Escolar Documentos
Profissional Documentos
Cultura Documentos
AC Grid
DC
DC
PV Array
Resonant
Inverter
Pan + Coil
S1
S2
S1
Lr
Vdc
VAB(t) Cr
S2
+
Ro Vo
-
Vdc
+
S2
S1
Cf
VCD(t) L r
Cr
Ro
Vo
-
(a)
(b)
Figure 2 Resonant inverter topologies: (a) full-bridge inverter and (b) class DE inverter
8 Vdc
Vo
2
( / o )
1 j
(1)
where Vo is the load voltage, Vdc is the dc-link voltage, Q is the quality factor of the resonant circuit, is the
angular switching frequency, and o is the angular resonance frequency.
B. Class DE Half-Bridge Inverter
The inverter topology shown in Figure 2 (b) was presented in [6], which can be promptly used in the induction
heating system. Analogously to (1), the expression for the rms load voltage is obtained as a function of the
resonance frequency, the switching frequency, and the dc-link voltage as in (2).
Cf
Vo
C f Cr
2
sin cos
j Vcc
j
1
2
2 j
Q
C
L
f r r o
f r r
(2)
where is the angle corresponding to the first harmonic component of vCD(t). According to [6], the duty cycle of
switches S1 and S2 can be fixed at 0.25, while the swirching frequency is varied.
III. Controller design
The proposed controller is based on neural networks ADALINE [7]. Each input Vdc(k) is expanded through a
tapped delay line (TDL) to provide a given number of delayed inputs, as the output fs(k) is introduced to the
input also passing through a TDL resulting in the inputs. This method is known as autoregressive model with
exogenous input (ARX), from which the output f(k) corresponding to the inverter switching frequency can be
determined to control the load voltage.
The forthcoming analysis uses time-series data to determine the structure of the mathematical model and
the parameters of the dynamic system. By taking into account a discrete-time linear system whose intervals are
set at 6 s, the angular switching frequency varies as a function Vdc, as the following discrete-time transfer
function can be obtained to describe the system in (3).
fs Z
b Z 1 b2 Z 2 . . . bm Z m
(3)
1
Vcc Z 1 a1 Z 1 a2 Z 2 . . . an Z n
Finally, by means of bilinear transform [8] [9], it is possible to represent the continuous-time transfer
function, which represents the proposed controller model for each type of resonant inverter. The transfer
functions of the controllers used in the full-bridge and class DE inverters are represented by (4) and (5),
respectively.
3867 s 2.56 105
(4)
H1s 41 2
s 5148 s 2.48 105
8607 s 3.123 107
(5)
H2 s 8 2
s 7215 s 4.012 107
IV. Simulation results
The inverter topologies are simulated in PSIM environment resulting in the waveforms presented in Figure
3 and Figure 4, where the rms load voltage is 220 V and the dc-link voltage is supposed to remain constant at
400 V. Considering that irradiance and temperature vary in Figure 1, both controllers are able to control both
aforementioned quantities as desired. The total harmonic distortion of the output voltage in Figure 3 (a) and
Figure 4 (a) is 2.89% and 4.11%, respectively.
load voltage
variation as function of
the dc-link voltage
Figure 3 Waveforms representing the operation of the class DE inverter.
load voltage
variation as function of
the dc-link voltage
Figure 4 Waveforms representing the operation of the full-bridge inverter.
Since the stability regarding the load voltage for both aforementioned inverters depends only on the
switching frequency, Figure 3 (d) and Figure 4(d) show how the switching frequency varies as a function of the
dc-link voltage. It can be seen that the standard deviation for the Class DE inverter is 450.74 kHz, while the one
of for the full-bridge Inverter is 1795.97 kHz.
In both cases, the proposed controllers are able to keep the dc-link voltage constant, whose response is
slow so that the current injected in the utility grid is not distorted. The settling time in Figure 3 (c) and Figure 4
(c) is 0.7 s and 0.9 s, respectively, thus demonstrating that the Class DE inverter response is faster. On the
other hand, overshoot in Figure 3 (c) and Figure 4 (c) is 430 V and 435 V, respectively. Even though the
topology in Figure 2 (b) presents superior performance in terms of the dc-link voltage, it can be stated that the
settling time for the load voltage in Figure 3 (b) and Figure 4 (b) is about 0.1 s and 0.4 ms, respectively.
V. Conclusion
The class DE resonant circuit shows superior stability performance when compared to the full-bridge one,
taking into account that the switching frequency standard deviation values for the Class DE inverter is smaller,
as consequence the load voltage would remained constant along the simulation.
Although the transfer functions in (4) and (5) are of second order with similar complexity, simulation results
have demonstrated the satisfactory performance of the class DE inverter, as this topology is recommended for
residential induction heating systems. The final version of this work is supposed to include further results
regarding the performance of the whole system represented in Figure 1.
VI.
REFERENCES
[1]
S. Llorente, F. Monterde, J. M. Burdio, and J. Acero, "A comparative study of resonant inverter
topologies used in induction cookers," in Applied Power Electronics Conference and Exposition, 2002.
APEC 2002. Seventeenth Annual IEEE, 2002, vol. 2, pp. 1168-1174 vol.2.
[2]
O. Godswill, O. A. Okiemute, and I. K. Charles, "Design of maximum power point tracker (MPPT) and
phase locked loop (PLL) in a PV-inverter," International Journal of Engineering Research and
Applications, vol. 2, no. 3, pp. 1014-1025, 2012.
[3]
A. Panda, M. Pathak, and S. Srivastava, "A single phase photovoltaic inverter control for grid
connected system," Sadhana, vol. 41, no. 1, pp. 15-30, 2016.
[4]
M. H. Rashid, Power electronics handbook: devices, circuits and applications. Academic press, 2010.
[5]
M. K. Kazimierczuk and D. Czarkowski, Resonant power converters. John Wiley & Sons, 2012.
[6]
T. Kondo and H. Koizumi, "Class DE voltage-source parallel resonant inverter," in Industrial Electronics
Society, IECON 2015 - 41st Annual Conference of the IEEE, 2015, pp. 002968-002973.
[7]
W. Zhang, "System identification based on an improved generalized ADALINE neural network," in 2011
Chinese Control and Decision Conference (CCDC), 2011, pp. 789-794.
[8]
D. C. Fielder, "Some Classroom Comments on Bilinear Transformations," IEEE Transactions on
Education, vol. 13, no. 2, pp. 105-107, 1970.
[9]
P. J. Pupalaikis, "Bilinear Transformation Made Easy," 2000: ICSPAT.