Вы находитесь на странице: 1из 6

A Novel Double-Loop Control Structure Based on

Fuzzy-PI and Fuzzy-PR Strategies for Single-Phase


Inverter in Photovoltaic Application
Parham Mohammadi Behrouz Azimian Amir Shahirinia
Department of Electrical Engeenering Inamori School of Engineering Center of Excellence for Renewable Energy
Amirkabir University of Technology Alfred University University of the District of Columbia
Tehran,Iran Alfred, USA Washington, D.C., USA
Parhammh.e@gmail.com mr.behrouz.azimian@IEEE.org amir.shahirinia@udc.edu

Abstract— Photovoltaic systems are on the way to becoming classic proportional integral (PI), and proportional-resonant
the main energy generating unit, in the future gird. Low voltage (PR) controllers. Unfortunately, these controllers lack certain
ride through (LVRT) capability is one of the most important features such as excellent dynamic behavior, stability, and static
features in the future grid-tied photovoltaic systems. In this paper,
a novel control strategy based on the fuzzy-PI controller (FPIC) tracking accuracy [1].
and fuzzy-PR controller (FPRC) is proposed. The FPRC is used Recently in [1], controller based on Model Predictive
for the inner current control loop, and the FPIC is used for outer
Controller (MPC) was proposed to remedy the aforementioned
power control loop. Unlike former FPRCs, the proposed inner
loop controller not only compensates the main frequency but also problems, in order to enhance the inverter performance in terms
uses self-tuned parameters for compensating selected low order of speed and overshoot. Unfortunately, the proposed control
harmonics, for enhancement of injected current to the grid. In the structure has two major drawbacks: 1) MPC controller needs
proposed structure, both inner and outer loop parameters are self- high computational power and therefore is not economical 2)
tuned on-line, hence the fast dynamic and low overshoot are The proposed structure performance was not compared with PR
achieved. The proposed structure has the following advantages: It
controller in term of Total Harmonic Dissertation(THD) index.
needs low computational power compared to other non-linear
controllers, therefore, it is more economical. Also, the fuzzy Fuzzy logic was used for enhancing the PI controller [9, 10] and
controller does not depend on complex mathematical models. Two PR controller [11, 12] performance. However, in all the former
different reactive power injection strategies are investigated in this
studies, fuzzy-PI controllers (FPICs) and fuzzy-PR controllers
paper. Then conventional control structure is compared with the
proposed control structure in terms of speed, overshoot, and total (FPRCs), were used as an inner control loop separately. In this
harmonic distortion (THD) index. Detailed simulations are paper, a double loop control structure is proposed and uses
conducted by MATLAB SIMULINK. Simulation results confirm FPRC as an inner current control loop, and FPIC as an outer
the superiority of the proposed control structure over conventional power control loop. Hence, all of the inner and outer loop
one in terms of speed, overshoot and THD index. parameters are self-tuned which lead to an enhanced dynamic
response. Furthermore, unlike former FPRCs [11, 12], the inner
Index Terms—Fuzzy-logic, LVRT, Photovoltaic, Reactive power
loop FPRC compensate most prominent harmonics in addition
I. INTRODUCTION to the main frequency, therefore the injected current THD index
is decreased. The proposed control structure has advantages
In recent years, the share of Photovoltaic(PV) systems in
generating power is increased by a steep rate [1-3]. Therefore, such as easy implantation, being independent of complex
future grid-tied photovoltaic systems should perform like mathematical models, and THD index decreasing compared to
traditional synchronous generators and have capabilities like MPC controller.
low voltage ride through (LVRT). A tremendous amount of The rest of the paper is organized as follows: A brief
study has been carried out for enhancing this capability in introduction of reactive power injection strategies is introduced
single-phase inverters [1, 4-8]. in section II. In section III, the conventional control structure is
Most of these studies [4-8] are focused on the reactive power investigated. FPRC and FPIC structures and fuzzy membership
injection strategies, inverter structure, and benchmarking of functions are investigated in section IV. Simulation results are
illustrated in section V and the conclusion is presented in
grid faults. Moreover, these studies used conventional double
section VI.
loop control structure, which is comprised of the inner current
control loop, and the outer voltage control loop (or power
control loop). These inner and outer control loops are based on

978-1-5386-7138-2/18/$31.00 ©2018 IEEE


II. REACTIVE POWER INJECTION STRATEGIES III. CONVENTIONAL CONTROL STRUCTURE
Reference [6] introduced three different reactive power Fig.1 depicts the conventional control structure [5-7, 9]. As it
injection strategy for single-phase grid-tied PV systems. can be seen in fig.1, the conventional control structure is
Among them, two of the most extensive strategies are selected comprised of two loops, the inner loop for controlling the
and will be explained in the following section. current, and the outer loop for controlling the injected active
and reactive power. DC/DC converter is used to keep the DC-
A. Constant Peak Current
link voltage constant, and its controller structure is out of this
In this strategy, reactive power is injected into the grid in a way paper scope.
so injected current peak will remain constant. Hence, the A. Inner Control Loop
reliability of the inverter is increased [6]. The injected reactive
current is proportional to the voltage sag depth. Under this Clarke transformation (abc→αβ), the instantaneous power
situation, the injected active and reactive currents can be theory, is used because of its benefits like direct and instinctive
calculated as active power and reactive power control [9]. Power calculation
scheme is based on the second-order generalized integer
= 1− (1 − ) (1) (SOGI) [10]. Also, the PR controller is used for the inner
current control loop, because of advantages like excellent static
= (1 − ) (2) accuracy for tracking sinusoidal reference, and its ability to
In which is grid voltage in per unit, is rated Grid current reject certain low order harmonics[11-13]. The PR transfer
and ≥ 2 . More detail of current injection equations is function with compensation of 3rd, 5th, and 7th can be given
as:
presented in [6].
B. Constant Active Power ( )= + + (5)
+ + (ℎ )
Another way to inject reactive power to the grid is injecting in , ,
a way that active power remains constant. Hence, the PV Whereas , , and represent proportional, resonant and
remains at the maximum power point and delivers maximum cascaded low order resonant blocks (harmonic compensator)
active power to the grid during a voltage sag. Under this coefficients.
situation, the injected active and reactive currents can be It is well known that high leads to an increased harmonic
calculated as impedance, which leads to lower THD [10]. On the other hand,
1 high causes high overshoots and stability problems. Also,
= (3)
high leads to faster dynamics in the cost of high THD [11].
= (1 − ) The harmonic compensator acts like proportional resonant
(4) converter for selected low order harmonics, with the reference
set to zero [12]. Similar to the resonant coefficient, harmonic
When ≤ 0.5, the injected active power should set to zero, compensator (HC) coefficients should be set high in order to
according to the grid codes [6]. achieve fast dynamics.

Figure 1. Conventional control structure.


Furthermore, like resonant coefficient, high HC coefficients
may lead to stability problem.
B. Outer Control Loop
Power control is used as an outer control loop; two PI
controllers are used for tracking the reference active and reactive
power. General PI controller relation can be expressed as
(a) (b)
( )= + (6)
Figure 3. Membership functions a. input variables b.
output variables.
Whereas and represent proportional and integral
coefficients respectively. Like PR controller, high results in equal. , ,.. are gains for normalizing the inputs and outputs
fast dynamic in the cost of high overshoot, and stability of the fuzzy logic controllers.
problems. High , also introduces fast dynamic and like it Fig.3 depicts membership functions of input and output
will cause high overshoot. parameters of fuzzy control. Seven triangular membership
IV. PROPOSED CONTROL STRUCTURE functions are used for each input.
Various off-line tuning methods, like the Ziegler–Nichols (NL, NM, NS, ZE, PS, PM, and PL stand for negative large,
method, can be used for tuning PI controllers. Also, reference negative medium, negative small, zero, positive small, positive
[13] introduces Ziegler–Nichols based tuning method for PR medium and positive large respectively.)
controllers. In [10], the design procedure for the PR controller As it is shown in fig.3, two membership functions are used for
in PV application was introduced. All of these works share the fuzzy outputs. S and B represent small and big respectively.
same disadvantage: based on the aforementioned reasons, the Mamdani type fuzzy logic system is used for designing FPIC
trade-off should be made among factors like overshoot, THD, and FPRC. Min-max compositions are used, and the output
fast dynamic, and stability margin. function, defuzzification, is done based on centroid method.
A. FPIC and FPRC Design Tables 1 and 2 show the rules for , , and
, , , respectively. In [13], six types of fuzzy-PID
As described in previous sections, to improve the performance structures are introduced. Two-input FLC structure with coupled
of conventional control structure, fuzzy logic is used in this rules is used for designing FPRC and FPIC in this study. The
paper. Both parameters of inner and outer control loops are self- basic rule base of two-input FLC structure can be expressed as
tuned based on fuzzy approach. Fig.2 illustrates the structure of [13] :
FPRC and FPIC. Error ( ) and change of error ( ) are
IF IS AND IS
sampled as inputs. As outputs of the fuzzy logic controller
, and , , are fed to outer control loop PI THEN IS
controllers . Also, and , are fed to inner control The most significant feature of the proposed structure is its fast
loop PR controller. Whereas , , and , represent dynamic since both of inner and outer loop parameters are tuned
proportional and integral coefficients of outer loop PI on-line.
controllers. Also, and , stand for proportional,
One can observe that the same structure and rule system is used
resonant and cascaded resonant blocks coefficient for the inner
for PI, PR and resonant blocks parameters, that is based on the
control loop.
following reasons:
The coefficient for main frequency, and cascaded resonant
1) Proportional coefficient has the same function in both
blocks for compensating low order harmonics, are considered
PI and PR controllers.

Figure 2. Structure of FPRC and FPRI.


TABLE I. simulations, and observing the effect of changing normalizing
FUZZY CONTROL RULES FOR , , ,
gains on the result. Also, in [16] evolutionary computing
(Genetic Algorithms) was used for obtaining optimal
NL NM NS ZE PS PM PL normalizing factor. In this paper normalizing gains were
NL B B B B B B B obtained based on computer simulations.
NM S B B B B B S
V. SIMULATION RESULTS
NS S S B B B S S
ZE S S S B S S S In this section, a comparative study is carried out between the
PS S S B B B S S conventional control structure and proposed control structure,
PM S B B B B B S in terms of overshoot, dynamic response, and THD. Two case
PL B B B B B B B studies are investigated: reactive power injection with constant
active power strategy, and constant current peak strategy.
TABLE II. Detailed simulations carried out with MATLAB SIMULINK.
FUZZY CONTROL RULES FOR , , Table III demonstrates the simulated system parameters.
NL NM NS ZE PS PM PL Figs 3.a and 3.b depict injected reactive and active power, with
NL B B B B B B B constant active power strategy for conventional and proposed
NM B S S S S S B control structure. In all of the case studies 0.4 p.u voltage sag
NS B B S S S B B occurs at t=0.5s, and at the t=1s the fault is cleared. As
ZE B B B S B B B mentioned before, in the classical PR and PI controllers, the
PS B B S S S B B trade-off should be made among factors like: overshoot and fast
PM B S S S S S B dynamic.
PL B B B B B B B One can observe, that the proposed control structure has better
performance in terms of overshoot and fast dynamic. At the
2) The resonant part of the PR controller introduces t=0s, PV system starts to inject active power at MPPT mode.
infinity gain at the main frequency as the integral part of the PI As it can be seen in Fig.3.a, proposed control structure tracks
controller introduces infinity gain at DC component. That is to reference active power faster, and with lower overshoot. Also,
say, conceptually, the resonant part of the PR controller can be one can observe, from fig.3.a, that proposed control structure
viewed as a general integrator of AC components. Therefore, it has a softer performance at the=1s. That is to say, the proposed
has the same function of integral part of PI controller[12]. structure tracks reference active power with lower overshoot.
3) Low order resonant blocks introduce infinity gain at As it can be seen in fig.3.b, injected reactive power, with the
selected harmonics. Hence, their combination with the proposed control structure, has lower oscillation compared with
proportion part acts like a PR controller for selected harmonics the conventional control structure. At t=0.5s, when voltage sag
with the reference set to zero. This leads to rejecting the low occurs, PV system starts to inject reactive power to the grid.
order harmonics. In other words, their parameters can be tuned
by the same logic that is used for tuning the PI and the PR TABLE III.
parameters. SIMULATED SYSTEM PARAMETERS
Rule base of FPRC and FPIC are based on the same
procedure that has been used in previous studies [13-15]. In the
beginning, both FPRC and FPIC should have a fast dynamic, 230V
therefore big control signal is needed. The differential of
proportional, integral, resonant ,and cascaded resonant blocks 50 Hz
coefficients is gradually decreasing with degrading of and . 1000W
Therefore, systems approach stable state with low overshoot. =3 , = 0.02
When the system reaches a stable state, a high proportional
coefficient is needed to obtain less static error [14], and in the ℎ 10
case of FPRC to obtain low THD [10]. On the other hand, in a =4 , =

stable state low integral, resonant, and low order cascaded 700μ , = 2.35μ
resonant blocks coefficients are needed to increase the 220 μ
robustness and stability of the system. − 400V
B. Normalizing Gains = 2, = 53 –Active
power
Normalizing gains ( , ,.. ) play an important role in the
= 1.5, = 51 –
performance of both FPRC and FPIC. According to [13] , a Reactive power
generic analysis for obtaining optimal normalizing gains is
extremely difficult, especially for a coupled double input fuzzy = 30, = , =

system, which is used in this study. The nonlinear tuning of 3000
normalizing gains can be achieved by means of computer
(a) (b)

(c) (d)
Figure 3. Comparison between proposed and conventional control structure.
a. constant active power strategy active power
b. constant active power strategy reactive power
c. constant current peak strategy active power
d. constant current peak strategy reactive power

It can be seen in fig.3.b that the proposed control structure has Therefore, it can be concluded that proposed control structure
faster dynamic at the start of voltage sag and has lower has better performance in both reactive power injection
overshoot when fault is cleared at t=1s. strategies Figs 4.a and 4.b show comparison between THD
Figs 3.c and 3.d illustrate the injected reactive and active indexes under proposed, and conventional control structure for
power with constant current peak strategy. As it can be seen in constant active power and constant current peak strategy. As it
figs 3.c and 3.d, the proposed control structure shows a better can be seen in both reactive power injection strategies, THD is
performance in terms of dynamic and overshoot. suppressed by the proposed control structure.

(a)
(b)
Figure 4. Comparison between proposed and conventional control structure.
a. constant active power strategy THD index
b. constant current peak strategy THD index
VI. CONCLUSION Transactions on Industry Applications, vol. 50, no. 3, pp. 1942-
1952, 2014.
LVRT capability is one of the most important features in future [6] Y. Yang, H. Wang, and F. Blaabjerg, "Reactive power injection
grid-tied photovoltaic systems. Fast dynamic, low overshoot strategies for single-phase photovoltaic systems considering grid
requirements," IEEE Transactions on Industry Applications, vol. 50,
and excellent power quality are the most critical aspects of
no. 6, pp. 4065-4076, 2014.
injecting reactive current during a grid fault. In this paper, a [7] Y. Yang, P. Enjeti, F. Blaabjerg, and H. Wang, "Wide-scale
novel double loop control structure was proposed. Because of adoption of photovoltaic energy: Grid code modifications are
the similar function of PI controller, PR controller, resonant explored in the distribution grid," IEEE Industry Applications
Magazine, vol. 21, no. 5, pp. 21-31, 2015.
blocks single fuzzy logic structure, and rule base were used for [8] P. Mohammadi, A. Eskandari, J. Milimonfared, and J. Moghani,
on-line self-tuning of their parameters. Furthermore, since both "LVRT capability enhancement of single-phase grid connected PV
the inner loop and outer loop parameters are self-tuned online, array with coupled supercapacitor," in Power Electronics, Drives
Systems and Technologies Conference (PEDSTC), 2018 9th
fast and soft (low over-shoot) dynamic response is achieved. Annual, 2018, pp. 193-198: IEEE.
Also, because of self-tuning parameters of cascaded resonant [9] G. Tsengenes and G. Adamidis, "Comparative evaluation of fuzzy-
blocks in addition to the resonant part parameter, THD index of PI and PI control methods for a three phase grid connected inverter,"
in Power Electronics and Applications (EPE 2011), Proceedings of
injected current is reduced. Several simulation scenarios were the 2011-14th European Conference on, 2011, pp. 1-10: IEEE.
carried out to validate the proposed control structure excellence [10] Z. Wang and T. Ahn, "Design of the optimal self-tuning fuzzy-PI
over the conventional one. Results from constant current and controller for rectifier current control in HVDC system," in Control
constant active power, reactive power injection strategies were Conference, 2006. CCC 2006. Chinese, 2006, pp. 1224-1227: IEEE.
[11] X. Duan and H. Zhang, "Quasi-proportional resonant control of the
presented for both, conventional and proposed control LCL-type grid-connected inverter based on fuzzy control and self-
structure. The presented results validated the superiority of the tuning of fuzzy parameters," in Control And Decision Conference
proposed structure over the conventional one, in terms of speed, (CCDC), 2017 29th Chinese, 2017, pp. 1793-1797: IEEE.
[12] H. Cai, W. Wei, Y. Peng, and H. Hu, "Fuzzy proportional-resonant
overshoot and THD index. In future work, obtaining optimal control strategy for three-phase inverters in islanded micro-grid with
normalizing gains for inner and outer loop based on nonlinear loads," in Fuzzy Systems (FUZZ-IEEE), 2014 IEEE
evolutionary algorithms will be investigated. Also, in this paper International Conference on, 2014, pp. 707-712: IEEE.
[13] Y. Yang, F. Blaabjerg, H. Wang, and M. G. Simoes, "Power control
distribution grid was modeled as a thevenin equivalent circuit, flexibilities for grid-connected multi-functional photovoltaic
since the aim of this paper is focused on developing novel inverters," IET Renewable Power Generation, vol. 10, no. 4, pp.
control structure for a grid-connected converter. However, in 504-513, 2016.
[14] M. Ciobotaru, R. Teodorescu, and F. Blaabjerg, "A new single-
future work, the results will be studied in the presence of the
phase PLL structure based on second order generalized integrator,"
IEEE test cases on radial feeders. in Power Electronics Specialists Conference, 2006. PESC'06. 37th
IEEE, 2006, pp. 1-6: IEEE.
REFERENCES [15] R. Teodorescu, F. Blaabjerg, M. Liserre, and P. C. Loh,
[1] E. Z. Bighash, S. M. Sadeghzadeh, E. Ebrahimzadeh, and F. "Proportional-resonant controllers and filters for grid-connected
Blaabjerg, "Improving performance of LVRT capability in single- voltage-source converters," IEE Proceedings-Electric Power
phase grid-tied PV inverters by a model-predictive controller," Applications, vol. 153, no. 5, pp. 750-762, 2006.
International Journal of Electrical Power & Energy Systems, vol. [16] H. Cha, T.-K. Vu, and J.-E. Kim, "Design and control of
98, pp. 176-188, 2018. Proportional-Resonant controller based Photovoltaic power
[2] M. K. Hossain and M. H. Ali, "Fuzzy logic controlled power conditioning system," in Energy Conversion Congress and
balancing for low voltage ride-through capability enhancement of Exposition, 2009. ECCE 2009. IEEE, 2009, pp. 2198-2205: IEEE.
large-scale grid-connected PV plants," in Power and Energy [17] A. Vidal et al., "Assessment and optimization of the transient
Conference (TPEC), IEEE Texas, 2017, pp. 1-6: IEEE. response of proportional-resonant current controllers for distributed
[3] A. Merabet, L. Labib, and A. M. Ghias, "Robust Model Predictive power generation systems," IEEE Transactions on Industrial
Control for Photovoltaic Inverter System with Grid Fault Ride- Electronics, vol. 60, no. 4, pp. 1367-1383, 2013.
Through Capability," IEEE Transactions on Smart Grid, 2017. [18] L. F. A. Pereira and A. S. Bazanella, "Tuning rules for proportional
[4] Y. Yang, F. Blaabjerg, and Z. Zou, "Benchmarking of grid fault resonant controllers," IEEE Transactions on Control Systems
modes in single-phase grid-connected photovoltaic systems," IEEE Technology, vol. 23, no. 5, pp. 2010-2017, 2015.
Transactions on Industry Applications, vol. 49, no. 5, pp. 2167- [19] G. K. Mann, B.-G. Hu, and R. G. Gosine, "Analysis of direct action
2176, 2013. fuzzy PID controller structures," IEEE Transactions on Systems,
[5] Y. Yang, F. Blaabjerg, and H. Wang, "Low-voltage ride-through of Man, and Cybernetics, Part B (Cybernetics), vol. 29, no. 3, pp. 371-
single-phase transformerless photovoltaic inverters," IEEE 388, 1999.

Вам также может понравиться