IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 22, NO.

3, JULY 2007 1841

Instantaneous Voltage Estimation for Assessment and
Monitoring of Flicker Indices in Power Systems
Mahmoud Mazadi, Seyed Hossein Hosseinian, William Rosehart, Senior Member, IEEE, and
David T. Westwick, Member, IEEE
AbstractIn a deregulated electricity industry, new concerns
have emerged regarding the quality of the power supply as compa-
nies tend to focus on their own objectives and interests. One main
concern regarding power-supply quality is voltage uctuation.
Voltage uctuations are dened as repetitive or random variations
in the magnitude of the supply voltage. This paper addresses a
method for monitoring the voltage icker using state estimation
(FSE). A test system is used to characterize the new algorithm
in which parameter inaccuracies have been considered. The
new algorithm can be used by operators as well as distribution
companies to monitor the icker severity throughout the network.
Index TermsPower quality (PQ), state estimation, voltage
icker.
I. INTRODUCTION
P
OWER-QUALITY (PQ) monitoring is necessary to char-
acterize electromagnetic phenomena at a particular loca-
tion in a power system. In some cases, the objective of the mon-
itoring is to diagnose PQ problems. In other cases, monitoring
may be used in planning the installation of PQ mitigating de-
vices. In addition, economic impacts of PQproblems are signi-
cant in many load centers. The effects on equipment and process
operation can include malfunction, damage, process disruption,
and premature aging. In addition to resolving equipment disrup-
tion, a database of equipment tolerances and sensitivities can be
developed from monitored data. Such a database can provide a
basis for developing equipment compatibility specications and
guidelines for future equipment enhancement. In most jurisdic-
tions, PQ indices, such as icker and harmonics, must be main-
tained in an acceptable range and are of signicant concern to
utilities [1], [2].
Sources of voltage icker are numerous, including arc fur-
naces and arc welders, motor starting, fans, pumps, and eleva-
tors and switching of power factor capacitors [3]. Cyclic voltage
icker exists when there is a slow change in the voltage mag-
nitude with frequencies between 0.5 to 30 Hz that appear as
a superimposed signal on the fundamental signal. The super-
imposed signal, which is generated due to the voltage icker,
Manuscript received June 12, 2005; revised March 6, 2006. This work was
supported in part by the NSERC Discovery Grant and in part by the Iranian
Ministry of Research and Science. Paper no. TPWRD-00414-2005.
M. Mazadi and S. H. Hosseinian are with the Department of Electrical
Engineering, AmirKabir University, Tehran 1978711167, Iran (e-mail:
[email protected]; [email protected]).
W. Rosehart and D. T. Westwick are with the Department of Electrical and
Computer Engineering, University of Calgary, Calgary, AB T2N 1N4, Canada
(e-mail: [email protected]; [email protected]).
Digital Object Identier 10.1109/TPWRD.2007.893188
appears as a change in the fundamental signal envelope and is
commonly known as the instantaneous icker level (IFL). The
IFL signal can be extracted from the voltage signal using an in-
ternational electrotechnical commission (IEC) ickermeter [4].
Flicker indices Pst (short-term icker level) and Plt (long-term
icker level) are then calculated from the statistics of the IFL
signal [4], [5]. In order to have an accurate and useful moni-
toring of PQ indices, two alternatives are available: First, using
ickermeters at all buses in the network and sending Pst and Plt
to the central control room and second, using existing measure-
ments of voltages or currents already being sent to the central
control room. Full measurement of Pst and Plt by the icker-
meter are prohibitive and uneconomical for a large system. This
is why the second alternative, together with state estimation, can
be applied to raw measurements to determine PQ indices. State
estimation has been used in power-ow studies since the late
1960s [6]. It is now an essential part in many energy-manage-
ment systems [7].
Recent contributions [8][11] have extended the concept to
harmonic state estimation (HSE) and identication of harmonic
sources. Many PQindices, such as Pst and Plt, are combinations
(often weighted) of individual (voltages, currents) states.
This paper presents a icker state estimation (FSE) technique
for an online assessment of voltage icker. An instantaneous
state estimator is formulated from the system admittance matrix
in the time domain and the placement of measurement points.
The measurement of the instantaneous voltage and current at se-
lected buses and lines is used to estimate all bus voltages. These
estimated voltages are then used to estimate icker levels. The
implementation of this technique can, in practice, be limited by
poor synchronization and accuracy of conventional instrumen-
tation; another source of inaccuracy is error in network parame-
ters, such as transmission lines and transformer admittances [7].
Therefore, numerical analysis is presented in this paper, exam-
ining the impact of clock synchronization error and parameter
error on the proposed approach. Noise is also introduced to the
measurement signal to model instrumentation accuracy.
The remainder of this paper is constructed as follows:
Section II presents a brief review of icker level evaluation.
Next, fundamental concepts of state estimation are provided
in Section III. The proposed generic icker state estimation
is detailed in Section IV. In Section V, a discussion of clock
synchronization and implementation challenges is given. Nu-
merical results are presented in Section VI. Finally, conclusions
are given in Section VII.
Throughout the paper, boldfaced uppercase letters will be
used to denote matrices, while boldfaced lowercase letters will
0885-8977/$25.00 2007 IEEE
1842 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 22, NO. 3, JULY 2007
denote vectors and functions with vector outputs. Symbols in a
regular typeface will denote scalar quantities.
II. FLICKER LEVEL EVALUATION
Many techniques are proposed in the literature for the evalua-
tion of icker level such as the fast Fourier transform technique
(FFT) technique [12], least absolute value (LAV) state estima-
tion [13], [14], and wavelet approaches [15], [16].
Leakage effect error is a major problem associated with FFT
applications. As the system frequency deviates from the nom-
inal value, the leakage effect error may be very serious [17].
To overcome this disadvantage, wavelet-transform-based ap-
proaches were proposed [15], [16]. The LAV method proposes
the application of an LAV state estimation to measure voltage
icker magnitude and its frequency [14]. These approaches
tend to use linear models for the voltage icker magnitude, its
phase angle, and frequency deviation. Linearizing the system
equation may lead to inaccurate results in environments with
high noise and stressed systems. Envelope tracking-based
approaches were proposed in [18][20] using neural networks
and genetic algorithms. These methods either tend to be very
sensitive to high-frequency components or require signicant
computational resources.
In [4], the short-term icker level (Pst) index and long-term
icker level (Plt) index have been introduced to indicate the
severity of icker in the voltage signal. These indices are ob-
tained from voltage signals that have been ltered to simulate
the lamp-eye-brain chain by eliminating the dc and rst har-
monic components in the squared input voltage signal [5], [12],
[21]. The ltered voltage signal is often referred to as the IFL
signal. The IFL signal is typically classied into different mag-
nitudes. Each magnitude is considered a class. The Pst and Plt
indices are based on the probability that the IFL signal is at or
above a given class. This method to dene classes and their cor-
responding probability, is referred to as the time at level classi-
cation [4].
Based on the time at level classication, the Pst index is cal-
culated using [4]
(1)
where the percentiles and are dened
as the maximum class that the IFL exceeds for 0.1%, 1%, 3%,
10%, and 50% of the time, respectively. The subscript indi-
cates that the values are averaged or smoothed. Detailed infor-
mation on the calculation of the Pst index can be found in [4]
and [21]
III. STATE ESTIMATION
State estimation is used to generate the best estimate of all bus
voltages from limited measurements of line currents or voltages
that are often corrupted with measurement noise. Three issues
involved are the choice of state variables, performance criteria,
and the selection of measurement points and quantities to be
measured. State variables are those that, if known, completely
specify the system. The instantaneous voltages at all buses are
chosen, since they allow all currents and power ows to be de-
termined. Various performance criteria are possible; the most
widely used being the weighted least square method (WLSM)
[21]. This method minimizes the weighted sum of the squares
of the residuals between the estimated instantaneous voltage and
the measured signals. In [22], it is shown that the WLSMmethod
tends to nd the best estimate of the system voltages when the
noise distribution is Gaussian. For a given measurement set and
system topology, basic circuit laws can be applied to nd a re-
lationship between the measured signals and the desired esti-
mated states. This relationship is often referred to the measure-
ment equation and can be expressed in general terms as
(2)
where and are vectors containing the measurements and
state variables, respectively, and is the measurement error
vector which is assumed to be made of independent random
variables with Gaussian distributions. The function relates
the measured signals to system states. This function could be
linear or nonlinear depending on the chosen and . Given
(2), the WLSM estimates the vector that minimizes the
weighted sum of the squares of the residuals [23]
(3)
where is a weighted sum of squared residuals, ,
computed as
(4)
Since the noise samples in the measurements are independent,
the matrix is strictly diagonal and contains the variances of
the measurements (if they are known). The parameters of this
matrix are effectively weightings that are used to increase the
contribution of measurements that are known to be more accu-
rate in determining the states.
In general, the measurement equation is nonlinear, when
choosing the bus voltages as state variables and real and
reactive power as measurements , However, if line currents
and some bus voltages are chosen as the measurements , (2)
becomes linear [21]. In this case, (2) can be expressed as
(5)
where is the measurement matrix [21] (i.e., it provides the
linear mapping between the measured signals and the states to
be estimated). State estimation can be classied as overdeter-
mined, completely determined, or underdetermined depending
on whether the number of independent measurement equations
is greater, equal to or less than the number of unknown state
variables [22]. Aunique solution to (4) assuming the linear rela-
tionship (5) is only possible for over determined and completely
determined systems and is given as [22]
(6)
The measurement matrix has more rows than columns but
(known as the gain matrix) is a square matrix. The
MAZADI et al.: INSTANTANEOUS VOLTAGE ESTIMATION FOR ASSESSMENT AND MONITORING 1843
solution for given by (6) can be easily determined using
standard linear techniques. If the problem is singular or has a
high condition number making the problem infeasible, this in-
dicates there is nonobservability for the given measurements.
IV. PROPOSED FLICKER STATE ESTIMATION APPROACH
An instantaneous state estimator is formulated from the
system admittance matrix in the time domain and the placement
of measurement points. Measurements of the instantaneous
voltage and current at selected buses are sent to a central
workstation for the estimation of bus voltages. These esti-
mated voltages are then used to determine the Pst values [5].
Placement of the measurement points is such that all buses are
observable.
In order to monitor the Pst at all buses in the network, the in-
stantaneous voltage at all buses is required. This paper proposes
to use a linear relationship between estimated bus voltages and
current/voltage measurements to estimate instantaneous volt-
ages. Current measurements are rarely used in power system
state estimation, due to the deterioration of the Jacobian con-
dition number [23]. This drawback is effectively eliminated by
using instantaneous currents as measurements, bus voltages as
states, and a time domain model instead of a steady-state phasor-
based model. In this case, is a linear function in (2) and con-
tains series and shunt admittances of the corresponding branch.
In order to build , a time-domain model of a branch is re-
quired. This model is obtained as outlined in the following two
subsections.
A. Time-Domain Model
Time-domain simulations are usually performed by simulta-
neously solving a set of rst-order differential [24]
(7a)
(7b)
(7c)
Numerical integration is carried out in order to nd for all
time samples in which
(8)
where is the time step and can be constant or variable
based on the requirement. The time-domain simulation is
implemented using [24]
(9)
where is a constant, referred to as the compensating factor.
If , (9) is referred to as a backward Euler approach and
gives the trapezoidal approach. Often a value of be-
tween 0 and 1 is chosen to take advantage of characteristics of
both approaches. In this paper, branches are modeled using a
lumped series resistance and inductance between buses, along
with a lumped capacitance element shunted to ground at each
terminal of the line. The standard differential voltagecurrent
relationships for inductor and capacitor elements can be applied
to (9) to obtain [25]
(10)
(11)
where is the voltage across the branchs series resistor/
inductor; and are the voltage and current of the capacitor,
respectively; and R, L, and C are the values of resistance, induc-
tance, and capacitance for the branch (line) under consideration.
B. Measurement Matrix
Each line current measurement adds a row to the measure-
ment matrix. Quantities being measured (line currents) are
themselves functions of other variables that are to be estimated
(bus voltages). For a line connected between buses and , the
sending-end line current is given as
(12)
where is the current through the shunt capacitance and
is the current through the series resistance/inductance.
The shunt capacitance is modeled as a capacitor along with a
history term which is associated with the numerical integra-
tion method. Similarly, the line section includes a series resis-
tance/inductance and a history term .
Substitution (10) and (11) into (12) gives
(13)
where are the history terms
of the capacitive and inductive components of the line respec-
tively, and are dened as (based on the remaining terms when
substituting (10) and (11) into (12)
(14)
(15)
Equation (13) can be written for all branches and collectively
written in matrix form as
(16)
1844 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 22, NO. 3, JULY 2007
where is dened as the history vector. Equa-
tion (16) can be rewritten as
(17)
Comparing (17) with (5) shows that the measurement vector
is dened as
(18)
and the state vector is dened as
(19)
Comparing (13) and (16), the th row of , corresponding to
the th branch between buses and is obtained as follows:
(20)
(21)
and if branch is not connected to bus . Assuming
there are measurements and states (bus voltages), then
is a vector of measurements, is the mea-
surement matrix, is a vector of bus voltages, and
is an vector of history terms. Instantaneous bus voltages are
then calculated using (6).
Once the instantaneous voltages are obtained, the Pst is cal-
culated using the approach outlined in Section II.
The monitoring of voltage icker severity in all buses gives
insight to existing PQ problems in the network. A nancial
penalty factor calculation can be determined for consumers with
loads responsible for system voltage icker.
V. CLOCK SYNCHRONIZATION AND REAL-TIME
IMPLEMENTATION CHALLENGES
In order to have an accurate and useful monitoring of PQ in-
dices, real-time measurements from the grid are required. In the
early 1980s, a measurement system with synchronized clocks
was developed to measure the voltage phase angle between two
buses in real time [26], and several applications of synchronized
measurement have been discussed in the literature, for example,
[27]. The major problem with these applications was the dif-
culty of achieving clock synchronization over a wide area. How-
ever, global positioning system(GPS) time signals can presently
be used as the source of synchronization, so that wide-area mon-
itoring of power systems based on synchronized measurements
is possible [28][31]. In this paper, it is initially assumed that
all clock signals are synchronized (i.e., using a GPS-based ap-
proach). Next, a normally distributed error of 2 ms is applied
to the measured signals. This error was chosen to approximate
delay uncertainty in a traditional power system communication
system, such as a microwave system.
VI. NUMERICAL ANALYSIS
A simulation based on the IEEE 14-bus test system [32] is
used to demonstrate the proposed approach. For the estimation
of the icker, the voltage at bus 1 in addition to 14 current
Fig. 1. Maximum per-unit error in voltage over the 10-min simulation, with
and without the noise signal introduced.
Fig. 2. Mean absolute error for per-unit voltage over the 10-min simulation,
with and without the noise signal introduced.
measurements is used. Although the number of real-time me-
ters is relatively high compared to the number of voltage mea-
surements to be estimated, the algorithm is proposed for sit-
uations where all voltage measurements are not available but
current measurements are available. Alternatively, the proposed
approach could be used as a backup to voltage meter readings.
Simulations of the grid in the time domain are conducted using
a time step of 1 ms for 10 min. An arc furnace, connected to bus
10, is modeled as a variable current source as
(22)
where and . Gaussian noise with zero
mean and % is applied to all measured quantities.
The measurement matrix is constructed by means of grid data
and . This factor shows that the integrating method is
closer to a Trapezoidal method rather than Euler. The estimated
instantaneous voltages are then used to obtain the IFL signals.
MAZADI et al.: INSTANTANEOUS VOLTAGE ESTIMATION FOR ASSESSMENT AND MONITORING 1845
Fig. 3. Cumulative probability function (CPF) for bus 6.
Fig. 4. Pst values for system buses.
Fig. 1 shows the maximum error in the estimated bus volt-
ages over the ten minute simulation with and without the noise
signal introduced. It was found that the maximum voltage error
is approximately 0.04 p.u. when considering noise. For the same
simulation, Fig. 2 shows the mean absolute error. It was found
that the mean error was less than 0.017 p.u.
The cumulative probability function (CPF) of the actual and
estimated IFL is shown in Fig. 3. For classes under 40, the pro-
posed approach effectively determines the correct probability. A
small error is seen for higher classes; however, its probability is
relatively small and should not signicantly affect the Pst value.
Fig. 4 shows the Pst for all system buses. In all cases, the pro-
posed approach calculated the Pst value with relatively small
error (less than 0.06 in the worst case). It should be noted that
the number of classes that were chosen is relatively high. Re-
duced error can be obtained if a smaller number of classes (or
levels) were chosen but the cost would be reduced resolution.
Fig. 5. Mean per-unit error in voltage over the 10-min. simulation, considering
parameter error with noise.
Fig. 6. Mean per-unit error in voltage over the 10-min simulation, considering
delay and noise.
Next, the impact of parameter errors was considered, since
the values of transmission line and transformer admittances used
in numerical models normally contain various inaccuracies due
to mathematical approximations, simplied modeling assump-
tions, and weather effects. Studies have concluded that errors of
the order of 5% of nominal values should be expected [7]. In-
cluding 5%parameter error as expected, leads to greater error in
estimated bus voltages, as shown in Fig. 5. However, the error
is still relatively small. These values were determined using the
worst mean errors when repeating the simulations 30 times.
Next, in order to examine the impact of clock synchroniza-
tion, the measurement signals where randomly shifted mil-
lisecond. This shift is not to model the communications delays,
but the uncertainty in these delays, since expected delays can
be incorporated into the data stream. Introducing this time shift
and noise resulted in increased error in the estimated voltages,
as shown in Fig. 6, but again, the error is relatively small (0.02
p.u. in the worse case).
1846 IEEE TRANSACTIONS ON POWER DELIVERY, VOL. 22, NO. 3, JULY 2007
VII. CONCLUSION
In this paper, a new technique for icker monitoring in power
systems has been proposed. By means of line current measure-
ments, instantaneous voltages were estimated. These estimated
values are then used to determine the Pst for all buses. The cur-
rent-based model allows for a linear mapping between the mea-
sured variables and the states to be estimated. Parameter error
and clock synchronization were both considered.
REFERENCES
[1] IEC Standard, Electromagnetic Compatibility (EMC), , 1991, Part 4:
Limits, Section 7: general guide on harmonics and inter-harmonics
measurements and instrumentation, for power supply systems and
equipment connected thereto.
[2] IEC Standard, Electromagnetic Compatibility (EMC), , 1997, Part 4:
Limits, Section 15: Flickermeter- Functional and Design Specica-
tions.
[3] R. C. Dougan, M. F. McGranaghan, and H. W. Beaty, Elect. Power
Syst. Qual. New York: McGraw-Hill, 1996.
[4] Flicker-Meter Functional and Design Sepecication, Int. Std., CEI,
IEC 868, 1990.
[5] M. Mazadi and S. H. Hosseinian, Flicker-meter simulation to use in
power systemanalysis programs, in Proc. 39th Int. Universities Power
Engineering Conf., UPEC Bristol UK, Proc., Sep. 2004, vol. 2, pp.
917923.
[6] F. C. Schweppe, J. C. Wildes, and D. Rom, Power system static state
estimation, part i, ii, ii, IEEE Trans. Power Syst., vol. PAS-89, no. 1,
pp. 120135, Jan. 1970.
[7] A. S. Debs, Modern Power Systems Control and Operation. Norwell,
MA: Kluwer, 1988.
[8] G. T. Heydt, Identication of harmonic sources by a state estimation
technique, IEEE Trans. Power Del., vol. 4, no. 1, pp. 569576, Jan.
1989.
[9] A. P. S. Meliopoulos, F. Zhang, and S. Zellingher, Power system har-
monic state estimation, IEEE Trans. Power Del., vol. 9, no. 3, pp.
17011709, Jul. 1994.
[10] Z. P. Du, J. Arrillaga, and N. R. Watson, Continuous harmonic state
estimation of power system, in Proc. Inst. Elect. Eng., Jul. 1996, vol.
4, pp. 329336.
[11] Z. P. Du, J. Arrillaga, and N. R. Watson, Implementation of har-
monic state estimation continuous harmonic state estimation of power
system, in Proc. 8th Int. Conf. Harmonic and Quality of Power,
Athens, Greece, 1998, pp. 273278.
[12] K. Srinivasan, Digital measurement of voltage icker, IEEE Trans.
Power Del., vol. 6, no. 4, pp. 15931598, Oct. 1991.
[13] A. Grigis, J. W. Stephan, and E. Makram, Measurement and predic-
tion of voltage icker magnitude and frequency, IEEE Trans. Power
Del., vol. 10, no. 3, pp. 16001605, Jul. 1995.
[14] S. A. Soliman and M. E. El-Hawary, Measurement of power system
voltage and icker levels for power quality analysis: A static lav state
estimation based algorithm, Elect. Power Energy Syst., vol. 22, no. 6,
pp. 447450, 2000.
[15] O. Poisson, P. Rioual, and M. Meuneir, Detection and measurement
of power quality disturbances using wavelet transform, IEEE Trans.
Power Del., vol. 15, no. 3, pp. 10391044, Jul. 2000.
[16] T. Zheng and E. Makram, Wavelet representation of voltage icker,
J. Power Syst. Res., vol. 48, pp. 133140, 1998.
[17] A. V. Oppenheim and R. W. Schafer, Discrete Time Signal Pro-
cessing. Upper Saddle River, NJ: Prentice-Hall, 1998.
[18] M. I. Marei, E. F. El-Saadany, and M. M. A. Salama, Estimation tech-
niques for voltage icker envelope tracking, J. Power Syst. Res., vol.
70, pp. 3037, 2004.
[19] W. M. A. Hasawi and K. M. El-aggar, A genetic based algorithm for
voltage icker measurement, J. Power Syst. Res., vol. 26, pp. 593596,
2004.
[20] T. K. A-Galil, E. F. El-Saadany, and M. M. Salama, Energy operator
for on-line tracking of voltage icker level, in Proc. IEEE Power Eng.
Soc. Winter Meeting, 2002, pp. 11531157.
[21] J. A. Arrillaga, N. R. Watson, and S. Chen, Power System Quality As-
sessment. New York: Wiley, 2000.
[22] A. J. Woods and B. F. Wollenberg, Power Generation, Operation and
Control. New York: Wiley, 1984.
[23] A. Abur and A. G. Expsito, Power System State Estimation Theory
and Implementation. New York: Marcel Dekker, 2004.
[24] J. H. Mathews, Numerical Methods for Mathematics, Science and En-
gineering. Englewood Cliffs, NJ: Prentice-Hall, 1992.
[25] S. H. Hosseinian, Improvements in simulation methods for power
system harmonic and transient studies, Ph.D. dissertation, Upon Tyne
University, Newcastle Upon Tyne, U.K., 1996.
[26] G. Missout and P. Girard, Measurement of bus voltage angle between
montreal and sept-iles, IEEE Trans. Power Syst., vol. 99, no. 2, pp.
536539, 1980.
[27] P. Bonanomi, Phase angle measurement with synchronized clock
principle and applications, IEEE Trans. Power App. Syst., vol. 100,
no. 12, pp. 50365043, Dec. 1981, .
[28] J. Balcells, V. Parisi, and D. Gonzalez, New trends in power mea-
suring instruments, in Proc. IECON, Ind. Electron. Soc., IEEE 28th
Annu. Conf., Nov. 2002, vol. 4, pp. 33443349.
[29] A. G. Phadke, Synchronized phasor measurement in power system,
IEEE Comput. Appl. Power, vol. 6, no. 2, pp. 1015, Apr. 1993.
[30] H. Saitoh, Trend in synchronized measurement technology applica-
tion in power systems, Trans. Inst. Elect. Eng. Jpn., vol. 119-B, no.
8/9, pp. 897900, 1999.
[31] H. Saitoh, Gps synchronized measurement application in japan, in
Proc. Transmission and Distribution Conf. Exhibit.: AsiaPacic. IEEE/
Power Eng. Soc., Oct. 2002, vol. 1, pp. 494499.
[32] Power Systems Test Case Archive, [Online]. Available: http://www.ee.
washington.edu/research/pstca., Dept. Elect. Eng., Unv. Washington,
Seattle.
Mahmoud Mazadi received the B.Sc. degree from Isfahan University of Tech-
nology, Isfahan, Iran, in 1996 and the M.S. degree from AmirKabir University,
Tehran, Iran, in 2000, where he is currently pursuing the Ph.D. degree.
From 2005 to 2006, he completed a research visit at the University of Cal-
gary, Calgary, AB, Canada. His research interests are power system operation,
state estimation, power quality, and identication of power systemdisturbances,
power system simulation, and modeling.
Seyed Hossein Hosseinian received the Ph.D. degree from Newcastle Upon
Tyne University, Newcastle Upon Tyne, U.K., in 1995.
Currently, he is an Assistant Professor of Electrical Engineering in the Depart-
ment of AmirKabir University. His research interests are power quality, power
system simulation and modeling, state estimation, and electricity marketing.
WilliamRosehart (M01SM06) received the Ph.D. degree in electrical engi-
neering from the University of Waterloo, Waterloo, ON, Canada, in 2001.
Currently, he is an Associate Professor in the Department of Electrical and
Computer Engineering at the University of Calgary. His research interests are
in optimization and stability analysis.
David T. Westwick (M97) received the Ph.D. degree in electrical engineering
from McGill University, Montreal, QC, Canada, in 1995.
Currently, he is an Associate Professor in the Department of Electrical and
Computer Engineering at the University of Calgary, Calgary, AB, Canada. His
research interests include system identication and parameter estimation.