Voltage Control In Power System Using Expert

System Based On SCADA System

C. Pimpa, and S.Premrudeepreechacharn, Member, IEEE

[6-8], other applications include bus load modeling [9] and

Abstract-- This paper present the type of expert system for implementation of a short term load forecast [10-12]. Some
controlling the 22kV voltage levels of power system in northern attempts have been reported in literature for reactive
region of Thailand based on the SCADA system. At present, for compensation made in this expert system, which can be easily
the operation, the operators have to make decision by their
knowledge and experience to control the voltage. This expert
used to form an expert system for real-time control.
system is obtained for alleviation of voltage violation in the day- This paper deals with the development of an expert system
to-day of distribution substation in the system and process the for voltage correction for normal case and contingency
data from the SCADA for helping the operator detect buses conditions by switching shunt reactive compensation and
experiencing abnormal conditions. This expert system focused transformer tap changer as shown in Fig. 1. In early for
on shunt reactive compensating devices and transformer tap
voltage controls, the operators have to make decision by their
change to select the most effective control for keeping the bus
voltage in the entire system within limits. This study selected 6 knowledge and experience to switching shunt reactive
distribution substations to be tested with this expert system. compensation and transformer tap change are controlled by
The performance of the expert system is compared with the computer from control center. This expert system solution
operator control in the terms of voltage deviation indices, methodology based on the voltage and load sensitivities of the
numbers of tap changing and minimize loss. load buses to controller devices. The performance of the
expert system is compared with the operator controlling
Index Terms-- Expert System, Power System Control, Var
voltage, which tested on voltage deviation indices, minimize
Control, Voltage Control.
numbers counter of tap changing and minimize loss. The
I. INTRODUCTION application of this expert system has been tested on 6
substations in power system in northern region of Thailand.
F or the purpose of economic, quality and security in power
services, the secure operation of a power system becomes
The system contains 11 of distribution transformers
(115/22kV), a tie transformer (230/115kV) and 28 steps of
essential to maintain network voltage profile within specified
shunt capacitors for 22kV, 6 steps of shunt capacitors for
limits. In the day-to-day operation, power systems experience
115kV. The power system network is shown in Fig. 2.
both over-voltage or under-voltage violations. These
violations occur due to inadequate reactive power support for
different loading conditions and network configurations. Voltage Control
These violations can be relieved by coordinated control and
switching of reactive power control devices [1].
The number of expert systems practically applied to electric
energy system operation is still low. But having solved the
Operator Computer Expert system
crucial integration of such systems into the given Control control by Control
environment of both technical process and human operating voltage limit
personnel, new interesting and advanced possibilities are Tap
opened [2]. Early applications of expert systems were in controlling Tap
Capacitors Tap Controlling.
reactive compensation for power system [3-5], and linear Controlling.
switching Capacitors
programming based optimal power flow for voltage control Inductors switching.
switching Inductors
switching.
This project was conducted with the support of Electricity Generating
Authority of Thailand (EGAT.), Thailand, in collaboration with Chiang Mai (Voltage control at present)
(In this paper)
University, Thailand.
C. Pimpa is with System Dispatch Section 2, Regional System Control and
Operation Division, Northern Regional Operation Department, EGAT,
Phitsanulok 65000 Thailand (e-mail: chakkrit@egat.or.th).
S. Premrudeepreechacharn is with the Department of Electrical Engineering,
Faculty of Engineering, Chiang Mai University, Chiang Mai 50200 Thailand Fig. 1. Block diagram of voltage control.
(e-mail: suttic@eng.cmu.ac.th).

0-7803-7322-7/02/$17.00 2002 IEEE

1282
0-7803-7322-7/02/$17.00 (c) 2002 IEEE
 CM3 115/22 kV
230/115 kV 2x50 MVA Knowledge base
2x200 MVA CM1

Data base Rule base

To network
230 kV

115/22 kV 115/22 kV
2x50 MVA 2x50 MVA

CM2 Inference
engine

115/22 kV 115/22kV
2x50 MVA 2x50MVA
To network
115 kV
LN1 User User
interface

115/22 kV LN2
50 MVA
CTG

To network
Fig. 4. Block diagram of expert system.
115 kV

III. KNOWLEDGE BASE

Fig. 2. Power system network considered in this paper.
The knowledge base of this expert system consists of
voltage sensitivities of load buses to different shunt capacitor
II. EXPERT SYSTEM compensating or on load transformer tap changing and status
The application of expert system for voltage control can be of measure device. The knowledge base consists of 2 parts;
data base and rules base, as described below.
described in a general block diagram form is shown in Fig. 3.
The SCADA system consists of two master computers in A. Data base
control center, which communicate with remote terminal The data base that obtained is arranged in the form of facts
units (RTUs.) in the substations. The power data of expert in the knowledge base of the proposed expert system.
system received data from SCADA system. The data of 1) The upper and lower limits of voltage at each bus, which
substation is sampled for MW/MVar, bus voltage and status difference value at any buses.
of devices, approximately every 5 to 10 seconds. Each system 2) The upper and lower limits of each control measure.
has knowledge in a particular domain and is capable of 3) Status of tap position and status (on/off) of each control
solving problems that require knowledge for expert system measures.
program. The structure of an expert system can be 4) The possible selection of control measures. (Tag for do
conveniently divided into three main modules in a general not control.)
block diagram form is shown in Fig. 4. 5) A sensitivity factor table of shunt capacitor
compensation list (SCL). The SCL list indicates in order of
preference the most suitable shunt capacitor compensation for
voltage correction at that load bus with the corresponding
Power system
Expert system sensitivities for one step switching. For example:
SCADA system
data of power
data for
expert system
program (
SCL N i , [ p, q,...], [ S cpi , S cqi ,...] ) (1)
system in northern This indicates that for node Ni , the most effective shunt
region of Thailand
capacitor compensation is located at node p with a sensitivity
Control action to power system of Sicp for one step switching, the next most effective shunt
capacitor compensation is located at node q with a sensitivity
of Sicq and so on.
6) A sensitivity factor table of transformer compensation
6 substations in
northern region of
list (TCL). The TCL list indicates in order of preference the
Thailand most effective transformer for voltage correction with the
corresponding sensitivities for one step tap change. For
example
Fig. 3 Block diagram of expert system for voltage control in power system
(
TCL N i , [t p , t q ,...], [ S tpi , S tqi ,...] ) (2)
This indicates that for node Ni , the most effective
transformer for voltage correction is located at node p with a

1283
0-7803-7322-7/02/$17.00 (c) 2002 IEEE
sensitivity of Sitp for one step tap change, the next most The load sensitivity ( SLi ) at each load buses, which using
effective transformer for voltage correction is located at node prediction equation of linear regression.
q with a sensitivity of Sitq for one step tap change and so on. L i = L0i + S Li (t i t ) (3)
7) The constraints of expert system are recommends
switching action for both shunt capacitors and transformer Vi = Vi 0 + S scli (C i ) (4)
tap for voltage correction. The expert system gives priority for
((ti t )( Li L ))
shunt capacitor compensation switching at the most effective S Li = (5)
for voltage correction. If this is not sufficient for voltage ( ti t ) 2
correction then it recommends transformer tap change at the
most effective transformer for each voltage violation node. S scli =
(V mes
vi Vvi0 ) (6)
These data base, except for the sensitivity table 5) and 6), C i
can be easily obtained for a power system. These data base where Li = Load at time sampling( i ).
can either be collected from the experts or from a numerical t = interval time sampling. (15 load sampling)
analysis using load flow calculations. SLi = load sensitivity.
B. Rules base Sscli = voltage sensitivity
3) For each bus voltage violation, sequentially search for
The rules base of an expert system describe relation or
the most effective control measure device by using the block
phenomena in the domain, and possible methods, heuristics
diagram to control device, which shown in Fig. 5 and
and ideas for solving problems in domain. The expert system
calculate the control action needed to recover the voltage
read the current voltage and load value and at each bus
violation.
through the monitor or input/output interface. Then, these
4) Check the control action from the view point of the
data compare the current voltage with their limits, and find
other bus voltage limits using the SCL or TCL ( Part 4 of Fig.
the buses with abnormal voltage and the magnitude of the
5 ). This is used to guarantee that recovering proper voltage
voltage violation. The expert system is implemented in the
level at one bus dose not cause bus voltage violations on other
form of a set of rules as described below:
buses.
1) If network voltage violation exists, then forms the follow
5) Output the search procedures and the recommended
list. Violation list, VL(Vv1 , Vv2 , ), in the order of most
control action through the I/O interface on monitor.
severe violation, where Vv1 is the most voltage violated node,
6) If voltage violation at node Vv is not completely relieved
Vv2 the next most voltage violated node and so on.
then pick the next most effective compensation from the SCL
2) Pick a node from the top of the violation list ( Vvi ).
list and go to rule 4).
- Check the most voltage violated node, it is lower or
7) Repeat the above procedure for each load bus of the
higher than voltage limits. (Part 1 of Fig. 5)
system until bus voltages are within limits.
- Check trend of load pattern at most voltage violated
The search procedure of the proposed expert system is
node.(increase or decrease, Part 2 of Fig. 5)
shown the detailed configuration of the developed expert
- Check trend of total load pattern in studies
system in Fig. 6.
network.(increase or decrease, Part 3 of Fig. 5)

Part 1. Part 2. Part 3. Part 4.

A1
Action 1 (A1)
A1 First, Search shunt reactive
compensation for optimum
A2
voltage to control.
A1 Second, If shunt reactive
compensation are not ready
A1
then it recommends
Voltage
Violation A1 transformer tap change.
Node (Vni)
A1

A2
Sensitivity A1
factor increase Action 2 (A2)
Sensitivity factor A2 For most effective control of
decrease transformer tap changed.
A2
Sensitivity
factor constant
A2

Fig. 5. Block diagram to control device.

1284
0-7803-7322-7/02/$17.00 (c) 2002 IEEE
 (Vdesired Vactual (i))
Input Interface. Vdeviation (i) = 100% (7)
Vdesired
The performance of the expert system is compared with the
Identify the Current voltage Current load operator control in the terms of voltage deviation indices,
operating at each buses at each buses counters of transformer tap changing. Typical results for the
configuration
No
expert system are presented in Fig. 8-12. The resulting are
presented by the average actual data from the expert system
Find buses Find load and operator for controlling the 22kV voltage levels based on
with sensitivity ( SLi ) at the SCADA system. The voltage of power system is in limits
Knowledge abnormal each load buses,
Base voltage which using range (-3%-7%) when operating by both operator and expert
prediction system. However, the results of the expert system can reduce
yes equation of linear the number of transformer tap change more than 50 percent
regression. and voltage deviation within limits as shown in Fig.12.
Form the follow list.
Violation list VL(Vv1 , Vv2 ,
), in the order of most
Load (MW)
severe violation Control action from the view 400
point of the controllers
constrains with load 350
sensitivity ( SLi ) 300
Pick a node from
the top of the 250
violation list ( Vv ).
200
No
Check control action 150
Check the measure from the view point
device already to 100
of other bus voltage
control limits 50
0
yes 5.00am. 10.00am. 3.00pm. 8.00pm. 1.00am.
Output Interface. Time

Fig. 7. The normalized total load curve of 6 substation in this case study.
Fig. 6. The search procedure for expert system.
Number (%)
The inference engine of the expert is used for control 60
action of shunt reactive compensation and transformer tap 50
change for voltage correction. All the fact of SCL and TCL
40
are combined to form the knowledge base and rules base. The
inference engine recommends switching at the most effective 30
of shunt reactive compensation and transformer tap for 20
voltage violated node.
10
IV. EXPERIMENTAL RESULTS 0
The approach proposed in this paper was implemented and -2 -1 0 1 2 3 4 5 6
tested on 6 substations in power system in northern of Voltage deviation(%) expert system operator
Thailand as shown in Fig. 1. The substations are Fig. 8. Load voltage deviation during 05.00-08.00.
Chiangmai1(CM1), Chiangmai2(CM2), Chiangmai3(CM3),
Number(%)
Chomthong(CTG), Lumphun1(LN1), Lumphun2(LN2), 60
which contained 11 of distribution transformers (115/22kV),
a tie transformer (230/115kV) and 28 steps of shunt 50
capacitors for 22kV, 6 steps of shunt capacitors for 115kV. 40
The normalized total load curve of networks is shown in Fig.
30
7. The normalized voltage deviation of the system was chosen
to be -3% and +7% for voltage limits in range (97-107%). 20
This value depends on the system operating conditions and is 10
set by operator based on experience with the system behavior.
In order to determine the voltage deviation is defined as (7) 0
and the time sampling (i) for each individual voltage bus is -2 -1 0 1 2 3 4 5 6
60 times/hour. Thus total numbers counter of voltage Voltage deviation(%)
expert system operator
deviation for 18voltage bus are 1080 times/hour.
Fig. 9. Load voltage deviation during 08.00-10.00.

1285
0-7803-7322-7/02/$17.00 (c) 2002 IEEE
 Number (%) VI. REFERENCES
50 [1] D. T. Bansilal and K. Parthasathy, An Expert System for Voltage Control
in a Power System Network, 1995 International Conference on Energy
40 Management and Power Delivery Vol.1, pp. 364-369.
[2] G. Krost and U. Spanel, Expert Systems Applied to Power System
30 Control, 12th Conference on the Electric Power Supply Industry 1998
(CEPSI 1998), Pattaya Thailand, Technical Paper Reference Book Vol.12,
pp. 22-30.
20 [3] S. J. Cheng, O.P. Malik and G.S. Hope, AN Expert System for Voltage
and Reactive Power Control of a Power System, IEEE Transaction on
10 Power System, Vol.3, No.4, pp. 1449-1455, November 1998.
[4] H. K. Clark and D. N. Ewart, Reactive Compensation in Power Systems,
0 First Symposium of Specialists in Electric Operational Planning
-2 -1 0 1 2 3 4 5 6 Electrobras-Rio de Janeiro, August 17-21,1987, pp. 344-353.
Voltage deviation(%) [5] T. B. Girotti, N. B. Tweed and N. R. Houser, Real-Time Var Control By
expert system operator SCADA, IEEE Transactions on Power Systems, Vol.5, No.1, pp. 61-64,
Fig. 10. Load voltage deviation during 17.30-18.00. February 1990
[6] D. S. Kirschen and H. P. Van Meeteren, MW/Voltage Control in a
Number(%) Linear Programming Based Optimal Power Flow, IEEE Transactions on
50 Power Systems, Vol.3, No.2, pp. 481-489, May 1988.
[7] S. K. Chang, G. E. Marks and K. Kato, Optimal Real-Time Voltage
40 Control, IEEE Transactions on Power Systems, Vol.5, No.3, pp. 750-
758, August 1990.
30 [8] K. H. Abdul-Rahman and S. M. Shahidehpour, A Fuzzy-Based Optimal
Reactive Power Control, IEEE Transactions on Power Systems, Vol.8,
20 No.2, pp. 662-670, May 1993.
[9] E. Handschin and C. Dornemann, Bus Load Modelling and Forecasting,
10 IEEE Transactions on Power Systems, Vol.3, No, pp. 627-633.2, May
1988.
0 [10] C. G. Kim, N. J. Woo, J. M. Lee and S. H. Kim, Implementation of a
-2 -1 0 1 2 3 4 5 6 Short-Term Load Forecasting Expert System, 12th Conference on the
Voltage deviation(%)
expert system operator Electric Power Supply Industry 1998 (CEPSI 1998), Pattaya Thailand,
Technical Paper Reference Book Vol.12, pp. 56-63.
Fig. 11. Load voltage deviation during 18.00-22.00. [11] S. Rahman and R. Bhatnager, An Expert System Based Algorithm for
Short Term Load Forecast, IEEE Transactions on Power Systems, Vol.3,
No.2, pp. 392-399, May 1988.
Tap changing [12] H. S. Jung, S. S. Kim, J. K Park, S. Dong and Kwanak-gu, DSM Effect
(times) Analysis on Hourly Load Forecast in Korea Using Engineering Estimation
35 Method, 12th Conference on the Electric Power Supply Industry 1998
30 (CEPSI 1998), Pattaya Thailand, Technical Paper Reference Book Vol.12,
25 pp. 134-141.
20
15 VII. BIOGRAPHIES
10
5
0 Chakkrit Pimpa was bon in Thailand in 1969. He received the B.S.degree in
electriccal engineering from the King Mongkut's
05.00- 08.00- 17.30- 20.00-
Institute of Technology North Bangkok, Thailand,
08.00 10.00 20.00 22.00 in 1994. In 1994 he worked in Electricity
Time
Generating of Thailand (EGAT), where he has
expert system operator been working on system dispatch section at
Regional System Control and Operation Division
Fig. 12. The average total numbers of transformer tap changing.

V. CONCLUSIONS Suttichai Premrudeepreechacharn (S91-M97)

An expert system for the power system voltage control is
developed in this paper. It has been tested with the actual data
of the power system in northern region of Thailand based on
the SCADA system. This expert system is to help the operator
detect buses experiencing abnormal conditions, select the
most effective shunt capacitor switching or transformer tap
change for each voltage violation. The results of the expert
system can reduce the number of transformer tap change
more than 50 percent and voltage deviation within limits.
was born in Chon Buri, Thailand in 1965. He
recieved B.Eng.in electrical engineering from Chiang Mai University Thailand
and M.S and Ph.D. in electric power engineering
from Rensselaer Polytechnic Institute, Troy, NY. He
is an assistant professor at Department of Electrical
Engineering, Chiang Mai University, Thailand. His
research interests include power quality, high quality
utility interface, power electronics and artificial
intelligent applied to power system.

1286
0-7803-7322-7/02/$17.00 (c) 2002 IEEE