Professional Documents
Culture Documents
Analysis of Predictive Control For Active Power Factor Correctio
Analysis of Predictive Control For Active Power Factor Correctio
Dept. of Electrical and Computer Engineering, University of Victoria, P.O. Box 3055. Victoria, B.C. CANADA V8W 3P6
phone: (604) 721-6028. fax: (6041 721-6052, email: kimj@sirius.uvic.ca
-
Abstrmt This paper describes a new active power laetor correction is a highly nonlinear system with the gains of the power
correction method based on predictive control. The predictive circuit and the location of its poles and zeros varying periodically.
control method generates a switching pattern with constant For this reason, the optimal design of average current controller
switching frequency and very little distortion in the input without introducing distortion and displacement into the input
current waveform. The designof predictive currentcontroller is current waveform is not trivial.
very simple without involving any current loop compensator.
This paper describes a new current programming technique for
The design equations and dynamic characteristics are developed
active power factor correction using predictive control. The
for a single-phase boost converter configuration.
proposed predictive current controller generates continuous input
current waveform with very little distortion and constant switching
INTRODUCTiON frequency. The design of current loop is simple at the expense of
In order to alleviatedisadvantagesand problems of conventional introducing additional analog divider. Fig. 1 illustrates a block
addc diode bridge rectifier, modem off-line dc power supplies are diagram of active power factor c o d o n circuit with predictive
i n c o m with active power factor m d o n circuits. These are control. The circuit diagram is very similar to the average current
based on programmingthe input current of the diode bridge rectifier control with the exception of the predictive controller in place of
using high-frequency dddc convertef stage. By programming the current loop compensator for average current controller.
input current proportional to the line voltage, the power supply
becomes a simple resistive load to the utility and the input power
factor is kept near unity.
In recent years, numetous papers have been published on the L D
active power factor correction methods. describing different dddc
converter topologies used. Buck converters[l] and buck-boost
converters[2] have been used, but boost regulators[3] are the most
a,4
popular choice because the input current can be continuousfor more
desirable waveform and the lowest high-frequency ripple content. If II A h
the output voltage has to be lower than the input peak voltage or
isolationis required, the Cuk converter[4]can be used while keeping
the input characteristicsof the boost converter.
As far as the programming the input current is concerned.
U I
however, the control methods are fairly limited. The most-
commonly used method is based on variable hysteresiscontrol.[3,5] lcom Voltage
With this method,the input current is controlled to stay within limits
of a current proportional to the line voltage, using variable hysteresis I Multioher
Compensator
SI-
controL The input current has very little distortion with power factor
close to unity. The main disadvantage of this method is that the
switching frequency is not only variable but also load dependent.
The other method available in the literame is the input current
shaper using input voltage integral.[2,4] With this method, the input
current is discontinuous and the control method becomes very
simple with a constant switching frequency. However, the main
disadvantageis the high input current ripple with magnitude at least Peak Voltage
twice the average value of the input current. Sample and Hold
Recently, an average control technique with constant switching
frequency has been introduced for easier design of Eh41 filters and
better control of the average input current.[6, 71 The average current
controller involves the design of current loop compensatorwhich is
based on some linearized and averaged representationof the power Fig. 1 Single-phase addc converter with active power factor
circuit. However, the power circuit used in active power factor correction circuit
(4)
447
The correction to this increase in the input power is made at the DYNAMIC RESPONSE ANALYSIS
next sampling incident. As the new value of the peak voltage is
The analysis of small-signal dynamic response for active power
sampled. the command current decreases due to the increase in the
factor Correction circuit should be done in two parts: one for the
peak voltage sampled. The inductor current also decreases and
cumnt loop with fast dynamic response and the other for the much
counter-balances the increase in the line voltage. The maximum
slower voltage loop response.
delay between the change in the line voltage and the necessary
correction is one sampling period which is the period of the second For the current loop analysis the quasi-static analysisis used for
harmonic. Due to the slow voltage loop, this delay does not produce the dynamic response much faster than the line frequency. On the
any significant fluctuations in the output voltage. other hand, the output voltage dynamics are much slower than the
line frequency and averaged dc-equivalent model of the power
circuit is used. In the voltage loop point of view the current loop
response can be considered as a pure gain block.
4.00 2.00
N
0.00
9
a00
I
-2.00
4.00
0.00
oa
4.00
0.00
Fig. 3 Simulation results of active power factor correction with Fig.4 Response of predictive current controller to a step
predictive control increase( 10%)in the input line voltage
448
Boost Converter
The small-signal dynamic response of the boost power circuit
shown in Fig. 1 can be analyzed using its canonical small-signal
model and assuming quasi staticcondition. The canonicalmodel and
its complete signal flow diagram are illustrated in Fig. 5,where
vs 1
J(S) = -.-
Dt3 R~
e(s) =
v, . ( 1 -sL,,/R,)
- (7)
D’
( 1 + sCR,) . R,
ZO(S> = ( 1 + sCR,)
where
Le, (14
A(s) = s 2 - L , , - C + s . ( - + C . (R,,+R,)) + 1
and Le, = UZ)” and Re, = R,DJ + RL DD’. V, is the quasi-static RO
input voltage and D’ is the correspondingcomplementary duty ratio.
R, and RL are parasitic resistances in filter inductor and capacitor,
respectively. Some important transfer functions are as follows; Predictive Cuwent Controller
The small-signal dynamic behaviour of predictive current
controller can be analyzd by giving small-signal perturbation to
Eg.(5). The small-signal perturbation is given around the quasi-
static operating condition which is given by,
449
10-
0,
%
3
I
' -10 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
-20
! o o 2
m I
a, a,
2 p!
m a, 0
=a, -50
......................................
3I
I
-1 oa -200'
1oo 1o5 10" 1o5
Fig.7 Bode plot of transfer function, icon/il Fig.8 Bode plot of transfer function, vJvC
450
Vdtage Loop Design REFERENCES
The voltage loop dynamic response is much slower than the [11 M. F. Schlecht, “Novel Topological Alternatives to the Design
current loop response. For less distorted input cumnt waveform of a Harmonic-Free UtilityiDC Interface”, 1983 IEEE-PESC
the bandwidth of the voltage loop must be designed to be much Rec.. pp. 206-216
lower than the second harmonics.Due to the slow dynamic
[2] A. R. Prasad, et al., “A New Active Power Factor Correction
response, the voltage loop characteristics can be analyzed Method for Singls-Phase Buck-Boost AC/DC Converter” Fbc.
consideringthe power circuit as dc-to& converter with the input
1992 IEEE-APEC.pp. 814-820
voltage replaced by its RMS value.[6]
[3] C. &U, R. B. Ridley and F. C. Lee, ‘’Design and Analysis of a
The dc operating conditions are described in terms of input
HystereticBoost Power Factor Correction Circuit”, 1990IEEE-
R M S voltage.: PESC Rec., pp. 800-807
[4] M. Brkovic and S. Cuk, “Input Current Shaper Using Cuk
Converter”,Proc of 1992 IEEE-INTELEC, pp. 532-539
[5] C. P. H e m and N. Mohan, “A Digitally Controlled ac to dc
Power Conditionerthat Draws SinusoidalInput Current”, 1986
IEEE-PESC, pp. 531-540
451