞ < y < +∞ σ > 0

estimated value of
the voltage vector

Probability density function Ф(y) for µ = 0 and σ = 0.25; 0.5; 1.0


4.2 State estimation 4.2.8

Objective function J(x) and minimization

J( x ) = ( z − h( x ))t R −1( z − h( x ))
m
( zi − hi ( x ))2
J( x ) = ∑
i=1 σi2
Necessary condition for the extremum of J(x):

∂J
= −2HtR −1( z − h( x )) = 0 (1)
∂x
 ∂h1 ∂h1 
 ∂x L
∂x n 
∂h( x )  1 
H= = M M 
∂x
 ∂hm L
∂hm 
 ∂x ∂x n 
 1

H Jacobian matrix of measurement functions.


4.2 State estimation 4.2.9

Case A: h(x) linear function of x

z = A⋅x+v

Objective function:

J( x ) = ( z − A ⋅ x )t R −1( z − A ⋅ x )

Necessary condition for extremum:

∂J
= −2A tR −1( z − A ⋅ x ) = 0
∂x

Solution:

x̂ = ( A tR −1A )−1 A tR −1 z
4.2 State estimation 4.2.10

Case B: h(x) nonlinear function of x

Taylor expansion of h(x) around a starting


point x0:

o o o
h( x ) = h( x ) + H( x )( x − x ) + ......

Introducing Taylor expansion in eq. (1) :

∂J
∂x
0
[
= − 2H tR −1 z − h( x ) − H( x )( x − x ) = 0
0 0
]
H t ( x )R −1H( x ) x − x
0 0
[ 0
]= H (x t 0
[
)R −1 z − h( x )
0
]
Iterative solution:
1 0
[
x = x + H ( x )R H( x ) Ht ( x )R −1 z − h( x )
t 0 −1 0
]
−1 0
[ 0
]
M
x
υ +1 υ
[ υ υ
= x + Ht ( x )R −1H( x ) Ht ( x )R −1 z − h( x ) ]
−1 υ
[ υ
]
4.2 State estimation 4.2.11

State estimation: Numerical example


V1 V2
Θ1 = 0
1 2

bus voltage measurement


active power measurement of line flow
reactive power measurement of line flow

Given: z1 = V1 = 1.02
z2 = V2 = 1.0
z3 = P12 = 3.0
z4 = Q12 = 0.3

σ 12 0 0 0 (0.1)2 0 0 0
0 σ 2
2 0 0 0 (0.1)2 0 0
R = =
0 0 σ 2
3 0 0 0 (0.05)2 0
0 0 0 σ 2
4 0 0 0 (0.05)2

Find: Weighted least squares estimates of Θ2; V1; V2


4.2 State estimation 4.2.12

State estimation: Numerical example


Solution:
h1(θ; V) = V1 h2(θ; V) = V2
h3(θ; V) = V12(g12+gs12)-V1V2[g12cos(θ1- θ2)+b12sin(θ1- θ2)]
g12=0; gs12=0; b12=-10; θ1=0 (Slack bus)
h3(θ; V) = -10 V1 V2 sin θ2

h4(θ; V) = -V12(b12+bs12) – V1V2[g12sin(θ1- θ2) – b12cos(θ1- θ2)]


bs12=0;
 ∂h1 ∂h1 ∂h1 
h4(θ; V) = 10V12- 10V1V2cos θ2  ∂x ∂x2 ∂x3 
 1 
 ∂h2 ∂h2 ∂h2 
 ∂x ∂x2 ∂x3 
Jacobian-Matrix H (θ ;V ) =  1
∂h ∂h3 ∂h3 
 3 
 ∂x1 ∂x2 ∂x3 
 ∂h4 ∂h4 ∂h4 
 ∂x ∂x2 ∂x3 
 1
4.2 State estimation 4.2.13

State estimation: Numerical example


 0 1 0 
 0 0 1 
H (θ ; V ) =  
 − 10 V1V 2 cos θ 2 − 10 V 2 sin θ 2 − 10 V1 sin θ 2 
 
 10 V1V 2 sin θ 2 20 V1 − 10 V 2 cos θ 2 − 10 V1 cos θ 2 

 x1o  θ 2o   0 
Iterative solution:  o  o  
 x 2  = V1  = 1 . 0 
 x 3o  V 2o  1 . 0 
   
1 . 02  1 . 0   0 . 02 
 1 . 0  1 . 0   0 
z − h( θ ; V ) = 
ο o − =  
 3 .0   0   3 .0 
     
 0 .3   0   0 .3 
 0 1 0 
 0 0 1 
H (θ ; V ) = 
ο o
 − 10 0 0 
 
 0 10 − 10 
4.2 State estimation 4.2.14

State estimation: Numerical example

0 0 -10 0 100 0 0 0 0.02


0 100 0 0
Ht R-1 (z-h) = 1 0 0 10 0
0 1 0 -10 0 0 400 0 3.0
0 0 0 400 0.3

-12000
Ht R-1 (z-h) = 1202
-1200

4.0 0 0
Ht R-1 H = 104 0 4.01 -4.0
0 -4.0 4.01

0.2500 0 0
(Ht R-1 H) -1 = 10-4 0 50.0624 49.9376
0 49.9376 50.0624

x1 = x0 + (Ht R-1 H) -1 Ht R-1 (z-h)


4.2 State estimation 4.2.15

State estimation: Numerical example


x11 0 0.2500 0 0 -12000
x2 1 = 1.0 + 10-4 0 50.0624 49.9376 1202
x31 1.0 0 49.9376 50.0624 -1200

x11 0 -0.3000
x21 = 1.0 + 0.0250
x31 1.0 -0.0050

x11 -0.3000
x21 = 1.0250
x31 0.9950

Result of the first iteration step:


θ21 = -0.3000
V11 = 1.0250
V21 = 0.9950
4.2 State estimation 4.2.16

State estimation: Numerical example

iteration θ2 V1 V2 max|z-h(θ;V)|

start 0 1.0 1.0 3.000


ν=1 - 0.300 1.025 0.995 0.463
ν=2 - 0.298 1.004 1.018 0.018
ν=3 - 0.298 1.003 1.018 0.018
ν=4 - 0.298 1.003 1.018 0.018

Final result after ν = 4 iterations:


^
θ2 = θ2 4 = -0.298
^ = V 4=
V 1.003
1 1
^
V = V 4= 1.018
2 2
4.2 State estimation 4.2.17

Input data and results of state estimation

Network topology State estimation: Estimated value of state


Transformer tap - Observability test vector
settings
- WLS-algorithm Other quantities derived
Measurements - Identification of bad data from state vector
Pseudo-measurements - Suppression of bad data ∧
Observed value J( x )
Variance of ∧
measurement errors Expected value E(J( x ))

network parameters Location of bad data


4.2 State estimation 4.2.17 /1

Observability
6 5
1 Slack

2 3 4
6 (Pij, Qij) measurements
nodes 4 , 5 not observable

1 Slack 6 5

4
2 3
7 (Pij, Qij) measurements
all nodes observable
4.2 State estimation 4.2.17 /2

Pseudo-measurements
Pi = 0 node i : passive
Qi = 0
i

Pi ≠ 0 node i : non passive


Qi ≠ 0
i
4.2 State estimation 4.2.17 /3

χ2, CHI-SQUARE DISTRIBUTION

Y1, Y2, ……, Yk independent random variables; with normal distribution;


mean value zero; variance one;

k degree of freedom

χ2 = Y12 + Y22 + …… + Yk2

Mean value µ χ 2 = k ; Variance σ 2χ 2 = 2k

ϕ(χ2) 0.1

k = 20

k = 50

0 20 40 60 80 χ2
4.2 State estimation 4.2.17 /4

Under the assumptions concerning the statistics of measurement


errors, J(x) is a random variable with χ2 distribution and degree of
freedom k = (m – n)

ϕ(J(x) )

µJ
µJ - 3σJ µJ + 3σJ

J(x)
Mean value µJ = (m – n)
Variance σJ2 = 2(m – n)
4.2 State estimation 4.2.18

Input data of state estimation

• On-Off status quantities of circuit breakers and isolators (network topology)

• Transformer tap settings

• Measurements (bus voltage magnitudes, active and reactive power flows,

active and reactive power injections)

• Pseudo measurements of passive nodes (Pi = 0, Qi = 0)

• Information regarding the measuring instrumentation (specification and

location of measurements, variance of measurement errors)

• Network parameters (lines: R, X, G, B; transformers: R, X, G, B, t)


4.2 State estimation 4.2.19

Results of state estimation


• Estimated value of the state vector x of all busses
∧ ∧
(voltage magnitude V i and voltage phase angle Θi)

• Other quantities derived from the state vector x
∧ ∧ ∧ ∧ ∧ ∧ ∧ ∧
( I ik ,Pik , Qik , I i,Pi, Qi,P v , Q v)

• Observed value of the minimization function J( x )

• Expected value of the minimization function E(J( x ))

• Location of bad data

• Covariance matrix of the state vector


4.2 State estimation 4.2.20

First implementations

1973 USA: American Electric Power


765 kV and 345 kV network
16 busses, 25 branches
measurements: V1, Pij, Qij

1976 Germany: RWE


380 kV and 220 kV network
150 busses, 250 branches
measurements: Vi, Pij, Qij, Pi, Qi
4.2 State estimation 4.2.21

Data for the application of state estimation (RWE network)


• Size of network
- 280 busses
- 360 transmission lines
- 40 transformers
• Total number of measurements
- 1010 measurements of active and reactive power flow
- 210 measurements of active and reactive power injections
- 140 measurements of active and reactive power injections
(pseudo measurements)
- 60 measurements of voltage mangnitude
• Redundancy:

m−n
r= = 1.54
n
4.2 State estimation 4.2.22

Application of state estimation (RWE network)


4.2 State estimation 4.2.23

Application of state estimation (RWE network)

State vector
380 kV State vector
380 kV

Wednesday, July, 19th 11:30 a.m. Wednesday, January 17th 3:00 a.m.
4.2 State estimation 4.2.24

State estimation: Sources of errors


• Interruption of data transmission
(measurements, on-off status quantities, transformer tap settings)

• Bad measurement data (defect of measuring instrumentation, wrong measurement


range, false sign of measurement)

• Network topology (incomplete representation of on-off status quantities)

• Variance of measurement errors (instrument transformer, measuring instrument, A/D-


converter)

• Network parameter (line length, transformer model)

• Delays in the transmission of data

Вам также может понравиться