Anda di halaman 1dari 9

5708

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 60, NO. 12, DECEMBER 2013

A Parameterization Approach for


Enhancing PV Model Accuracy
Yousef A. Mahmoud, Weidong Xiao, Member, IEEE, and Hatem H. Zeineldin, Member, IEEE

AbstractReliable and accurate photovoltaic (PV) models are


essential for simulation of PV power systems. A solar cell is typically represented by a single diode equivalent circuit. The circuit
parameters need to be estimated accurately to get an accurate
model. However, one circuit parameter was assumed because of
the limited information provided by commercial manufacturing
datasheets, and thus the model accuracy is affected. This paper
proposes a parameterization approach for PV models to improve
modeling accuracy and reduce implementation complexity. It develops a method to accurately estimate circuit parameters, and
thus improving the overall accuracy, relying only on the points
provided by all commercial modules datasheet. The proposed
modeling approach results in two simplified models demonstrating the advantage of fast simulation. The effectiveness of the
modeling approach is thoroughly evaluated by comparing the
simulation results with experimental data of solar modules made
of mono-crystalline, multi-crystalline, and thin film.
Index TermsEquivalent circuit, modeling, parameterization,
photovoltaic (PV), solar cell.

I. I NTRODUCTION

HOTOVOLTAIC (PV) system studies need a reliable and


accurate mathematical model to predict energy production
from the PV resource under various irradiance and temperature
conditions [1], [2]. PV models can be categorized into two main
types, double diode models and single diode models, which are
illustrated in Fig. 1. The double diode model is characterized
by its high accuracy [3][12]; however, it is relatively complex
and suffers from low computational speed [13]. The second
type, single-diode model, is the most commonly used model
in power electronics simulation studies, because it offers a
reasonable tradeoff between simplicity and accuracy [6], [14]
[19]. Another advantage of using the single diode model is the
possibility to parameterize it based only on provided information by datasheet as presented in [1], [2], [14], [16], [20], [21].
Modeling PV partial shading has been investiaged in [22], [23].

Manuscript received August 18, 2012; revised October 17, 2012; accepted
November 23, 2012. Date of publication November 30, 2012; date of current
version June 21, 2013.
Y. A. Mahmoud is with the Department of Electrical and Computer Engineering, University of Waterloo, Waterloo, ON N2L 3G1, Canada (e-mail:
ymahmoud@uwaterloo.ca).
W. Xiao is with the Electrical Power Engineering Program, Masdar Institute
of Science and Technology, Abu Dhabi, UAE (e-mail: mwxiao@masdar.ac.ae).
H. H. Zeineldin is with the Masdar Institute of Science and Technology,
Masdar City, UAE, and also with the Electrical Power and Machines Department, Cairo University, 12613 Giza, Egypt (e-mail: hzainaldin@masdar.ac.ae).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TIE.2012.2230606

Fig. 1.

Equivalent circuit of double-diode and single-diode model.

There are various methods presented in literature, for extracting parameters of single diode PV models. One approach would
be to use the device physics to develop expressions for the IV
curve parameters, but such parameters would be in terms of
semiconductor material constants and manufacturing variables
such as doping densities [5], [13], [24]. Most semiconductor
constants vary considerably with production spread and are not
provided in a manufacturers data sheet [5], [13], [24]. A lookup table-based model to approximately represent the IV curve
of the PV cell is presented in [25], but it is highly simplified and
results in a high modeling error. Another approach is to use error minimization optimization analytical techniques [26], [27]
or numerical techniques [13], [21], [28] to estimate parameters
of the PV equation from a measured IV curve. Such approach,
in spite of its high accuracy, requires a measured curve which
may not be available for users. A common approach, which is
followed in this paper, is to parameterize the PV model based
only on information provided by product datasheets [1], [2], [8],
[14], [16], [29][34]. Although the extracted parameters do not
necessarily correspond to the PV module physical parameters
(for example, negative values for the series resistance might be
resulted) as explained in [6], [12], [35], [36], the simulated IV
and P V curves can highly match the real curves.
PV manufacturing datasheets provide only four information
about the output electrical characteristics of their PV modules at
standard test conditions (STC), which are short-circuit current
Isc , open circuit voltage Voc , operating voltage and current at
max power point (Vm , Im ), and the implicit information that the
peak of P V curve occurs at the voltage point (Vm ). Thus, only
four equations can be written accurately relying on datasheet
information. However, single diode PV models have five unknown parameters which need to be estimated. To compensate,
the parameterization in [6] starts with one predefined parameter,
the ideality factor, and then derives the rest four parameters accordingly. Similarly, the parameterization in [2] approximates
that one parameter, the shunt resistance Rsh , equals to inverse
of the slope at short-circuit point, and then derives the rest
four parameters accordingly. As a result, the accuracy of the
modeled curves is affected, although they exactly pass through
the four points provided by datasheets. Furthermore, they might
lead to a singular solution, if the approximated parameter is
significantly different from the exact value.

0278-0046/$31.00 2012 IEEE

MAHMOUD et al.: PARAMETERIZATION APPROACH FOR ENHANCING PV MODEL ACCURACY

This paper proposes a parameterization approach that does


not presume parameters and thus enhances modeling accuracy.
It develops a method to determine the best estimation of one
parameter Rsh independently in order to solve the set of four
equations accurately without any approximation or assumption.
In addition to improving the model accuracy, the proposed
modeling approach reduces the model complexity demonstrating the advantage of fast simulation. The parameterization
approach utilizes measured IV curves to deduce a systematic
approach for estimating the PV parameters. However, users are
not required to have a measured IV curve to implement the
proposed parameterization approach.
The developed method is derived as follows: an optimization
model is formulated to find the upper and lower bounds of Rsh
which produce a non-singular system of equations. By using
each Rsh value, in the determined range, the remaining PV
parameters are determined and the root mean square deviation
(RMSD) of the resulted modeled curves is plotted against the
corresponding Rsh values, and a relation between both of them
is found. This relation will provide a rule of thumb to choosing
the best value of Rsh and Rs without the need from users to
perform the optimization or acquire measured IV curves.
The paper is organized as follows: Section II reviews the
limitations in current PV parameterization approaches and
describes how the proposed parameterization approach meets
them. Then, it estimates the parameters of the PV model at STC
which includes photon current, diode saturation current, diode
ideality factor, shunt, and series resistances. Section III includes
the effect of atmospheric conditions variation, temperature, and
irradiance, on the profound STC parameters. A comprehensive experimental evaluation is conducted in Section IV to
demonstrate the improvement and verify the effectiveness of
the proposed approach over the existing approaches based on
three different solar cell materials in terms of mono-crystalline,
poly-crystalline, and thin film.
II. PV M ODELING
A single diode model consists of a current source, diode,
series, and shunt resistances, as shown in Fig. 1. The
currentvoltage (IV ) characteristic can be expressed as follows [37], [38]:

 

q(V +IRs )
(V + IRs )
(1)
I = Iph Is e Ns KT A 1
Rsh
where q, K, T , and Ns are the electron charge, Boltzmann
constant, module temperature, and number of series connected
cells, respectively. The values of the electron charge q and
Boltzman constant are 1.6 1019 C and 1.38 1023 J/K,
respectively. The parameters Iph , Is , A, Rs , and Rsh are the
photon current, diode saturation current, ideality factor, series
resistance, and shunt resistance, respectively. Therefore, five
unknown parameters need to be estimated from the available
information given in product datasheets or from experimental
measurements. PV manufacturers typically provide values of
open-circuit voltage (Voc ), short-circuit current (Isc ), and the
maximum power point (Vm , Im ) at STC. Temperature coefficients are also given with respect to voltage and current.

5709

A. Model Parameters
As shown in (1), there are five unknown parameters which
are A, Rs and Rsh , Iph , and Is . Estimating these parameters
will be based completely on datasheet information to avoid
the need of measurements which are not always available.
Substituting short-circuit point (0, Isc ) at STC into (1) gives
 qIsc Rs
 I R
sc s
(2)
Isc = Ipho Irs e( Ns KTo A ) 1
Rsh
where Ipho and Irs are the specific values of the photon current
and saturation current at STC. To is the STC temperature. The
second and third terms of (2) can be neglected resulting in the
following estimation of the photon current at STC:
Ipho
= Isc .

(3)

Substituting open circuit voltage point (Voc , 0) in (1) gives


Irs =

Ipho Voc /Rsh


.
qVoc
e( Ns KTo A ) 1

(4)

Substituting the operating voltage and current at maximum


power point (Vm , Im ) in (1) at STC produces

 

q(Vm +Im Rs )
(Vm + Im Rs )
Ns KT oA
1
Im = Ipho Irs e
.
Rsh
(5)
Implicit information is also available at the max power point
because the power equation derivative is equal to zero at max
power point. Therefore, substituting the max power point to the
derivative of the power equation (dP/dV = 0) gives


 


q(Vm +Im Rs )
(Vm + Im Rs )
dP
Ns KTo A
1
= Isc Irs e
dV
Rsh

 



q(Vm +Im Rs )
q
1
Irs e Ns KTo A
+Vm
= 0. (6)

Ns KTo A
Rsh
Substituting the variables Ipho from (3) and Irs from (4) into
(5) and (6) produces (7) and (8)


qVoc
Isc Rsh Vm Im Rs Im Rsh ( N sKT
)
oA
0=
e
Isc Rsh Voc

 q(Vm +Im Rs )  
Voc Vm Im Rs Im Rsh

e Ns KTo A
(7)
Isc Rsh Voc
Ipho RVoc
sh
0 = Isc
qVoc
e( Ns KTo A ) 1


 

q(Vm +Im Rs )
Ns KTo A qVm
Ns KTo A

1
e
Ns KTo A


2Vm + Im Rs

.
Rsh

(8)

The information provided in the PV datasheet has been


utilized to derive the four (3), (4), (7) and (8) as in [6], [13].
However, there are five unknowns which include Ipho , Rsh , Rs ,
A, and Irs . These parameters cannot be calculated from the four
derived (3), (4), (7), and (8). An approach would be to assume

5710

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 60, NO. 12, DECEMBER 2013

the value of one parameter (for example, the ideality factor as


in [6], or the shunt resistance as in [2]) and find the remaining
parameters accordingly. Although the IV curve passes exactly
through the datasheet points, intermediate points on the IV
and P V curves (the curvatures) can exhibit inaccuracies as
will be shown in later sections. Therefore, this paper provides
a method to determine the best estimation of one parameter
Rsh independently in order to solve the set of four equations
accurately without any approximation or assumption, and hence
improving the modeling accuracy in the intermediate points of
the IV curve.

TABLE I
S HUNT R ESISTANCE B OUNDARY OF D IFFERENT PV M ODULES

B. Modeling Performance Indices


The RMSD of the PV current will be used to evaluate the
modeling accuracy. The current RMSD is a measure of the
differences between values predicted by a model and the values
actually measured from real systems. For this specific study, the
RMSD of the model-generated PV current array I with respect
to the measured values I is defined as the square root of the
mean square error and can be expressed as follows:

n
2

j=1 (ij ij )
RM SD(I) =
(9)
n


i1
i1
i2
i2

where, I =
... and I = ... .
in
in
The normalized RMSD (NRMSD) of the PV current is the
RMSD divided by the range of PV current values which can be
expressed as
N RM SD =

RM SD
Isc

(10)

where Isc is the short-circuit current at STC and represents the


reference base of the PV current. In this paper, the RMSD and
NRMSD results will be used to verify and deduce the proposed
parameterization approach.
C. Shunt Resistance Evaluation
The process of shunt resistance estimation involves determining its upper and lower bounds within which a solution to the
set of equations provided in Section II can be found as well
as utilizing IV curves to determine the best Rsh value. First,
an optimization problem is formulated to determine Rsh upper
and lower bounds. Then, the corresponding PV parameters for
each Rsh value in the determined range are calculated, and
the IV curves are utilized to determine the model accuracy
at each value of Rsh . As will be seen latter, the results of this
procedure will be used only to deduce a systematic approach for
estimating the PV parameters which will not rely on acquiring
IV curves.
As mentioned earlier, the set of four equations provided in
(3), (4), (7), and (8) are not sufficient to estimate the five PV
parameters. Thus, one parameter needs to be estimated independently in order to solve the set of equations. An approach

would be to assume the value of one parameter (ideality factor)


and calculate the rest as in [6]. Another approach approximates
that the shunt resistance Rsh equals to the inverse of the slope
at short-circuit point [2]. Unfortunately, these approximations
affect the accuracy of the modeled curve, although it exactly
passes through the points provided by datasheets. To improve
the accuracy, this paper proposes a method to determine the
best estimation of shunt resistance Rsh . The proposed method
is derived as follows: first, (7) and (8) are utilized to find the
entire range of Rsh which produce a non-singular solution
when solving (7) with (8). Then, each value of Rsh in the
determined range is substituted in the equation set (3), (4), (7),
and (8) to find rest of parameters at each point of Rsh . Finally,
the RMSD of the resulted modeled curves, at each value of Rsh
inside the range, are plotted against corresponding Rsh values,
and a relation between the Rsh and resulted modeling RMSD is
found. The relation will provide a rule of thumb to choosing the
best value of Rsh which results in minimum RMSD. Following
paragraphs in this section presents the method explicitly.
There exists an upper and lower bound on the shunt resistance for which the solution of (7) and (8) is non-singular. The
upper bound is determined by maximizing the value of Rsh
subject to (7) and (8). The problem can be formulated in an
optimization framework where the main objective J is
J = Maximize Rsh .

(11)

The constraints involved in the optimization include (7) and


(8). The same constraints will apply to determine the lower
bound on Rsh but with an objective of minimizing Rsh . The
optimization problem is solved using the reduced gradient
approach. Table I summarizes the lower and upper bounds
of Rsh for the various solar modules under study. Within
the determined range of Rsh , the other parameters can be
derived accordingly from (3), (4), (7) and (8). As a result, a
model can be constructed with the known values of Rsh , Rs ,
A, Iph , and Is . The modeling accuracy can be evaluated in
terms of the RMSD, defined in (9). The RMSD samples are
demonstrated in Table II for the various values of Rsh within
the predetermined range. The analysis has been conducted on
four mono-crystalline, six poly-crystalline, and two thin film
solar panels. Fig. 2 shows the characteristics of the NRMSD
versus normalized values of Rsh based on four solar panels.

MAHMOUD et al.: PARAMETERIZATION APPROACH FOR ENHANCING PV MODEL ACCURACY

5711

TABLE II
ROOT M EAN S QUARE D EVIATION V ERSUS VARIATION
OF S HUNT R ESISTANCE

Fig. 3. Characteristics of the series resistance and the shunt resistance for
different PV modules.

bound on Rsh for the case of E19/320 and PS_P36, and thus the
maximum value of Rsh is equal to infinity which corresponds
to a non-zero Rs value. On the other hand, for the case of JAM5
and JAP6, Rsh has a finite upper bound which corresponds to
Rs equal to zero.
Thus, the solution of (7) and (8) either results in a solution
where Rs is equal to zero or Rsh is equal to infinity. The finding
is verified for the different types of commercial PV panels under
study: single crystalline, multi-crystalline, and thin film. Using
the above, a simple procedure can be implemented to determine
the best parameters for PV models.
Finally, it is worth to mention that it is not required, from
users, to utilize measured IV curves to estimate the value
of Rsh . Measured curves are used in this paper only for the
purpose of showing that Rsh results in the lowest RMSD at its
maximum value.
D. Modeling Parameterization

Fig. 2. Characteristics of the normalized root-mean-square deviation


(NRMSD) and the normalized Rsh .

The Rsh values are normalized to the value of a reference


current. The results show that setting Rsh equal to the upper
bound gives relatively low NRMSD values. Thus, the first step
in the proposed modeling approach is calculating Rsh upper
bound and setting Rsh to its maximum value.
One important observation, resulting from the optimization
model, is that for cases where the upper bound on Rsh is not
infinity (as for JAM and PS-M36S models), the optimal value
for Rs is zero. On the contrary, for cases where the upper bound
of Rsh is infinity, the optimal value of Rs is non-zero. Using the
above, a simple procedure can be implemented to determine
the best parameters for PV models. This will be explained
in the next subsection. That means the solution of (7) and (8)
always results either for zero series resistant or infinite shunt
resistant.
To demonstrate the result graphically, the range of Rsh is
plotted against its corresponding Rs values for four different
PV modules as shown in Fig. 3. Two characteristics can be
distinguished from Fig. 3(a) and (b). There is no finite upper

A simple and general approach is proposed to estimate


the best parameters for various PV models. Fig. 4 presents
a flowchart highlighting the procedure for extracting the five
parameters in a single diode model. The modeling process starts
with the approximation of the photon current Ipho according
to (3), where the short-circuit current is given in the product
datasheets. Then, to find the max value of Rsh , the method
starts by assigning Rsh = infinity( 107 ) and attempting to
solve (7) and (8) simultaneously. If a solution for Rs and A,
within the feasible region, is reached, then the next step would
be to calculate Irs using (4). Therefore, all five unknowns are
estimated, and these parameters are determined at STC.
For cases where a solution is not reached, this signifies that
Rsh has finite upper bound which needs to be calculated. As
highlighted in the previous subsection, the upper bound on Rsh
can be determined simply by setting Rs = 0. Thus, Rs is set
to zero, and (7) and (8) are solved to estimate A and Rsh .
Similarly, the last step would be to calculate Irs using (4).
Fig. 5 shows the simplified single diode model. Fig. 5(a) would
be used for models, such as JAP6-60/240, where the value of
Rsh is finite and Rs is zero. On the other hand, Fig. 5(b) is
considered the best representation for PV models such as PSP60, where Rsh is infinity and Rs has a non-zero value. Thus,
the PV equivalent circuit representation will depend on the PV
model type.

5712

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 60, NO. 12, DECEMBER 2013

Fig. 6. Block diagram of simulation model implementation: (a) Rs = 0;


(b) Rsh = .

is described to evaluate Is according to the meteorological


features. The open-circuit voltage varies with temperature and
can be represented by (14) [40]
VOC (G, To ) VOC (G, T ) = ||T

Fig. 4. Proposed parameterization flowchart to estimate the five parameters in


a single-diode equivalent circuit.

Fig. 5. Simplified single-diode models resulting from the proposed parameterization: (a) with Rsh only; (b) with Rs only.

where Voc (G, To ) and Voc (G, T ) are the open circuit voltages
at a certain temperature (T ) and the STC temperature (To ),
respectively, for a specific irradiance level. || is the absolute
value of the voltage temperature coefficient given by the product datasheets. T is the temperature deviation from the STC,
which is calculated as T = T To . Assigning I = 0 into (1),
the open circuit formula is approximated as in (15)


Iph
Ns KT A
ln
Voc
+1 .
(15)
q
Is
By substituting (14) into (15), Is can be expressed as
||T q

III. PARAMETER VARIATIONS AND S IMULATION


M ODEL C ONSTRUCTION

Is =

A PV simulation model is ready for construction when all


parameters at STC are derived. All the profound parameters
vary with meteorological conditions, majorly, temperature and
irradiance. The photon current Iph is proportional to the irradiance and affected slightly by temperature. It can be expressed in
(12), where T is the temperature deviation from the temperature at STC, and is temperature coefficient, which is given by
manufacturer datasheet
Iph = G(Ipho + T ).

(14)

(12)

The saturation current Is depends on the temperature and is


usually shown as [39]
 3
qEg
1
1
T
Is = Irs
(13)
e KA ( To T ) .
To
As shown in (13), Is depends on the value of Eg , which is
not given in the product datasheet. The symbol To refers to
the cell temperature equal to 298 K. An alternative approach

e Ns KT A G(Isc + T )
To

||T q

(GIsc /Irs + 1) T e Ns KT A

(16)

In (16), it is shown that the saturation current, Is , is a function


of the solar irradiance and the variation of cell temperature from
the STC. The effect of meteorological conditions on the rest of
parameters is slight and can be neglected [35]. The simulation
model can be constructed using the estimated parameters of
Rsh , A, Rs , Ipho , and Irs as well as the impact of irradiance
and temperature, which are expressed in (12) and (16). The
PV module can be modeled by a dependent current source and
mathematic blocks as shown in Fig. 6. In the case of Rs = 0,
the PV circuit model is implemented using a controlled current
source and a simple computational block shown in Fig. 6(a),
where I = f (V ). This model avoids the use of a nonlinear
solver, and thus simple mathematical blocks are required. In
case of Rsh = , the implementation requires a numerical
solver, which is illustrated in Fig. 6(b), since I = f (V, I). Both
models are simpler than the practical model shown in Fig. 1.
The modeling accuracy and simulation speed will be presented
in the next section.

MAHMOUD et al.: PARAMETERIZATION APPROACH FOR ENHANCING PV MODEL ACCURACY

5713

Fig. 7. Comparison plots between model output and measured data of


KC200GT with (a) variation of temperature and (b) variation of irradiance.

IV. E VALUATION
Fig. 7 demonstrates the modeled IV curves with respect
to the experimental data for the KC200GT poly-crystalline PV
model, for various temperatures and irradiance levels. It can
be seen that simulation results of the proposed model coincides closely with the experimental measurements. In addition,
the accuracy of the proposed approach was evaluated experimentally and compared against previous approaches. Fig. 8
presents the IV curves, P V curves, and absolute error for
two modeling approaches with respect to the experimental data
of the JAM5(L)-72/165 mono-crystalline PV module. By comparing with the modeling approach in [6], the proposed method
demonstrates small modeling errors in the span of the PV voltage. Similar evaluation is conducted for poly-crystalline and
thin film modules. Fig. 9 illustrates the smaller modeling error
of the proposed approach in comparison with the modeling approach in [6]. Fig. 10 demonstrates the modeling performance
of a thin film solar module with respect to the IV and P V
curves. The results show that the proposed modeling approach,
for all cases, outperforms the method presented in [6].
The modeling accuracy is compared with respect to the
NRMSD, which is defined in (8). The comparison includes nine
solar panels from different manufacturers, built with different
materials in term of mono-crystalline, poly-crystalline, and thin

Fig. 8. Modeling performance comparison between the proposed parameterization and the modeling process in [6] regarding (a) IV curves, (b) P-V
curves, and (c) absolute modeling errors; based on the JAM5(L)-72/165 monocrystalline PV module at STC.

film. Table III summarizes the modeling accuracy comparison


for four mono-crystalline, three poly-crystalline, and two thin
film modules. With regard to the NRMSD, the proposed modeling approach shows on average a 1.8% error reduction in
comparison with the method proposed in [6]. The extracted
model parameters are also listed in Table III for comparison.
The results show that the ideality factor varies with module
type and model. Assigning a constant ideality factor limits the
modeling accuracy and results in potential mismatched values
of Rsh and Rs .

5714

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 60, NO. 12, DECEMBER 2013

Fig. 10. Modeling performance comparison between the proposed parameterization and the modeling process in [6] regarding (a) IV curves, (b) P-V
curves, and (c) absolute modeling errors; based on the MPT3.6-75 thin film PV
module at STC.
Fig. 9. Modeling performance comparison between the proposed parameterization and the modeling process in [6] regarding (a) IV curves,
(b) P-V curves, and (c) absolute modeling errors; based on the JAP6-60/240
poly-crystalline PV module at STC.

In addition to the modeling accuracy, the proposed parameterization approach provides benefits, which include simplified
structure and fast simulation. As described in Section II, the
parameterization leads to the equivalent circuits without either
Rsh or Rs , as shown in Fig. 5(a) and (b) respectively. The
simulation study shows that the simplified models contribute
less computation time compared to the practical model per-

taining to both Rsh and Rs , which is presented in [6]. For


the simplified model shown in Fig. 4(a), the simulation time
is reduced by 34.72% in favor of the direct computational
form, I = f (V ). For another model (Fig. 5(b)), the simulation
time is also 8.89% faster than the practical model because
of the ignored value of Rsh . The simulation shows that the
model configuration (refer to Fig. 6(a)) reduces the computer
simulation time thanks to the absence of numerical solver. Although the proposed method was tested on nine PV samples, we
envisage that the method would be applicable to the majority of
PV samples.

MAHMOUD et al.: PARAMETERIZATION APPROACH FOR ENHANCING PV MODEL ACCURACY

5715

TABLE III
PARAMETERIZATION R ESULT AND M ODELING D EVIATION C OMPARISON

V. C ONCLUSION
This paper has proposed an effective approach to improve the
accuracy of the single diode PV model. The model parameters
are extracted according to the output electrical characteristics
given by manufacturing datasheets, which include the shortcircuit current, the open circuit voltage, and the maximum
power point. Although the manufacturer datasheets provide
four information about the electrical characteristics, while the
model has five unknowns, the proposed approach provides a
method to accurately estimate the values of five parameters
without any approximations. This achieves the best accuracy of
PV model as well as avoiding reliability problems involved in
PV parameter identification. It ensures high accuracy through
the three characteristic points in the PV datasheet (the opencircuit voltage, short-circuit current, and the maximum power
point), guarantees that the maximum point generated by the
mathematical model coincides with the datasheet, and achieves
the best curvature. The proposed modeling approach results
in two simplified models demonstrating the advantage of fast
simulation. The method is also easier to be implemented in
various simulation platforms for solar power systems studies.
Experimental measurements validated and proved the effectiveness of the generalized modeling approach for three types
of PV materials, made of mono-crystalline, multi-crystalline,
and thin film.
R EFERENCES
[1] D. Dondi, A. Bertacchini, D. Brunelli, L. Larcher, and L. Benini, Modeling and optimization of a solar energy harvester system for self-powered
wireless sensor networks, IEEE Trans. Ind. Electron., vol. 55, no. 7,
pp. 27592766, Jul. 2008.
[2] A. Chatterjee, A. Keyhani, and D. Kapoor, Identification of photovoltaic
source models, IEEE Trans. Energy Convers., vol. 26, no. 3, pp. 883
889, Sep. 2011.
[3] R. C. Campbell, A circuit-based photovoltaic array model for
power system studies, in Proc. 39th North Amer. Power Symp., 2007,
pp. 97101.
[4] Z. Salam, K. Ishaque, and H. Taheri, An improved two-diode Photovoltaic (PV) model for PV system, in Proc. Joint Int. Conf. PEDES,
2010, pp. 15.
[5] J. Hyvarinen and J. Karila, New analysis method for crystalline silicon
cells, in Proc. 3rd World Conf. Photovoltaic Energy Convers., 2003,
pp. 15211524.
[6] M. G. Villalva, J. R. Gazoli, and E. R. Filho, Comprehensive approach
to modeling and simulation of photovoltaic arrays, IEEE Trans. Power
Electron., vol. 24, no. 5, pp. 11981208, May 2009.
[7] F. Adamo, F. Attivissimo, and M. Spadavecchia, A tool for photovoltaic
panels modeling and testing, in Proc. IEEE Instrum. Meas. Technol.
Conf., 2010, pp. 14631466.

[8] F. Adamo, F. Attivissimo, A. Di Nisio, and M. Spadavecchia, Characterization and testing of a tool for photovoltaic panel modeling, IEEE Trans.
Instrum. Meas., vol. 60, no. 5, pp. 16131622, May 2011.
[9] F. Attivissimo, A. Di Nisio, M. Savino, and M. Spadavecchia, Uncertainty analysis in photovoltaic cell parameter estimation, IEEE Trans.
Instrum. Meas., vol. 61, no. 5, pp. 13341342, May 2012.
[10] J. A. Gow and C. D. Manning, Development of a photovoltaic array
model for use in power-electronics simulation studies, Proc. Inst. Elect.
Eng.Elect. Power Appl., vol. 146, no. 2, pp. 193200, Mar. 1999.
[11] L. Sandrolini, M. Artioli, and U. Reggiani, Numerical method for the extraction of photovoltaic module double-diode model parameters through
cluster analysis, Appl. Energy, vol. 87, no. 2, pp. 442451, Feb. 2010.
[12] J. P. Charles, I. Mekkaoui-Alaoui, G. Bordure, and P. Mialhe, A critical
study of the effectiveness of the single and double exponential models for
the I-V characterisation of solar cells, Solid State Electron., vol. 28, no. 8,
pp. 807820, Aug. 1985.
[13] D. S. H. Chan and J. C. H. Phang, Analytical methods for the extraction
of solar-cell single- and double-diode model parameters from I-V characteristics, IEEE Trans. Electron Devices, vol. ED-34, no. 2, pp. 286293,
Feb. 1987.
[14] R. Kadri, J. P. Gaubert, and G. Champenois, An improved maximum
power point tracking for photovoltaic grid-connected inverter based on
voltage-oriented control, IEEE Trans. Ind. Electron., vol. 58, no. 1,
pp. 6675, Jan. 2011.
[15] V. Scarpa, S. Buso, and G. Spiazzi, Low-complexity MPPT technique
exploiting the PV module MPP locus characterization, IEEE Trans. Ind.
Electron., vol. 56, no. 5, pp. 15311538, May 2009.
[16] A. Yazdani, A. R. Di Fazio, H. Ghoddami, M. Russo, M. Kazerani,
J. Jatskevich, K. Strunz, S. Leva, and J. A. Martinez, Modeling guidelines
and a benchmark for power system simulation studies of three-phase
single-stage photovoltaic systems, IEEE Trans. Power Del., vol. 26,
no. 2, pp. 12471264, Apr. 2011.
[17] D. Sera, R. Teodorescu, and P. Rodriguez, PV panel model based
on datasheet values, in Proc. IEEE Int. Symp. Ind. Electron., 2007,
pp. 23922396.
[18] C. Carrero, J. Amador, and S. Arnaltes, A single procedure for helping
PV designers to select silicon PV modules and evaluate the loss resistances, Renew. Energy, vol. 32, no. 15, pp. 25792589, Dec. 2007.
[19] J. J. Soon and K.-S. Low, Optimizing photovoltaic model parameters for simulation, in Proc. IEEE Int. Symp. Ind. Electron., 2012,
pp. 18131818.
[20] F. A. F. Adamo, A. Di Nisio, A. M. L. Lanzolla, and M. Spadavecchia,
Parameters estimation for a model of photovoltaic panels, in Proc. 19th
IMEKO World Congr. Fundam. Appl. Metrol., 2009, pp. 964967.
[21] W. D. Soto, S. A. Klein, and W. A. Beckman, Improvement and validation of a model for photovoltaic array performance, Solar Energy,
vol. 80, no. 1, pp. 7888, Jan. 2006.
[22] L. F. L. Villa, T.-P. Ho, J.-C. Crebier, and B. Raison, A power electronics
equalizer application for partially shaded photovoltaic modules, IEEE
Trans. Ind. Electron., vol. 60, no. 3, pp. 11791190, Mar. 2013.
[23] G. Petrone and C. A. Ramos-Paja, Modeling of photovoltaic fields in
mismatched conditions for energy yield evaluations, Int. J. Elect. Power
Syst. Res., vol. 81, no. 4, pp. 10031013, Apr. 2011.
[24] D. S. H. Chan, J. C. H. Phang, J. R. Phillip, and M. S. Loong, A
comparison of extracted solar cell parameters from single and double
lumped circuit models, in Proc. 1st Int. Photovoltaic Sci. Eng. Conf.,
1984, pp. 151153.
[25] M. E. Ropp and S. Gonzalez, Development of a MATLAB/simulink
model of a single-phase grid-connected photovoltaic system, IEEE
Trans. Energy Convers., vol. 24, no. 1, pp. 195202, Mar. 2009.

5716

IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS, VOL. 60, NO. 12, DECEMBER 2013

[26] J. A. Jervase, H. Bourdoucen, and A. Al-Lawati, Solar cell parameter


extraction using genetic algorithms, Meas. Sci. Technol., vol. 12, no. 11,
pp. 19221925, Nov. 2001.
[27] L. Xue, L. Sun, W. Huang, and C. Jiang, Solar cells parameter extraction using a hybrid genetic algorithm, in Proc. 3rd ICMTMA, 2011,
pp. 306309.
[28] M. A. D. Blas, J. L. Torres, E. Prieto, and A. Garcia, Selecting a suitable
model for characterizing photovoltaic devices, Renew. Energy, vol. 25,
no. 3, pp. 371380, Mar. 2002.
[29] G. R. Walker, Evaluating MPPT converter topologies using a matlab PV
model, J. Elect. Electron. Eng., vol. 21, no. 1, pp. 4955, 2001.
[30] A. N. Celik and N. Acikgoz, Modeling and experimental verification
of the operating current of mono-crystalline photovoltaic modules using
four- and five-parameter models, Appl. Energy, vol. 84, no. 1, pp. 115,
Jan. 2007.
[31] Y.-C. Kuo, T.-J. Liang, and J.-F. Chen, Novel maximum-power-pointtracking controller for photovoltaic energy conversion system, IEEE
Trans. Ind. Electron., vol. 48, no. 3, pp. 594601, Jun. 2001.
[32] N. D. Benavides and P. L. Chapman, Modeling the effect of voltage
ripple on the power output of photovoltaic modules, IEEE Trans. Ind.
Electron., vol. 55, no. 7, pp. 26382643, Jul. 2008.
[33] W. Xiao, M. G. J. Lind, W. G. Dunford, and A. Capel, Real-time identification of optimal operating points in photovoltaic power systems, IEEE
Trans. Ind. Electron., vol. 53, no. 4, pp. 10171026, Jun. 2006.
[34] C. Liu, K. T. Chau, and X. Zhang, An efficient wind-photovoltaic hybrid generation system using doubly excited permanent-magnet brushless machine, IEEE Trans. Ind. Electron., vol. 57, no. 3, pp. 831839,
Mar. 2010.
[35] W. Xiao, W. G. Dunford, and A. Capel, A novel modeling method for
photovoltaic cells, in Proc. 35th Annu. IEEE Power Electron. Spec.
Conf., 2004, vol. 3, pp. 19501956.
[36] W. Xiao, W. G. Dunford, P. R. Palmer, and A. Capel, Regulation of photovoltaic voltage, IEEE Trans. Ind. Electron., vol. 54, no. 3, pp. 1365
1374, Jun. 2007.
[37] S. Liu and R. A. Dougal, Dynamic multiphysics model for solar array,
IEEE Trans. Energy Convers., vol. 17, no. 2, pp. 285294, Jun. 2002.
[38] J. J. Soon and K.-S. Low, Photovoltaic model identification using particle
swarm optimization with inverse barrier constraint, IEEE Trans. Power
Electron., vol. 27, no. 9, pp. 39753983, Sep. 2012.
[39] P. Maffezzoni, L. Codecasa, and D. DAmore, Modeling and simulation of a hybrid photovoltaic module equipped with a heat-recovery
system, IEEE Trans. Ind. Electron., vol. 56, no. 11, pp. 43114318,
Nov. 2009.
[40] Y. Mahmoud, W. Xiao, and H. H. Zeineldin, A simple approach to modeling and simulation of photovoltaic modules, IEEE Trans. Sustainable
Energy, vol. 3, no. 1, pp. 185186, Jan. 2012.

Yousef A. Mahmoud received the B.Sc. degree in


electrical power engineering from Albalqaa Applied University, Amman, Jordan, in 2009, and the
M.Sc. degree from Masdar Institute of Science and
Technology, Abu Dhabi, UAE, in 2012. Currently,
he is working toward the Ph.D. degree in electrical
engineering at the University of Waterloo, Waterloo,
ON, Canada.
His research interests are mainly in control and
operation of grid-connected photovoltaic power systems including photovoltaic modeling, maximum
power point tracking, inverter control, and partial shading impact.
Weidong Xiao (M07) received the M.Sc. and Ph.D.
degrees from the University of British Columbia,
Vancouver, BC, Canada, in 2003 and 2007,
respectively.
He is a Faculty Member with the electric power
engineering program at the Masdar Institute of
Science and Technology, Abu Dhabi, UAE. In
2010, he spent eight months working as a visiting
scholar at the Massachusetts Institute of Technology,
Cambridge, MA, USA. Prior to his academic career,
he was with MSR Innovations Inc. in Canada as
an R&D Engineering Manager focusing on projects related to integration,
research, optimization, and design of photovoltaic power systems. His research
interests include photovoltaic power systems, dynamic systems and control,
power electronics, and industry applications.
Hatem H. Zeineldin (M06) received the B.Sc. and
M.Sc.degrees in electrical engineering from Cairo
University, Cairo, Egypt, in 1999 and 2002, respectively. In 2006, he received the Ph.D. degree in electrical and computer engineering from the University
of Waterloo, Waterloo, ON, Canada.
He worked for Smith and Andersen Electrical Engineering Inc. where he was involved with projects
involving distribution system design, protection, and
distributed generation. He then worked as a Visiting
Professor at the Massachusetts Institute of Technology, Cambridge, MA, USA. He is currently an Associate Professor with Masdar
Institute of Science and Technology, Masdar City, UAE, and a Faculty Member
at Cairo University, Cairo, Egypt. His research interests include power system
protection, distributed generation, and deregulation.

Anda mungkin juga menyukai