2013 IEEE International Conference on Control Applications (CCA)

Part of 2013 IEEE Multi-Conference on Systems and Control

Hyderabad, India, August 28-30, 2013

Sliding mode control of a buck converter for

maximum power point tracking of a solar panel
Amir Hussain, Avanish Kumar, Laxmidhar Behera
Department of Electrical Engineering,
Indian Institute of Technology Kanpur
Email: [email protected], [email protected], [email protected]

Abstract—This paper presents a novel sliding mode control cycle of converter in order to achieve MPPT. In incremental
(SMC) approach for maximum power point tracking (MPPT) conductance method, change in terminal voltage and array
of photovoltaic (PV) panel using buck converter. Since classical current are measured to predict the effect of change of voltage
hill climbing and incremental conductance MPPT methods can
not differentiate change in power due to perturbed voltage on power output. The above MPPT algorithms can be realized
and changing weather condition, both algorithms are prone with guaranteed convergence stability. However, both of the
to failure in case of rapidly varying solar radiation. Most of methods need large step of perturbation for fast convergence
the maximum power point tracking approaches presented in which results in more chattering around MPPT. Also, the
literature are based on finding desired PV voltage for MPPT change in output power does not purely reflect the effect of
and tracking the actual PV voltage to desired one. On contrary,
in this paper, SMC approach is applied to control inductor change in perturbed voltage because power may be changed
current and converter output voltage for MPPT and then by changing weather condition too.
convergence and tracking of PV voltage is achieved. Incremental In this paper, we propose SMC to regulate duty cycle of
conductance method is used to find desired value of PV voltage dc-dc buck converter in order to force the PV module to
and steady state analysis is performed on state equation to find operate at its maximum power point. In a typical sliding
out desired inductor current and output voltage of converter.
The proposed controller is able to bring the system on sliding mode control [12] approach, the first objective is to choose
surface in finite time and force it to remain on sliding surface. a proper sliding surface (SS) such that if system state is on
It has been found that exact MPPT is achieved even when we sliding surface, it is driven to zero. The designed controller
consider varying atmospheric condition and disturbances. The must bound the system to remain on SS. Second objective
feature of the proposed algorithm are supported by theoretical is choose a controller which force the system to reach SS.
analysis and simulation results.
The sliding mode controller has robust control property under
Index Term—photovoltaic system, DC-DC converters, sliding the presence of parameter variations and can achieve the tight
mode control, steady state error, MPPT. regulation of the states for all operating points [15].

I. I NTRODUCTION II. S YSTEM MODELING

Photovoltaic system is the technology that uses solar cells A. PV characteristic
or an array of them to convert solar light directly into
electricity. The power produced by the array depends on A PV cell can be represented by current source with
cells temperature and solar irradiation. Usually the energy diode and two resistances connected as shown in Fig.1.
generated by these solar cells is used to provide electricity Where Rs is relatively small and Rp is relatively large,
to a load and the remaining energy is saved into batteries. A which are neglected in the equation in order to simplify
major challenge in the use of PV is posed by its nonlinear the simulation.The mathematical expression of the equivalent
currentvoltage (IV ) characteristics, which result in a unique model can be written as (1)-(3).
maximum power point (MPP) on its powervoltage (PV ) Io = Ig − Is (e(βvo ) − 1) (1)
curve. The challenge becomes more difficult because panel
voltage and current corresponding to MPP keeps changing Ig = (Isc + kI (T − T r))λ/λr (2)
with variation in cell temperature and solar radiation.
Several methods have been designed and implemented  3
T 1 1
to search for this operation point among which common Is = Ir eqEgp ( Tr − T )/pK (3)
tracking schemes are perturb and observe (P&O) [1]-[3], Tr
incremental conductance [4]-[7], short circuit current [8], where, Io is output current from single PV cell, Ig is
open circuit voltage [9]. Some modified techniques have also light generated current, Id the current through diode which
been proposed, with the objective of minimizing the hardware is replaced by diode equation and Is the reverse saturation
or improving the performance [10]-[13]. In P&O method, current of diode, Ir is the reverse saturation current at the
the panel voltage is constantly perturbed by varying duty reference temperature Tr (298 K), Egp = 1.1 eV is the
978-1-4799-1559-0/13/$31.00 ©2013 IEEE 661
Fig. 1. Single diode model of PV cell.

Fig. 3. IV characteristic of solar panel at different cell temperature and

solar radiation.

Fig. 2. PV array in conjunction with a buck converter.

bandgap energy of the semiconductor used in the cell, Isc

is the short-circuit cell current at reference temperature and
solar irradiation, kI (in milliamperes per kelvin) is the short-
circuit current temperature coefficient, λ is the irradiance Fig. 4. PV characteristic of solar panel at different cell temperature and
(in milliwatts per square centimeter) and λr is reference solar radiation.
radiation level (100 mW/cm2). β is inverse thermal voltage
given by,
  
q 1 Rb
β= f1 (x) = − 1 x3 − (Rb + Rl )x1 − Vd
pKT L Rld
where, q = 1.6 x 1019 C is charge of an electron, K = 1.3805
x 1023 J/K is Boltzmanns constant, T cell temperature, and 1  
f2 (x) = np Ig − np Is (e(βx2 /ns ) − 1)
p is ideality factor. Now if PV module is made up of ns cells Ca
connected in series and np cells connected in parallel then
 
output current of PV module is given by, 1 x3
f3 (x) = x1 −
Cb Rld
ipv = np Ig − np Is (e(βvpv /ns ) − 1) (4)
(Vd + x2 )
B. Power Converter g1 (x) =
L
Fig. 2 shows a PV array connected to a dc-dc buck
converter. Power output from PV module is controlled by x1
g2 (x) = −
varying the duty cycle of buck converter so as to operate at Ca
a voltage at which Power output can be maximized. The state
space model of buck converter can be given by the following g3 (x) = 0 (6)
equation,
where
ẋ = f(x) + g(x)u (5) x1 = Inductor current(A)
x2 = Panel voltage(V)
x = [x1 , x2 , x3 ] x3 = Output voltage(V)
Ipv = Current drawn from solar panel(A)
f(x) = [f1 (x), f2 (x), f3 (x)] u = Duty cycle
Ca = Input capacitor(F)
g(x) = [g1 (x), g2 (x), g3 (x)] Cb = Output capacitor(F)

662
 III. D ESIGN OF SLIDING MODE CONTROLLER
While applying sliding mode control, the most important
concern is selection of sliding surface. There may exist
infinite number of sliding surfaces for a system but all
of them do not brings the system to origin. Furthermore,
if sliding surface comprises lesser number of states than
the order of system, the controller must stabilize all those
states which are not included in sliding surface. A sliding
surface is presented in [15] which comprises error in PV
voltage and inductor current as sliding states considering
ipv as a constant. For smaller change in solar radiation ipv
may be assumed constant but for large change in irradiance,
ipv must be considered as variable for choosing sliding
surface and controller design. Considering above constraints,
a sliding surface is chosen taking error in inductor current
and converter output voltage as sliding states. The chosen
surface needs desired value of inductor current and converter
output voltage which are calculated in the sequel.

A. Calculation of desired states

Incremental conductance method with an added inner loop
is used to calculate desired value of PV voltage. Calculation
is made fast by taking larger step size in inner loop when
error is more and step size is decreased gradually when it
reaches sufficiently close to the desired value. Since, calcu-
lation of desired value does not involve controller (which
is fixed between 0 to 1), the system is always stable and
calculation is made fast without any significant chattering.
Hence application of incremental conductance method for
calculating desired value is free from the demerits of chatter-
ing or slow convergence in case of rapid change in weather
condition. Calculation of desired value of inductor current Fig. 5. Flow chart for calculation of desired panel voltage.
and converter output voltage is performed as follows
In desired condition Now we have eliminate x1 and its derivative in order to
get direct relation between x2r and x3r and their derivatives.
e1 = e2 = e3 = 0, ė1 = ė2 = ė3 = 0
using steady state for capacitor output voltage,
Assuming x1r , x2r , x3r to be desired value of inductor
x3r
current, PV voltage and converter output voltage respectively x1r = + Cb ẋ3r (10)
and using above equation for inductor current, Rld
1 Rb (VD + x2 ) and
[( − 1)x3 − (Rb + Rld )x1 − VD ] + u = ẋ1r
L Rld L ẋ3r
ẋ1r = + Cb ẍ3r (11)
Rb Rld
(VD + x2r )u = Lẋ1r + VD + (Rb + Rld )x1r − ( − 1)x3r
Rld substituting (10) and (11) in equation (9)
(7)
1
similarly for converter output voltage, following equation can ẍ3r = (A/B − C) (12)
be derived LCb
βx2r where
βx2r
ẋ1r u = np Ig − np Is (e ns − 1) − Ca ẋ2r (8)
A = (VD + x2r )(np Ig − np Is (e ns − 1) − Ca ẋ2r )
Dividing (8) by (9) to eliminate u
x3r
Rb B = Cb ẋ3r +
Lẋ1r + VD + (Rb + Rld )x1r − ( − 1)x3r Rld
VD + x2r Rld Lẋ3r Rb
= C= + (Rb + Rld )B + VD − ( − 1)x3r
x1r βx2r Rld Rld
np Ig − np Is (e ns − 1) − Ca ẋ2r Once ẍ3r is known, x3r and ẋ3r can be updated. Also,
(9) using (11) x1r will be updated.
663
B. Choosing sliding surface −f1 (x) + ẋ1r − k(f3 (x) − ẋ3r )
û =
g1 (x)
Rewriting dynamic equation for converter output voltage
as Above controller bounds the system on sliding surface once
1 x3 it is reached on sliding surface. If system is not on sliding
ẋ3 = (x1 − ) (13) surface, above controller can not be applied. To bring system
Cb Rld
on sliding surface, controller can be modified as
when desired condition is reached, ẋ3 = ẋ3r
k1 sgn(s)
1 x3r u = û − (20)
ẋ3r = (x1r − ) (14) g1 (x)
Cb Rld
Using above two equations, the error dynamics can be found Controller in (20) brings the system on sliding surface with
as a constant rate which makes convergence very slow. To make
1 e3 it fast, (20) can again be modified as
ė3 = (e1 − ) (15)
Cb Rld
k1 sgn(s) + k2 s
u = û − (21)
now choosing sliding surface as g1 (x)

s = e1 + ke3 (16)
D. Convergence and tracking of PV voltage
when system is on sliding surface, Writing state equation for PV voltage as
s = 0 ⇒ e1 = −ke3
1 np Is (βx2 /ns )
putting value of e1 in (15) x˙2 = (np Ig + np Is − x1 u) − (e
Ca Ca
1 1
ė3 = − (k + )e3 (17)
Cb Rld
x˙2 = k3 − k4 (e(βx2 /ns ) (22)
similarly, using sliding condition and (15) the dynamics of
e1 can be written as where, k3 is bounded quantity and k4 is a positive constant.
1 1 Above equation can be proved for x2 to be bounded if both
ė1 = − (k + )e1 (18) k3 and k4 are strictly positive. Noting that u is between 0
Cb Rld
to 1 and inductor current is always less than the generated
Above two equations show that e1 and e3 converges to zero
current in PV system, k3 is always positive. Hence x2 is also
if
bounded and stability of whole system is guaranteed with
1 1
(k + )>0 (19) SMC. Now, performing steady state analysis for tracking of
Cb Rld x2 to its desired value. In steady state
Choosing k > 0 above condition is easily satisfied and time
taken for e1 and e3 to converge to a small value ε is  
1 x3
x˙3 = x1 − =0
  Cb Rld
Cb |e(0)|
tε = ln Therefore x1 and x3 are bounded by
1 |ε|
K+
Rld x3 = x1 Rld (23)
where, tε is time taken for e1 and e3 to converge to small
Similarly, performing steady state analysis on inductor cur-
value ε and e0 is initial value of e1 and e3 . It can be observe
rent and capacitor output voltage results in,
from (17) and (18) that time constant for converge for e1
and e3 is same. Therefore, both error converges together if
system is on sliding surface. Also, Cb is in order of μF, time βx2
constant is very low even for smaller value of k. (Vd + x2 )np Ig + np Is − np Is e ns + (24)
(Rld + Rl )x21 + VD x 1 = 0
C. Sliding mode Controller
Above relation represent a quadratic equation in x1 which
When system is on sliding surface, the equivalent control has one positive and one negative solution for a particular
input is obtained by substituting ṡ = 0 value of x2 . Since inductor current can not flow in opposite
e˙1 + k e˙3 = 0 direction in converter, negative solution of x1 is not possible.
Hence, there is only one possible solution for x1 exists for
one value of x2 in steady state. Also, from (23), x3 has only
ẋ1 − ẋ1r + k(ẋ3 − ẋ3r ) = 0
one solution for a particular x1 . Hence all the three states
of PV system are bounded by one to one mapping in steady
f1 (x) + g1 (x)û − ẋ1r + k(f3 (x) − ẋ3r ) = 0 state.
664
Fig. 6. MPPT control response at 298K and 100mW/cm2 (a)Red line: Fig. 7. MPPT control response at 298K and 100mW/cm2 (a) Red line:
MPP voltage, blue dash: obtained voltage by SMC,(b) Red line :Power at Desired , Blue dash: obtained inductor current (b) Red line:Desired, Blue
MPP Blue dash: obtained power by SMC and (c) control input u(t). dash: obtained PV array voltage and (c) Red line: Desired , Blue dash:
obtained output capacitor voltage.

In PV system, for a particular cell temperature and solar

radiation, there exists only one desired value of panel voltage
for MPPT. Since all the states are bounded by one to one
relation, there exists only one value of desired inductor
current and converter output voltage for a particular weather
condition. Therefore, if inductor current and converter output
voltage are converging to their desired value, PV voltage must
reach to its desired value provided system remains stable
during the course of convergence.

IV. SIMULATION RESULTS

The simulation of this solar power system has been carried
out to verify the theoretical results. The specifications of
the Solar Panel are as follows: Number of cells in parallel,
np = 1 and the number of cells in series, ns = 36. Short
circuit current, Isc = 4.8A. Short circuit current temperature
coefficient, kI = 2.06mA/0 C. The specifications of the Dc-
Dc converter are as follows: Storage Inductance (L) of 95μH,
Capacitance’s Ca = 200μF and Ca = 300μF . The internal
resistances Rb and Rl of capacitance Cb and Inductance Fig. 8. MPPT control response at 298K and 100mW/cm2 (a) error in
L are chosen as 162mΩ and 1Ω respectively. The forward inductor current e1 , error in PV array voltage e2 and error in output capacitor
voltage e3 (b) waveform of sliding surface.
voltage drop of the power diode D, vD is taken as 0.7 Volts.
Incremental value of each step is calculated from the flow
chart. Frequency of searching algorithm is set at 10KHz. sufficiently close to zero. Next, we have simulated our system
Fig.6-8 shows simulation of the above solar panel using SMC under varying temperature and solar radiation as shown in
when we consider fixed temperature and solar radiation at Fig. 9. Tracking of output capacitor voltage, output power
298K and 100mW/cm2 respectively. The controller gains and control input are depicted in Fig. 10. Error in inductor
are chosen as k1 = 3000 and k2 = 300. These high values current, panel voltage and output capacitor voltage are shown
of k1 and k2 is justified as value of g1 (x) is 2 order higher in Fig. 11 under varying cell temperature and solar radiation.
than controller gains. Fig. 6. shows that MPPT voltage and
extracted power from PV array reach their desired values with
V. C ONCLUSION
in 3-4 ms. In Fig. 8. convergence of errors of system states
in shown. It has been found that all errors converges to zero In this paper, MPPT control strategy based on sliding mode
asymptotically and with in 2-3 ms and they are found to be theory has been presented for PV solar power generation
665
 system . A dc-dc buck converter is used to control the
voltage so as to ensure MPPT. Error dynamics has been
derived for the system and from that a sliding surface is
chosen to ensure the convergence of all system states to
desired values. The control diagram proposed is very simple,
using a global model of the nonlinear system, and from
which a nonlinear control law directly applied to the power
switch. In the presence of the disturbance and uncertainty,
the robust MPPT is also assured while the maximum power
tracking error is attenuated to a prescribed level. Compared
with the traditional PI controller, proposed SMC achieves
fast tracking response with precise control and significantly
Fig. 9. variation in temperature and solar radiation. less chattering. Thus, proposed SMC control assures better
tracking performance with high robustness.

R EFERENCES
[1] Z. Salameh and D. Taylor, “Step-up maximum power point tracker
for photovoltaic arrays”, Solar Energy, vol. 44, no. 1, pp. 57-61,
1990.
[2] G. Carannante, C. Fraddanno, M. Pagano, and L. Piegari,“ Exper-
imental performance of MPPT algorithm for photovoltaic sources
subject to inhomogeneous insolation,” IEEE Trans. Ind. Electron.,
vol. 56, no. 11, pp. 4374-4380, Nov. 2009.
[3] H. Patel and V. Agarwal, “MPPT scheme for a PV-fed single-phase
singlestage grid-connected inverter operating in CCM with only one
current sensor,” IEEE Trans. Energy Convers., vol. 24, no. 1, pp.
256-263, Mar. 2009.
[4] K. H. Hussein, I. Muta, T. Hoshino, and M. Osakada, “Maximum
photovoltaic power tracking: An algorithm for rapidly changing
atmospheric condition,” Proc. Inst. Electr. Eng., Gen. Transmiss.
Distrib., vol. 142, no. 1, pp. 59-64, 1995.
[5] K. H. Hussein and I. Muta, “Maximum photovoltaic power tracking:
An algorithm for rapidly changing atmospheric conditions”, Proc.
Inst. Electr. Eng.Generation, Transmission Distribution, vol. 142, no.
1, pp. 59-64, Jan. 1995.
[6] T. Noguchi, S. Togashi, and R. Nakamoto, “Short-current pulse-based
maximum-power-point tracking method for multiple photovoltaic and
converter module system,” IEEE Trans. Ind. Electron., vol. 49, no.
1, pp. 217-223, Feb. 2002.
[7] C. Dorofte, U. Borup, and F. Blaabjerg, “A combined two-method
MPPT control scheme for grid-connected photovoltaic systems,” in
Proc. Eur. Conf. Power Electron. Appl., Sep. 1114, 2005, pp. 1-10.
Fig. 10. MPPT control response under varying temperature and solar [8] D. Sera, T. Kerekes, R. Teodorescu, and F. Blaabjerg, “Improved
radiation ((a) Red line: MPP voltage, blue dash : obtained voltage by MPPT method for rapidly changing environmental conditions,” in
SMC,(b) Red line :Power at MPP Blue dash : obtained power by SMC Proc. IEEE Int. Ind. Electron. Symp., Jul. 2006, vol. 2, pp. 1420-
and (c) control input u(t). 1425.
[9] N. Kasa, T. Iida, and H. Iwamoto, “Maximum power point tracking
with capacitor identifier for photovoltaic power system,” Proc. Inst.
Electr. Eng.Electr. Power Appl., vol. 147, no. 6, pp. 497-502, Nov.
2000.
[10] N. Kasa, T. Iida, and L. Chen, “Flyback inverter controlled by
sensorless current MPPT for photovoltaic power system,” IEEE Trans.
Ind. Electron., vol. 52, no. 4, pp. 1145-1152, Aug. 2005.
[11] M. Veerachary, T. Senjyu, and K. Uezato, “Maximum power point
tracking control of IDB converter supplied PV system,” Proc. Inst.
Electr. Eng.Electr. Power Appl., vol. 148, no. 6, pp. 494-502, Nov.
2001.
[12] J.J.E.Slotine and W.Li. Applied Nonlinear Control. Prentice hall, New
Jersey, 1991.
[13] Stanishlaw H. Zak. System and Control. Oxford University Press,
New York, 2003.
[14] Chian-Song Chiu,Ya-Lun Ouyang, Chan-Yu Ku, “Terminal sliding
mode control for maximum power point tracking of photovoltaic
power generation systems”, Solar Eneargy., Vol. 86, no. 10, pp.
2986-2995, Oct. 2012.
[15] Domingo, Biel, Enric, Fossas, et al., 2001. “Application of sliding-
Fig. 11. Error in inductor current e1 , error in PV array voltage e2 and error mode control to the design of a buck-based sinusoidal generator”,
in output capacitor voltage e3 . IEEE Trans. Ind. Electron. Vol. 48, no. 3, pp. 563-571, Jun. 2001.

666