Professional Documents
Culture Documents
www.jatit.org
ABSTRACT
The paper describes a variable speed wind generation system where fuzzy logic principles are used for
light load efficiency improvement and optimization. A squirrel cage induction generator feeds the power
to an improved topology of matrix converter which pumps power to a utility grid or can supply to an
autonomous system. The power factor at the interface with the grid is controlled by the matrix converter
to ensure purely active power injection into the grid for optimal utilization of the installed wind turbine
capacity. Furthermore, the reactive power requirements of the induction generator are satisfied by the
matrix converter to avoid self-excitation capacitors. The generation system has fuzzy logic control with
vector control in the inner loops. Fuzzy controller tracks the angular frequency with the wind velocity to
extract the maximum power and programs the machine flux for light load efficiency improvement. The
complete control system has been developed, analyzed, and validated by simulation study. Performances
have then been evaluated in detail.
Index Terms- Wind energy, squirrel cage Induction generator, matrix converter, fuzzy control, wind
power generation
79
Journal of Theoretical and Applied Information Technology
www.jatit.org
electronics and control, the life-cycle cost is lower. installed at additional cost. Again, the conventional
Many different configurations of variable-speed control principles used in these systems make the
wind turbines have been introduced in the response sluggish and give nonoptimum
literature [4], [5], [6]. The Authors of [4] performance. Very recently, a matrix converter
introduces a high- performance configuration, based WECS has been proposed to overcome some
commonly known as Scherbius drive, composed of of the above problems.
a doubly-fed induction generator (DFIG) and a Matrix Converter (MC) provides direct AC-AC
PWM AC/DC/AC converter connected between conversion and is considered an emerging
the stator and rotor terminals to implement variable alternative to the conventional two-stage AC-DC-
speed operation. Another configuration for AC converter topology [18],[19] as it provides a
variable-speed wind turbines has been introduced large number of control levers that allows for
by Zhang, Watthanasarn, and Shepherd [5], where independent control on the output voltage
system is composed of a DFIG with a matrix magnitude, frequency and phase angle, as well as
converter connected between the stator control the input power factor. When compared with the
winding and the main stator terminals. Variable AC-DC-AC converter system, the bold feature of
speed is implemented through control of the matrix MC is elimination of the DC-link reactive
converter. But these configurations suffer from the elements, e.g., bulky capacitors and/or inductors. A
disadvantages of having a wound rotor and a four-output leg improved MC topology with
brush-slip ring arrangement, and high cost of the advantages over the conventional nine-
doubly fed induction generator. bidirectional-switch topology has been developed
There have been active studies regarding the in which voltage gain is improved, control is
topologies of variable-speed turbines including simplified, free from commutation problems, and
those using cage induction[7] forth leg creates a three-phase plus neutral utility
and slow rotating, direct-drive machines [8]. Such power supply which provides the facility for
topologies usually require 100% rated, four- unbalanced and single-phase loads.
quadrant power-electronic converters. As the This paper, a complete simulation study to
converter cost decreases, such topologies will no validate the theoretical concepts, describes a
doubt become increasingly more popular. So, variable speed wind turbine system with a squirrel
recently, the use of squirrel-cage induction cage induction generator and a four-output leg
generator (SCIG) for grid-connection of WECS commutation free matrix converter where fuzzy
has been well established, due to its simplicity,
AC-AC
robustness, small size per generated kilowatt, and
low cost in comparison with other types of electric Gear
SCIG
Box
machines [9], [10], [11]. However, the major
drawbacks of the IGs are reactive power Matrix
Converter Grid
consumption and poor voltage regulation under Wind
varying speed, but the development of static power Turbine
converters has facilitated the control of the output
voltage of IG [12]–[16]. Fig. 1. Schematic diagram of proposed wind
The most common configuration of power energy conversion scheme
converters for WECS based on variable-speed logic control has been used extensively light load
wind turbine and SCIG is that composed of two efficiency improvement. All the control algorithms
back-to-back voltage source converters with a have been validated by simulation study and
large capacitor on the dc-link but it suffers from system performance has been evaluated in detail.
the demerit of bulky capacitor and having
relatively reduced lifetime [10]. THE PROPOSED WIND GENERATION SYSTEM
DESCRIPTION
The major problems in above mentioned
traditional power conversion schemes are the poor Fig. 1 shows a block diagram of the proposed
line power factor and harmonic distortion in line system where four-output leg improved topology
and machine currents. The IEEE Standard 519 [17] of matrix converter is used for interfacing WECS
severely restricts line harmonic injection. with the grid and fuzzy logic control has been used
Therefore, to satisfy the stringent harmonic for performance enhancement and efficiency
standard and poor power factor problem, active optimization. As the entire power generated by the
type VAR and harmonic compensators can be wind turbine is transferred through the matrix
80
Journal of Theoretical and Applied Information Technology
www.jatit.org
converter, this work targets low-to-medium-power and Vw the wind speed in m/s. The power
wind turbines. For medium-to-high-power wind coefficient is related to the tip speed ratio and
turbines, doubly-fed induction generator with a rotor blade pitch angle according to Equation (3)
pilot converter connected to the auxiliary winding [20].
will be more appropriate. The wind turbine is
⎛ 151 ⎞ − 18.4 / λi
followed by a gear box which steps up the shaft C p (λ , θ ) = 0.73 ⎜⎜ − 0.58θ − 0.002θ 2.14 − 13.2 ⎟⎟ e
speed. Note that working at a low shaft speed ⎝ λi ⎠
translates into a low induction generator terminal (3)
frequency which can result in core saturation
unless a low terminal voltage is imposed. At low where
terminal voltages, the operating current will be
high, making the scheme impractical. The matrix 1
λi =
converter interfaces the SCIG with the grid and 1 0.003
− 3
implements shaft speed control to achieve λ − 0.02θ θ +1
maximum power point tracking at varying wind (4)
velocities. It also performs power factor control at
the grid interface and satisfies the Var demand at and
the induction generator terminals. The proposed
scheme allows for connecting individual wind ω r Rr
turbines to the grid. It also permits paralleling the λ=
outputs of several wind turbine generation units at Vw
the grid interface. The power handling capability
of the system can be enhanced by adopting a multi- (5)
converter approach. In the following sub-sections,
In (5), r is the angular speed of the turbine
different elements of the system will be described.
shaft. The theoretical limit for Cp is 0.59 according
A. Wind Turbine Characteristics to Betz's Law [21], but its practical range of
Both horizontal and vertical axis wind turbines variation is 0.2-0.4. In this paper, the rotor pitch
are used in wind generation systems. The vertical angle is assumed to be fixed. Fig. 2 shows a typical
Darrieus (egg beater) type has the advantages of Cp versus curve. As Equations (1)-(5) suggest,
being located on the ground and accepting wind the aerodynamic torque(Tm=Kwr2) and the
from any direction without any special yaw mechanical power (P0 =Kwr3) generated by the
mechanism. It is, therefore, preferred for high wind turbine at a given wind velocity is a function
power output. The disadvantages are that the of the shaft speed.
turbine is not self-starting and there is a large
pulsating torque which depends on wind velocity, 0.4
Power coefficient Cp(λ)
⎣ η gear ⎦ 0.1
(1) 0.05
0
0 1 2 3 4 5 6 7 8 9 10
Speed ratio (λ)
1
P0 = ρC p ArVw3 Fig. 2. Polynomial function curve fitting of
2
(2) turbine power coefficient (Cp)
where P0 is the power in W, p the air density in
g/m3, Cp a dimensionless factor called power
coefficient, Ar the turbine rotor area in m2 (Ar =
Rr2, where Rr is the rotor blade radius), η gear is
81
Journal of Theoretical and Applied Information Technology
www.jatit.org
Gating Pulses n
iA iB iC
Fig. 4. Typical P0 versus r curves for different Controller
wind velocities (* = PMax). N
Side 2
This means that, at reduced speed light load φ1 V2 f2 θ2
steady state conditions, generator efficiency can be
improved by programming the flux [22] and [23], Fig. 5. Schematic diagram of the improved matrix
which will be discussed later. converter topology
In this paper, a wind turbine model has been
created in MATLAB simulation package based on
the equations (1)-(5). The block diagram of the
82
Journal of Theoretical and Applied Information Technology
www.jatit.org
V2 Measurement
fPWM
2 K of
P0 , Vac, i
generation
m δ
Fig. 6. Fuzzy logic based control block diagram of a wind generation system.
83
Journal of Theoretical and Applied Information Technology
www.jatit.org
with imprecise or noisy data; thus, it is able to ∆δ phase shift of the phase angle δ
extend its control capability even to those (positive for increment
operating conditions where linear control of δ and negative for
techniques fail (i.e., large parameter variations). decrement of δ);
The main drawback of an FL-based control system RF regulating factor of FC1 MFs;
is that the tuning of its membership functions ∆RF step change of RF;
(MFs) needs too much “trial and error.” In order to σ variance of e1 ;
reduce the time-consuming process of the MFs ωe1 − ωref absolute mean deviation of
tuning or to ameliorate the performance when it we1 from ref .
does not satisfy the specification, we can apply an The frequency at the point, e1 is compared to
online-tuned adaptive fuzzy control system ref, and the error is passed through FC1, which
(AFCS). AFCSs can adapt to their environment produces a signal ∆δ (Fig.7). By accumulating the
and acquire new knowledge by themselves through successive values of ∆δ, the value of δ comes up,
learning. A possible arrangement of such a system according to (6)
is the implementation of a fuzzy controller (FC) to δnew = δold + ∆δ (6)
adjust the parameters of another FC. This where δnew is the new value of the phase angle δ
adjustment is accomplished online. The main FCs and δold is the previous value of the phase angle δ.
MFs are tuned online through the supervisor-FC, As was previously mentioned, ref is the angular
which follows the reasoning of an expert which frequency, when the wind turbine operates at the
would manually tune the MFs. This type of control maximum aerodynamic efficiency for a specific
offers, except for the automatic MFs adjustment, wind speed. The value of ref is dynamically
an adapting tuning of the MFs, according to the approached in real time from FC2 (Fig.7), using a
behavior of the system. The system can be maximum power point tracking (MPPT) technique.
satisfactorily controlled for start-up and This is achieved by changing the reference value of
regenerative braking shutdown modes besides the the frequency by ∆ ref and monitoring the
usual generating mode of operation. AFCs are thus corresponding change of the output power, P0.
very suitable for the control of systems which are With an incrementation (or decrementation) of
strongly fluctuating, such as wind turbine reference frequency, the corresponding
generation system. incrementation (or decrementation) of output
power P0 is estimated. If ∆P0 is
III. CONTROL SYSTEM DESCRIPTION
The system has following fuzzy logic
controllers.
A. Angular Frequency Regulator (AFCS)
In order to achieve maximum wind power
absorption according to the wind speed, matrix
converter regulates the angular frequency. This is
managed through an AFCS (Fig.6). This system
[25] consists of FC1, which is the main controller
of the AFCS, FC3, whose main role is to fine-tune
FC1 and FC2, which dynamically detects online the
angular frequency that corresponds to the Fig. 7. Angular frequency regulator control system
maximum aerodynamic efficiency of the wind AFCS
turbine for a specific wind speed re f (Fig.7). positive with last positive ∆ ref , in per-unit value
The meaning of the symbols shown at Fig.7 is by L∆ ref , the search is continued in the same
the following: direction.
P0 output real power; If,o n the other hand, +∆ ref causes ¯∆P0, the
e1 angular frequency at the terminal of direction of search is reversed. The variables ∆P0,
generator; ∆ ref and L∆ ref are described by membership
ref angular frequency for maximum functions and rule table. In the implementation of
aerodynamic efficiency; fuzzy control, the input variables are fuzzified, the
δ phase angle of PWM firing pulse valid control rules are evaluated and combined,
generation, and finally the output is defuzzified to convert to
the crispy value. The wind vortex and torque ripple
84
Journal of Theoretical and Applied Information Technology
www.jatit.org
Mambership Functions
D.V.F. D.F. D.S. D.V.S. OK I.V.S. I.S. I.F. I.V.F.
1
which is not global, so the output ∆ r is added to
some amount of L∆ ref in order to give some
for δ
momentum to continue the search and to avoid 0.5
such local minima. The controller operates on a
per-unit basis so that the response is insensitive to 0
system variables and the algorithm is universal to -100 -80 -60 -40 -20 0 20 40 60 80 100
δ
any system. The advantages of fuzzy control are
obvious. It provides adaptive step size in the search
Fig. 10. Membership functions for the fuzzy set
that leads to fast convergence, and the controller
phase angle δ
can accept inaccurate and noisy signals. The AFCS
operation does not need any wind velocity TABLE I
information, and its real time based search is FUZZY RULES FOR FC1
insensitive to system parameter variation.
¯ ref \ d( ¯ PB NB PS NS
The structure of the FCs is described in detail in ref)/dt
the following paragraphs. PB IVF IVF IVF IVF
1) Structure of FC1: FC1 is the main FC of the PS IF OK IS OK
angular frequency regulator control system. Its OK IVS DVS OK OK
inputs are the deviation of we1 from its reference NB DVF DVF DVF DVF
e1 ¯ ref and its derivative. Its output is the phase NS OK DF OK DS
shift of δ, ∆δ.
The initial MFs for the input are e1 ¯ ref shown TABLE II
in Fig.8. The term initial is used because these FUZZY RULES FOR FC2
MFs are tuned online by FC3. These MFs are the S M B
initial MFs before their tuning. Of course, the
we1 − wref \σ
general form of the final MFs will remain the S OK NS NB
same. As shown in Fig.8, five fuzzy subsets are M PS PM PM
needed for the input e1 ¯ ref: negative big (NB), B PB PB PB
negative small (NS), (OK), positive small (PS),
and positive big (PB). TABLE III
FUZZY RULES FOR FC2
∆P \ L∆ P ZE N
Mambership Functions
PB PB PVB NB
0.5 PM PM PB NM
PS PS PM NS
0 ZE ZE ZE ZE
-100 -80 -60 -40 -20 0 20 40 60 80 100 NS NS NM PS
(ω - ωref)
NM NM NBV PM
NB NB NVB PB
Fig. 8. Membership functions for the fuzzy set ¯
NVB NVB NVB PVB
ref
N.B. N.S. P.S. P.B.
Mambership Functions
needed are NB, NS, PS, and PB, and they are
0.5
shown in Fig.9.
The fuzzy sets required for the phase angle ∆δ
are decrease very fast (DVF), decrease fast (DF),
0 decrease slowly (DS), decrease very slowly (DVS),
-100 -80 -60 -40 -20 0 20 40 60 80 100
d(ω - ωref)/dt OK, increase very slowly (IVS), increase slowly
(IS), increase fast (IF), and increase very fast
Fig. 9. Membership functions for the fuzzy set (IVF), and they are shown in Fig.10.The
derivative of ¯ ref corresponding outputs are selected as given in
Table I.
85
Journal of Theoretical and Applied Information Technology
www.jatit.org
86
Journal of Theoretical and Applied Information Technology
www.jatit.org
optimum speed ref established by FC2 (which output of the regulator is the modulation index “m”
operates at rated flux ΨRrated), the rotor flux ΨR is of the sinusoidal PWM reference signal and,
reduced by decreasing the magnetizing current ids. consequently, the modulation signal of the
This causes increasing torque current iqs by the magnitude of the ac voltage generated by the MC.
speed loop for the same developed torque. As the The structure of the FC5 is similar to that of FC1.
flux is decreased, the machine iron loss decreases IV. SIMULATION RESULTS AND
with the attendant increase of copper loss. DISCUSSSION
However, the total system (converters and
machine) loss decreases, resulting in an increase of Simulations were performed in
total generated power P0. The search is continued MATLAB/SIMULINK. Through C++
until the system settles down at the maximum programming, the design of AFCs was
power point M, as indicated in Fig.11. Any attempt accomplished, converting the simple fuzzy
to search beyond point M will force the controller controllers of the fuzzy logic toolbox into AFCs,
to return to the maximum power point. The which can be self-tuned online. The objective of
principle of fuzzy controller FC4 is somewhat simulation is to illustrate interacting phenomena
similar to that of FC2 and is explained in Fig.12. which have not been understood before and verify
The system output power P0 is sampled and the control strategy [Fig. 6] proposed in this study.
compared with the previous value to determine the The system is simulated with a wind velocity
increment ∆P0. In addition, the last excitation which fluctuates with a mean value near 10m/s and
current decrement (L∆ids) is reviewed. On these at 25 s, following an almost step change, rises to a
bases, the decrement step of ids is generated from mean value near 12.5m/s (Fig. 14). No local load is
fuzzy rules through fuzzy inference and connected. The generator speed and air-gap torque
defuzzification, as indicated. It is necessary to both increase [Fig. 15(a)].
process the inputs of FC4 in per-unit values.
16
Therefore, the adjustable gains KP and KIDS wind speed (m/s)
14
convert the actual variable to variables with the
12
following expressions
KP = awr + b 10
(5) 8
KIDS = c1wr ¯ c2 iqs + c3 6
(6) 4
25 50 75 100 125 150
Where a, b, c1, c2 and c3 are derived from
time (s)
simulation studies.
The current iqs is proportional to the generator
torque, and ∆ ref is zero because the fuzzy Fig. 14. Wind velocity
controller FC4 is exercised only at steady-state
conditions. The FC4 controller operation starts air gap torque (pu)
when FC2 has completed its search at the rated flux
condition. If wind velocity changes during or at the
end of FC4, its operation is abandoned, the rated
rotor speed (pu)
flux is established, and AFCS control is activated.
87
Journal of Theoretical and Applied Information Technology
www.jatit.org
frequency (Hz)
the control effect of fuzzy logic controller, which
adjusts the firing delay angle at the rectifier end
according to the change of system frequency.
While the increase of output real power is caused
by the increase of wind speed.
output real power
Figure 15 and 16 show the response of the system
for the above-mentioned disturbance. When the
25 50 75 100 125 wind velocity increases, output power starts to
150
time (s) increase and then control system gives an order to
increase ref (Fig.16). We can observe that e1 is
continuously attached to ref. ref keeps
(b)
increasing, trying to reach its optimal value for the
Fig. 15. Response of WECS with the Grid—wind
speed change. specific wind velocity. During the disturbance, e1
stays close to ref and increases until it reaches its
1.88 optimal value. When this happens, output power
ωe1 – ωref - -
1.73 and e1 reach a new steady state, which
1.58 corresponds to the maximum absorption of real
(pu)
88
Journal of Theoretical and Applied Information Technology
www.jatit.org
efficiency. The matrix converter also implements stator orientated field control,” in Proc. IEEE
APEC, Miami Beach, FL, 2003, vol. 2, pp.
unity power factor at the interface with the grid for 846–852.
optimal utilization of the installed wind turbine [14] M. Naidu and J. Walters, “A 4-kW 42-V
power and satisfies the reactive power demand of induction-machine-based automotive power
generation system with a diode bridge
the induction generator to avoid self-excitation rectifier and a PWM inverter,” IEEE Trans.
capacitors. The system was analyzed and designed, Ind. Appl., vol. 39, no. 5, pp. 1287–1293,
and performances were studied extensively by Sep./Oct. 2003.
[15] R. Leidhold, G. Garcia, and M. I. Valla,
simulation to validate the theoretical concepts. “Induction generator controller based on the
instantaneous reactive power theory,” IEEE
Trans. Energy Convers., vol. 17, no. 3, pp.
REFERENCES 368–373, Sep. 2002.
[16] T. Ahmed, K. Nishida, and M. Nakaoka, “A
[1] “Time for action: Wind energy in Europe,” novel induction generator system for small-
European Wind Energy Asso., Rome, Italy, scale AC and DC power applications,” in
Oct. 1991. Proc. IEEEPESC, Jun. 12–16, 2005, vol. 1,
[2] L.L. Freries, Wind Energy Conversion pp. 250–256.
Systems, Prentice Hall, 1990. [17] “IEEE recommended practices and
[3] S. Heier, Grid Integration of Wind Energy requirements for harmonic control in electric
Conversion Systems, New York: Wiley, power systems,” Project IEEE- 519, Oct.
1998. 1991.
[4] R.Pena, J.C.Clare, and G.M.Asher, " A doubly [18] M. Venturini and A. Alesina, "The
fed induction generator using back-to- back generalized transformer: A new
PWM converters supplying an isolated load bidirectional sinusoidal waveform frequency
from a variable speed wind turbine" IEE Proc., converter with continuously adjustable input
Electric Power Applications, Volume: 143 , power factor " in Proc. IEEE PESC’80, 1980,
Issue: 5 , Sept. 1996 pp.:380 – 387. pp. 242– 252.
[5] L Zhang, C Watthanasarn, and W. Shepherd, [19] P. W. Wheeler, IEEE, J. Rodríguez, J.
"Application of a matrix converter for the Clare,L. Empringham, and A. Weinstein, "
power control of a variable-speed wind- Matrix Converters: A Technology Review " ,
turbine driving a doubly-fed induction IEEE Trans. On Industrial elctronicas, Vol.
generator" IECON 97,Vol.2, pp.906 – 911, 49, NO. 2, APRIL 2002,pp 276-289.
Nov. 1997. [20] J. G. Slootweg, S. W. H. de Haan, H.
[6] R. Spee, S. Bhowmik, J.H.R. Enslin, “Adaptive Polinder, and W. L. Kling, "General Model
control strategies for variable-speed doubly- for Representing Variable SpeedWind
fed wind power generation systems ", IEE Turbines in Power System Dynamics
Industry Applications Society Annual Simulations", IEEE Trans. On Power Systems,
Meeting, Vol.1, Oct. 1994 pp. 545 - 552 Vol. 18, NO. 1, FEB.2003, pp. 144-151.
[7] W. Lu and B. T. Ooi, “Optimal acquisition and [21] L.S.T. Ackermann, "Wind Energy
aggregation of offshore wind power by Technology and current status. A Review",
multiterminal voltage-source HVDC,” IEEE Renewable and Sustainable Energy Review,
Trans. Power Del., vol. 18, no. 1, pp. 201– 4:315-375, 2000.
206, Jan. 2003. [22] G. C. D. Sousa, B. K. Bose, and J. G. Cleland,
[8] E. Spooner, P. Gordon, J. R. Bumby, and C. D. “Fuzzy logic based on-line efficiency
French, “Lightweight ironless-stator PM optimization control of an indirect vector
generators for direct drive wind turbines,” controlled induction motor drive,” in Proc.
Proc. Inst.Elect. Eng., Electr. Power Appl., IEEE-IECON Conf., Maui, HI, pp. 1168–
vol. 152, no. 1, pp. 17–26, 2005 1174, Nov. 1993.
[9] E. Suarez and G. Bortolotto, “Voltage- [23] G. C. D. Sousa and B. K. Bose, “A fuzzy set
frequency control of a self-excited induction theory based control of a phase controlled
generator,” IEEE Trans. Energy Convers., converter dc machine drive,” in IEEE-IAS
vol. 14, no. 3, pp. 394–401, Sep. 1999. Annu. Meeting Conf. Rec., Dearborn, MI, Oct.
[10] R. Teodorescu, and F. Blaabjerg, "Flexible 1991, pp. 854–861
Control of Small Wind Turbines With Grid [24] L. Wei, T.A. Lipo, “ A Novel Matrix
Failure Detection Operating in Stand-Alone Converter Topology with simple
and Grid- Connected Mode",IEEE Trans. On Commutation".
Power Electronics, Vol. 19, NO. 5, SEP. 2004, [25] Vinod Kumar and R.R. Joshi, “Investigating
pp 1323, 1332. AC/DC Interactions in Advanced Converter
[11] W. Lu and B. T. Ooi, "Multiterminal LVDC Based WECS Connected to a Weak AC Grid,”
system for optimal acquisition of power in
wind-farm using induction generators", IEEE Proceedings of International Conference on
Tran. On Power Electronics Vol. 17, NO. 4, Recent Advancements and Applications of
JULY 2002, pp. 558-563. Computer in Electrical Engineering (RACE),
[12] O. Ojo and I. Davidson, “PWM-VSI inverter
assisted stand-alone dual stator winding Govt. Engineering College, Bikaner (India),
induction generator,” in Conf. Rec. IEEE-IAS March 24-25, 2007, pp. 626-632.
Annu. Meeting, Oct. 1999, pp. 1573–1580.
[13] D. Seyoum, F. Rahman, and C. Grantham,
“Terminal voltage control of a wind turbine
driven isolated induction generator using
89