Real-Time Implementation of ANFIS Control For Renewable Interfacing Inverter in 3P4W Distribution Network
Real-Time Implementation of ANFIS Control For Renewable Interfacing Inverter in 3P4W Distribution Network
Real-Time Implementation of ANFIS Control For Renewable Interfacing Inverter in 3P4W Distribution Network
1, JANUARY 2013
121
AbstractPower electronics plays an important role in controlling the grid-connected renewable energy sources. This paper presents a novel adaptive neuro-fuzzy control approach for
the renewable interfacing inverter. The main objective is to
achieve smooth bidirectional power flow and nonlinear unbalanced load compensation simultaneously, where the conventional
proportional-integral controller may fail due to the rapid change
in the dynamics of the highly nonlinear system. The combined
capability of neuro-fuzzy controller in handling the uncertainties
and learning from the processes is proved to be advantageous
while controlling the inverter under fluctuating operating conditions. The inverter is actively controlled to compensate the harmonics, reactive power, and the current imbalance of a three-phase
four-wire (3P4W) nonlinear load with generated renewable power
injection into the grid simultaneously. This enables the grid to
always supply/absorb a balanced set of fundamental currents at
unity power factor even in the presence of the 3P4W nonlinear
unbalanced load at the point of common coupling. The proposed
system is developed and simulated in MATLAB/SimPowerSystem
environment under different operating conditions. The digital
signal processing and control engineering-based laboratory experimental results are also provided to validate the proposed control
approach.
Index TermsDistributed generation, grid interconnection,
neuro-fuzzy control, nonlinear load, power quality, renewable
energy, unbalanced load.
I. I NTRODUCTION
Manuscript received September 18, 2009; revised April 27, 2011 and
September 14, 2011; accepted November 20, 2011. Date of publication
January 26, 2012; date of current version September 6, 2012.
M. Singh is with the Department of Electrical Engineering, Deenbandhu
Chhotu Ram University of Science and Technology, Murthal, Haryana 131039,
India (e-mail: smukhtiar_79@yahoo.co.in).
A. Chandra is with the Department of Electrical Engineering, Ecole de
technologie superieure, Universite du Quebec, Montreal, QC H3C 1K3, Canada
(e-mail: ambrish.chandra@etsmtl.ca).
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.2012.2186103
122
Fig. 2.
Fig. 3.
Fig. 1. Schematic and control description of proposed renewable-based distributed generation system.
SINGH AND CHANDRA: IMPLEMENTATION OF ANFIS CONTROL FOR RENEWABLE INTERFACING INVERTER
123
Fig. 4.
membership functions are assigned to each input. The trapezoidal and triangular membership functions are used to reduce
the computation burden as shown in Fig. 4, and their corresponding node equations are given as follows:
b1
1
a1
b
(1)
A1 () =
1 a1
b1 a1
0
a1
a2
1 0.5b
| a2 | 0.5b2
2
A2 () =
(2)
0
| a2 | 0.5b2
a3
0
a3
A3 () =
a
(3)
3 b3
b3 a3
1
b3
where the value of the parameters {ai , bi } changes with the
change in error and accordingly generates the linguistic value of
each membership function. Parameters in this layer are referred
as premise parameters or precondition parameters.
Layer 2: Every node in this layer is a circle labeled as
which multiplies the incoming signals and forwards it to the
next layer
i = Ai (1 ) Bi (2 ) ,
i = 1, 2, 3.
(4)
(7)
124
Fig. 6. Simulation results: Phase a grid voltage, grid current, load current,
inverter current, load neutral current, and inverter neutral current.
Fig. 7. Simulation results. (a) Phase a grid voltage and current. (b) Load
current. (c) Inverter current.
Fig. 5. Simulation results. (a) Grid voltages. (b) Grid currents. (c) Unbalanced
load currents. (d) Inverter currents.
SINGH AND CHANDRA: IMPLEMENTATION OF ANFIS CONTROL FOR RENEWABLE INTERFACING INVERTER
Fig. 8. Experimental results: (a) Grid currents and (b) inverter currents, just
before and after compensation.
125
Fig. 9. Experimental results: Inverter performance under the renewable interfacing mode of operation.
126
TABLE I
I NVERTER P ERFORMANCE AS A C OMPENSATING D EVICE
A PPENDIX I
S YSTEM PARAMETERS
Three-phase supply (rms)
: Vg = 30 V, 60 Hz
Three-phase nonlinear load
: R = 26.66 , L = 10 mH
One-phase linear load(AN)
: R = 26.66 , L = 10 mH
One-phase nonlinear load(BN) : R = 56 , L = 10 mH
DC-link capacitance and voltage : Cdc = 3000 F, Vdc = 75 V
Coupling inductance
: Lsh = 2.0 mH
Source impedance ratio
: X/R = 7
A PPENDIX II
it can be noticed that the dc-link voltage is almost constant
irrespective of any kind of variation in injected inverter current.
A comparative table showing the total harmonic distortions
(THDs) and unbalance factor (UF) before and after compensation is given in Table I, where the percentage UF is calculated
separately for each phase using
%U Fabc =
|iabc iavg. |
100.
iavg.
(8)
Vdc )2 .
2 = (Vdc
Precondition Parameter Tuning: The precondition parameters are required to update the fuzzy membership functions as
discussed in the previous section for Layer1. To minimize the
error function 2 by the gradient descent method, the change in
each precondition parameter must be proportional to the rate of
change of the error function w.r.t. that particular precondition
parameter, i.e.,
aAi =
V. C ONCLUSION
This paper has presented a novel adaptive neuro-fuzzy
control algorithm for the renewable interfacing inverter. The
controller works satisfactorily under the dynamic operating
conditions. It has also been shown that the inverter is able
to perform all the duties of the shunt APF while maintaining the smooth bidirectional power flow simultaneously. The
simulation results supported by the experimental results are
provided to validate the fact that the renewable interfacing
inverter can act as a multioperation device in order to utilize
its maximum rating. The current unbalance, current harmonics,
and load reactive power demand of an unbalanced nonlinear
load at PCC are compensated effectively such that the gridside currents are always maintained as a balanced set (0% UF)
of sinusoidal current (2.7% THD) at UPF. Moreover, the load
neutral current is restricted to flow toward the grid side (almost
zero) by supporting it locally from the fourth leg of the inverter.
When the power generated from the renewable is more than
the total load power demand, the grid-interfacing inverter with
the proposed control approach successfully fulfills the total
load demand (active, reactive, and harmonics) and delivers the
remaining active power to the main grid at UPF operation.
(A1)
2
aAi
= 1, 2, 3
(A2)
where is the constant of proportionality defined as the learning rate. Therefore, the new value of the consequent parameter
is given as
aAi (n + 1) = aAi (n) + aAi ,
i = 1, 2, 3
(A3)
i = 1, 2, 3.
(A4)
or
aAi (n + 1) = aAi (n)
2
,
aAi
aA1
Vdc i
aA1
A1
1
d
(A5)
where
2
= 2 (Vdc
Vdc ) = 2
Vdc
Vdc
=J
i
d
(A5a)
(A5b)
SINGH AND CHANDRA: IMPLEMENTATION OF ANFIS CONTROL FOR RENEWABLE INTERFACING INVERTER
i
d
=
( f1 + 2 f2 + 3 f ) = f1
(A5c)
1 1 1
1
A1
A2 + A3
=
=
A1 A1 A1 + A2 + A3
(A1 + A2 + A3 )2
A2
1
=
(A1 + A2 + A3 ) (A1 + A2 + A3 )
A3
+
(A1 + A2 + A3 )
2 + 3
(A5d)
=
A1 + A2 + A3
aA1
A1
bA 1
=
=
aA1 aA1 bA1 aA1
(bA1 aA1 )2
bA1 + aA1 aA1
=
(bA1 aA1 )2
aA1
bA1 aA1
1
=
(bA1 aA1 ) (bA1 aA1 ) (bA1 aA1 )
A1 1
(A5e)
=
bA1 aA1
where J is Jacobian matrix, which can be taken as constant,
being a single-inputsingle-output ANFIS architecture, and can
be included in the learning rate. In computing all the terms of
(A5) and putting in (A4), we can find the updated value of the
parameter aA1 as follows:
aA1 (n + 1) = aA1 (n) + 2 (n) f1 (n)
A1 (n) 1
2 (n) + 3 (n)
. (A6)
A1 (n)
2 (n) + 3 (n)
. (A7)
A1 (n) + A2 (n) + A3 (n) bA1 (n) aA1 (n)
In the same manner, the precondition parameters for the remaining fuzzy membership functions can be derived as follows:
127
Consequent Parameter Tuning: To tune the consequent parameters as discussed in Layer 4, the following updated laws
are developed:
2
,
a0i
2
a1i (n + 1) = a1i (n) c
,
a1i
i = 1, 2, 3
(A11)
i = 1, 2, 3
(A12)
d
,
a0i
Vdc id fi a0i
2
2 Vdc i
fi
=
d
,
a1i
Vdc id fi a1i
i = 1, 2, 3
(A13)
i = 1, 2, 3.
(A14)
i = 1, 2, 3
(A15)
(A16)
(A17)
, i = 1, 2, 3
(A18)
A1 + A2 + A3
a1i (n + 1) = a1i (n) + 2 c
i
, i = 1, 2, 3. (A19)
a1 + a2 + a3
1 (n) + 3 (n)
A1 (n) + A2 (n) + A3 (n)
1 A2 (n)
bA2 (n)
R EFERENCES
(A8)
1 (n) + 2 (n)
A1 (n) + A2 (n) + A3 (n)
A3 (n) 1
bA3 (n) aA3 (n)
(A9)
1 (n) + 1 (n)
A1 (n) + A2 (n) + A3 (n)
A3 (n)
.
bA3 (n) aA3 (n)
(A10)
128