Professional Documents
Culture Documents
Authorized licensed use limited to: Bapatla Engineering College. Downloaded on October 21, 2009 at 23:15 from IEEE Xplore. Restrictions apply.
2
this time, the power to the load is supplied temporarily from figure. The DSTATCOM has 12 IGBT switches with anti-
the DSTATCOM. This leads to a decrease in dc link voltage if parallel diodes(not shown in the figure for the sake of clarity),
the load is increased or an increase in capacitor voltage if the a dc storage capacitor, three isolation transformers and three
load is reduced. For good compensation, it is important that interface inductors. In modeling the system, the inductance
the capacitor voltage remains as close to the reference value as and resistance of the isolation transformers can be included in
possible. After a load change has occurred, depending on the Lf and Rf, the interface inductance and resistance. Since we
values of Kp and Ki, the capacitor voltage takes 6-8 cycles to assume the source voltages to be balanced, sinusoidal and
settle. Most of the times, the gains are chosen by trial and stiff, Ls and Rs, the source impedances, are assumed to be
error. A method to obtain good Kp and Ki values for the negligible. Hysteresis band current control has been used for
DSTATCOM application is given in [7]. This has been used as tracking the reference currents. As seen in Fig. 1, the
the base values during steady operation. DSTATCOM injects the reference filter currents generated
However, during transient operation, it is possible to using the symmetrical component theory at the Point of
improve the performance of the dc link by varying the gains of Common Coupling (PCC). For good current injection and thus
the PI controller using a set of heuristic rules based on expert compensation, maintaining the dc capacitor value close to the
knowledge. Also, improvements in technology such as faster reference value is very important.
DSPs allow us to increase the sampling rate for better
feedback as to how the system responds to changes. For III. DC LINK PI CONTROL AND FUZZY SUPERVISION
nonlinear systems, like the DSTATCOM, fuzzy based control During load changes, there is some active power exchange
has been proved to work well [8]. In this paper, it has been between the DSTATCOM and the load. This leads to a
shown that fuzzy logic based supervision of the dc link PI reduction or an increase in the dc capacitor voltage. Using a PI
controller gains improves the transient and settling controller, the Ploss term in (1) is controlled to ensure that the
performance of dc link voltage control. Hence, the use of
dc capacitor voltage does not deviate from the reference value.
fuzzy logic for this application is justified.
The control output of a PI controller is given by (2).
This paper has been organized in the following manner.
First an explanation of the VSI topology for the DSTATCOM Ploss = K p (vdcref − vdc ) + K i ∫ (vdcref − vdc ) dt (2)
used is given and then the state space modeling used to
simulate the working of the DSTATCOM is explained. The The input to the PI controller is the error in the dc link voltage
design of the fuzzy supervisor for this system is elucidated. and the output is the value of Ploss . The value of Ploss depends
The methodology and results of the simulation are shown in on the value of K p , K i and the error in dc link voltage. Thus, it
the final section, proving improved dc link performance.
is important to tune K p and K i properly. Because of the
II. H-BRIDGE VSI TOPOLOGY BASED DSTATCOM inherent non-linearity and complexity of the system, it is
Many voltage source inverter (VSI) topologies have been difficult to tune the gains of the controller. It is usually done
used for DSTATCOM applications [9]. The three phase H- by trial and error. The base values of K p and K i have been
bridge VSI topology has been chosen because of the presence designed using the energy concept proposed in [7]. Also, it has
of a single dc link capacitor (unbalance in capacitor voltages been shown in literature that fuzzy supervision can improve
which may occur in topologies with two dc capacitors need the performance of PID controllers in nonlinear systems [10]-
not be considered), simplicity and independent current [12].However, these mostly deal with set-point changes in
tracking in each phase. Fig. 1 shows a three phase, four wire control applications. The derivative control term is not used
compensated system using H-bridge VSI based DSTATCOM. because improvement in stability may or may not be obtained
when used only with proportional control and if it is used with
vsa Ls Rs isa PCC ila integral control as well, tuning for good performance is
Za
v sb isb ilb difficult [13]. The design of a fuzzy system is highly system
Zb specific and requires in-depth knowledge of the system and
vsc isc ilc n
Lf Rf
Zc the various parameters that can be controlled for good
N ifa ifb ifc Unbalanced performance. The design of a fuzzy supervisor for dc link PI
load
A phas e
control in a DSTATCOM is given in the next section.
H-Bridge
DC Link
if i Cd c
S1a S3a S1b S3b S1c S3c
IV. DESIGN OF THE FUZZY LOGIC SUPERVISOR FOR PI
Cdc vdc CONTROLLER
S4a S2a S4b S2b S4c S2c L nl R nl
PID controllers are extensively used in industry for a wide
i0 range of control processes and provide satisfactory
n' Nonlinear load
performance once tuned when the process parameters are well
known and there is not much variation. However, if operating
Fig. 1. A Three phase, four wire compensated system using H-bridge VSI
based DSTATCOM. conditions vary, further tuning may be necessary for good
The unbalanced and non-linear load used is also shown in the performance. Since many processes are complicated and non-
linear, fuzzy control seems to be a good choice. Literature
Authorized licensed use limited to: Bapatla Engineering College. Downloaded on October 21, 2009 at 23:15 from IEEE Xplore. Restrictions apply.
3
(a)
Input Inference Output
Fuzzification Defuzzification
Mechanism
μ
Rule Base NL NM NS Z PS PM PL
1
The inputs of the fuzzy supervisor have been chosen as the (b)
error in dc link voltage and the change in error in dc link Fig. 3(a) Membership functions for error input. (b) Membership functions for
change in error input
voltage.
C. Inference Mechanism
err (i ) = vdcref − vdc (i) (3)
The two main functions of the inference mechanism are:
derr (i ) = err (i ) − err (i − 1) (4) a) Based on the active membership functions in error
and the change in error inputs, the rules which apply
In (3) and (4) above, e(i) is the error and e (i ) is the change in for the current situation are determined.
error in the ith iteration. vdcref is the reference dc link voltage b) Once the rules which are on are determined, the
and vdc (i ) is the dc link voltage in the ith iteration. The outputs certainty of the control action is ascertained from the
membership values. This is known as premise
of the fuzzy supervisor are chosen as the change in Kp value
quantification.
and the change in Ki value.
Thus at the end of this process, we shall have a set of rules
K p = K p ref + ΔKp (5) each with a certain certainty of being valid. The database
containing these rules is present in the rule base from which
K i = K i ref + ΔKi (6) the control action is obtained. The rule base will be discussed
in the next section. An example of a rule is given in (7). The
K p ref and K i ref are the steady state values determined by the
terms PL and PM are the membership functions for error and
method specified in [7]and ΔKp and ΔKi are the outputs of the for change in error respectively.
fuzzy logic supervisor.
Authorized licensed use limited to: Bapatla Engineering College. Downloaded on October 21, 2009 at 23:15 from IEEE Xplore. Restrictions apply.
4
IF
NM L L M S S Z S
"error" is PL (positive large)
"change in error" is PM (positive medium) NS L M S S Z Z Z
(7)
THEN Z M Z Z Z Z Z M
"ΔKp" is L ( Large Kp )
PS Z Z Z S S M L
"ΔKi" is SKi ( Small Ki )
PM S Z S S M L L
The minimum operation is used to determine the certainty
called μpremise of the rule formed by their combination. PL Z S S M L L L
D. The Rule Base (a)
Authorized licensed use limited to: Bapatla Engineering College. Downloaded on October 21, 2009 at 23:15 from IEEE Xplore. Restrictions apply.
5
Voltage in volts
bridge rectifier feeding an inductive R-L load. The parameters 0
of the system are given in Table 1. Since the average load -100
To compute Plavg a moving average filter for half a cycle has -400
0 0 .0 1 0 .0 2 0 .0 3 0 .0 4 0 .0 5 0 .0 6 0 .0 7 0 .0 8 0 .0 9 0 .1
been used. In steady state, at each zero crossing of the phase a Time in s e c o n d s
(a)
voltage, Ploss is generated using the PI controller. No n lin e a r Unb a la nc ed L o a d Cu r r e n ts
15 phas e-a phas e-b phas e-c
TABLE I
SIMULATION PARAMETERS 10
Current in Amperes
Supply voltage 220V (phase-rms), 50 Hz 0
Rla = 50 Ω, Lla = 20 mH -5
Unbalanced load Rlb = 35 Ω, Llb = 40 mH
Rla = 70 Ω, Lla = 20 mH -10
Three-phase full wave rectifier drawing
Non-linear load
a dc current of 5 A -15
0 0 .0 1 0 .0 2 0 .0 3 0 .0 4 0 .0 5 0 .0 6 0 .0 7 0 .0 8 0 .0 9 0 .1
DC capacitor C dc = 2200 µF Time in s e c o n d s
(b)
Interface inductor Lf = 20 mH, Rf = 5 Ω Fig. 7(a). Balanced sinusoidal source voltages (b) Nonlinear unbalanced load
currents.
Reference dc link voltage V dc ref = 500 V Re f e r en c e a n d A c tu a l Filte r Cu r r e n ts in Ph a s e A
5
Referenc e c ompens ator c urrent ac tual c urrent
3
Arbitrarily tuned controller
gains
K p = 40, K i = 20 2
Current in Amperes
-1
state space modeling and analysis of the DSTATCOM can be 0 .1 0 .10 2 0 .1 0 4 0 .1 0 6 0 .1 0 8 0 .11 0 .1 1 2 0 .11 4 0 .1 1 6 0 .1 1 8 0 .1 2
-10
The source voltages and the load currents have been plotted
-15
in Figs. 7(a)-(b). The load consists of an unbalanced R-L load
-20
and a diode rectifier supplying a highly inductive load as -25
given in Table 1. The THDs for the load currents for phase a, 0 .0 1 0 .0 1 5 0 .0 2 0 .0 2 5 0 .0 3 0 .0 3 5
Time in s e c o n d s
0 .0 4 0 .0 4 5 0 .0 5
phase b and phase c are 12.3%, 10.5% and 14.5% respectively. (b)
Phase-a compensator current showing the hysteresis action is Fig. 8(a). Phase-a reference and actual compensator currents (b) Balanced
shown is shown in Fig. 8(a) and the compensated source and sinusoidal source currents.
currents are shown in Fig. 8(b). The source currents have a unity power factor and are
balanced sinusoids. The THDs for the source currents for
phase a, phase b and phase c are 1.63%, 1.77% and 1.58%
respectively. Thus, we can clearly observe the compensating
Authorized licensed use limited to: Bapatla Engineering College. Downloaded on October 21, 2009 at 23:15 from IEEE Xplore. Restrictions apply.
6
action of the DSTATCOM. Fig. 9 shows the voltage [3] H. Akagi, Y. Kanazawa, and A. Nabae, “Instantaneous reactive power
compensators comprising switching devices without energy storage
waveforms when a normally tuned PI (using the fast acting dc components,” IEEE Trans. on Ind. Appl. Vol. 20, no. 3, 625-630, 1984.
concept[7]) and a fuzzy supervised PI controller are used. [4] F. Z. Peng and J. S. Lai, “Generalized instantaneous reactive power
Clearly, the fuzzy PI controller gives a better performance of theory for three-phase power systems,” IEEE Trans. on Instrumentation
and Measurement. Vol. 45, no. 1, 293-297, 1996
the dc link with lesser settling time and nearly 50% lesser
[5] H. Kim, F. Blaabjerg, B. B. Jemsen and J. Choi, “Instantaneous power
deviation of the dc link during load change. The load has been compensation in three-phase systems by using p-q-r theory,” IEEE
doubled at 0.2 seconds and at 0.4 seconds, it is brought back to Trans. on Power Electronics, Vol. 17, no 5, 701-709, 2002.
the original value. A clear dip in the dc link voltage is seen at [6] A. Ghosh and A. Joshi, “A new approach to load balancing and power
factor correction in power distribution system,” IEEE Trans. on Power
0.2 seconds as the compensator supplies the load during the Delivery, Vol. 15, no. 1 pp. 417-422, 2000.
transient period after the increase in load. [7] K. Karthikeyan and Mahesh Kumar, “A three phase DSTATCOM
compensating AC and DC loads with Fast Dynamic Response,” IEEE
Canadian Conf. on Elec. and Comp. Engineering, pp. 1199 – 1202, 2008
DC link Voltages - Normally tuned and Fuzzy tuned [8] D. Abramovitch, “Some crisp thoughts on fuzzy logic,” American
540
Control Conference, Vol. 1 pp. 168-172, 1994
530
Fuzzy supervised PI
[9] S. Iyer, A. Ghosh and A. Joshi, “Inverter Topologies for DSTATCOM
Normally tuned PI
applications - a simulation study,” Elsevier, Vol. 75 pp. 161 175, 2005
520 [10] R. P. Copeland and K. S. Rattan, “A fuzzy logic supervisor for PID
control of unknown systems,” Proc. of the 1994 IEEE International
510 Symposium on Intelligent Control, pp. 22-26, 1994.
DC Link Voltage in
500
371, 1990.
490 [12] Zhen-Yu Zhao, M. Tomizuka and S. Isaka, “Fuzzy gain scheduling of
PID controllers”, IEEE Tran. on Systems, Man and Cybernetics, Vol. 23,
480 pp. 1392-1398, 1993.
[13] K. H. Ang, G. Chong and Y. Li, “PID control system analysis, design
470 and technology,” IEEE Trans. on Control Systems Technology, Vol. 13,
no. 4, 559-576, 2005
460
0 0.1 0.2 0.3 0.4 0.5 0.6 [14] B. N. Singh, A. Chandra and K. Al-Haddad, “DSP-based indirect-
Time in
seconds
current-controlled STATCOM. I. Evaluation of current control
techniques,” IEE Proc. on Electric Power Applications, Vol. 147, pp.
Fig. 9. DC link voltages using fuzzy supervised and normally tuned PI 107-112, 2000
controllers. [15] A. Ajami and H.S. Hosseini, “Application of a Fuzzy Controller for
An increase in voltage at 0.4 seconds when the compensator Transient Stability Enhancement of AC Transmission System by
takes power from the load during the transient period after the STATCOM,” International Joint Conference SICE-ICASE, pp. 6059-
6063, 2006
load has been brought back to the initial value. From Fig. 9, [16] H.R. Van Lauta Nemke and Wang De-zhao, “Fuzzy PID Supervisor,”
we clearly see an improvement in dc link performance. 24th IEEE Conf. on Decision and Control, Vol. 24, pp 602-608, 1985
[17] K. M. Passino and S. Yurkovich; “Fuzzy Control” Addison-Wesley,,
1998
VI. CONCLUSIONS
VIII. BIOGRAPHIES
A fuzzy logic supervisor to the dc link PI controller of the
DSTATCOM has been proposed. The supervisor varies the Harish Suryanarayana is a graduate student in the
department of Electrical Engineering at the Indian
gain of the PI controller during the transient period in a way Institute of Technology – Madras, Chennai, India.
that improves performance. The system has been modeled and His reseach interests are Power Electronics, Power
simulated in the MATLAB technical environment. The Quality and Drives.
performance of the dc link and compensation were observed
with and without the fuzzy supervisor. Simulation results
show a 50% reduction in voltage deviation of the dc link
voltage with faster settling time. Instantaneous symmetrical
Mahesh K. Mishra (S’2000-M’02) received his
component theory has been used for load compensation. Good Bachelor of Technology from College of
compensation has been observed as source current THDs for Technology, Pantnagar, India and M.E. from
each phase is 1.63%, 1.77% and 1.58% while the load THDs University of Roorkee, India in 1991 and 1993
respectively. In Feb. 2002, he received the Ph.D. in
are 12.37%, 10.5% and 14.54% respectively. Thus, through Electrical Engineering from Indian Institute of
simulation studies, the implementation of a fuzzy supervisor Technology, Kanpur, India. Currently he is Assistant
for DC link voltage control in a DSTATCOM using Professor in the Electrical Engineering Department
at IIT-Madras. His interests are in the areas of
instantaneous symmetrical component theory for load Power Distribution Systems, Power Electronics
compensation has been demonstrated. and Control Systems.
VII. REFERENCES
[1] A. Ghosh and G. Ledwich; “Power Quality enhancement using custom
power devices,” Kluwer Academic Publishers, Boston, 2002
[2] N. Hingorani, “Introducing Custom Power ,” IEEE Spectrum. Vol. 32,
no. 6 pp. 41-48,1995.
Authorized licensed use limited to: Bapatla Engineering College. Downloaded on October 21, 2009 at 23:15 from IEEE Xplore. Restrictions apply.