State Feedback Control of DC-DC Power Converters
Uploaded by
santhoshkumar.eeeState Feedback Control of DC-DC Power Converters
Uploaded by
santhoshkumar.eeeREPUBLIQUE ALGERIENNE DEMOCRATIQUE ET POPULAIRE
MINISTERE DE L’ENSEIGNEMENT SUPERIEUR ET DE LA RECHERCHE SCIENTIFIQUE
UNIVERSITE LARBI BEN M’HIDI OUM EL BOUAGHI
FACULTE DES SCIENCES ET DE LA TECHNOLOGIE
DEPARTEMENT DES SCIENCES ET DE LA TECHNOLOGIE
Mémoire présenté pour l’obtention du diplôme de Master
en Informatique Industrielle
THEME
State Feedback Control of DC-DC
Power Converters
Présenté par: Encadré par :
Lalouani Abderehmane Dr. Goléa Noureddine
Promotion 10/11
ACKNOWLEDGEMENTS
It is my great pleasure to thank the many people who enabled me to perform such piece of work.
First, I am extremely grateful to my supervisor Professor N. Goléa, it was an ever such a
great honor being so fortunate enough to work under his supervision, I would love to have the
grace to thank him for his intelligent outstanding guidance, support and patience throughout the
course of this work. His enthusiasm, commitment, able guidance, knowledge and creative thinking
have been invaluable source of inspiration and motivation for me during my entire master’s course.
It made me feel very comfortable when I knew where to find an answer for all my questions.
Sincere appreciation also extends to Mr. Bara and Mr. Barkat, the honor the jury
members thank you. My special thanks to all my teachers during the last five years of study
especially within the Master course level. Their kindness has made my study in the M I.I
program an enjoyable one.
My special thanks to the real friend who walks in when the rest of the world walks out and to all
my fellow students for their dazzling sense of humor and respect have made such a wonderful time
of joy, to Mohamed, Adel, Rafik, Imed, Yacine, Omar, Nacer, Havith… thanks to
every one of you each one by his name.
To my lovely dear parents who made me who I am to day, to my mother for her love that
reckons hours for months, and days for years and very little absence of her love is an age to me, to
my father for his support, for his care and his belief without which none of this would have been
possible. To all my dear sisters and to all my adorable nieces and dear nephews, to my brothers:
Fiyçal, Yaçine and Takiéddinne. For their support and encouragement during the entire
course of my graduate study
i
Abstract
This work investigates the state feedback control of certain DC-DC power converters. As a first step,
the models of the considered converters are established; their properties are analyzed in both dynamic
and static regimes. As a second step, the linear state feedback, based on the poles placement, is
designed and tested on the modelled DC-DC converters. For the case of unavailable states, a linear full
order observer is introduced to estimate the missing variables. As a third step, the passivity based
control (PBC) is considered. The PBC exploits the energetic structure of the DC-DC converters to
achieve the desired performance. Also, a passivity based observer is introduced to estimate the
unmeasured variables. The simulations tests were conducted in the PSIM software environment. The
obtained results were effective and confirm the theoretical predictions.
ii
Contents
Contents
INTRODUCTION ...........................................................................................................1
CHAPTER 1: OPEN-LOOP MODELING OF DC-DC CONVERTERS…................. 2
1.1 Introduction......................................................................................................2
1.2 Buck Converter….............................................................................................3
1.2.1 Switched Converter Model.................................................................3
1.2.2 Average Model....................................................................................4
1.2.3 Equilibrium Point................................................................................4
1.2.4 Desired Equilibrium Point..................................................................5
1.3 Boost Converter…............................................................................................5
1.3.1 Switched Converter Model.................................................................6
1.3.2 Average Model….................................................................................6
1.3.3 Equilibrium Point…............................................................................6
1.3.4 Desired Equilibrium Point..................................................................7
1.3.5 Linearization….....................................................................................7
1.4 Buck Boost Converter....................................................................................8
1.4.1 Switched Converter Model.................................................................9
1.4.2 Average Model….................................................................................9
1.4.3 Equilibrium Points…...........................................................................9
1.4.4 Desired Equilibrium Point................................................................10
1.4.5 Linearization.....................................................................................10
1.5 Cuk Converter…............................................................................................11
1.5.1 Switched Converter Model...............................................................12
1.5.2 Average Model…...............................................................................13
1.5.3 Equilibrium Point..............................................................................13
1.5.4 Desired Equilibrium Point................................................................13
1.5.5 Linearization......................................................................................14
1.6 Zeta Converter................................................................................................15
1.6.1 Switched Converter Model...............................................................15
iii
Contents
1.6.2 Average Model...................................................................................16
1.6.3 Equilibrium Point...............................................................................17
1.6.4 Desired Equilibrium Points...............................................................17
1.6.5 Linearization.......................................................................................17
1.7 Conclusion.....................................................................................................19
CHAPTER 2: LINEAR STATE FEEDBACK OF DC-DC CONVERTERS.…………..20
2.1 Introduction……………………………………………………………………………………………….20
2.2 Poles Placement………………………………………………………………………………….…….20
2.2.1. Substitution………………………………………………………………………...….…….21
2.2.2. Controllable Form…………………………………………………….......…….……..21
2.2.3. Ackermann Formula………………………………………………….....…….……….21
2.3 State Feedback with Integral Action……..…………………………………………………22
2.4 Observer Based State Feedback…...………………………………………………………….22
2.4.1 Introduction………….……………………………………………………………………….22
2.4.2. Substitution………………………………………………………………………………..23
2.4.3. Observable Form……..................………………………………………………..24
2.4.4. Ackermann Formula……………………………………………………………….24
2.5 Application…………………………………………………………………………..…………………..25
2.5.1 Buck Converter……………………………………………………..………………………27
a. State Feedback……………………………………………………………………..28
b. State Feedback with Integral Action………………..………………….29
c. Observer Based State Feedback Control….………..……………….30
2.5.2 Boost Converter……………………………………………………………..…………….31
a. State Feedback Control…….…………………………………….……………32
b. State Feedback with Integral Action……….…….…………………….33
c. Observer Based State Feedback Control….…………..………………35
2.5.3 Buck Boost Converter…………………….……………………………..…………….36
a. State Feedback Control…………..……………………………..……………..37
b. State Feedback with Integral Action…….………………….………….38
iv
Contents
c. Observer Based State Feedback Control…….……………….……….39
2.5.4 Cuk Converter………………………………………………….……………….………….40
a. State Feedback Control……………………………….………………….……41
b. State Feedback with Integral Action………….…………….………….43
c. Observer Based State Feedback Control …..…………….….……….44
2.5.5 Zeta Converter…………………………………………………………………….…………46
a. State Feedback………………………………………………………………………47
b. State Feedback with Integral Action…………………………………….48
c. Observer Based State Feedback Control …0..……………………….50
2.6 Conclusion……………………………………………………………………………………………….52
CHAPTER 3: PASSIVITY BASED CONTROL OF DC-DC CONVERTERS……….53
3.1 Introduction……………………………………………………………………………….……….53
3.2 DC-DC Converters Energetic Models…………………………………………….53
3.3 Control Design……..…………………………………….…………………………………….53
3.4 Non Linear Observer………………………………………………………….……………..55
3.5 Application………………………………………………………………………….…………….56
3.5.1 Buck Converter………………………………………………………….………….57
a. Passivity Based Control…………………………………….…………..57
b. Non Linear Observer….…….……………………………….…………..58
3.5.2 Boost Converter……………………….…………………………………………..59
a. Based Passivity Control….……………………………………………..59
b. Non Linear Observer….….………………………………………………61
3.5.3 Buck Boost Converter…………….…………………………………………….63
a. Passivity Based Control.…….……………………………………………63
b. Non Linear Observer...…………………………………………………….64
3.5.4 Cuk Converter…………….………………………………………………………….66
a. Passivity Based Control………………………………………………….66
b. Non Linear Observer……..……………………………………………….68
3.5.5 Zeta Converter……………………………………………………………………….70
a. Passivity Based Control………………………………………………….70
v
Contents
b. Non Linear Observer…….………………………………………………….73
3.6. Conclusion………………………………………………………………………………………………….75
CONCLUSION…………………………………………………………………………………………………………..76
BIBLIOGRAPHY……………………………………………………………………………………………………….77
vi
Introduction
Introduction
Power supply technology is an enabling technology that allows us to build and operate
electronic circuits and systems. All active electronic circuits, both digital and analog, require
power supplies. Many electronic systems require several DC supply voltages. Power supplies
are widely used in computers, telecommunications, instrumentation equipment, aerospace,
medical, and defense electronics. An DC supply voltage is usually derived from a battery or
an ac utility line using a transformer, rectifier, and filter. The resultant DC voltage is not
constant enough and contains a high ac ripple that is not appropriate for most applications.
Voltage regulators are used to make the DC voltage more constant and to attenuate the ac
ripple [1].
Power electronics devices are physical devices that can be mathematically modeled as
controlled dynamic systems and, hence, they are suitably conformed for the application of
existing control theories. Specifically, control theory is mainly concerned in the design of the
regulating subsystem in a power electronics device for enhancing its overall performance in
accordance with the prescribed objective. Although difficult, the objectives behind the design
of a certain power electronics device can usually be translated into a rather concrete “control
objective” for which an arsenal of techniques exist nowadays [2-3].
The basic aim of this work is to present two control techniques relevant to the design of
feedback controllers for DC-DC power converters, namely, the poles placement based state
feedback and passivity based control.
The present manuscript is organized as follows.
1. In the first chapter, we develop the switched models of the five DC-DC converters
under study. Since we are using continuous-time control techniques, the switched
models are converted into continuous averaged models. The study of the steady state
averaged dynamic permits the assessment of the converters input–output behaviors.
Linearized models are established and their proprieties analyzed for the linear state
feedback purpose.
2. In chapter two, the linearized models, from the first chapter, are used for the design of
linear state feedback control of the DC-DC converters, in order to achieve a particular
desired behavior. To guarantee a zero steady state error, we introduce an integral action,
which will work out this problem by assuring that the steady state error will end up to
zero. If it is supposed that both the voltage(s) and current(s) are measured, so much more
sensors are needed then and consequently causing a high cost, so that to estimate the
current(s) with a low cost and less complexity it is preferred to introduce a state
observer, considering that the voltage(s) is the only measurable variable(s).
3. The passivity based control (PBC) approach has been also applied to DC-DC
converters, and then a passivity based non-linear observer will be used for the estimation
of the state variables. The PBC exploits useful energetic properties of the DC-DC
converters to get a simpler and stable control structures.
The conclusion gives some final remarks on the realized work.
1
Chapter 1 Open-loop modeling of DC-DC converters
Chapter 1
Open Loop Modeling of DC-DC Converters
1.1. Introduction
A fundamental step for the implementation of control techniques is the representation of
converter in a form of dynamic system, with an appropriate model for the proposed control
technique. This chapter will be focused on the modeling of the DC-DC converters for the
application of the selected control techniques. In what follows, it is assumed that the DC-DC
converters operate in the continuous conduction mode, i.e. neither of the inductor currents are
identically zero on an open interval of time.
The modeling has to go through the following steps:
1. Determine the switched or the topological converter model. The switched model usually
takes the following form:
x f (x , u )
y Cx
With u 0,1 is the switching function (or switching functions, according to the number of
the independent switches), x Rn is the state vector (composed of the inductors currents and
capacitors voltages), and y is the output (or outputs) that will be controlled.
2. Obtain the average model (that is almost always non-linear). This is realized by replacing
the switching function u by its average value uav 0,1 , which yields the average model as
follows:
x f (x , uav )
y Cx
3. Define the constant desired operating values (current, voltage) around on which we want to
control the converter, this process is the action of fixing the constant desired values for the
output y (usually the reference value for the output voltage(s); then resolve the average
model equations within the steady state regime:
0 f (x , uav )
y Cx
To find back the reference values for the remaining states and control inputs (x , uav )
4. Once we have obtained the average model, we proceed to perform the linearization of the
model around the desired equilibrium point. The linearization of the average model (1.2),
around the desired equilibrium point x , uav , yields the following state equations:
.
x Ax Buav
y C x
With x x x , uav uav uav , y y y , and the matrices A and B are obtained via the
Taylor first order development:
2
Chapter 1 Open-loop modeling of DC-DC converters
f (x , uav ) f (x , uav )
A ,B
x x x ,uav uav uav x x ,u
av uav
1.2 Buck converter
The Buck converter is used for step down operation. A DC-DC buck converter with its output
filter arrangement is shown in figure 1.1.
Fig. 1.1 Practical Buck converter realization.
1.2.1 Switched Converter model
In order to obtain the Buck converter switched model, all we have to do is to apply the
Kirchhoff’s laws to fig. 1.1 circuit, and to combine the different topologies (fig. 1.2) into a
switched state-space model.
The two states are given by:
Fig.1.2 (a) u 1 , (b) u 0 .
When the ideal switch is ON as shown in the figure 1.2 (a) the dynamics of the inductor
current i (t ) and the capacitor voltage v (t ) are both given by;
di
L v E
dt
(1.1)
dv v
C i
dt R
And when the switch is OFF as shown in the figure 1.2 (b) the dynamics are presented by:
di
L v
dt
(1.2)
dv v
C i
dt R
After doing a comparison between the two states (ON and OFF) we get the following unified
model:
3
Chapter 1 Open-loop modeling of DC-DC converters
di
L v uE
dt
(1.3)
dv v
C i
dt R
So, when u 1 or u 0 we find back the model (1.1) or the model (1.2). The model (1.3) is
often called switched model with the binary function of the switching u 0,1 .
1.2.2 Average model
The average converter model is exactly the same as (1.3), except that the control variable u is
replaced by its continuous average variable uav that takes its values in the interval [0, 1].
The average model of the Buck converter is found to be given by:
.
L x 1 x 2 uav E
. x2 (1.4)
C x 2 x1
R
With: x 1 i, x 2 v
In the matrix form, we get:
x Ax Buav
0 1/ L E / L
x x
u (1.5)
1/C 1/ RC 0 av
The average model is obviously linear; in addition it is controllable and observable for each of
the two output states.
1.2.3 Equilibrium point
The main purpose of the control will be often to regulate the output voltage to a desired
average value. This is fulfilled by choosing the input u that will control the switch state
(position) according to the average value (reference), which is interpreted as the average duty
cycle in PWM techniques, and can be interpreted as an equivalent control.
Generally, within equilibrium, it is a likely linking together, the average values of the system
state, and the constant average values corresponding to the control input.
These relations, within the equilibrium state, are useful in defining the static converter
characteristics.
In equilibrium state, the derivation of the mean (average) states is zero, and the average
control input uav takes a constant value u av . Then, using the representation (1.4) and noting
the average equilibrium as x 1 and x 2 , we get at equilibrium:
0 x 2 uav E
x2 (1.6)
0 x1
R
After solving the system equations (1.6), we get the equilibrium states of the system as next:
4
Chapter 1 Open-loop modeling of DC-DC converters
x 2 Eu av ,
E (1.7)
x1 u av
R
This parameterization of the equilibrium point by the average control is useful to determine
the converter damping character.
Relation (1.7) shows that the average output voltage is a fraction of E , and the converter can
not amplify the input voltage since u av is restricted in [0, 1].
x2
1.0
E
0.8
0.6
0.4
0.2
u av
0.0
0.0 0.2 0.4 0.6 0.8 1.0
Fig. 1.3: Static transfer function
1.2.4 Desired equilibrium point
Supposing that the desired average voltage at equilibrium is x 2 EVd , so we get at
equilibrium:
EVd
x1 , x 2 EVd , uav Vd (1.8)
R
Hence, the desired average voltage has to satisfy 0 < Vd < 1.
1.3 Boost converter
The boost converter has earned its name grace to its ability of producing an DC output voltage
greater in magnitude than the DC input voltage (step-up). The circuit topology for the boost
converter is as shown in figure 1.4
Fig.1.4 Practical Boost converter realization
When the transistor Q is ON (fig.1.5 (a)), The current in inductor L, rises linearly and at this
time capacitor C, supplies the load current, and it is partially discharged. During the second
interval when transistor Q is OFF (fig.1.5 (b)). The diode D is ON and the inductor L,
supplies the load and, additionally, recharges the capacitor C.
5
Chapter 1 Open-loop modeling of DC-DC converters
Fig.1.5 (a) u 1 , (b) u 0 .
1.3.1. Switched Converter model:
When the switching function is u=1, the following dynamic is obtained:
di
L E
dt
(1.9)
dv v
C
dt R
When the switching function is u=0, the following dynamic is obtained:
di
L v E
dt
(1.10)
dv v
C i
dt R
So the dynamic of the converter is as follows:
di
L (1 u )v E
dt
(1.11)
dv v
C (1 u )i
dt R
1.3.2 Average model
So, we get the following unified average model:
di
L (1 uav )v E
dt
(1.12)
dv v
C (1 uav )i
dt R
Or:
.
L x 1 (1 uav )x 2 E
. x (1.13)
C x 2 (1 uav )x 1 2
R
With x 1 i, x 2 v
The variable x 1 is the average current of the inductor and x 2 is the output average voltage.
1.3.3 Equilibrium point
In equilibrium state, the derivation of the mean (average) states is zero, and the average
control uav takes a constant value u av , as a result, we get the linear system equations for the
6
Chapter 1 Open-loop modeling of DC-DC converters
values in steady state of average states, noting the equilibrium average values of the current
and voltage as x 1 and x 2 , so we get at equilibrium:
0 1 u av E
x 1
1 (1.14)
1 u av x 2 0
R
After solving the system equations (1.14), we get the system equilibrium states as follows:
E 1 E
x1 2 , x2 (1.15)
R 1 u av 1 u av
x2
10
E
2 u av
0.0 0.2 0.4 0.6 0.8
Fig. 1.6: Static transfer function.
1.3.4 Desired equilibrium point
The average model of this converter is found to be given by:
.
L x 1 (1 uav )x 2 E
. x (1.16)
C x 2 (1 uav )x 1 2
R
If the desired average voltage at equilibrium is x 2 EVd , then, we get at equilibrium:
E 2 1
x1 Vd , u av 1 (1.17)
R Vd
E
As if >0 then x 1 0 and the desired average voltage has to satisfy 1 < Vd < .
R
1.3.5 Linearization
The linearization of the average model is found to be given by:
. 1
L x1 x 2 EVd uav
Vd
(1.18)
. 1 1 E
C x 2 x1 x 2 Vd2 uav
Vd R R
With
x1 x 1 x 1
7
Chapter 1 Open-loop modeling of DC-DC converters
x 2 x 2 x 2 (1.19)
uav uav uav
In the matrix form, we get:
1 EVd
0
. LVd L
x x 2
uav (1.20)
1 1 EVd
CVd RC RC
The linearized model is controllable as being proved next:
EVd EVd
L RLC EVd
M 2 , det M (R 2C LVd Vd2 ) 0 (1.21)
EVd E EVd L(RC )2
RC CL (RC )2
Our model is also observable for the two states variables
For y x1 , comes up
1 0 1
N
, det N 0 (1.22)
0 1/ LV LVd
d
For y x 2 , we get then:
1
0
CVd 1
N , det N 0 (1.23)
1 CVd
1
RC
1.4 Buck-Boost Converter
The buck-boost converter is capable of producing an DC output voltage which is either
greater or smaller in magnitude than the DC input voltage. The arrangement for the buck-
boost converter is as shown in figure (1.7)
Fig.1.7 Practical Buck-Boost converter realization
When the transistor Q is ON fig (1.8(a)), input voltage is applied across the inductor and the
current in inductor L rises linearly. At this time the capacitor C , supplies the load current, and
it is partially discharged. During the second interval when the transistor is OFF fig.1.8 (b), the
voltage across the inductor reverses in polarity and the diode conducts. During this interval
the energy stored in the inductor supplies the load and, additionally, recharges the capacitor.
8
Chapter 1 Open-loop modeling of DC-DC converters
Fig.1.8 (a) u 1 , (b) u 0 .
1.4.1. Switched Converter model
By applying the Kirchoff’s laws on the two previous circuits, we get the following dynamics:
When the switching function is u=1, the following dynamic is obtained:
di
L E
dt
(1.24)
dv v
C
dt R
When the switching function is u=0, the following dynamic is obtained:
di
L v
dt
(1.25)
dv v
C i
dt R
So the dynamic of the converter is found to be given as:
di
L (1 u )v uE
dt
(1.26)
dv v
C (1 u )i
dt R
1.4.2 Average model
The average model of the Buck-Boost converter is found to be given as:
.
L x 1 (1 uav )x 2 uav E
. x (1.27)
C x 2 (1 uav )x 1 2
R
With x 1 i, x 2 v
1.4.3 Equilibrium point
At equilibrium, we get:
0 (1 u av ) x 1 Eu av
(1 u av ) 1/ R x 2 0 (1.28)
After solving the system equations (1.28), we get the system equilibrium states as follows:
E u av Eu av
x1 2
, x2 (1.29)
R (1 u av ) (1 u av )
9
Chapter 1 Open-loop modeling of DC-DC converters
0.0 0.2 0.4 0.6 0.8
uav
0
-2
x2
E
-4
-6
-8
Fig.1.9 Static transfer function.
1.4.4 Desired equilibrium point
The average model of this converter is found to be described by:
Lx1 (1 uav )x 2 Euav
x (1.30)
Cx 2 (1 uav )x 1 2
R
If the desired average voltage at equilibrium is x 2 EVd , so we have at equilibrium:
EVd (Vd 1) Vd
x1 , u av (1.31)
R Vd 1
E
As >0 then x 1 0 and the desired average voltage has to satisfy Vd <0.
R
1.4.5 Linearization
The average model linearization through the equilibrium points is as below:
. 1 E (1 Vd )
x1 x 2 uav
L(Vd 1) L
(1.32)
. 1 1 EVd (Vd 1)
x 2 x1 x 2 uav
C (Vd 1) RC RC
With:
x1 x 1 x 1
x 2 x 2 x 2 (1.33)
uav uav uav
In the matrix form, we get:
1 E (1 Vd )
0
. L(Vd 1) L
x x uav (1.34)
1 1 EVd (Vd 1)
C (Vd 1) RC RC
The controllability of the linearized model is fulfilled for the reason that:
10
Chapter 1 Open-loop modeling of DC-DC converters
E (1 Vd ) EVd
L RLC E 2(Vd 1) Vd 1
M , det(M ) 2 0 (1.35)
EVd (Vd 1) E EVd (1 Vd ) CL R C C
RC CL (RC )2
In the other hand our model is also observable for all the two states variables
We get for y x1 :
1 0
1
N 1 , det N 0
0 L(Vd 1) (1.36)
L(Vd 1)
The following is gotten for y x 2 :
1
0
C (Vd 1) 1
N , det N 0 (1.37)
1 C (Vd 1)
1
RC
1.5 Cuk converter
Cuk converter is derived from the cascading of buck and boost converters. The buck, boost
and buck-boost converters all transfer energy between input and output using the inductor and
analysis is based on voltage balance across the inductor. This converter is shown in fig. 1.10.
The input circuit in the Cuk converter is clearly a Boost circuit converter, and the output
circuit is seen to be a Buck converter. Thus, we may also think of the Cuk converter as a
“Boost-Buck” converter.
Fig.1.10 Practical Cuk converter realization.
The Cuk converter operates in two steps. The first one is obtained when the transistor is ON
in the same time the diode D is reversely polarized giving the circuit topology in fig. 1.11(a).
During this period, the inductor current i1 is drawn from the voltage source E . This mode
represents the charging mode. The second step starts up when the transistor is OFF and the
diode D is directly polarized giving the circuit topology in fig. 1.11(b). This mode of
operation is known as the discharging mode since all the energy stored in L1 is now
transferred to the load R .
11
Chapter 1 Open-loop modeling of DC-DC converters
Fig 1.11 (a) u 1 , (b) u 0 .
1.5.1 Switched Converter model
The derivation of the dynamics of the Cuk converter is carried out in the same manner in
which we analyzed the topologies of the previous basic DC-DC power converters.
When u 1 , we obtain the following equations for i1 and i2 in the obtained circuit topology.
di1
L1 E
dt (1.38)
di
L2 2 v1 v2
dt
And the following equations for the capacitor voltages v1 and v2 ,
dv1
C1 i2
dt
(1.39)
dv v
C 2 2 i2 2
dt R
When u 0 , we obtain the following equations for i1 and i2
di1
L1 v1 E
dt
(1.40)
di
L2 2 v2
dt
The capacitor voltages v1 and v2 are described by:
dv1
C1 i1
dt
(1.41)
dv v
C 2 2 i2 2
dt R
The Cuk switched model is then given by combining the previous partial models, that is:
di1
L1 (1 u )v1 E
dt
dv
C 1 1 (1 u )i1 ui2
dt
(1.42)
di2
L2 uv1 v2
dt
dv v
C 2 2 i2 2
dt R
12
Chapter 1 Open-loop modeling of DC-DC converters
1.5.2 Average model
Replacing u by uav , we get the following average model:
.
L1 x 1 (1 uav )x 2 E
.
C 1 x 2 (1 uav )x 1 uav x 3
. (1.43)
L2 x 3 uav x 2 x 4
. x4
C2 x 4 x3
R
With: x 1 i1, x 2 v1, x 3 i2, x 4 v2
1.5.3 Equilibrium point
At equilibrium, with the average control uav taking a constant value u av , we get as a result,
the system equations for the steady state regime:
0 (1 u av ) 0 0 x1 E
(1 u av ) 0 u av 0 x 2 0
x 3 (1.44)
0 u av 0 1 0
x
0 0 1 1/ R 4 0
After solving the system equations (1.44), we get the system equilibrium states as follows:
E u av 2 E E u av Eu av
x1 2
, x2 , x3 and x 4 (1.45)
R (1 u av ) (1 u av ) R (1 u av ) (1 u av )
x4
E
Fig 1.12 Static transfer function
1.5.4 Desired equilibrium point
If the desired average voltage at equilibrium is x 4 EVd , so we have at equilibrium:
E 2 E
x1 Vd , x 2 E (1 Vd ) , x 3 Vd et u av Vd /(Vd 1) (1.46)
R R
E
As >0 then x 1 >0, x 3 <0 and the desired average voltage has to satisfy Vd <0.
R
13
Chapter 1 Open-loop modeling of DC-DC converters
1.5.5 Linearization
The linearization of the model is found to be given as below:
. 1 E (1 Vd )
x1 x 2 uav
L1(Vd 1) L1
. 1 Vd EVd
x 2 x1 x 3 uav (1.47)
C 1(Vd 1) C 1(Vd 1) RC 1
. Vd 1 E (Vd 1)
x 3 x 2 x 4 uav
L2(Vd 1) L2 L2
. 1 1
x 4 x 3 x 4
C2 RC 2
In which: x1 x 1 x 1 , x 2 x 2 x 2 , x 3 x 3 x 3 , x 4 x 4 x 4 , uav uav uav .
In the matrix form, we get:
0 1/L1(Vd -1) 0 0 E (1-Vd )/L1
1/C (V 1) 0 Vd /C 1(Vd 1) 0 EV /RC
. 1 d d 1 (1.48)
x x uav
0 Vd / L2(Vd 1) 0 1/ L2 E (1-Vd )/L2
0 0 1/C 2 1/ RC 2 0
On the one hand we got det M 0 so our model is controllable, in which that:
1 1 1 1
E (1-Vd )/L1 EVd /(Vd -1)RL1C1 (E / L1C 1(Vd 1))( Vd ) (EVd /RL1 )( Vd2 )/ C12 (1-Vd )3
L1 L2 L1 L2
1 1 1 1 1 1 1 1 1
EVd /RC1 (E /C 1 )( Vd ) (EVd /RC1 )( Vd2 )/C 1(1-Vd )2 (E ( Vd2 )/ L1C12 (1-Vd )2 ) (EVd /L2C1 )(( Vd2 )/C 1(Vd -1)2
C 1 L2 L1 L2 L1 L2 L1 L2 L2C 2
M
1 L2 Vd2 1
E (1-Vd )/L2 EVd /(Vd -1)RL2C1 (EVd /L1L2C1 ) (E / L22 )((Vd2/C1(Vd -1))+ ) EVd2 ( ) E (1-Vd )/RL22C22
C1 2
L1C1(Vd 1) C1(1-Vd ) C 1
(1-Vd ) 1 EVd EVd2 EVd2 1 1
0 E (1- V )/L C ( E/RL C )( V 2
) E (1-V )( )
d 2 2 2 2
C2 C 1(Vd 1)
d 2
L2L1C1(Vd 1) 2 2
RL2C2C 1(Vd 1) L2C2C1(Vd 1)
d
RC 2 L2
(1.49)
On the other hand we got det N 0 for all the two mean state variables the model we go t
is observable, given that:
For y x 2 comes
1 1
0 1/ L1(Vd 1) 0 ( Vd2 )/ L1C1(1-Vd )3
L1 L2
1 1 Vd2
1 0 ( ) 0 (1.50)
2
(1-Vd ) C 1 L1 L2
N , det N 0
Vd L2 Vd2
0 Vd / L2(Vd 1) 0 ( 1)
C 1L2(Vd 1) L1(Vd 1) (1-Vd )2
Vd
0 0 Vd / L2C 2(Vd 1)
2
RL2C2(Vd 1)
For y x 4 comes
14
Chapter 1 Open-loop modeling of DC-DC converters
Vd
0 0 0
L2L1C1(Vd 1)2
1 1 2 1
0 0 Vd / L2C 1(Vd 1) (Vd /C1(Vd 1))(( Vd )/C 1(Vd -1)2 ) (1.51)
L1 L2 L2C 2
N , det N 0
Vd2 1
0 1/ L2 1/ RL2C 2 (( 2 2 2
)/ L2 ) (1/R L2C2 )
C1(1-Vd ) C 2
1 1 1 1 1
1 1/ RC 2 (1/C 2 )( 2 2 ) (1/RC22 )( )
R C 2 L2 RL2 RC 2 L2
1.6 Zeta converter
The Zeta converter can both amplify and reduce, without polarity inversions, the value of the
input source voltage E. We briefly summarize next the most important features involved in
the modeling of the Zeta converter. Figure 1.13 depicts a semiconductor realization of a Zeta
DC-to-DC power converter. The ideal switch based realization of the Zeta converter is
depicted in Figure 1.13.
Figure 1.13 The Zeta DC-DC converter realization
The Zeta converter operates in two steps. The first one is obtained when the transistor is ON
and instantaneously, the diode D is reversely polarized giving the circuit shown in fig.
1.14(a). During this period, the inductors currents i1 and i2 are drawn from the voltage source
E . This mode is the charging mode. The second mode starts up when the transistor is OFF
and the diode D is directly polarized giving the circuit shown in fig. 1.14(b). This mode of
operation is known as the discharging mode, i.e., all the energy stored in L2 will be
transferred now to the load R .
Fig 1.14 (a) u 1 , (b) u 0 .
1.6.1 Switched Converter model
By applying the Kirchoff’s laws on the two previous circuits, we get the following dynamics:
When the switching function is u=1, the following dynamic is obtained:
15
Chapter 1 Open-loop modeling of DC-DC converters
di1
L1 E
dt
dv
C 1 1 i2
dt
(1.52)
di
L2 2 v1 v2 E
dt
dv v
C 2 2 i2 2
dt R
When the switching function is u 1 , the following dynamic is obtained:
di1
L1 v1
dt
dv
C 1 1 i1
dt
(1.53)
di2
L2 v2
dt
dv v
C 2 2 i2 2
dt R
The Zeta converter dynamic is then described by combining the previous partial models. We
obtain the following system of differential equations:
di1
L1 (1 u )v1 uE
dt
dv
C 1 1 (1 u )i1 ui2
dt
(1.54)
di2
L2 uv1 v2 uE
dt
dv v
C 2 2 i2 2
dt R
1.6.2 Average model
Replacing u by uav , we get the following average model:
.
L1 x 1 (1 uav )x 2 uavE
.
C 1 x 2 (1 uav )x 1 uav x 3
. (1.55)
L2 x 3 uav x 2 x 4 uav E
. x
C2 x 4 x3 4
R
With: x 1 i1, x 2 v1, x 3 i2, x 4 v2
16
Chapter 1 Open-loop modeling of DC-DC converters
1.6.3 Equilibrium point
Within the equilibrium state, with uav u av , we get:
0 (1 u av ) 0 0 x1 uav
(1 u av ) 0 u av 0 x 2 0
- E
x 3 uav (1.56)
0 u av 0 1
x 4 0
0 1/ R
0 1
After solving the system equations (1.59), we get the system equilibrium states as follows:
E u av 2 u av E u av u av
x1 2
, x2 E , x3 and x 4 E (1.57)
R (1 u av ) (1 u av ) R (1 u av ) (1 u av )
x4
E
8
u av
0
0.0 0.2 0.4 0.6 0.8
Fig. 1.15 Static transfer function
1.6.4 Desired equilibrium point
If the desired average voltage at equilibrium is x 4 EVd , then we get:
E 2 E Vd
x1 Vd , x 2 EVd , x 3 Vd et u av (1.58)
R R Vd 1
E
As >0 then x 1 >0, x 3 >0 and the desired average voltage has to satisfy Vd >0.
R
1.6.5 Linearization
The linearization of the model is found to be given as below:
1
L1x1 x2 E (1 Vd )uav
Vd 1
1 Vd V (1 Vd )
C 1x2 x1 x3 d uav (1.59)
Vd 1 Vd 1 R
Vd
L2x 3 x2 x4 E (Vd 1)uav
Vd 1
x4
C 2x 4 x3
R
17
Chapter 1 Open-loop modeling of DC-DC converters
With: x1 x 1 x 1 , x2 x 2 x 2 , x3 x 3 x 3 , x4 x 4 x 4 , uav uav uav .
In the matrix form, we have:
0 1/L1(Vd 1) 0 0 E (1 Vd )/ L1
1/C (V 1) 0 V / C V
( 1) 0 EV ( 1 V )/RC
x d 1 (1.60)
1 d d 1 d d
x u
0 Vd / L2 (Vd 1) 0 1/ L2 E(Vd 1) / L2 av
0 0 1/C 2 1/ RC 2 0
It happened that the linearized model is controllable according to the following:
V 1 V2 1
E (1 Vd )/ L1 EVd /RL1C 1 (E / L1C 1 )( d ) E ( d Vd2 )/ L1C 1(Vd 1)
L2 L1(Vd 1) L2 L1(Vd 1)
1 1 2 1 1 2 1 1 2
EVd( 1 Vd )/RC 1 E / L1C 1 EVd / L2C 1 2
(EVd /RC1(Vd 1))( Vd ) 2 2
(E ( Vd )/ L1C1(Vd 1)) (EVd /L2C1 )(( Vd )/(Vd 1) 2
L1 L2 L1 L2 L1 L2
M 2 2
E(V 1) / L 2 EV EV V 1 2 1 V 1 2 2 2
d 2 EV d /RC L
1 2 d
(V d 1 ) d
( d
2
+ ) EVd ( 2
(( d
2
)/ R L C
2 1 ) E(Vd 1) /R L C C
2 2 1
L1L2C1(Vd 1) RL2C1 C1(Vd 1) C 2 RL2L1C1(Vd 1) C1(Vd 1) C2
Vd2 (Vd 1) EVd 1 Vd Vd 1 1
0 E(V 1) / L C (E/RL C C )( ) ( ) E ( V +1)( )/ L C 2
d 2 2 2 1 2
C C L C L C R 2 2
C L2
V
( 1)
d
L R 2
C
2 2
1 2 2 1 1 2 2 2 d 2 2
det M 0 (1.61)
Our model is observable too at the tow states variables as is proven below:
For y x 2 , gives
1 1
0 1/ L1(Vd 1) 0 ( Vd2 )/ C12(Vd +1)3
L1 L2
(1.62)
1 1 2 2
1 0 -( Vd )/C 1(Vd +1) 0
L1 L2
N , det N 0
L2 1 1 2 Vd
0 Vd / L2(Vd 1) 0 Vd ( 2 3
( Vd )
C1 (Vd 1) L1 L2 L2C 2C 1(Vd 1)
Vd
0 0 Vd / L2C 1(Vd 1)
RL2C 1C 2(Vd 1)
For y x4 , gives
Vd
0 0 0
L2C 1C2(Vd 1)
Vd (1.67)
0 0 Vd / L2C 2(Vd 1) 2
RL2C2(Vd 1)
N , det N 0
V 2
1 1
0 1/C 2 1/ RC2 2
2 d
(1/C 2 )( 2 )
2
L2C1(Vd +1) 2
L2 R C 2
1 1 1 1 1
1 1/ RC 2 (1/C 2 )( 2 ) (1/RC 2
)( )
L2 R C 2 RL2C1 2 2 2
L2 R C 2
18
Chapter 1 Open-loop modeling of DC-DC converters
1.7 Conclusion
The modeling has shown that the DC-DC converters are nonlinear systems (except for the
Buck). The study in equilibrium state has established the converters static transfer functions,
and permitted the understanding of the converters input-output relationships. Linearization
around an equilibrium point has established the characteristics of local controllability and
observability and thus prepared the ground for the application of linear control techniques.
19
Chapter 2 Poles Placement Based State Feedback Control of DC-DC Converters
Chapter 2
Linear State Feedback Control of DC-DC Converters
2.1. Introduction
In this chapter, we investigate the application of linear state feedback control techniques to
improve the dynamic behavior of DC-DC converters. Sections 2.2, 2.3 and 2.4 present the
necessary theoretical elements for the use of these techniques, namely, the state feedback by
pole placement, the state feedback with integral action and the state feedback with observer.
Section 2.5 presents the application of these techniques to different DC-DC converters
modeled in the first chapter, and section 2.6 concludes the chapter. The underlying theory for
this chapter is summarized from [4].
2.2. Poles placement
Considering the DC-DC converter linearized model:
x Ax Bu
(2.1)
y Cx
If it is considered that the above system is completely controllable, then the control law is as
follows:
u Kx (2.2)
Hence, the matrix K of (1 n ) dimension is called the gain matrix.
Figure 2.1 shows the system defined by equation (2.1), open loop then closed loop mode, using
the control law (2.2).
u x x x x
u
a b
Fig 2.1: (a) open loop system, (b) closed loop system with u Kx .
So the system becomes a closed loop control system as shown in Figure 2.1.b,
x (A BK )x (2.3)
Noting that the eigenvalues of the matrix A BK ( 1, 2,.., n ) are the desired closed loop
poles.
In order to place the system poles via the state feedback, the system has to verify a necessary
condition, which is the controllability, that is, the matrix
20
Chapter 2 Poles Placement Based State Feedback Control of DC-DC Converters
M B AB ... An 1B should be a full rank matrix.
In order to calculate the gain matrix K , which is necessary for the poles placement, several
ways are possible. Among them we cite:
2.2.1 Substitution
The gain K k1 k2 ... kn can be substituted directly into the characteristic polynomial of
sI A BK , equalizing this latter with the closed loop desired polynomial of converter
model.
2.2.2 Controllable form
The transformation matrix T MW , where W is defined by:
an 1 an 2 a1 1
a a1 1 0
n 2
W . (2.4)
a1 1 . 0
1 0 0 0
The gain matrix is then:
K n an n 1 an 1 ... 1 a1 T 1 (2.5)
Where desired characteristic polynomial:
s 1 s 2 ...s n s n 1s n 1 2s n 2 ... n 1s n (2.6 )
And the characteristic polynomial:
sI A s n a1s n 1 ... an 1s an (2.7)
2.2.3 Ackermann formula
To get the gain matrix K , it is possible using the Ackermann formula which is given as the
following equation:
K 0 ... 0 1 M 1(A) (2.8)
Hence
(A) An 1An 1 ... n 1A n I (2.9)
21
Chapter 2 Poles Placement Based State Feedback Control of DC-DC Converters
2.3. State feedback with integral action
To assess more realistic situations, we suppose that the converter is subject to static disturbance
d due to load variations, that is:
x Ax Bu
(2.10)
y Cx d
Then, the state feedback can't drive the output y to zero. It is necessary to introduce an integral
action in order to eliminate the disturbance effect. To this end, we introduce the additional state
variables:
xI C x (2.11)
Or equivalently:
x I C x (2.12)
Then, the augmented model is given by:
z A*z B *u (2.13)
With z T x xI and
A 0 B
A
* , B* (2.14)
C 0 0
The new state feedback control law is given by:
x
u K *z K K I (2.15)
x I
Which can be interpreted as a state feedback control law involving n m dimensional
augmented state vector formed by the state vector x and the integrator state vector xI .
The matrix K * of (1 (n m )) dimensional
The closed loop system is given by:
A BK BkI
z A* B *K * z z (2.16)
C 0
If the desired behavior is specified through the selection of n m desired eigenvalues
1, 2,..., n m , then, the gain vector K * can be computed using one the previously mentioned
methods.
2.4. Observer based state feedback
2.4.1 Introduction
In the above part of this chapter, it is considered that all the state variables are available
for the feedback. However, practically, some variables may not be available. therefore, the
estimate phase of those variables is definitely an essential step, means, that we must estimate
the non-measurable state variables so as to elaborate the control signals.
22
Chapter 2 Poles Placement Based State Feedback Control of DC-DC Converters
In case if the state observer estimates all the system state variables, even if some of those
variables are accessible to direct measurement, this observer type is known as a full order state
observer (fig. 2.2).
xx y
B
C
uu A
K
xx̂ˆ yˆŷ
B
C
A
Ke
Fig. 2.2: Observer based state feedback.
Considering the DC-DC converter linearized model:
.
x Ax Bu
(2.17)
y Cx
A necessary condition to the realization of a state observer is that the converter linearized
model is observable. This condition is realized if the observability matrix
N C T ATC T ... (AT )n 1C T (2.18)
has a rank (N ) n .
The full order observer is defined by:
xˆ Axˆ Buav Ke (y yˆ)
(2.19)
yˆ Cxˆ
The estimation (or observation) error e x xˆ is defined by:
e (A KeC )e (2.20)
Where Ke is the observer gain vector.
If Ke is chosen such as the matrix (A KeC ) is stable, then, the estimation error
e(t ) e (AKeC )te(0)
converges exponentially to zero. That is, e(t ) 0 x xˆ .
If the desired observer behavior is specified through the selection of n desired eigenvalues
1, 2,..., n , then, the gain vector Ke can be computed using one the following methods.
2.4.2 Substitution
Supposing x is a vector of the dimensional 3, under circumstances, the gain matrix is drawn
out as following:
KeT ke1 ke 2 ... ken (2.21)
23
Chapter 2 Poles Placement Based State Feedback Control of DC-DC Converters
The matrix substitution Ke into the desired characteristic polynomial is found to be given by:
sI A KeC s 1 s 2 s 3 (2.22)
By doing an equalization of the coefficients of the same power on s of the two sides of the
latter equation, then the values ke1, ke 2 and ke 3 can be easily determined.
2.4.3 Observable form
The transformation matrix T is described as:
1
T WN T , N is the matrix of the observability
N C T ATC T ... (AT )n 1C T (2.23)
Where W is defined by:
Hence the characteristic polynomial is:
sI A s n a1s n 1 ......... an 1s an (2.24)
And the desired characteristic equation for the error dynamic:
s 1 s 2 ...s n s n 1s n 1 2s n 2 ... n 1s n 0 (2.25)
So:
n an n an
n 1 an 1 n 1 an 1
Ke T WN T
1
(2.26)
a a
1 1
1 1
The equation (2.26) gives directly the gain matrix Ke of our observer.
2.4.4 Ackermann formula
So as to calculate the observer gain, we can use the Ackermann formula of which its form is
given as follows:
1
0 C 0
CA
1
Ke (A) N T (A) (2.27)
0 0
1 CAn 1 1
Where s is the desired characteristic polynomial of the observer :
s s 1 s 2 ... s n s n 1s n 1 2s n 2 ... n 1s n (2.28)
With 1, 2,..., n are the desired eigenvalues, whereas we have:
A An 1An 1 2An 2 ... n 1A n I n (2.29)
24
Chapter 2 Poles Placement Based State Feedback Control of DC-DC Converters
2.5. Application:
In this section, we’re about studying the application of the linear state feedback control on
the DC to DC converters that have been analyzed within the first chapter, starting by the
application of poles placement without and with integral action, then the application of the full
order observer.
For each converter, the calculation steps are as follows:
I) Feedback
1. Fixing the desired capacitor voltage.
2. Calculating the equilibrium points as shown within chapter 1.
3. Linearizing the converter model at the desired operating points, to have
.
x Ax Bu
y Cx
4. Fixing the specification of the closed loop response according to a specified overshoot or a
settling time or both. Those specifications are defined by two parameters n and , such
that:
1 2
a) For the overshoot specification: M p e
b) For the settling time specification we use : e wnts
The damping ratio is usually taken as 0.7 2 .
5. Calculating the gain matrices K and K I using the follows Matlab command:
K = place (A, B, P)
Where P is the vector of the desired eigenvalues
Then, the average control is given by:
uav uav Kx uav K (x x )
In order to apply uav , it must be converted to gate signal u 0,1 , this function is realized via
an PWM technique as shown is the figure 2.3. The carrier signal has a triangular waveform
with 40 kHz frequency. The control signal uav is compared with the carrier signal, and the
comparator output is switched at every intersection instant, so the gate signal u is of constant
frequency and variable duty ratio (fig. 2.4).
In all simulations the MOSFET is taken as open circuit when it is off and it is equivalent to
0.7V voltage source in series with 50m resistance, when it is on. Similarly, the Diode is
taken as open circuit when it is off and it is equivalent to 0.4V voltage source in series with
25m resistance, when it is on [1]. The converters structures and parameters values are taken
from [2].
25
Chapter 2 Poles Placement Based State Feedback Control of DC-DC Converters
PWM u Convertisseur DC-
DC
uav
State Feedback x Voltage and current
control law measurement
Fig.2.3. DC-DC converter state feedback control scheme.
carrier
uav
u
u
uav
Fig.2.4. Carrier based PWM technique.
II) Observers
We suppose that only the voltages are measurable, and build an observer for estimating the
currents.
Full-order observer form:
xˆ Axˆ Buav Ke (y yˆ)
yˆ Cxˆ
The calculation of the gain matrix is as follows:
1. Choosing n and .
2. Using the place Matlab function: Ke= place (A', C', P), P the desired poles.
PWM
u Converter DC-DC
Voltage
uav
State Feedback
x
control law State observer
Fig 2.5 State Feedback with observer
26
Chapter 2 Poles Placement Based State Feedback Control of DC-DC Converters
2.5.1. Buck converter:
The assigned values for the buck converter are given by: E 24V , L 15.91mH ,C 50F
and R 25
The desired average voltage is: x 2 Vd E 12V , which yields x 1 0.4800 and uav 0.5 .
The linearized model is given as follows:
0 62.854 1508.5
x x
0 uav
20000 800
This average model is controllable as proven within the first chapter.
20
15
10
0.8
0.6
0.4
0.2
uav
0.54
0.52
0.5
0.48
0.46
0 0.005 0.01 0.015 0.02
Time (s)
Fig. 2.6: Buck open loop response
From the open loop characteristic, we get:
n 1121.2rad / sec , 0.35676 .
It is concluded from the above curves that the response is slower and present 25% overshoot
peak for the voltage and 62.5% for the current. Noting the time scale of the graphs, settling
time in this graph is about 15 msec.
For better performance, the use of state feedback control will greatly reduce both the peak
overshoots and deviations of the two i and v , with acceptable settling time in the millisecond
range.
27
Chapter 2 Poles Placement Based State Feedback Control of DC-DC Converters
a. State feedback
The average law control:
uav Kx k1x1 k2x2
The chosen parameters 1, n 2000rad / sec
The gain matrix of those parameters is:
K 2.1213 0.006 064
uav v
uav u av
Fig.2.7: Buck state feedback control
v v
12 12
10 10
8 8
6 6
4 4
2 2
0 0
i i
0.8 0.8
0.6 0.6
0.4 0.4
0.2 0.2
0 0
Uav uav
1 1
0.5 0.5
0 0
0 0.004 0.008 0 0.004 0.008
Time (s) Time (s)
(a) (b)
Fig. 2.8: Buck closed loop response. (a) R=25, (b) resistance variation R=20.
After the application of feedback control we obviously notice the improvement on the buck
converter performance, i.e. faster response, less deviations and low amplitude peaks. Even
though this positive effect on the system response still has the inefficiency of the non-achieved
zero static error that is approximated around 12.5% (fig. 2.8) .
A zero static error dynamic controller for buck converter is designed to achieve zero voltage
steady-state error, thus it is proposed adding an integrator to state feedback control that will
serve eliminating non-zero error as will be demonstrated below.
28
Chapter 2 Poles Placement Based State Feedback Control of DC-DC Converters
b. State feedback with Integral action:
The augmented Buck converter model is:
62.854 0 1508.5
0
.
z 20000. 800 0 z 0 uav
0 1 0 0
With
z x1 x2 x
2
The augmented control law is given by:
uav k1x1 k2x2 kI x2
According to augmented Buck converter model, the desired polynomial is:
s p s 2 2ns n2 s 3 p 2n s 2 n2 2pn s p n2
The chosen parameters: 2, n 1000rad / sec, p 400.
Then the augmented gain matrix is: K I 2.3865 -0.0509475 13.2583
uav v
uav v v
u av
Fig.2.9. Buck state feedback + I control
29
Chapter 2 Poles Placement Based State Feedback Control of DC-DC Converters
v v
14 14
12 12
10 10
8 8
6 6
4 4
2 2
0 0
i i
0.6 0.8
0.6
0.4
0.4
0.2
0.2
0 0
Uav Uav
1 1
0.5 0.5
0 0
0 0.01 0.02 0.03 0.04 0 0.01 0.02 0.03 0.04
Time (s) Time (s)
(a) (b)
Fig.2.10: Buck closed loop response. (a) R=25, (b) resistance variation R=20.
The output voltage integral has been introduced as additional state variable as shown in Fig.
2.9, this integral action has successfully achieved a zero steady-state error that wasn’t be
possible relying just on state feedback control. This approach has well proven its efficiency
when having load variation, particularly, although the load is down to 20 ohms, the integrator
is still effective on the converter response by providing an achieved zero static error
c. Observer based state feedback control
The observability condition has been verified in the first chapter.
The chosen parameters for the observer: 1, n 4000rad / sec . The appropriate gain
matrix for those parameters is given by:
737.1
Ke
7200
The full order observer structure:
0 800 1508.5 737.1
xˆ xˆ
0
u
7200 x2
20000 8 000 av
The average law control is:
uav u av k1
x1 k2x2
30
Chapter 2 Poles Placement Based State Feedback Control of DC-DC Converters
i v
v
v
uav
i
uav Observer
v
u av
Fig. 2.11: Buck state feedback with full order observer.
v
12
10
8
6
4
2
0
i i_est
0.8
0.6
0.4
0.2
0
Uav
1
0.5
0
0 0.01 0.02 0.03
Time (s)
Fig. 2.12: Buck closed loop response with full order observer.
The result gotten from the above graphs shows that the estimated current converges to the
inductor current.
2.5.2 Boost converter
The assigned values for the boost converter are given by:
E 24V , L 15.91mH ,C 50F and R 52
The desired average voltage is: x 2 Vd E 36V , then x 1 1.0385 and uav 0.3333
The linearized model is given as follows:
0 42 2263
x x
-20769 uav
13333 385
31
Chapter 2 Poles Placement Based State Feedback Control of DC-DC Converters
V
60
40
20
0
1
0
uav
0.4
0.36
0.32
0 0.01 0.02 0.03
Time (s)
Fig. 2.13: Boost open loop response
From the open loop characteristic, we get:
n 2242.38(rad / sec) and 0.1784.
It is obviously shown, reading the slower response boost converter with about 30 msec settling
time range. The response presents overshoot peaks of 25% for voltage and 188.8% for the
current, which causes undesirable deviations before the steady-state.
For high efficiency, state feedback control will handle such mentioned problems by providing
a faster response, and reducing the overshoots and deviations occurred within open loop
application
a. State feedback control:
The boost model is controllable as shown in the first chapter
The average law control:
uav Kx k1x1 k2x2
The chosen parameters 1.8, n 2000rad / sec
The gain matrix of those parameters is:
K 2.2809 0.0570004
32
Chapter 2 Poles Placement Based State Feedback Control of DC-DC Converters
uav v
uav u av
Fig.2.14: Boost state feedback control
V v
40 40
30 30
20 20
10 10
0 0
I I
1.5 1.5
1 1
0.5 0.5
0 0
Uav uav
1 1
0.5 0.5
0 0
0 0.004 0.008 0 0.004 0.008
Time (s) Time (s)
(a) (b)
Fig.2.15: Boost closed loop response. (a) R=52, (b) resistance variation R=45.
With no question, state feedback has well enhanced the boost converter response giving an
improved closed-loop behavior upon open-loop behavior, by making it faster, faint oscillations
and a bit low overshoots. In addition the output voltage seems having a problem in achieving a
reachable desired voltage due to steady-state error which is about 9.7%, therefore an integrator
has to be added to state feedback for make sure providing a zero steady-state error.
b. State feedback with integral action:
The boost converter expanded model is:
0 42 0 2263
z 13333 385 0 z 20769 uav
0 1 0 0
With z x1 x2 x .
2
33
Chapter 2 Poles Placement Based State Feedback Control of DC-DC Converters
The augmented control law is given by:
uav k1x1 k2x2 kI x2
According to augmented Boost converter model, the desired polynomial is:
s p s 2 2ns n2 s 3 p 2n s 2 n2 2pn s p n2
The chosen parameters: 1.8, n 1000rad / sec, p 100.
Then the augmented gain matrix is: K I 0.2965 -0.0122 4.1432
i
v
uav
v
u av v
uav
Fig.2.16: Boost state feedback + I control
V V
40 40
30 30
20 20
10 10
0 0
I I
2 2
1.5 1.5
1 1
0.5 0.5
0 0
Uav Uav
1 1
0.5 0.5
0 0
0 0.02 0.04 0 0.02 0.04
Time (s) Time (s)
(a) (b)
Fig.2.17: Boost closed loop response. (a) R=52, (b) resistance variation R=40.
From the above graphs we can well see that the static error that was occurred within state
feedback control is now completely eliminated, which means that the output voltage is exactly
the desired one even with load variation. The desired voltage is now fully reachable, but this
going to cause some overshoots and a bit slow settling time.
34
Chapter 2 Poles Placement Based State Feedback Control of DC-DC Converters
c. Observer based state feedback control
The boost model is defined as follows:
0 42 2263
x x
13333 385 -20769 uav
The observability condition has fulfilled within the first chapter
The average law control:
uav Kx k1xˆ
1 k2x2
The chosen parameters for the observer: 1.8, n 4000rad / sec
The appropriate gain matrix for those parameters is given by:
1158.1
Ke
13900
The full order observer structure:
0 1200. 1 2263 1158.1
xˆ xˆ
u
-20769 av 13900 x2
13333 14285
v
v
i
uav v
uav
v
Observer
u av
Fig.2.18: Boost state feedback with full order observer.
35
Chapter 2 Poles Placement Based State Feedback Control of DC-DC Converters
40
30
20
10
0
I IL_est
1.4
0.7
-0.7
-1.4
Uav
1
0.5
0
0 0.01 0.02 0.03
Time (s)
Fig.2.19: Boost closed loop response
Using the results of our graphs, we can compare the estimated current with the real current of
the inductor of the yielded curve by approximating the estimation error that is 4.14% although
the mentioned estimation error still has clear advantage estimating the real inductor current.
2.5.3 Buck-Boost converter
The assigned values for the buck-boost converter are given by:
E 25V , L 15.91mH ,C 470F and R 25
The desired average voltage is: x 2 Vd E 25V where x 1 0.9615 and uav 0.5
The linearized model is given as follows:
0 31.4 3142.7
x x
2045.8 uav
-1063.8 40.9
This model is controllable and observable as proven within the first chapter
36
Chapter 2 Poles Placement Based State Feedback Control of DC-DC Converters
0
-20
-40
4
2
0
uav
0.52
0.48
0 0.1 0.2
Time (s)
Fig. 2.20: Buck-boost open loop response
Within the open loop characteristic, we get: n 182.846(rad / sec) , 0.1119
The full state feedback control will assure enhancing the slow response buck-boost converter
by improving the settling time (15msec), which means a faster response, and reducing the
overshoots that are about 54% for voltage 330% for current.
a. State feedback control:
0 31.4 3142.7
x x
2045.8 uav
-1063.8 40.9
The average law control:
uav Kx k1x1 k2x2
The chosen parameters 1, n 300rad / sec
The gain matrix of those parameters is:
0.1821
K
-0.0064
uav v
uav u av
Fig.2.21: Buck Boost state feedback control
37
Chapter 2 Poles Placement Based State Feedback Control of DC-DC Converters
V V
0 0
-5 -5
-10 -10
-15 -15
-20 -20
-25 -25
-30 -30
I I
3 3
2 2
1 1
0 0
Uav Uav
1 1
0.5 0.5
0 0
0 0.02 0.04 0 0.02 0.04
Time (s) Time (s)
(a) (b)
Fig. 2.22.Buck-Boost closed loop response. (a) R=52, (b) resistance variation R=40.
As mentioned before in buck and boost converters parts, state feedback has proven once more
again its efficiency by improving the buck-boost response, this latter is now more faster with
lower overshoots and less oscillations. Besides to these advantages still there is the
disadvantage of static error that is about 2.4% which makes a problem to achieve the desired
voltage. So for more improved performance, a state feedback with integrator added has been
designed to work out such problem of non-zero static error.
b. State feedback with Integral action
The buck-boost converter error model is:
0 31.4 0 3142.7
z 1063.8 40.9 0 z 2045.8 uav
0 1 0 0
With z x1 x2 x .
2
The augmented control law is given by:
uav k1x1 k2x2 kI x2
According to augmented Buck-boost converter model, the desired polynomial is:
s p s 2 2ns n2 s 3 p 2n s 2 n2 2pn s p n2
The chosen parameters: 1, n 200rad / sec, p 400 .
Then the augmented gain matrix is: K I 0.2659 -0.0374 -4.7857
38
Chapter 2 Poles Placement Based State Feedback Control of DC-DC Converters
i
uav
uav u av v v
Fig. 2.23 Buck Boost state feedback + I control
V V
0 0
-5 -5
-10 -10
-15 -15
-20 -20
-25 -25
-30 -30
I I
4 4
3 3
2 2
1 1
0 0
Uav Uav
1 1
0.5 0.5
0 0
0 0.04 0.08 0 0.04 0.08
Time (s) Time (s)
(a) (b)
Fig. 2.24. Buck-Boost closed loop response. (a) R=52, (b) resistance variation (R=40).
As show above the desired voltage buck-boost converter has been perfectly reached, despite
the load variation the steady-state error is a proven zero. Still there are a noticeable overshoots
and slower settling time than the one who’s from state feedback control.
c. Observer based state feedback control
The buck-boost model is defined as follows:
0 31.4 3142.7
x x
2045.8 uav
-1063.8 40.9
The observability condition has fulfilled within the first chapter
The average law control:
uav Kx k1xˆ
1 k2x2
The chosen parameters for the observer: 1, n 1000rad / sec
The appropriate gain matrix for those parameters is given by:
39
Chapter 2 Poles Placement Based State Feedback Control of DC-DC Converters
-908.6
Ke
1959.1
The full order observer structure:
0 940 3142.7 -908.6
xˆ xˆ
u
2045.8 av 1959.1 x2
-1063.8 2000
i v
v
v
uav
i
uav
Observer
u av
v
Fig. 2.25 Buck Boost state feedback with full order observer.
V
0
-5
-10
-15
-20
-25
-30
I IL_est
2
0
-2
-4
-6
Uav
1
0.5
0
0 0.02 0.04
Time (s)
Fig. 2.26 Buck Boost closed loop response.
From the above graphs, it’s obviously noticed that the estimation error has successfully
converged to zero, means that i iˆ
2.5.4. Cuk converter
The assigned values for the Cuk converter are given by:
E 100V , L1 L2 30mH ,C 1 150F ,C 2 50F and R 10
40
Chapter 2 Poles Placement Based State Feedback Control of DC-DC Converters
The desired average voltage is: x 4 Vd E 150V , then x 1 22.5,x 2 250,x 3 15 and
uav 0.6
The linearized model is given as follows:
0 -13.3333 0 0 8333.3
2666.7 0 4000 0 -100000
x x uav
0 -20 0 -33.3333 -8333.3
0 0 20000 -2000 0
This model is controllable and observable as proven within the first chapter
V1 V2
300
200
100
0
-100
-200
I1 I2
30
20
10
0
-10
-20
uav
0.648
0.624
0.6
0.576
0.552
0 0.02 0.04
Time (s)
Fig. 2.27: Cuk open loop response
From the figure 2.27 we notice, that in comparison with the desired tensions and currents
values, the steady state errors are very important. Moreover, the responses present very large
oscillations and rather slow times of establishment.
By applying state feedback command it is possible getting a faster response, less oscillations
and a reduced overshoots.
a. state feedback control:
The average law control:
uav Kx k1x1 k2x2 k 3x3 k 4x4
According to Cuk converter model, the desired polynomial is:
s 2 21n1s n21 s 2 22n 2s n22
The chosen parameters 1 2 1, n 1 500rad / sec, n 2 1000rad / sec
The gain matrix of those parameters is:
41
Chapter 2 Poles Placement Based State Feedback Control of DC-DC Converters
K 0.0960 -0.0040 0.0239 -0.0015
v1
i2
i1
uav v2
uav u av
Fig.2.28: Cuk state feedback control
V1 V2 V1 V2
300 300
200
v1 200
v1
100 100
0 0
v2
-100 v2 -100
-200
I1 I2 I1 I2
30 30
20
i1 20
i1
10 10
0
i2 0
-10 -10 i2
-20 -20
Uav Uav
1 1
0.5 0.5
0 0
0 0.02 0.04 0 0.02 0.04
Time (s) Time (s)
(b) (b)
Fig. 2.29: Cuk closed loop response. (a) R=10, (b) resistance variation R=8.5.
The closed-loop response Cuk converter has been effectively enhanced from a slow response to
a faster one and from high oscillated overshoots to less oscillations and low peaks. Although
state feedback doesn’t provide an achieved zero steady-state error, but in case adding an
integrator to state feedback this error will be completely removed.
42
Chapter 2 Poles Placement Based State Feedback Control of DC-DC Converters
b. State feedback with integral action
The Cuk converter error model is:
0 -13.3333 0 0 0 8333.3
2666.7 0 4000 0 0 -100000
z 0 -20 0 -33.3333 0 z -8333.3 uav
0 0 20000 -2000 0 0
0 0 0 1 0
With z x1 x2 x3 x4 x .
4
The augmented control law is given by:
uav k1x1 k2x2 k3x3 k3x4 kI x4
According to augmented Cuk converter model, the desired polynomial is:
(s p ) s 2 21n 1s n21 s 2 22n 2s n22
The chosen parameters: 1 2 1, n 1 500rad / sec, n 2 500 rad / sec, p 40rad / sec
Then the augmented gain matrix is:
K I 0.0253 -0.0011 0.0339 -0.0036 -0.1688
v1
i2
i1 v2
v2 v2
uav
uav
u av
Fig.2.30: Cuk state feedback + I control
43
Chapter 2 Poles Placement Based State Feedback Control of DC-DC Converters
300
200
v1 300
v1
200
100 100
0 0
-100 v2 -100
v2
-200 -200
40
30 i1
30
20
i1
20 10
10
0
0
-10 i2 -10
-20
i2
-20
0.9
uav 0.9 uav
0.8 0.8
0.7 0.7
0 0.04 0.08 0 0.04 0.08
Time (s) Time (s)
(b) (b)
Fig.2.31: Cuk closed loop response. (a) R=10, (b): resistance variation R=8.5.
The output voltage integral is treated as additional state feedback for reason to remove the
static error, so from the Cuk response it can be easily seen that this error has ended up to zero,
even though load variation both desired voltages are now greatly reachable. But this
improvement will induce a slower settling time if it is compared with the one gotten from state
feedback control.
c. Observer based state feedback control
The Cuk model is defined as follows:
0 -13.3333 0 0 8333.3
2666.7 0 4000 0 -100000
x x uav
0 -20 0 -33.3333 -8333.3
0 0 20000 -2000 0
The observability condition has fulfilled within the first chapter
The average law control:
uav Kx k1xˆ
1 k2xˆ
2 k 3xˆ
3 k 4x4
The chosen parameters for the observer:
1 2 1, n1 500rad / sec, n 2 1000rad / sec
The appropriate gain matrix for those parameters is given by:
-129.9
-2883.3
Ke
123.4
1000
44
Chapter 2 Poles Placement Based State Feedback Control of DC-DC Converters
The full order observer structure:
0 -13 0 130 8333.3 -129.9
2667 0 4000 2883 -100000 -2883.3
ˆ
xˆ x uav x4
0 -20 0 -157 -8333.3 123.4
0 0 20000 -3000 0 1000
v
v2
v2
uav i 1
uav v1
uav
i 2
Observer
v2
Fig. 2.32. Cuk state feedback with full order observer.
Voltage_VC1 Voltage_VC2
300
200
100
-100
-200
i1 i2 i1_est i2_est
20
10
-10
-20
Uav
1
0.5
0
0 0.02 0.04
Time (s)
Fig. 2.33: Cuk closed loop response with full order observer.
From the above graphs, it’s obviously noticed that the estimation error has successfully
converged to zero
45
Chapter 2 Poles Placement Based State Feedback Control of DC-DC Converters
2.5.5. Zeta converter:
The assigned values for the Zeta converter are given by:
E 100V , L1 600mH , L2 10mH ,C 1 C 2 10F and R 40
The desired average voltage is: x 4 Vd E 200V , then x 1 10,x 2 200,x 3 5 and
uav 0.666
The linearized model is given as follows:
0 -555.5556 0 0 500000
33333 0 -66667 0 -1500000
x x uav
0 66.6667 0 -100 30000
0 0 100000 -2500 0
This model is controllable and observable as proven within the first chapter
V1 V2
200
100
0
I1 I2
20
10
uav
0.68
0.64
0.6
0 0.004 0.008
Time (s)
Fig. 2.34: Zeta open loop response
From the open loop response Zeta converter we can well notice that in comparison with the
desired voltages and currents values, the static errors are very important. Moreover, the
responses present very large oscillations and rather slow settling times.
The application of state feedback control will definitely improve the Zeta performance by
providing faster response, less deviations and low overshoots
46
Chapter 2 Poles Placement Based State Feedback Control of DC-DC Converters
a. State feedback control
0 -555.5556 0 0 500000
33333 0 -66667 0 -1500000
x x uav
0 66.6667 0 -100 30000
0 0 100000 -2500 0
The average law control:
uav Kx k1x1 k2x2 k 3x3 k 4x4
According to Zeta converter model, the desired polynomial is:
s 2 21n1s n21 s 2 22n 2s n22
s 4 2 1n 1 2n 2 s 3 n22 n21 412n 1n 2 s 2 2n 1n 2(1n 2 2n 1 )s (n 1n 2 )2
The chosen parameters 1 2 0.9, n 1 n 2 3000rad / sec
The gain matrix of those parameters is:
K 0.0136 0.0007 0.0142 -0.0003
i2
i1 v1
uav uav v2
u av
Fig.2.35: Zeta state feedback control
47
Chapter 2 Poles Placement Based State Feedback Control of DC-DC Converters
V1 V2 V1 V2
200
200
150 v1 150 v1
v2 100
100
50
v2
50
0 0
I1 I2 I1 I2
12 i1 14
12
10
8
10 i1
8
6
6 i2
4
i2 4
2 2
0 0
uav uav
1 1
0.5 0.5
0 0
0 0.02 0.04 0 0.02 0.04
Time (s) Time (s)
(a) (b)
Fig.2.36: Zeta closed loop response. (a) R=40, (b) resistance variation (R=25).
The application of feedback command greatly improved the Zeta converter performance,
means faster response, less deviations and low overshoots, despite this improvement, it is still
affected by the inefficiency of the steady-state error, a zero static error dynamic controller for
Zeta converter is designed to achieve zero voltage steady-state error, thus an integrator will be
added to state feedback control that will effectively eliminate the non-zero error as will be
shown next.
b. State feedback with Integral action:
The Zeta converter error model is:
0 -555.5556 0 0 0 500000
33333 0 -66667 0 0 -1500000
z 0 66.6667 0 -100 0 z 30000 uav
0 0 100000 -2500 0 0
0 0 0 1 0 0
with
z x1 x2 x3 x4 x
4
The augmented control law is given by:
uav k1x1 k2x2 k3x3 k3x4 kI x4
48
Chapter 2 Poles Placement Based State Feedback Control of DC-DC Converters
According to augmented Buck converter model, the desired polynomial is:
(s p ) s 2 21n 1s n21 s 2 22n 2s n22
The chosen parameters: 1 2 1, n 1 n 2 3000rad / sec, p 500rad / sec
Then the augmented gain matrix is: K I 0.0152 0.0012 0.0173 -0.0005 0.243
v1
i2
v2
i1
v2 v2
uav uav
u av
Fig.2.37: Zeta state feedback + I control
V1 V2 V1 V2
v2 250
v2
200 200
150
100
v1
100
0 50
v1
0
I1 I2 I1 I2
20
20
15 i1 15
i1
10 10
5
0
i2 5
i2
0
uav uav
0.8 uav 1
uav
0.6
0.5
0.4
0
0 0.02 0.04
0 0.02 0.04
Time (s) Time (s)
(a) (b)
Fig.2.38: Zeta closed loop response. (a) R=40, (b) resistance variation (R=25).
As shown, the Zeta performance has well improved after the application of the same control
approach but with integral action. With or without load variation, a zero steady-state error is
actually obtained with an acceptable settling time and less oscillation have proven too.
49
Chapter 2 Poles Placement Based State Feedback Control of DC-DC Converters
c. Observer based state feedback control
The Zeta model is defined as follows:
0 -555.5556 0 0 500000
33333 0 -66667 0 -1500000
x x uav
0 66.6667 0 -100 30000
0 0 100000 -2500 0
The observability condition has fulfilled within the first chapter
The average law control:
uav Kx k1xˆ
1 k2xˆ
2 k 3xˆ
3 k 4x4
The chosen parameters for the observer: 1 2 1, n1 n 2 8000rad / sec
The appropriate gain matrix for those parameters is given by:
-11650
196980
Ke
3510
29500
The full order observer structure:
0 -560 0 11650 500000 -11650
33330 0 -66670 -196980 -1500000 196980
ˆ
xˆ x uav x4
0 70 0 -3610 30000 3510
0 0 100000 -32000 0 29500
v
v2
v2 i 1
uav v1
uav
i 2
u av
Observer
v2
Fig. 2.39. Zeta state feedback with full order observer.
50
Chapter 2 Poles Placement Based State Feedback Control of DC-DC Converters
V1 V2
400
300
200
100
i1 i2 i1_est i2_est
60
40
20
0
-20
-40
-60
Uav
0.6
0
0 0.002 0.004
Time (s)
Fig. 2.40: Zeta closed loop response with full order observer.
From the above graphs, it’s obviously noticed that the estimation error has successfully
converged to zero
51
Chapter 2 Poles Placement Based State Feedback Control of DC-DC Converters
2.6. Conclusion:
This chapter has demonstrated the process of applying modern control design methods to
regulator design for DC to DC converters. Full state feedback control for pole placement was
applied – first without integral action, then with an integrator added, underling the clear
advantages of state feedback, that has a positive effect on response settling time, reducing the
undesirable peak overshoots and serve having a less oscillated performance, referring that this
approach doesn’t provide a zero static error, this latter has been solved by adding an integral
action to state feedback control, that has proven its efficiency working out the steady-state
error then the use of state estimation technique with Full-Order Estimator was discussed and
simulated, that comes out with the advantages of the observer in estimating those non
measurable state variables basically the current(s).
52
Chapter 3 Passivity based control of DC-DC converters
Chapter 3
Passivity based control of DC-DC converters
3.1 Introduction
Within this chapter, the behavior of DC-DC converters is being controlled this time with
another control approach so-called based passivity control. In this part, we’ll try pursuing the
application of passivity based control technique to improve the dynamic behavior of DC-DC
converters. Sections 3.2, 3.3 and 3.4 present the theoretical elements necessary for the use of
these techniques, namely, the based passivity and the based passivity control with non-linear
observer. Section 3.5 presents the application of these techniques to different DC-DC
converters modeled in the first chapter, and section 3.6 concludes the chapter. The PBC theory
is taken from [2] and [3].
3.2 DC-DC Converters energetic models
The modeled DC-DC converters show a clear “Energy Management” structure. It is considered
in this section that the general model of the static converters is given by:
Dx J (uav )x Rx Buav (3.1)
Where x Rn , u Rm , B Rnm matrix, D is a symmetric definite positive matrix, J (uav )
is an anti-symmetric matrix (J (uav ) J (uav ) 0) , it is also a function of the variable uav :
m
J (uav ) J 0 J iuiav
i 1
The matrix R is a symmetric semi-definite positive, i.e. R RT 0 . The term Rn
represents the constant output voltage sources.
The term J (uav )x represents the conservative (workless) forces, and the term Rx represents
the dissipative forces in the converter. The term Buav represents the acquisition of the energy.
1
An explanation of this terminology is that, if V x T Dx is the "energy state", so, its
2
derivative with respect to time
V x TJ (uav )x x T Rx x T Buav x T (3.2)
exhibits a zero term x TJ (uav )x , i.e., J (uav )x provides invariant part of the energy, a semi-
definite negative term x T Rx , and the term x T Buav responsible of the acquisition or the
subtraction of the energy of the converter. Whereas the passive output is defined as y BT x ,
the latter term is simply the product yuav so-called the acquisition rate.
3.3 Control design
For the desired equilibrium point (x , uav ) , and using the fact that the matrix J (uav ) is affine in
the variable uav , yields the follows relation:
53
Chapter 3 Passivity based control of DC-DC converters
dJ (uav )
J (uav ) J (uav ) (uav uav ) (3.3)
duav u
av uav
We get at equilibrium:
0 J (uav )x Rx Buav (3.4)
Defining the errors as next: e x x and eu uav uav obviously e x .
The subtraction of (3.1) and (3.4), we get the following error dynamic:
dJ (u )
De J (uav )e R.e av
x eu Beu (3.5)
duav uav uav
Then we simply get:
dJ (u )
De J (uav )e R.e av
x B eu (3.6)
duav uav uav
T
dJ (u )
If the output error is defined as follows: ey av
x B e , which is the passive
duav uav uav
output of the system (3.6), it can be written:
dJ (u )
De J (uav )e R.e av
x B eu (3.7)
duav uav uav
T
dJ (u )
ey av
x B e (3.8)
duav uav uav
In order to stabilize the error dynamic, the control law is chosen as follows:
T
dJ (u )
eu ey av
x B e (3.9)
duav uav uav
Where 0 of dimension (m m ) .
The closed-loop error dynamic is then as following:
dJ (u ) dJ (u )
T
De J (uav )e R av
x B av
x B e (3.10)
duav u u duav u u
av av av av
For demonstrating the stability of the error dynamic (3.10), we evaluate the total derivation of
1
the energy function V (e ) eT De that makes:
2
54
Chapter 3 Passivity based control of DC-DC converters
dJ (u ) dJ (u )
T
V e J (uav )e e R
T T av
x B av
x B e
duav u u duav u u
av av av av
(3.11)
dJ (u ) dJ (u )
T
eT R av
x B av
x B e 0
duav u u duav u u
av av av av
The derivative V is negative definite just if the following condition holds
dJ (u )
T
R dJ (uav )
x B av
x B 0 (3.12)
du duav u u
av uav uav av av
That implies the following control law:
T
dJ (u )
uav uav ey uav av
x B e (3.13)
duav uav uav
makes the origin globally asymptotically stable, if the condition (3.12) called “dissipativity
matching” is verified.
3.4 Non-linear observer
We consider a non-linear observer structure for the estimation of the non measurable states. Far
from the proposed linear observer within the second chapter, the non-linear observer uses
directly the non-linear model of the converter, which means that there is no need for the
linearizing step.
Here we consider the converter model given by:
Dx J (uav )x Rx Buav
(3.14)
y CTx
Where x Rn , uav Rm and y R p .
Since the converter is observable, the non-linear observer is given by:
Dxˆ J (uav )xˆ Rxˆ Buav K (y yˆ)
(3.15)
yˆ C T xˆ
Then the estimation error is as follows:
Deˆ J (uav )
e R.
e Key
(3.16)
ey C T
e
Where
e x xˆ is the state estimation error,
e y y yˆ is the output estimation error.
If the proposed gain is of the form: K CL , with the matrix L R pp is positive definite, the
error dynamic is then:
Deˆ J (uav )
e R CLC T
e
(3.17)
ey C T
e
55
Chapter 3 Passivity based control of DC-DC converters
To demonstrate the stability of the estimation error dynamic (3.17), we evaluate the total
1
derivation of the Lyapunov function V (eˆ) eˆT Deˆ that makes:
2
V
eTJ (uav ) eT R CLC T
e e
(3.18)
eT R CLC T
e0
The derivative V is negative definite just if the matrix R CLC T is definite positive. This
condition is called "dual dissipativity matching".
3.5 Application
a. Passivity based control
1. Determine the converter energetic average model. This model usually takes the form (3.1).
2. Define the constant desired operating values (x , uav ) (current, voltage).
3. Choose in the control law (3.9).
4. Verify the dissipativity matching condition (3.12).
5. In order to apply uav , it must be converted to gate signal u 0,1 , this function is realized
via an PWM technique as shown within chapter 2.
PWM
u Converter DC-DC
uav
Passivity based x Voltage and current
control law measurement
Figure 3.1: DC-DC converters Passivity based control scheme.
b. Non-linear observer
1. Choose the observer gain K CL .
2. Verify the dual dissipativity matching condition.
3. Implement the observer as in (3.15).
PWM u Converter DC-DC
y
uav
Passivity Based
x Non-linear
control law observer
Fig 3.2: Passivity based control with observer
56
Chapter 3 Passivity based control of DC-DC converters
3.5.1 BUCK CONVERTER
a. Passivity based control
The Buck average model:
Lx1 x 2 uav E
1
Cx 2 x 1 x2
R
This model can be written as follows:
L 0 0 1 0 0 E
x x
x
0 C 1 0 0
1 0 uav
R
E
The desired voltage x 2 EVd then x 1 Vd , and uav Vd .
R
The error dynamic: e1 x 1 x 1 e2 x 2 x 2 is given by:
L 0 0 1 0 0 E
e
e e
0 C 1 0 1 e 0 u
0
R
ey E 0 e
The control law:
uav uav eu uav ey Vd E 0 e Vd Ee1
Since the matrix:
0 0 E 2 0
E
0
0 1
0 E 0 1 0
R R
The dissipativity matching condition is verified.
The PBC scheme is given in figure 3.3.
uav u av
Figure 3.3: Buck passivity based control.
57
Chapter 3 Passivity based control of DC-DC converters
12
10
8
6
4
2
0
i
0.6
0.4
0.2
0
Uav
1
0.5
0
0 0.004 0.008
Time (s)
Figure 3.4: Buck closed loop response 0.1 .
From the buck open loop response, the buck performance was slower, sensitive and had
followed by undesirable deviations, but after the application of the passivity based control, the
performance we get has a faster response, less deviations, low amplitude start-up peaks and
almost achieved zero steady-state error, i.e. the buck behavior has been perfectly improved.
b. Non-linear observer
The non-linear observer form:
L 0 0 1 0 0 u E k
xˆ av 1 x xˆ
0 C xˆ 1 0 1 ˆ
x 0 k 2
0 2
2
R
The dynamic of the estimation error is given by:
eˆT eˆ1 eˆ2 , with eˆ1 x 1 xˆ1, eˆ2 x 2 xˆ2
L 0 0 1 0 0 k1
eˆ
eˆ
0 C eˆ 1 0 0
1 ˆ
e k 2
2
R
1 1
Considering the Lyapunov function; V (Leˆ12 Ceˆ22 ) Then V k1eˆ1eˆ2 (k2 )eˆ22 .
2 R
1 2
If k1 0 and k2 0 we get (k2 )eˆ2 0 , that is negative semi-definite.
R
The observer:
Lxˆ1 xˆ2 Euav
1
Cxˆ 2 xˆ1 xˆ2 k2 x 2 xˆ2
R
58
Chapter 3 Passivity based control of DC-DC converters
i
v
i
uav
Observer
v̂
Fig. 3.5: Buck non-linear observer.
v
12
10
8
6
4
2
0
i i_est
0.8
0.6
i
0.4 i
0.2
0
Uav
1
0.5
0
0 0.01 0.02
Time (s)
Fig. 3.6: Buck closed loop response with non-linear observer 0.02,k2 0.01 .
It is clear that the observer performance is not good enough. There is a residual error on the
estimation of the current, which induces an error on the tracking of the reference voltage.
3.5.2 Boost converter
a. Passivity based control
The Boost average model:
Lx1 (1 uav )x 2 E
1
Cx 2 (1 uav )x 1 x2
R
This model can be written as follows:
59
Chapter 3 Passivity based control of DC-DC converters
L 0 0 (1 uav ) 0 0 E
x x
x
0 C (1 u ) 1 0
0 0
av
R
E 2 1
The desired voltage x 2 EVd , then x 1 Vd and uav 1
R Vd
The error dynamic can be written as next
L 0 0 (1 uav ) 0 0 Vd E
e e
0 C (1 u ) 1 e E 2 eu
0 0 Vd
av
R R
E
ey Vd E Vd2 e
R
The control law:
E E
eu ey Vd E Vd2 e Vd Ee1 Vd2e2
R R
The closed loop dynamic is then as follows:
Vd E
L 0 0 (1 uav ) 0 0 E
e
0 C e (1 u ) 1 E 2 Vd E Vd e
2
0 0 V R
R R
av d
Since the matrix:
E2 3
0 0 Vd E Vd2E 2 Vd
E R
1 E 2 Vd E Vd2 0
0 Vd R E2 3 1 E2 4
R R Vd 2 Vd
R R R
The dissipativity matching condition is verified.
The control law will be then as follows:
E 2
uav uav ey uav Vd Ee1 Vd e2
R
E 2
With e1 x 1 Vd e2 x 2 EVd
R
uav u av
Figure 3.7: Boost passivity based control.
60
Chapter 3 Passivity based control of DC-DC converters
V
40
30
20
10
0
i
1.5
0.5
0
Uav
1
0.5
0
0 0.004 0.008
Time (s)
Figure 3.8: Boost closed loop response 0.05 .
It is obviously had been shown reading the open loop slower response of boost converter with
large settling time range, the response was also sensitive and was followed by start-up peak
overruns and noticeable oscillations before the steady-state. Now the previous drawbacks are
greatly reduced, means the boost closed-loop behavior has became faster, with reduced start-up
peaks overshoots and no occurred oscillations, and last a nearly achieved zero static error .
Otherwise the passivity based control has given absolutely an improved closed loop
performance than the one we get from open loop characteristic.
b. Non-linear observer
The non-linear observer form
L 0 0 (1 uav ) 0 0 E k1
xˆ
x xˆ
0 C xˆ (1 u ) 0 0
1 ˆ
x 0 k 2 2
av
R
2
The dynamic of the estimation error is given by:
L 0 0 (1 uav ) 0 0 k1
eˆ
0 C eˆ (1 u ) 0 0
1 e k e2
ˆ
2
R
av
with eˆT eˆ1 eˆ2 , eˆ1 x 1 xˆ1, eˆ2 x 2 xˆ2 .
1 1
Considering the Laypunov function V (Leˆ12 Ceˆ22 ) , then V k1eˆ1eˆ2 (k2 )eˆ22 .
2 R
1
For k1 0 and k2 0 , we get V (k2 )eˆ22 0 , that is negative semi-definite.
R
The observer is then:
61
Chapter 3 Passivity based control of DC-DC converters
Lxˆ1 (1 uav )xˆ2 E
1
Cxˆ 2 (1 uav )xˆ1 xˆ2 k2 x 2 xˆ2
R
i
uav Observer
v̂
Fig. 3.9: Boost non-linear observer.
V
40
30
20
10
0
i IL_est
2
i
1.5
1 i
0.5
0
Uav
1
0.5
0
0 0.01 0.02
Time (s)
Fig. 3.10: Boost closed loop response with non-linear observer 0.01,k2 0.02 .
According to the gotten results from the above graphs, the convergence of the estimated
current towards its true value is perfectly achieved
62
Chapter 3 Passivity based control of DC-DC converters
3.5.3 Buck-Boost converter
a. Passivity based control
The Buck-Boost average model:
L 0 0 (1 uav ) 0 0 E
x x
x u
0 C (1 u ) 1 0 av
0 0
av
R
E Vd
The desired voltage is x 2 EVd , then x 1 Vd (vd 1) and uav .
R Vd 1
The error dynamic can be written as follows;
0 0 (1 Vd )E
L 0 0 (1 uav )
e e
e E eu
0 C (1 u ) 0 1
0
av
R R Vd (Vd 1)
E
ey (1 Vd )E Vd (Vd 1) e
R
The proposed control law:
E E
eu ey (1 Vd )E Vd (Vd 1) e (1 Vd )Ee1 Vd (Vd 1)e2
R R
The closed loop dynamic will be then:
0 0 (1 Vd )E
L 0
e
0 (1 uav )
e E
0 C 1 E (1 Vd )E Vd (Vd 1) e
(1 u ) 0 0 R
av
R
V (
R d dV 1)
Since the matrix
E2 3
0 0 (1 Vd )E Vd2E 2 Vd
E R
1 E (1 Vd )E Vd (Vd 1) 0
0 V (V 1) R E2 3 1 E 2
4
R R Vd 2 Vd
d d
R R R
Is positive definite, the dissipativity matching condition is verified.
The control law is as next:
E
uav uav ey uav (1 Vd )Ee1 Vd (Vd 1)e2
R
E
Where e1 x 1 Vd (vd 1) et e2 x 2 Vd E
R
63
Chapter 3 Passivity based control of DC-DC converters
uav u av
Figure 3.11: Buck-boost based passivity control
v
0
-10
-20
-30
-40
i
2
1.5
0.5
0
Uav
1
0.75
0.5
0.25
0
0 0.05 0.1 0.15 0.2
Time (s)
Figure 3.12: Buck-boost closed loop response
The passivity based control has positively enhanced the slow open loop response buck-boost
converter by making a faster response, removing peaks overruns and no deviations have been
occurred.
b. Non-linear observer:
The non-linear observer form
L 0 0 (1 uav ) 0 0 u E k
xˆ
av 1 x xˆ
0 C xˆ (1 u ) 0 0 1
ˆ
x 0 k 2 2
av
R
2
The dynamic of the estimation error is given by:
L 0 0 (1 uav ) 0 0 k1
eˆ
eˆ
0 C eˆ (1 u ) 0 0
1
ˆ
e k 2
av 2
R
with eˆT eˆ1 eˆ2 , eˆ1 x 1 xˆ1, eˆ2 x 2 xˆ2 .
64
Chapter 3 Passivity based control of DC-DC converters
1 1
Considering the Laypunov function V (Leˆ12 Ceˆ22 ) , then V k1eˆ1eˆ2 (k2 )eˆ22 .
2 R
1
For k1 0 and k2 0 , we get V (k2 )eˆ22 0 , that is negative semi-definite.
R
The observer is given as follows:
Lxˆ1 (1 uav )xˆ2 Euav
1
Cxˆ 2 (1 uav )xˆ1 xˆ2 k2 x 2 xˆ2
R
i
i v
i
uav uav v̂ Observer
Fig. 3.13: Buck-boost non-linear observer.
V
0
-10
-20
-30
-40
i i_est
2
1.5
0.5
0
Uav
1
0.5
0
0 0.05 0.1 0.15 0.2
Time (s)
Fig. 3.14: Buck-boost closed loop response with non-linear observer.
From the above graphs, it is obviously seen that the convergence of the estimated current
towards its true value is a bit achieved, mentioning that the error estimation is not completely
zero.
65
Chapter 3 Passivity based control of DC-DC converters
3.5.4 Cuk converter
a. Passivity based control
The Cuk average model:
L1 0 0 0 0 (1 uav ) 0 0 0 0 0 0 E
(1 u ) 0 0 0 0
0 C
0 0 0 uav 0 0
1
x
av
x 0 x
0 0 0
0 0 L2 0 0 uav 0 1 0
1
0 0 0 C2 0 0 1 0 0 0 0 0
R
E 2 E Vd
If the desired voltage x 4 EVd , then x 1 Vd , x 2 E (1 Vd ) , x 3 Vd et uav
R R Vd 1
The error dynamic can be written as follows:
0 0 0 0 0 E (1 Vd )
L 0 0 0 (1 uav ) 0 0
1 E
(1 u ) 0 0 0 0
0 C
0 0 0 uav 0 V (1 V )
1
e
av
e 0 e R d d
0 0 0 eu
0 0 L2 0 0 uav 0 1 )
E Vd
(1
0 0 0 C2 1
0 0 1 0 0 0 0
R 0
E
ey E (1 Vd ) Vd (1 Vd ) E (1 Vd ) 0 e
R
E 2 E
with e1 x 1 x 1 x 1 Vd , e2 x 2 x 2 x 2 E (1 Vd ) , e3 x 3 x 3 x 3 Vd ,
R R
Vd
e4 x 4 x 4 x 4 Vd E and eu uav uav uav Vd ( uav eu ).
1 Vd
The proposed control law:
E
eu ey E (1 Vd ) Vd (1 Vd ) E (1 Vd ) 0 e
R
E
E (1 Vd )e1 Vd (1 Vd )e2 E (1 Vd )e3
R
The closed loop dynamic will be then:
0 0 0 0 E (1 Vd )
0 0 0 0 E
V (1 V ) E
De J uav e 0 d d E (1 V ) Vd (1 Vd ) E (1 Vd ) 0 e
0 0 0 R d
E (1 Vd ) R
1
0 0 0
R 0
Since the matrix
66
Chapter 3 Passivity based control of DC-DC converters
E2
E 2(1 Vd )2 Vd (1 Vd )2 E 2(1 Vd )2 0
R
E2 E E2
Vd (1 Vd )2 ( Vd (1 Vd ))2 Vd (1 Vd )2 0
R R R
2 0
2 2 E
E (1 Vd ) Vd (1 Vd )2 E 2(1 Vd )2 0
R
1
0 0 0
R
Is positive definite, the dissipativity matching condition is verified.
The control law is given by:
Vd V
uav E (1 Vd ) e1 d e2 e3
Vd 1 R
v1
i1
uav u av
Figure 3.15: Cuk based passivity control
67
Chapter 3 Passivity based control of DC-DC converters
V1 V2
300
200
100
-100
-200
I1 I2
20
10
-10
-20
Uav
1
0.5
0
0 0.02 0.04
Time (s)
Figure 3.16: Cuk closed loop response.
The closed-loop response of Cuk converter has now the following properties; faster response,
less oscillations and low start-up overruns peaks, so the passivity control has given a closed
loop performance with an acceptable efficiency
b. Non-linear observer:
The non-linear observer form
L1 0 0 0 0 (1 uav ) 0 0 0 0 0 0 E k1
0 C (1 u ) 0 0 0 0 0 k
0 0 0 uav 0
1
xˆ
av
x 0
ˆ xˆ 2 x xˆ
0 0 0 2 2
0 0 L2 0 0 uav 0 1 0 k 3
1
0 0 0 C2 0 0 1
0 0 0 0 0 k 4
R
The dynamic of the estimation error is given by:
eˆT eˆ1 eˆ2 eˆ3 eˆ4 , With eˆi x i xˆi , i 1, 2, 3, 4 .
L 0 0 0 0 (1 uav ) 0 0 0 0 0 0 k
1 1
(1 u ) 0 0 0 0
0 C 0 0
0 uav 0 k
1
eˆ
av
ˆ
e eˆ 2 eˆ
0 0 L 0 0 uav 0 1 0 0 0 0
k 3
2
2
0 0 0 C2 0 1 k
0 1 0 0 0 0
R 4
Considering the Lyapunov function
1
V (L1eˆ12 C 1eˆ22 L2eˆ32 C 2eˆ42 )
2
then
68
Chapter 3 Passivity based control of DC-DC converters
1
V (k1eˆ1 k3eˆ3 k 4eˆ4 )eˆ2 (k2eˆ22 eˆ42 )
R
1
If it is supposed that k1 k 3 k 4 0 and k2 0 makes V (k2eˆ22 eˆ42 ) 0 is
R
negative semi-definite.
The observer is given as follows:
L1xˆ1 (1 uav )xˆ2 E
C 1xˆ 2 (1 uav )xˆ1 uav xˆ3 k2(x 2 xˆ2 )
L2xˆ1 uav xˆ2 xˆ4
1
C 2xˆ 2 xˆ3 xˆ4
R
v
v1
Observer
iˆ1
v̂1
uav
iˆ2
v̂2
Figure 3.17: Cuk non-linear observer.
V1 V2
300
200
100
-100
-200
I1 i1est I2 i2est
20
10
-10
-20
Uav
1
0.5
0
0 0.02 0.04
Time (s)
Figure 3.18: Cuk closed loop response with non-linear observer.
69
Chapter 3 Passivity based control of DC-DC converters
The convergence of the estimation of the state variables towards their true values is
approximately achieved via the nonlinear observer given.
3.5.5 Zeta converter
a. Passivity based control
The Zeta average model:
L1 0 0 0 0 (1 uav ) 0 0 0 0 0 0 Euav
0 C (1 u ) 0 0 0 0
1 0 0 0 uav 0 0
x 0
av
x 0 0 0 x Eu
0 0 L2 0 0 uav 0 1 av
1
0 0 0 C2 0 0 1 0 0 0 0 0
R
E 2 E Vd
If the desired voltage x 4 EVd , then x 1 Vd , x 2 EVd , x 3 Vd and uav .
R R Vd 1
The error dynamic is given by:
L1e1 (1 uav )e2 E (Vd 1)eu
E
C 1e2 (1 uav )e1 uave3 Vd (Vd 1)eu
R
,
L2e3 uave2 e 4 E (Vd 1)eu
1
C 2e4 e3 e4
R
this can be written as follows;
0 0 0 0 0 E (Vd 1)
L 0 0 0 (1 uav ) 0 0
1
0 C
0 0 (1 u )
0 uav 0 0 0 0 0 E V (V 1)
1
e
av
e 0 e R d d
0 0 0 eu
0 0 L2 0 0 uav 0 1 E (Vd 1)
1
0
0 0 0 C 2 0 0 1
0 0 0
R
0
E
ey E (Vd 1) Vd (Vd 1) E (Vd 1) 0 e
R
The proposed control law is:
E
eu ey E (Vd 1) Vd (Vd 1) E (Vd 1) 0 e
R
The closed loop dynamic will be then:
70
Chapter 3 Passivity based control of DC-DC converters
E (Vd 1)
0 00 0
0 E
0 0 0 V (V 1)
d d E
De J (uav )e 00 0 0 R E (Vd 1) R Vd (Vd 1) E (Vd 1) 0
e
1
E (Vd 1)
0
0 0
0
R
1
Evaluating the total derivation of energy function V (e ) eT De that makes:
2
E (Vd 1)
0 0 0 0
0 E
0 0 0 V (V 1)
E
V eT 0 d d
0 0 0 R E (Vd 1) R Vd (Vd 1) E (Vd 1) 0e
E (Vd 1)
1
0 0 0
0
R
Where
0 0 0 0 E (Vd 1)
0 0 0 E
0 V (V 1)
d d E (V 1) E V (V 1) E (V 1) 0
0 0 0 0 e R d d d d
E (Vd 1) R
1
0 0 0 0
R
E2
E 2(Vd 1)2 Vd (Vd 1)2 E 2(Vd 1)2 0
R
E 2
E E2
Vd (Vd 1)2 ( Vd (Vd 1))2 Vd (Vd 1)2 0
R R R
2
2 2 E
E (Vd 1) Vd (Vd 1)2 E 2(Vd 1)2 0
R
1
0 0 0
R
Which is semi positive definite so V is semi negative definite
The control law is given by:
Vd E
uav E (Vd 1)e1 Vd (Vd 1)e2 E (Vd 1)e3
Vd 1 R
Vd V
E (Vd 1) e1 d e2 e3
Vd 1 R
71
Chapter 3 Passivity based control of DC-DC converters
v1
i1
uav u av
Figure 3.19: Zeta based passivity control
v1 v2
200
150
100
50
0
i1 i2
12
10
8
6
4
2
0
Uav
1
0.8
0.6
0.4
0.2
0
0 0.004 0.008
Time (s)
Figure 3.20: Zeta closed loop response.
The closed-loop response of Zeta converter has now the following properties; faster response,
less oscillations and low start-up overruns peaks, so the passivity command has given a closed
loop performance with an acceptable efficiency
72
Chapter 3 Passivity based control of DC-DC converters
b. Non-linear observer:
The non-linear observer form
L1 0 0 0 0 (1 uav ) 0 0 0 0 0 0 E k1
0 C (1 u ) 0 0 0 0 0 k
0 0 0 uav 0
1
xˆ
av
xˆ 0 xˆ 2 x xˆ
0 0 0 4 4
0 0 L2 0 0 uav 0 1 0 k 3
1
0 0 0 C2 0 0 1
0 0 0 0 0 k 4
R
The dynamic of the estimation error is given by:
L1 0 0 0 0 (1 uav ) 0 0 0 0 0 0 k1
(1 u ) 0 0 0 0
0 C 0 0
0 uav 0 k
1
eˆ
av
eˆ 0 eˆ 2 eˆ
0 0 0 4
0 0 L2 0 0 uav 0 1 k3
1
0 0 0 C2 0 0 1 0 0 0 0 k 4
R
With : eˆT eˆ1 eˆ2 eˆ3 eˆ4 , With eˆi x i xˆi , i 1, 2, 3, 4 .
Considering the Lyapunov function
1
V (L1eˆ12 C 1eˆ22 L2eˆ32 C 2eˆ42 )
2
then
1
V (k1eˆ1 k3eˆ3 k 4eˆ4 )eˆ2 (k2eˆ22 eˆ42 )
R
1
If it is supposed that k1 k 3 k 4 0 and k2 0 makes V (k2eˆ22 eˆ42 ) 0 is
R
negative semi-definite.
The observer is given as follows:
L1xˆ1 (1 uav )xˆ2 uav E
C 1xˆ 2 (1 uav )xˆ1 uav xˆ3
L2xˆ1 uav xˆ2 xˆ4 uavE
1
C 2xˆ 2 xˆ3 xˆ4 k 4 (x 4 xˆ4 )
R
73
Chapter 3 Passivity based control of DC-DC converters
v2
Observer
iˆ1
uav v̂1
iˆ2
v̂2
Figure 3.21: Zeta non-linear observer.
VC1 VC2
200
150
100
50
0
IL1 IL2 i1est i2est
14
12
10
8
6
4
2
0
Uav
1
0.8
0.6
0.4
0.2
0
0 0.01 0.02
Time (s)
Figure 3.22: Zeta closed loop response with non-linear observer.
The convergence of the estimation of the state variables towards their true values is
approximately achieved via the nonlinear observer given.
74
Chapter 3 Passivity based control of DC-DC converters
3.6 Conclusion:
Another control technique has been demonstrated through this chapter, underling the advantage
using directly the non-linear model of the DC-DC converter, so there is no more need to the
linearization for the application of passivity based control approach and its non-linear observer,
this approach has proven its efficiency in improving the performance of dc-dc converters,
which has a perfect effect on the behavior response by making it faster which means settling
time in milliseconds range, and help getting closed loop response with a low peak overruns and
lesser deviations.
75
Conclusion
Conclusion
DC-DC simple PWM converters controllers and linear observers are derived in this humble
work, with a proof of controllability for linear state feedback, and observability for state
feedback estimation, in addition of passivity based control and non-linear observers are then
derived, each with a Lyapunov proof of stability, all these approaches have been used whether
to regulate the output voltage or estimate the non-measurable currents. The simulation study
well illustrates the theoretical results pointing out a perfect robustness properties of such
mentioned controllers. Note that these PWM feedback regulation schemes have been applied
on many converter models taking into account the nature of nonlinear parameterization
problem of an dc-dc converter, that has been studied within chapter one. The voltage mode
control full state feedback by pole placement and PBC work well when the load is constant, in
case if the load is changed, the static error is one drawback of the control circuit, means, even
though the remarkable effects of those techniques on the converter behavior, by making the
converter performance more faster, reduced start-up overruns and less oscillations as possible,
still, when it comes to realistic situations the converters are absolutely subject to static
disturbance due to load variation, so, when the load is up or down than the proposed one, the
control circuit cannot attain the desired output voltage and the difference between the desired
and true voltage can be forced to zero by adding an integrator to the average linear control
law, i.e. the state variable that appears directly in the output of the converter is introduced as
additional integral state variable, which will make sure having an achieved zero static error,
unfortunately this will sometimes cause some overshoots and a bit slow settling time as
shown within the Boost and Buck-Boost application. The use of linear state observer has been
built for reason to sense the non-measurable currents of dc-dc converters, hence, the estimated
current and the measured output voltage have been used to elaborate the linear state feedback
control law, within steady state, the observers do well estimating the inductor currents, by
providing a completely achieved estimated currents, means that the estimation error is fully
converged to zero, other than the transient regime of the converter response, it’s an obviously
shown that the circuit control has a failed estimation action. As for the non-linear observer
based on PBC once shares some properties of the linear observer in what concerns the Zeta
converter, means within steady state a perfect estimation is provided and a failed estimation
within transient regime, on the other hand this observer is sometimes not efficient as proven
within Buck and Boost converters.
So as to conclude, at it has been shown through this piece of work that the dc-dc
converters have a high properties after the control application, means that these converters
work more better within the closed loop response than the ones within open loop response.
Furthermore the techniques control developed during this work could be applied to many
other identifiable nonlinear converters. In addition to an advantageous improvement generally
good, robustness and simplicity properties additionally convergence of state variables (relying
on Linear State Feedback & PBC Nonlinear Observers) are generally achieved.
76
Bibliography
Bibliography
[1] M. K. Kazimierczuk, Pulse-width Modulated DC–DC Power Converters. John
Wiley & Sons, Chichester. UK 2008.
[2] H. Sira-Ramerez and R. Silva-Ortigoza, Control Design Techniques in Power
Electronics Devices, Springer-Verlag London 2006.
[3] R. Ortega, A.Loria, P. J. Nicklasson, H. Sira-Ramerez, Passivity-based Control of
Euler-Lagrange Systems: Mechanical, Electrical and Electromechanical
Applications. 1998.
[4] R. L. Williams II and D. A. Lawrence, Linear state space control systems. John
Wiley & Sons, Hoboken, New Jersey. 2007.
77
You might also like
- ECE311S Dynamic Systems & Control: Course Notes by Bruce Francis January 2010No ratings yetECE311S Dynamic Systems & Control: Course Notes by Bruce Francis January 2010116 pages
- ECE311S Dynamic Systems & Control: Course Notes by Bruce Francis January 2010No ratings yetECE311S Dynamic Systems & Control: Course Notes by Bruce Francis January 2010116 pages
- A Modified Method For Tuning PID Controller For Buck-Boost ConverterNo ratings yetA Modified Method For Tuning PID Controller For Buck-Boost Converter7 pages
- Multiple Controllers For Boost Converters Under Large Load Range: A GA and Fuzzy Logic Based ApproachNo ratings yetMultiple Controllers For Boost Converters Under Large Load Range: A GA and Fuzzy Logic Based Approach6 pages
- On DC/DC Voltage Buck Converter Control Improvement Through The QFT ApproachNo ratings yetOn DC/DC Voltage Buck Converter Control Improvement Through The QFT Approach8 pages
- Informe IEEE - Baltazar - Palma - ParedesNo ratings yetInforme IEEE - Baltazar - Palma - Paredes12 pages
- Pole Placement by State Feedback in DC Motor by MatlabNo ratings yetPole Placement by State Feedback in DC Motor by Matlab13 pages
- Optimal Control of A High Gain DC - DC ConverterNo ratings yetOptimal Control of A High Gain DC - DC Converter11 pages
- Design and Analysis of Controllers For BNo ratings yetDesign and Analysis of Controllers For B66 pages
- Model Predictive Control of A DC-DC Buck ConverterNo ratings yetModel Predictive Control of A DC-DC Buck Converter7 pages
- Voltage Mode SMC of DC-DC Buck Converters: I Nigo Larrea Sridhar SeshagiriNo ratings yetVoltage Mode SMC of DC-DC Buck Converters: I Nigo Larrea Sridhar Seshagiri6 pages
- Modelling and Model Predictive Control of A DC-DC Boost ConverterNo ratings yetModelling and Model Predictive Control of A DC-DC Boost Converter6 pages
- Comparative Studies of PID Controller Tuning Methods On A DC-DC Boost ConverterNo ratings yetComparative Studies of PID Controller Tuning Methods On A DC-DC Boost Converter5 pages
- 2010 - State Space Controller For Bidirectional DC-DC Converter Buck Mode - OcilkaNo ratings yet2010 - State Space Controller For Bidirectional DC-DC Converter Buck Mode - Ocilka3 pages
- Design and Control of A Buck-Boost DC-DC Power Converter100% (1)Design and Control of A Buck-Boost DC-DC Power Converter65 pages
- LQR-PID Control Design of A Buck ConverterNo ratings yetLQR-PID Control Design of A Buck Converter5 pages
- Fuzzy Logic Control of A Switched-Inductor DC-DC Buck Converter in CCMNo ratings yetFuzzy Logic Control of A Switched-Inductor DC-DC Buck Converter in CCM64 pages
- Digital Control Algorithm For Two-Stage DC-DC Converters: Energy ProcediaNo ratings yetDigital Control Algorithm For Two-Stage DC-DC Converters: Energy Procedia7 pages
- LMI Robust Control Design For Boost PWM Converters: C. Olalla R. Leyva A. El Aroudi P. Garce S I. QueinnecNo ratings yetLMI Robust Control Design For Boost PWM Converters: C. Olalla R. Leyva A. El Aroudi P. Garce S I. Queinnec11 pages
- Advanced Controllers For Quadratic Boost Converter - A Case StudyNo ratings yetAdvanced Controllers For Quadratic Boost Converter - A Case Study6 pages
- Enumeration-Based Predictive Control For Buck DC-DC ConverterNo ratings yetEnumeration-Based Predictive Control For Buck DC-DC Converter5 pages
- LMI-Fuzzy Control Design For Non-Minimum-Phase DC-DC Converters An Application For Output RegulationNo ratings yetLMI-Fuzzy Control Design For Non-Minimum-Phase DC-DC Converters An Application For Output Regulation19 pages
- LMI-Based Robust Controllers For DC-DC Cascade Boost ConvertersNo ratings yetLMI-Based Robust Controllers For DC-DC Cascade Boost Converters10 pages
- Design of Fractional-Order PI Controller For DC-DC Power Converters PDFNo ratings yetDesign of Fractional-Order PI Controller For DC-DC Power Converters PDF6 pages
- Designing A Model Reference Adaptive ConNo ratings yetDesigning A Model Reference Adaptive Con5 pages
- A Comparative Study of Several Control Techniques Applied To A Boost ConverterNo ratings yetA Comparative Study of Several Control Techniques Applied To A Boost Converter8 pages
- Comparative Study of Two Sliding Surfaces To Control A Double Boost ConverterNo ratings yetComparative Study of Two Sliding Surfaces To Control A Double Boost Converter11 pages
- Control Theory - Robust Systems, Theory and ApplicationsNo ratings yetControl Theory - Robust Systems, Theory and Applications258 pages
- Switching Function Design For Output Voltage Tracking in The DC DC Buck Power ConverterNo ratings yetSwitching Function Design For Output Voltage Tracking in The DC DC Buck Power Converter14 pages
- Sliding Mode Control of DC-to-DC Power Converters Using Integral ReconstructorsNo ratings yetSliding Mode Control of DC-to-DC Power Converters Using Integral Reconstructors14 pages
- French - Index / Class Ix / English MediumNo ratings yetFrench - Index / Class Ix / English Medium5 pages
- Controlof DCMotor Using Integral State Feedbackand Comparisonwith PIDSimulationand Arduino ImplementationNo ratings yetControlof DCMotor Using Integral State Feedbackand Comparisonwith PIDSimulationand Arduino Implementation7 pages
- Optimal Control of Buck Converter by State Feedback LinearizationNo ratings yetOptimal Control of Buck Converter by State Feedback Linearization6 pages
- State-Space Average Modeling Based PID Control For Three Novel Topologies of DC-DC Boost ConverterNo ratings yetState-Space Average Modeling Based PID Control For Three Novel Topologies of DC-DC Boost Converter6 pages
- Addis Ababa Science and Technology UniversityNo ratings yetAddis Ababa Science and Technology University19 pages
- Buck Boost Converter Small Signal ModelNo ratings yetBuck Boost Converter Small Signal Model14 pages
- DC Voltage Regulator Using Buck-Boost Converter Based PID-Fuzzy ControlNo ratings yetDC Voltage Regulator Using Buck-Boost Converter Based PID-Fuzzy Control5 pages
- The Linux Terminal for Advanced Users - The Command Line Made Easy: First EditionFrom EverandThe Linux Terminal for Advanced Users - The Command Line Made Easy: First EditionNo ratings yet
- An Efficient Supervisory Power Management Scheme For A WindBattery-Assisted Hybrid Autonomous SystemNo ratings yetAn Efficient Supervisory Power Management Scheme For A WindBattery-Assisted Hybrid Autonomous System12 pages
- A Common Ground Switched-Quasi-Z - Source Bidirectional DCDC Converter With Wide-Voltage-Gain Range For EVs With Hybrid Energy SourcesNo ratings yetA Common Ground Switched-Quasi-Z - Source Bidirectional DCDC Converter With Wide-Voltage-Gain Range For EVs With Hybrid Energy Sources13 pages
- Design of Power System Stabilizer For DFIG-based Wind Energy Integrated Power Systems Under Combined Influence of PLL and Virtual Inertia ControllerNo ratings yetDesign of Power System Stabilizer For DFIG-based Wind Energy Integrated Power Systems Under Combined Influence of PLL and Virtual Inertia Controller11 pages
- Slide Bearings Type M: in Mechanical EngineeringNo ratings yetSlide Bearings Type M: in Mechanical Engineering10 pages
- 207 Contemporary Framework in ManagementNo ratings yet207 Contemporary Framework in Management111 pages
- Electrical Noise and Transients: Application NoteNo ratings yetElectrical Noise and Transients: Application Note5 pages
- HAP and ASHRAE Standard 90.1 Appendix GNo ratings yetHAP and ASHRAE Standard 90.1 Appendix G10 pages
- Prepared by F. S. Salla Ag. Commissioner of Fire Public Safety Division Fire and Rescue ForceNo ratings yetPrepared by F. S. Salla Ag. Commissioner of Fire Public Safety Division Fire and Rescue Force19 pages
- CS5500: Reinforcement Learning Assignment 3: Additional GuidelinesNo ratings yetCS5500: Reinforcement Learning Assignment 3: Additional Guidelines7 pages
- Fractional-Order Control: Theory and Applications in Motion ControlNo ratings yetFractional-Order Control: Theory and Applications in Motion Control11 pages
- Computer Networking Notes For Tech PlacementsNo ratings yetComputer Networking Notes For Tech Placements16 pages
- Water Supply and Wastewater Engineering - CE3303 - Hand Written Notes - Unit 1 - Water SupplyNo ratings yetWater Supply and Wastewater Engineering - CE3303 - Hand Written Notes - Unit 1 - Water Supply58 pages
- Avo, Inversion AND Seismic Attributes: Poisson's Ratio VolumetricsNo ratings yetAvo, Inversion AND Seismic Attributes: Poisson's Ratio Volumetrics2 pages
