Jayachandran 2018

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

MPC based Secondary Control Strategy for an

Islanded AC Microgrid under Linear Loads


Jayachandran.M Ravi.G
Department of Electrical and Electronics Engineering Department of Electrical and Electronics Engineering
Pondicherrry Engineering College Pondicherrry Engineering College
Puducherry, India 605014 Puducherry, India 605014
Email: jayachandran.escet@pec.edu Email: ravig@pec.edu

Abstract—The dynamic response of the converters are much


faster than the load frequency dynamics. Usually, droop control
method is employed to represent converters for active and
reactive power sharing in microgrids. The drawback of this
method is that voltage and frequency deviates from its nominal
value under steady state. A distributed secondary control based
on Model predictive control with state space approach is proposed
for load frequency control and voltage restoration of islanded
AC microgrid. Hence, This secondary controller slowly add
increments to the primary level until the frequency and voltage
deviation gets eliminated. To demonstrate the effectiveness of
the proposed controller is carried out by simulation under
mismatched feeder impedance with balanced and unbalanced
linear load conditions.

I. I NTRODUCTION
With high penetration of distributed generator (DG) units
on the electrical grid systems, renewable energy source (RES)
Fig. 1. Structure of microgrid with power electronic inter-
have been widely used. The advantages of DG units incorpo-
faced parallel connected DG units
rated to power grid system are reducing pollution, decreasing
transmission losses and improving utilization of local RESs
[1]. In recent years, research on advanced distributed control
in microgrids such as graph theory, predictive control and strains. This control strategy does not include modulators
multi-agent systems in order to improve the robustness and and current controllers, and therefore completely reduces the
reliability of the system, optimize powerflow simultaneously complexity [5]–[8].
and realize accurate real and reactive power sharing among In general, DGs operate with inductive feeder, the Pω-QV
multiple DGs [2]. Hierarchical control scheme has commonly droop control is employed to attain real and reactive power
employed for energy management system in microgrids. it sharing and PV-Qω droop control have applied for resistive
comprises of three levels: 1) the primary control includes feeder impedance. In order to prevent power coupling, proper
power droop control and virtual impedance loop to achieve real control of VSI with virtual impedance loop is employed.
and reactive power sharing, 2) the secondary control which Droop controller is employed to share active and reactive
restores the frequency and voltage deviation caused by the power sharing in microgrids. The drawback of this method is
primary control, and 3) the tertiary control which manages that voltage and frequency deviates from its nominal value
the powerflow between microgrid and grid [3]. In the inner under steady state and it can be restored using secondary
loop controller design, traditional linear PI controller may not control scheme (SCS) [9]. it is often employed to MG for
suitable for control of VSI due to slow response and create accurate power sharing, voltage unbalance and harmonic com-
instability caused by integral term. Proportional resonant (PR) pensation etc [10] . Conventional centralized secondary control
voltage and current controller has been widely used to generate system requires point to point low bandwidth communication,
voltage reference applied to PWM [4]. However, this approach which adds complexity of the system. Alternately, distributed
has a few drawbacks such as it requires modulator and operates secondary control avoids a single centralized control and
with fixed frequency. MPC with finite control set scheme has improves reliability of the MG. It exchanges information
several advantages than other predictive control techniques between neighbouring DG units to adjust the global system
owing to its rapid dynamic response, operates in variable variables [11].
frequency, and successful insertion of nonlinearities and con- In this paper, FCS-MPC based primary control strategy is

978-1-5386-3695-4$31.00 2018
c IEEE 644
DG Power Stage Sensitive
P DG Feeder
Load
Bus
Impedance
Rf, Lf
a
b
Vdc
c
n Cf Measurement
iabc voabc ioabc Block

in vabc
N

abc ioab
ab Fundamental
abc voab Powers
ab Calculation
ioab Secondary Control
P Q
FCS-MPC Virtual
Minimization Power MPC based
Voltage Impedance
g-function Droop Frequency
Controller Loop
Control Secondary wn
Control
V vab w
E
dw
V *abc ab
 MPC based
abc Voltage
 Secondary
Es ws dE Control
V vabc En
Three Phase
- Sinusoidal
+ V Reference
DG Primary Controller gabc Generator

Fig. 2. Block diagram of the Hierarchical predictive control scheme for an islanded AC microgrid.

developed for three-phase four-leg VSI to adjust the output current and capacitor voltage of RLC filter are measured and
voltage and current that makes the system stable, and enhance controlled by the predictive controller, which is performed in
the power quality at PCC. This scheme comprises, droop abc frame, to generate gating signals to be applied to the VSI.
control to achieve stability and virtual impedance loop to
A. Power Droop Control
prevent real and reactive power coupling by offering inductive
impedance. Moreover, secondary control based on model The instantaneous powers are are computed from the αβ-
predictive control (MPC) algorithm with state space approach axis and passed through 1st order LPFs with 4π (rad/s) to get
is proposed to eliminate voltage and frequency deviations. fundamental real and reactive powers. To ensure PQ flow con-
trol, obtaining good power sharing and avoid communication
II. S TRUCTURE OF M ICROGRID S YSTEM wires, the droop control is utilized for regulating frequency and
A common MG system with distribution generation (DG) the amplitude of reference voltage in accordance with real and
units, distortion feeders and complex loads are depicted in reactive powers [12]. P-ω and Q-E power droop controllers can
Fig. 1. For each DG unit consists of RES which provides DC be expressed (1) and (2) as
power, integrated to microgrid with RLC filter and connected ω = ωn − kp · P (1)
to PCC through distribution feeder. Complex loads are also
E = En − kq · Q (2)
connected to PCC. In this paper, Islanded mode microgrid op-
eration is considered, therefore, the main grid is disconnected E and ω are voltage amplitude and frequency generated by
from microgrid using the static transfer switch (STS). Besides, droop control. En and ωn are the nominal voltage amplitude
RES with DC link capacitor is considered as fixed DC Voltage and angular frequency. Droop coefficients kp and kq defined
to simplify the theoretic analysis. by kp = Δf /pmax and kq = ΔV /Qmax . where Δf and
ΔV are the maximum frequency and voltage deviation, Pmax
III. P ROPOSED P REDICTIVE H IERARCHICAL C ONTROL and Qmax are the nominal real and reactive power. The droop
S TRUCTURE FOR M ICROGRID characteristics (1) and (2) guarantee to bring the magnitudes of
The arrangement of DG power unit with decentralized voltage and phase angles to stability and eradicate circulating
primary predictive controller is illustrated in Fig. 2. Power current completely at steady state. The value of virtual DG
droop controller and virtual impedance loop in DG local impedance can be obtained based on [13] and its correspond-
controller are performed in αβ reference frame that provides ing voltage drop Vv can be calculated in stationary reference
the reference voltage for predictive control model. With the frame from DG line current i0 as,
    
help of Clarke transformation, the three phase sinusoidal Vvα Rv −ωLv ioα
signals are transformed between abc and αβ frames. Inductor = (3)
Vvβ ωLv Rv ioβ

4th International Conference on Electrical Energy Systems (ICEES) 645


w 
B. Predictive Voltage Controller
w0 
To improve reference tracking, enhance stability and dis-

  

  



w1   
dw dE
w2 
turbance rejection at the fundamental frequency, FCS-MPC
 
scheme is utilized [14] The discrete time model of DG unit  
 
 
 

obtained from continuous time equations, is applied to predict P Q

the future behaviour of the system for each of the sixteen (24 ) P1 P2 Q1 Q2

possible states. The control method selects the switching state ωs = ω0 − kp · P + δω Es = E0 − kq · Q + δE


that minimizes the error (cost function) between the output (a) (b)
voltages and their references. The voltage measured at each
leg of the inverter with reference to the negative terminal of Fig. 3. Secondary control (a) Frequency restoration in MG.
DC link (N) can be represented in terms of switching states (b) Voltage restoration in MG.
as,
ViN = Si Vdc, i = a, b, c, n (4) IV. S ECONDARY MPC C ONTROL FOR R EGULATING
M ICROGRID VOLTAGE AND F REQUENCY
The feasible switching states with equivalent line and phase
voltages of the inverter are found in [15]. The inverter voltage Fig 3 shows how secondary control eliminates frequency
applied to RLC filter is, and voltage deviations from its nominal value caused by
P-ω and Q-E droop control algorithms respectively. Fig 3a
Vjn = VjN − VnN = (Sj − Sn )Vdc, j = a, b, c (5) demonstrates that The secondary control output signal (cor-
recting frequency) (δω) influences the frequency reference
RLC filter equations in differential form, regarding voltage and (ω) of primary control by shifting the droop lines up, so
current vectors are formulated as: that frequency can finally reach the nominal value (ωn ). The
frequency applied to signal generator (ωs ) can be represented
di
Vo = V − Lf − Rf i (6) by,
dt ωs = ω + δω (10)
dVo Similarly, The voltage applied to signal generator (Es ) as,
i = io + Cf (7)
dt
Es = E + δE (11)
Filter current i(k), output current io (k) and output voltage
(k+1) The model algorithmic control (MAC) the dynamic matrix
Vo (k) are applied to predict the output voltage Vo at next
control (DMC) are two predictive control schemes can be used
sampling instant. The system in (6) and (7) can be expressed
for stable system based on step response models of the plant.
in state space form [8] as,
On the other hand, the generalized predictive control (GPC) is
     
(k+1) developed for unstable system based on the transfer function
Voj Voj (k) Vj (k)
(k+1) = Aq + Bq , j = a, b, c (8) model of the plant [16]. The discrete model of the system is
ij ij (k) ioj (k)
used in order to predict the future behaviour of the system
and set the future values of control actions are calculated by
where, Aq = eATs , Bq = B
A (Aq − I) optimizing cost function. The DG system is assumed to be
    linear with no constrains. The following transfer function of
0 1/Cf 0 −1/Cf
and A = ,B= the system in Z-domain is obtained from state space equation
−1/Lf −Rf /Lf 1/Lf 0
to predict the future behaviour of the system as,
where, A and B are state-space matrices. 0.9332Z −1 + 0.01252Z −2
The optimization algorithm successively selects one of the H(Z) = (12)
1 − 0.1364Z −1 + 0.08208Z −2
16 feasible voltage vectors V(k) and applies it to equation
Generally, the model of the distributed generator sys-
(8), to attain the voltage prediction at next time instant for
tem adopted with Controlled autoregressive moving average
all switching states of each inverter. In order to obtain the
(CARMA) model [17] has the following form,
lowest voltage error, the cost (objective) function is used in
(9), to calculate the error between the reference voltage and A[Z −1 ]y(k) − B[Z −1 ]u(k) − C[Z −1 ]v(k) = 0 (13)
the predicted voltage for each voltage vector.
where, y(k), u(k) and v(k) are output, control input and
g= (Va∗ − Voa
(k+1) 2 (k+1) 2
) + (Vb∗ − Vob ) + (Vc∗ − Voc
(k+1) 2
) (9) disturbance input sequences respectively. A[Z −1 ], B[Z −1 ] and
C[Z −1 ] are polynomial matrices in Z −1 . It can be represented
Then, the best possible values of voltage vector and cost as,
function are selected. Finally, the output switching states
A[Z −1 ] = I + A1 Z −1 + A2 Z −1 + ... + An Z −n (14)
corresponding to the optimum voltage vector are achieved and
applied to VSI. B[Z −1 ] = B1 Z −1 + B2 Z −1 + ... + Bn Z −n (15)

646 4th International Conference on Electrical Energy Systems (ICEES)


C[Z −1 ] = I + C1 Z −1 + C2 Z −1 + ... + Cn Z −n (16) Start

The state space realization of the CARMA model as follows,


Identify Plant Model
x(k + 1) = Φx(k) + Γu(k) + Λv(k) (17) eq(12)

y(k) = Hx(k) + v(k) (18)


Initial values: Initial state, output, controls,
tuning parameters, set point etc
where, Φ, Γ, Λ, H are m × m, m × p, m × q, q × m matrices.
⎡ ⎤ ⎡ ⎤
−A1 I B1 Constant parameters: weight matrices,
⎢ . . ⎥ ⎢ . ⎥ prediction matrices etc
⎢ ⎥ ⎢ ⎥
⎢ . . ⎥ ⎢ ⎥
Φ=⎢ ⎥,Γ = ⎢ . ⎥,
⎢ . . ⎥ ⎢ ⎥
⎢ ⎥ ⎢ . ⎥ Perform control law
⎣−An−1 I ⎦ ⎣ Bn−1 ⎦ u(k)=u(k-1)+Δu(k)

−An 0 . . . 0 Bn
⎡ ⎤ (19)
C1 − A1 Calculate state and output
eq(20) and eq(21)
⎢ . ⎥
⎢ ⎥
⎢ . ⎥
Λ=⎢ ⎢ ⎥,H = I 0 . . . 0
. ⎥ Compute cost function to find Δu(k)
⎢ ⎥ eq(23) or eq(24)
⎣Cn−1 − An−1 ⎦
Cn − An
The state observer of the plant and the j-step ahead predictor
represented in equation (20) and (21) as, No
Required update
x(k + 1) = (Φ − ΛH)x(k) + Γu(k) + Λy(k) (20) number of cycles u=u+Δu(k+j-1)

y(k + j|k) = HΦj x(k) + HΦj−1 Λ[y(k) − Hx(k)]



j
(21) Yes
+ HΦj−1 Γu(k + j − i)
i=1 Stop

The equation (21) represents j-step ahead predictor in the state-


space domain. It is assumed that the future Δu = 0, then the Fig. 4. Flowchart of proposed state space GPC algorithm
becomes,
PCC
DG1 DG Feeder 1

y(k + j|k) = HΦj x(k) + HΦj−1 Λ[y(k) − Hx(k)]


Rf, Lf
(22) DC
PoC1

Power
Supply
The algorithm for state space GPC as illustrates in Fig 4. The Cf
a
frequency can be predicted based on equation (22) and applied DG Local
Controller-1
b
c
n

to equation (23) to generate control action ωs by minimizing DG2 DG Feeder 2


Rf, Lf
cost function. In order to mitigate the stability issues the state PoC2 Linear Load
DC
space GPC is proposed for MIMO systems based on Z-domain Power
Supply
Cf
analysis of the predictor and the cost function is given by [9],
DG Local abcn
Controller-2


p 
m
δω = [ωP LL (k+j|k)−ωn ]2 +λ [Δu(k+j −1)]2 (23) Fig. 5. MG test system
j=1 j=1

Where, λ is the weighting factor. The first term minimizes


the tracking error between the prediction of the measured V. S IMULATION R ESULTS
system frequency and its reference frequency (nominal value) Fig. 5 illustrates the simulation test system for islanded
and second term minimizes the control action. it produces the microgrid. In order to meet the load demand, two equal DGs
optimal control action δω at the secondary control. p and m are used. The parameters of MG test system are presented in
are prediction and control horizons respectively. In the same Table. I. An impedance matrix of a distribution test feeder Z606
way control action δE can be obtained for voltage regulation (underground feeder) and Z601 (overhead feeder) in Ω/km are
and the objective function as, used according to [18], [19].

p 
m Case 1: The three phase balanced linear load of impedance
δE = [E0 (k + j|k) − En ]2 + λ [Δu(k + j − 1)]2 (24) ZL = 12+j13Ω is applied at t=0.5sec to PCC. The total power
j=1 j=1 required to meet the load demand is 8.97KVA. The current

4th International Conference on Electrical Energy Systems (ICEES) 647


Table I. DG Power Stage and Control System Parameters
6A 6B 6C

6ABC 6
Power system & 
Value
control parameters 
DC link voltage Vdc = 650 V 
Rf =0.05 Ω
)A )B )C
RLC filter Lf =2 mH   

)ABC !
Cf =80μF

Simulation Ts = 1μs
Sample period Primary controller Ts = 33μs 
Secondary controller Ts = 0.05s    
Kp = 0.0001 rad/sec/W 4IME 3EC
Droop coefficients
Kq = 0.0001 V/Var
Rv = 0.2 Ω (a)
Virtual impedance
Lv = 4 mH 6A 6B 6C
DG feeder1 = Z601

6ABC 6
DG feeder 
DG feeder2 = Z606

Prediction horizon p = 10
Secondary control 
Control horizon m= 5
tuning parameters
Weighting factor λ = 1  )A )B
 )C


)ABC !
Case 1: 6.0847KW, 6.5907KVAr
Case 2: 6.0847KW, -6.5907KVAr 
Load
Case 3: 13.225KW, 0VAr 
Case 4: 7.098KW, 7.6891KVAr
   
4IME 3EC

6A 6B 6C
(b)
6ABC 6


Fig. 7. Voltage and current waveforms of the MPC based


controller in an islanded MG with 6.0847KW, -6.5907KVAr
)A ) )C load. Voltage THD: 3.43% (a) DG1. (b) DG2.
   B
)ABC !


6A 6B 6C

6ABC 6

    


4IME 3EC 

(a)
)A ) )C
  B

)ABC !

6A 6B 6C
6ABC 6

 
 

   
)A )C
   )B 4IME 3EC
)ABC !


(a)
6A 6B 6C

6ABC 6

    


4IME 3EC 

(b)
)A )C
 )B

)ABC !

Fig. 6. Voltage and current waveforms of the MPC based



controller in an islanded MG with 6.0847KW, 6.5907KVAr

load. Voltage THD: 3.60% (a) DG1. (b) DG2.
   

(b)
lags voltage due to inductive load and power factor is 0.6783
lagging. Thus inductive load consumes reactive power. As a Fig. 8. Voltage and current waveforms of the MPC based
result voltage sag occurs. The performance of the proposed controller in an islanded MG with 13.225KW load. Voltage
controller as presented in Fig 6. It is observed from the THD: 3.43% (a) DG1. (b) DG2.
simulation result that the output voltage is not affected by
change in load by proposed control scheme.
Case 2: The three phase balanced linear load of impedance leads voltage due to capacitive load and power factor is 0.6783
ZL = 12 − j13Ω is connected at t=0.5sec to PCC. The leading. Thus capacitive load injects reactive power. As a result
total power required to meet the load demand is 8.97KVA. voltage swell occurs. However, the proposed control scheme
When load changes from no load to full load the current effectively tracks output voltage when change in load as shown

648 4th International Conference on Electrical Energy Systems (ICEES)


6A 6B 6C
6ABC 6 



0 7

 $'

$'
 
 )A ) )C      
)ABC !

1 6!2
B

 $'

$'
 
         
4IME 3EC

(a) (a)
6A 6B 6C

6ABC 6



0 7

 $'

 $'

)A     
  )B )C $' 

1 6!2
)ABC 6

 $'



      
   
4IME 3EC 4IME 3EC

(b) (b)
Fig. 9. Voltage and current waveforms of the MPC based 

0 7
controller in an islanded MG with unbalanced load. Voltage 
$'
THD: 3.63% (a) DG1. (b) DG2. 
$'

     
1 6!2

in Fig 7.  $'
Case 3: The three phase balanced linear resistive load of  $'

impedance ZL = 12Ω is attached at t=0.5sec to PCC. The      
4IME 3EC
total power required to meet the load demand is 13.225KVA.
As observed from the Fig 8 that the current and voltage (c)
waveforms are inphase due to resistive load and power factor
is 1. Hence reactive powers are not present. The output voltage 
0 7


not affected with respect to change in load by the proposed 
$'
control scheme as illustrated in Fig 8. $'

Case 4: The three phase unbalanced linear RL load of      
1 6!2

impedance ZLa = 6 + j6.5Ω, ZLb = 12 + j13Ω and 


$'
ZLc = 24 + j26Ω are applied to each phase at t=0.5sec 
$'

to PCC. The total power required to meet the load demand      
is 10.465KVA. Unbalanced load causes unbalanced voltage. 4IME 3EC
However, It is observed from the Fig 9 that the output voltages (d)
balanced by incorporated by the proposed control strategy.
Thus in all four cases the proposed control scheme effectively Fig. 10. Power sharing performance of proposed controller in
tracks the voltage. an islanded MG with Linear load. (a) Balanced RL load. (b)
The real and reactive power sharing performance of pro- Balanced RC load. (C) Balanced R load (D) Unbalanced RL
posed control scheme under balanced and unbalanced load load
conditions as depicted in Fig 10. The real and reactive power
demand is presented in Table I. It is observed from the
simulation results that the active and reactive powers are load and slight deviations present in PCC as depicted in Fig
shared properly by virtual impedance method as illustrated 11b. when balanced R load is suddenly applied to PCC, there
in Fig 10. is a voltage drop at both DGs and frequency not maintained
When a load is connected at t=0.5sec to PCC, there is a its nominal value Fig 11c. Further, unbalanced load creates
small voltage and frequency deviation is observed as depicted voltage and frequency variation as shown in Fig 11d.
in Fig 11. In Fig 11a the voltage sag is observed due to in- In order to validate the effectiveness of the proposed state
ductive nature of load and frequency deviates from its nominal space MPC based secondary control scheme, the load is
value. On the other hand, voltage swell occurs by capacitive activated from t = 1sec, voltage and frequency gradually

4th International Conference on Electrical Energy Systems (ICEES) 649




6OLTAGE 6
&REQUENCY (Z 6OLTAGE 6 $'
 $' 
$'
$'
 

&REQUENCY (Z
      
 $' 
$'
 $'
 $'
         
4IME SEC 4IME 3EC
(a) (a)

&REQUENCY (Z 6OLTAGE 6

$'
$'


    
 $'
$'


     
4IME SEC

(b) (b)
6OLTAGE 6

$'
 $'


    $' 
&REQUENCY (Z


$'


     


4IME SEC

(c) (c)
6OLTAGE 6


$'
 $'


    $' 
&REQUENCY (Z

 $'


     
4IME 3EC

(d) (d)
Fig. 11. Voltage and frequency performance of proposed con- Fig. 12. Voltage and frequency performance of proposed
trol scheme without secondary controller in an islanded MG. control scheme with secondary controller in an islanded MG.
(a) Balanced RL load. (b) Balanced RC load. (C) Balanced R (a) Balanced RL load. (b) Balanced RC load. (C) Balanced R
load (D) Unbalanced RL load load (D) Unbalanced RL load.

move towards its nominal value within 0.4 s. It can be Moreover, RL load is suddenly applied at t=1sec, a small
seen that the proposed secondary control strategy eliminate voltage and frequency deviation is observed. If DG1 is turned
the voltage and frequency deviation produced by the droop off at t=1.5s the frequency of DG2 further deviates from its
control, and promptly regulate to its nominal value in an nominal value and the voltage drop occurs due to an increase
islanded microgrid. If the balanced or unbalanced linear load of load as depicted in Fig 13. Simulation results shows that the
applied to the system, both frequency and voltage remains to proposed MPC based Secondary control scheme restores the
the nominal values, causing negligible transients as shown in voltage and frequency deviations of DGs as shown in Fig 14.
Fig 12. It can be seen from the Fig 14 that the secondary controller

650 4th International Conference on Electrical Energy Systems (ICEES)


.ORMAL LOAD $' R EFERENCES
&REQUENCY (Z 6OLTAGE 6 /PERATION 3TEP $ISCONNECTION
 [1] C. Wang, X. Yang, Z. Wu, Y. Che, L. Guo, S. Zhang, and Y. Liu,
 $' “A highly integrated and reconfigurable microgrid testbed with hybrid
 $' distributed energy sources,” IEEE Transactions on Smart Grid, vol. 7,
 no. 1, pp. 451–459, Jan 2016.
    [2] F. Garcia-Torres, C. Bordons, and S. Vazquez, “Voltage predictive
 control for microgrids in islanded mode based on fourier transform,” in
$'
$' 2015 IEEE International Conference on Industrial Technology (ICIT).

IEEE, 2015, pp. 2358–2363.
[3] J. M. Guerrero, J. C. Vasquez, J. Matas, L. G. De Vicuña, and
     
4IME 3EC M. Castilla, “Hierarchical control of droop-controlled ac and dc micro-
grids - A general approach toward standardization,” IEEE Transactions
Fig. 13. Voltage amplitude and frequency in an islanded MG on Industrial Electronics, vol. 58, no. 1, pp. 158–172, 2011.
without secondary control [4] M. Savaghebi, A. Jalilian, J. C. Vasquez, and J. M. Guerrero, “Au-
tonomous voltage unbalance compensation in an islanded droop-
controlled microgrid,” IEEE Transactions on Industrial Electronics,
vol. 60, no. 4, pp. 1390–1402, 2013.
.ORMAL LOAD $'
[5] P. Cortés, M. P. Kazmierkowski, R. M. Kennel, D. E. Quevedo, and
&REQUENCY (Z 6OLTAGE 6

 /PERATION 3TEP $ISCONNECTION


J. Rodrı́guez, “Predictive control in power electronics and drives,” IEEE
Transactions on industrial electronics, vol. 55, no. 12, pp. 4312–4324,
 $' 2008.
$' [6] S. Kouro, P. Cortés, R. Vargas, U. Ammann, and J. Rodrı́guez, “Model

    predictive controla simple and powerful method to control power con-
 verters,” IEEE Transactions on Industrial Electronics, vol. 56, no. 6, pp.
$' 1826–1838, 2009.
 $' [7] J. Rodriguez, J. Pontt, C. A. Silva, P. Correa, P. Lezana, P. Cortés, and
U. Ammann, “Predictive current control of a voltage source inverter,”
     
4IME 3EC IEEE Transactions on Industrial Electronics, vol. 54, no. 1, pp. 495–503,
2007.
Fig. 14. Voltage amplitude and frequency in an islanded MG [8] V. Yaramasu, J. Rodriguez, B. Wu, M. Rivera, A. Wilson, and C. Rojas,
with secondary control “A simple and effective solution for superior performance in two-level
four-leg voltage source inverters: Predictive voltage control,” in 2010
IEEE International Symposium on Industrial Electronics. IEEE, 2010,
pp. 3127–3132.
 [9] C. Ahumada, R. Cardenas, D. Saez, and J. M. Guerrero, “Secondary
3ECONDARY

$'
INPUT Δ U

control strategies for frequency restoration in islanded microgrids with


VOLTAGE

 $'
consideration of communication delays,” IEEE Transactions on Smart
 Grid, vol. 7, no. 3, pp. 1430–1441, May 2016.
[10] M. Savaghebi, A. Jalilian, J. C. Vasquez, and J. M. Guerrero, “Secondary
     control scheme for voltage unbalance compensation in an islanded
3ECONDARY
FREQUENCY
INPUT Δ U

 $' droop-controlled microgrid,” IEEE Transactions on Smart Grid, vol. 3,


 $' no. 2, pp. 797–807, June 2012.
[11] R. Olfati-Saber and R. M. Murray, “Consensus problems in networks

of agents with switching topology and time-delays,” IEEE Transactions
      on Automatic Control, vol. 49, no. 9, pp. 1520–1533, Sept 2004.
4IME 3EC
[12] Y. Han, P. Shen, X. Zhao, and J. M. Guerrero, “Control strategies for
Fig. 15. Difference between desired and predicted signal of islanded microgrid using enhanced hierarchical control structure with
secondary control multiple current-loop damping schemes,” IEEE Transactions on Smart
Grid, 2015.
[13] J. He and Y. W. Li, “Analysis and design of interfacing inverter output
virtual impedance in a low voltage microgrid,” in 2010 IEEE Energy
effectively tracks the voltage and frequency reference signal Conversion Congress and Exposition. IEEE, 2010, pp. 2857–2864.
[14] G. Mirzaeva, G. C. Goodwin, B. P. McGrath, C. Teixeira, and M. E.
with zero steady state error as depicted in Fig 15 and the Rivera, “A generalized MPC framework for the design and comparison
system is stable with desirable power sharing. of VSI current controllers,” IEEE Transactions on Industrial Electronics,
vol. 63, no. 9, pp. 5816–5826, 2016.
[15] V. Yaramasu, M. Rivera, B. Wu, and J. Rodriguez, “Model predictive
VI. C ONCLUSION current control of two-level four-leg inverters-Part I: Concept, algorithm,
and simulation analysis,” IEEE Transactions on Power Electronics,
State space MPC based secondary control scheme has been vol. 28, no. 7, pp. 3459–3468, 2013.
[16] D. Clarke, C. Mohtadi, and P. Tuffs, “Generalized predictive control-
presented for an islanded microgrid system to improve output Part I. The basic algorithm,” Automatica, vol. 23, no. 2, pp. 137 – 148,
voltage quality and share load demand among several parallel 1987.
connected DGs. The effectiveness of the proposed strategy [17] Y. Cheng, “CARMA-model-based j-step-ahead prediction for MIMO
systems,” Frontiers of Electrical and Electronic Engineering in China,
have been validated for an islanded AC microgrid system. The vol. 2, no. 1, pp. 99–103, 2007.
simulation results reveals that the proposed scheme effectively [18] R. G. Ferraz, R. J. Cabral, A. S. Bretas, A. S. Bazanella, and R. C.
track their reference voltages with zero steady state error and Leborgne, “System unbalance effect on faulted distribution systems: A
numerical study,” in 2010 IEEE Power and Energy Society General
also regulates voltage and frequency to its nominal value. Thus Meeting. IEEE, 2010, pp. 1–6.
it is concluded that the proposed control strategy enhances [19] W. H. Kersting, “Radial distribution test feeders,” in 2001 IEEE Power
the power quality for the balanced and unbalanced linear load Engineering Society Winter Meeting, vol. 2. IEEE, 2001, pp. 908–912.
under mismatched feeder impedance.

4th International Conference on Electrical Energy Systems (ICEES) 651

You might also like