Modeling Under MATLAB by ANFIS of Three-Phase Tetrahedral Transformer Using in Microwave Generator for Three Magnetrons Per Phase

TELKOMNIKA, Vol. 16, No. 5, October 2018, pp.2406~2414
ISSN: 1693-6930, accredited First Grade by Kemenristekdikti, Decree No: 21/E/KPT/2018
DOI:10.12928/TELKOMNIKA.v16i5.10003  2406
Received May 25, 2018; Revised August 15, 2018; Accepted September 8, 2018
Modeling Under MATLAB by ANFIS of Three-Phase
Tetrahedral Transformer Using in Microwave Generator
for Three Magnetrons Per Phase
Mouhcine Lahame*
1
, Mohammed Chraygene
2
, Hamid Outzguinrimt
3
, Redouane Batit
4
,
Rajae Oumghar
5
, Mohamed Ferfra
6
1,2,3,4,5
Materials, Systems and Information of Technology Team (MSTI), High School of technology, Ibn
Zohr University, BP 33/S 80000 Agadir, Morocco
6
Research Team in Power and Control, Mohammadia’s School of Engineering, Mohamed University, BP:
765, Ibn Sina, Agdal-Rabat, Morocco
*Corresponding author, e-mail: mouhcine.lahame@edu.uiz.ac.ma
Abstract
This work deals with the modeling of a new three-phase tetrahedral transformer of HV power
supply, which feeds three magnetrons per phase. The design of this new power supply is composed of
three single-phase with magnetic shunt transformers coupling in star; each one is size to feed voltage-
doubling cells, thereby feeds a magnetron. In order to validate the functionality of this power supply, we
simulate it under Matlab-Simulink environment. Thus, we modeled nonlinear inductance using a new
approach of neuro-fuzzy (ANFIS); this method based on the interpolation of the curve B(H) of
ferromagnetic material, the results obtained gives forms of both voltages and currents, which shows that
they are in accordance with those of experimental tests, respecting the conditions recommended by the
magnetron manufacturer
Keywords: modeling, tetrahedral, three magnetrons, matlab-simulink, neuro-fuzzy (ANFIS)
Copyright © 2018 Universitas Ahmad Dahlan. All rights reserved.
1. Introduction
Current power supply systems for microwave application use a single-phase
transformer to feed one magnetron, so to meet certain high energy needs, it is necessary to
have several single-phase power supplies; The development and innovation of these systems
consists tofinding more efficient and optimized solutions that can feed several magnetrons at a
time, to reduce cost, volume and ease of maintenance.
In order to provide these criteria, in this article we modeled a new Three-phase HV
power supply, for three magnetrons per phase as shown in Figure 1. This model is a design of
tetrahedral magnetic shunt type transformer, which sized to feed three double cells each one is
composed of a capacitor and a diode; the doublers feed their turns a single magnetron [1-6].
The advantage of this coupling is to assure the stability of the current in each magnetron
respecting the characteristics imposed by the manufacturer. Unlike the ordinary transformer, the
leakage flux in the shunts is the same order as that in primary and secondary core.
The modeling of this power supply is essentially to model its own special shunt
transformer. It seems necessary to develop a new model in appropriate π-model, formed of
non-linear passive electrical components (inductance) [7-9]. We take into account the actual
state of saturation of the transformer, during its nominal operation in the supply circuit
magnetron HV 800 Watts at 2450 MHz.
This work is divided into two parts; first part presents the equivalent model, and in the
second part, we implemente and simulate this model, based on fuzzy logic network (ANFIS)
available in library of Matlab-Simulink [15-19], which interpolate the curve B(H) of material used
SF19 to model the different non-linear inductances. Thus, the simulation results obtained were
comparing with those of the experimental tests by checking the regulation process of the current
in each magnetron.

TELKOMNIKA ISSN: 1693-6930 
Modeling Under MATLAB by ANFIS of Three-Phase Tetrahedral Transformer… (Mouhcine L.)
2407
Phase 3
Phase1
Phase 2
Phase 2 Phase 3
Phase1
Three phase
source
1
2
3 Primary Secondary
D2
C3
D1
C1
D5
Voltage
doubler
Magnetron 5
WaveguideC5
D6
C6
D7
Voltage
doubler
Magnetron 7
Waveguide C7
D8
C8
Voltage
doubler
Magnetron 6
Waveguide
Voltage
doubler
Magnetron 8
Waveguide
Voltage
doubler
Magnetron 1
Waveguide
Voltage
doubler
Magnetron 2
Waveguide
D3
C3
Voltage
doubler
Magnetron 3
Waveguide
D4
C4
Voltage
doubler
Magnetron 4
Waveguide
D9
C9
Voltage
doubler
Magnetron 9
Waveguide
Figure 1. Three-phase transformer HV power supply for three magnetrons per phase
2. Modeling of Three-Phase Transformer Power Supply for Three Magnetrons per Phase
used in Industrial Microwave
Figure 2 shows the structure of a three-phase tetrahedral shunt transformer, which
consists in modeling a magnetic shunt transformer adapted to the constraints imposed by the
manufacturer for the operation of these magnetrons.
The aim of this section is to model each transformer composed of three-limb, the
windings (primary and secondary) wound around the midlle limb, and the outer limb used to
close the magnetic circuit. Magnetic shunts integrated to deflect a portion of the flow between
the two windings, in order to adjust the anode current in the magnetron.
Figure 2. Three-phase magnetic tetrahedral shunt transformer structure
Here are the different dimensions of a phase the tetrahedral transformer.

 ISSN: 1693-6930
TELKOMNIKA Vol. 16, No. 5, October 2018: 2406-2414
2408
- The width of the non-wound core: mm75a 
- The width of the magnetic circuit: mm20b 
- Number of stacked sheets of the shunt: 18n3 
- Number of turns in the primary: 224n1 
- Number of secondary turns: 2400n2 
- Height of the sheet stack of shunts: 3n5.0h 
- Surface of the core: baSS 21 
- Thickness of the air gap: mm75.0e 
Figure 3. Geometry of a phase the tetrahedral transformer
Taking into account the saturation of the material in the modeling, we consider that the
leakage flux in gap air is negligible compared to the flow in the shunts. We apply the two-ohm’s
law and Hopkinson to define different electrical and magnetic equations of each phase:
i11i
1i
2i1 ir+
dt
d
nu

 (1)
i22i
2i
2i2 ir-
dt
d
nu

 (2)
i22i11i2sii1pi ininR+R  (3)
i11i3shii1pi inR+R  (4)
i22i3shii2si inRR  (5)
i3i2i1 +  (6)
The equations obtained present a model of a quadruple in π-model referred to the
secondary, by multiplying the equations (1) (4) (5) (6) in a transformation ratio (m). The
advantage of this model, which contains non-linear inductances (primary, secondary and
shunts) for each phase presented in Figure 4, is to model each of the inductances associated
with each portion of the transformer, which is impossible in case of a T-model [7].
dt
)iL(d
iru
'
pi
'
pi'
i1
'
1i
'
i1  (7)
dt
)iL(d
ir-u sisi
i22ii2  (8)
sii2
'
shi iii  (9)
'
shi
'
pi
'
i1 iii  (10)
dt
)iL(d
dt
)iL(d
dt
)iL(d '
shi
'
shisisi
'
pi
'
pi
 (11)
3. Simulation and Expermental Results
We integrate under Matlab-Simulink the equivalent electric model of tetrahedral
transformer in the global circuit of the Three-phase HV power supply. Figure 4 present the three

2409
magnetrons of each phase by its equivalent diagram, composed by a diode connected in series
with a dynamic resistance of 350 ohms, and a voltage source of a threshold E of 3800 volts.
The different nonlinear inductances studied are represented by the characteristic
resulting from the relation,   i)i(niL 2  that can be determined from the magnetization
curve B(H) of the material, and the geometrical dimensions of the transformer [11-14].
The relations characterizing these inductances are:
 SBnin 22 
 
2n
H
i


(i’sh3)f
i’ r’31 i r
u’ u
(ish3)e
(L’sh3)e
(L’sh3)f
Magnétron 8
Magnétron 7
3800 V
3800 V
C8
D8
D7
C7
Source
Triphasée
1
2
3
Primaire Secondaire
Phase1
Phase1
L’ L
i’p33 iS33
S33P33
Magnétron 9
3800 V
C9
D9
(i’sh3)f
i’ r’ i r
u’ u23
(i’sh3)e
(L’sh3)e
(L’sh3)f
Magnétron 5
Magnétron 4
3800 V
3800 V
C5
D5
D4
C4
L’ L
i’p23 iS23
S23P23
Magnétron 6
3800 V
C6
D6
(i’sh3)f
i’ r’31 i r
u’ u13
(i’sh3)e
(L’sh3)e
(L’sh3)f
Magnétron 2
Magnétron 1
3800 V
3800 V
C2
D2
D1
C1
L’ L
i’p13 iS13
S13P13
Magnétron 3
3800 V
C3
D3
23
23 23 23 23
13
13
13
33
3333
33
33
13
Figure 4. Global equivalent model of eletrical circuit for three-phase transformer power supply
three magnetron per phase
These relations allowing obtaining a subsystem as shown in Figure 5 contain:
1. A voltage integrator to derive the flow  as a function of time to deduce the magnetic field B
using the relation  SBnin 22 
2. A Simulink block called Fuzzy Logic Controller Block that implements a fuzzy inference
system (FIS) in Simulink, to interpolate the B(H) curve.
)i(

 ISSN: 1693-6930
2410
3. A current source imposed to derive the current from the relation
2n
).H(
i


Applying a method based on using of neurofuzzy inference (ANFIS) to interpolate the
B(H) curve, taking into account the saturation phenomenon. We give the architecture in five
layers (fuzzification, Multiplication of the entries between them, Normalization of the input data,
Adaptive layer with modulation of the weights, recombination of the outputs) of the ANFIS model
presented in Figure 6, making it possible to build a fuzzy system thanks to neural networks.
Figure 5. Subsystem of a nonlinear inductance
To develop the ANFIS model, we follow these steps [17].
a. Step 1: The definition of the inputs/outputs of the ANFIS model, which are the parameters
nB and nH of the material used for the construction of our transformer.
b. Step 2: Find fuzzy partitions of input data by ANFIS In this step; we choose the Grid partition
technique to generate the initial model in which we define the number and type of
membership function for the entrees.
c. Step 3: Training input/output data via the neural network, choosing a hybrid algorithm with
the following conditions, an interaction number of 300 and an error (tolerance) of 0.005.
Once the network is traine, we test the system by different data values to check its good
functionality. The Figure 7 shows the fuzzy neuro network setup and training window
(ANFIS) in which we visualize the interpolation results of the B(H) curve.
Figure 6. Architecture of the ANFIS model
+
-
V 1/S x
÷ x S
-
n2.S l÷n2
Input
OutputVoltage
measurement
Integrator
Divide Product
Current
source
Fuzzy logic controller
Ф B H I

2411
Figure 7. Setup and training window of neuro-fuzzy network
After having executed and tested the ANFIS model, we load the results obtained under
the "fuzzy logic controller" block present in the model of the non-linear inductance in Figure 5.
To simulate our power supply under Matlab-Simulink.
0,60 0,61 0,62 0,63 0,64
-1,25
-1,00
-0,75
-0,50
-0,25
0,00
0,25
0,50
0,75
1,00
1,25
Magnétron 1
Magnétron 2
Magnétron 3
Magnetronscurrents(A)
Time (s)
0,60 0,61 0,62 0,63 0,64
-6000
-4000
-2000
0
2000
4000
6000
Magnetronsvoltages(V)
Time (s)
Magnétron 1
Magnétron 2
Magnétron 3
0,60 0,61 0,62 0,63 0,64-1,25
-1,00
-0,75
-0,50
-0,25
0,00
0,25
0,50
0,75
1,00
1,25
Magnétron 1 Magnétron 2 Magnétron 3
Diodescurrents(A)
Time (s)
0,60 0,61 0,62 0,63 0,64
-6000
-4000
-2000
0
2000
4000
6000
Capacitorsvoltages(V)
Time (s)
Figure 8. Simulation waveforms of theoretical currents and voltages obtained

 ISSN: 1693-6930
2412
0,60 0,61 0,62 0,63 0,64
-3
-2
-1
0
1
2
3
Secondaryscurrents(A)
Time (s)
0,60 0,61 0,62 0,63 0,64
-6000
-4000
-2000
0
2000
4000
6000
Secondarysvoltages(V)
Time (s)
Figure 8. Simulation waveforms of theoretical currents and voltages obtained
Figure 9. Experimental curves of the voltage and current waveforms of the magnetron HV
supply
4. Results and Discussions
After having simulated our model under Matlab-Simulink, the time curves of currents
and voltages (secondary, magnetrons and capacitor) represented in Figure 8 are in agreement
with those obtained by experimental test Figure 9 of a single-phase power supply [2-5], which
presents a phase among the Three-phase of our model. The error rate between the theory and
the experiment is of the order of 6%.
To validate the operation of the model under Matlab-Simulink with respecting the
maximum current 1.2A and an average current of 300mA, for the nominal operation of
magnetron. During this simulation, we checked the current stabilizing effect in each magnetron
for a variation of ± 10% of the primary voltage around its nominal value 220V as shown in
Figure 10. As a result, for both 200V and 240V cases, the maximum amplitude and the average
value of the current in each magnetron do not exceed the imposed limit.

2413
0,60 0,61 0,62 0,63 0,64
-1,25
-1,00
-0,75
-0,50
-0,25
0,00
0,25
0,50
0,75
1,00
1,25
Magnétron 1
Magnétron 2
Magnétron 3
Time (s)
200 V
0,60 0,61 0,62 0,63 0,64
-1,25
-1,00
-0,75
-0,50
-0,25
0,00
0,25
0,50
0,75
1,00
1,25
Magnétron 1
Magnétron 2
Magnétron 3
Temps (s)
240 V
Figure 10. Stabilization of the anode current in each of the three magnetrons with respect to the
variations of the source voltage (± 10%)
5. Conclusion
In this article we studied the validity of a new three-phase HV power supply with
tetrahedral transformer feeding three magnetrons per phase, this study was simulated using a
new method of interpolation of the curve B(H) based on fuzzy standards (ANFIS) to model the
nonlinear inductance under Matlab-Simulink. The different curves (voltages, currents) obtained
compared with those of experimental shows a coordination between them.
The modeling of the power supply makes it possible to validate the constraints imposed
by the manufacturer on the level of the magnetron current due to the effect of saturation of the
transformer, which ensures the protection of the magnetron against any variation of the input
voltage. For perspective of this work, we study the optimization in order to reduce the volume of
this transformer, and consequently gain of cost and congestion.
References
[1] Theocharis AD, Milias-Argitis J, Zacharias T. Single-phase transformer model including magnetic
hysteresis and eddy currents. Electrical Engineering. 2008; 90(3): 229-241.
[2] El Ghazal N, Chraygane M, Ferfra M, Belhaiba A, Fadel M, Bahani B. Modeling of New Three-phase
High Voltage Power Supply for Industrial Microwave Generators with One Magnetron per phase.
International Journal of Electrical and Computer Engineering (IJECE). 2013; 3(2): 270-278.
[3] Bassoui M, Ferfra M,Chraygane M, Ould Ahmedou M, Elghazal N, Bahani B. Modeling of new high
voltage power supply with threephase character for microwaves generators with one magnetron by
phase under MATLAB SIMULINK code. ARPN J Eng Appl Sci. 2014; 9(12): 2559-68.
[4] Bahani B, Bouzit A, Chraygane M, Ferfra M, El Ghazal N, Belhaiba A. Modeling of a New High
Voltage Power Supply for Microwave Generators with Three Magnetrons. International Journal of
Electrical & Computer Engineering (IJECE). 2013; 3(2): 2088-8708.
[5] Batit R, Chraygane M, Ferfra M, Fadel M, Bahani B :Failures’ study of a new character three-phase
high voltage supply for industrial microwave generators with one magnetrons per phase. Journal Of
Engineering Science And Technology Review (JESTR). 2016; 6(3): 1248-59.
[6] Lind M G, Xiao W, Dunford W G. Modeling of a constant voltage transformer. IEEE Transactions on
Circuits and Systems I: Regular Papers. 2006; 53(2): 409-418.
[7] De Leon F, Farazmand A, Joseph P. Comparing the T and π Equivalent Circuits for the Calculation of
Transformer Inrush Currents. IEEE Transactions on Power Delivery. 2012; 27(4): 2390-2398.
[8] David Greene J, Gross C.A., Non linear modelling of transformers. IEEE transactions On Industry
Applications.1988; 24(3): 434-438.
[9] Ferfra M, Chraygane M, Fadel M, Ould Ahmedou M. Non-Linear Modelling of an Overall New High
Voltage Power Supply for N=2 Magnetrons for Industrial Microwave Generators. Physical and
Chemical News. 2010; 54: 17-30.
[10] Vakilian M, Degeneff RC. A method for modeling nonlinear core characteristics of transformers during
transients. IEEE transactions on power delivery. 1994; 9(4): 1916-1925.
[11] Liu G, Xu X. Improved modeling of the nonlinear B-H curve and its application in power cable
analysis. IEEE Trans Magn. 2002; 38(4): 1759-63.

 ISSN: 1693-6930
2414
[12] Zhao Z, Liu F, Cheng Z, Yan W, Liu L, Zhang J, Fan Y. Measurements and Calculation of Core-
Based B-H Curve and Magnetizing Current in DC-Biased Transformers. IEEE Transactions on
Applied Superconductivity. 2010; 20.3: 1131-1134.
[13] Kawkabni B. Simond J J.Improved modeling of three-phase transformer analysis based on magnetic
equivalent circuit diagrams and taking into account nonlinear BH curve. In Electromotion 2005.
2005; (No. LME-CONF-2008-034).
[14] Hwang J H, Lord W. Exponential series for B/H curve modelling. In Proceedings of the Institution of
Electrical Engineers.1976; 123 (6): 559-560.
[15] Tutmez B, Hatipoglu Z,Kaymak U ,Modelling electrical conductivity of groundwater using an adaptive
neuro-fuzzy inference system. Comput. Geosci. 2006; 32: 421-433.
[16] Jang JSR, SunC, Mizutani E, Neuro-Fuzzy and Soft Computing: A Computational Approach to
Learning and Machine Intelligence. Englewood Cliffs, NJ: Prentice-Hall. 1997.
[17] Bassoui M, Ferfra M, Chraygane M. Improved modeling of a single-phase high voltage power supply
for microwave generators for one magnetron. J Electr Eng. 2016; 16(4).
[18] Buragohain M, Mahanta C. (2008). A novel approach for ANFIS modelling based on full factorial
design. Applied soft computing. 2008; 8(1): 609-625.
[19] Kisi O, Suspended sediment estimation using neuro-fuzzy and neural network approaches.
Hydrological Sciences Journal. 2005; 50(4): 683-696.

Modeling Under MATLAB by ANFIS of Three-Phase Tetrahedral Transformer Using in Microwave Generator for Three Magnetrons Per Phase

More Related Content

What's hot (19)

Similar to Modeling Under MATLAB by ANFIS of Three-Phase Tetrahedral Transformer Using in Microwave Generator for Three Magnetrons Per Phase (20)

More from TELKOMNIKA JOURNAL (20)

Recently uploaded (20)

Modeling Under MATLAB by ANFIS of Three-Phase Tetrahedral Transformer Using in Microwave Generator for Three Magnetrons Per Phase