THE INTERNATIONAL JOURNAL OF ENGINEERING AND INFORMATION TECHNOLOGY (IJEIT), VOL. 3, NO.

1, Dec 2016 36

Stability Analysis of Controlled DC Motor

A. Elbkosh I. Farhat
Faculty of Engineering Department of electrical and computer Faculty of Engineering Department of electrical and computer
engineering, Alasmrya University engineering, Alasmrya University
[email protected] [email protected]

Abstract—Controlled dc motors are nonlinear systems, that The second method is based on deriving the
show a nonlinear action in their operation including, sub- monodromy matrix of the system, which is the
harmonics and chaos when they work outside their design fundamental solution matrix of the system for one
specifications. This nonlinearity forces the motor changing complete cycle, and finding the Floquet multipliers. The
its normal operation to a random-like behaviour; In this
Floquet multipliers are the eigenvalues of the monodromy
paper, the nonlinear dynamics of DC motors are
investigated. It is shown that the concept of the Poincaré matrix.
map approach and the monodromy matrix method can be This paper is constructed as follows: first a general DC
successfully applied to determine the stability of DC motors. motor driven by a chopper circuit is modeled using a
sampled data model. Based on this map, the existence and
Index Terms: DC motor, Monodromy matrix, nonlinear location of a periodic orbit are obtained. Next the
behaviour, Poincaré map. nonlinear phenomena in the system including bifurcation
and chaos are shown by simulation. Finally the stability
analysis of the system is considered using the two
I. INTRODUCTION methods mentioned above.

T he performance of any system is generally evaluated

by its steady-state and dynamic behaviour. When
performing a steady-state analysis, the existence and
II. TOPOLOGY AND OPERATION OF
CONTROLLED DC MOTOR

location of periodic solutions are of concern and can be A simplified block diagram of a speed controller DC
eased by deriving a discrete map that describes the motor is shown in Figure 1. It consists of a feedback loop
dynamics of the system, and finding its fixed point. When which observes the speed variation and adjusts the duty
performing a dynamic analysis however, stability and cycle d. The switch S is controlled by a comparator which
transient response are greatly focused on and can be compares a control signal Vcon with a periodic saw-tooth
studied using the closed loop eigenvalues of the system. waveform Vramp. Switch S is open when Vcon > Vramp and is
In this paper, the stability of the controlled DC motor closed when Vcon < Vramp as shown in Figure 2.
will be studied using two different approaches. The first
Vramp
is the conventional Poincaré map approach for studying ωref
stability of any periodic system, based on: Kp +-
Vcon ω
 Deriving the Poincaré map that describes the
dynamics of the system and finding its fixed points. L J
 Linearizing the map around the fixed points and S

finding the eigenvalues of the jacobian of this map.

 If all eigenvalues have a magnitude less than unity, B
R
the system is stable, otherwise the system is
E D KT i
unstable and nonlinear behaviour may exist.
KE ω TL
The essence of this method lies in the capture of the
dynamics in the small neighbourhood of a periodic orbit.
One drawback of using this method in electrical circuits Figure 1. DC Motor With Speed Control
is that sometimes it is difficult to derive the Poincaré map
of the system analytically because the equations of the The control signal Vcon is derived from the speed signal
system are transcendental. Therefore this map can only be through a standard error amplifier. Using a simple
calculated numerically [1-4]. proportional feedback controller, the control signal can be
written as:

Received 27 April 2016; revised 30 April 2016; accepted 12 August Vcon t   K p  ωt   ω ref  (1)
2016.

Available online 16 August 2016.

www.ijeit.misuratau.edu.ly ISSN 2410-4256 Paper ID: EN028

A. Elbkosh and I. Farhat / Stability Analysis of Controlled DC Motor 37

where ωref is the reference speed, the desired speed, and  1  1

0  0
Kp is the gain of the feedback amplifier. The ramp signal  J   J 
can be expressed as: B ON   and B OFF    (5)
 t 
 0  1/L   0 0
Vramp t   VL  ΔV  mod 1  (2)  
 T 
The switching instant occurs when the control signal is
where ΔV  VU  VL  , VL , VU are the lower and upper equal to the ramp signal; thus the switching condition
voltages of the ramp signal respectively and T is the h x, t  is defined through feedback proportional control
period of one cycle. as:

h x, t   K p ωt   ω ref   Vramp  0 (6)

VU Vramp Vcon
In general, the circuit gives an average speed close to
the desired value with a periodic ripple equal to the
period of the driving clock as shown in Figure 3. The
output speed shows a repetitive oscillation with a fixed
VL speed ripple, also called period-1 operation. However,
nonlinear phenomena such as bifurcation and chaos will
appear when one of the circuit parameters is varied, the
t
d circuit parameters is chosen as L=53.7e-3 H; R=2.8 ohm
; Kp=0.7; B=0.000275; J=0.000557; Tl=0.38; ωref=100
t rad/sec ; Ke=0.1356;Kt=0.1324; [1-5].
103
Figure 2. Typical Operation Waveforms of dc Motor with Speed Mode
Control
102.5

The system is governed by two sets of linear 102

Speed (rad/sec)

differential equations related to the ON and OFF states of

101.5
the controlled switch. The inductor current iL and the
speed of the motor ω are taken as state variables. 101

The equations that represent the dynamics of the

system in state space form are: 100.5

f  xt , t 
100
0.6 0.65 0.7 0.75 0.8
S : ON Time (sec)


xt   f xt , t    (3) Figure 3. Simulation Results, Period -1 Operation
f xt , t 
  S : OFF
III. NONLINEAR PHENOMENA IN DC MOTOR
where f  xt , t  , f  xt , t  are the two smooth vector
fields when the switch is ON and OFF respectively, DC motors driven by a chopper circuit are nonlinear
defined as: dynamic systems. The nonlinearities arise mainly due to
switching power devices in control circuits, and
f  x t , t   A ON x t   B ON U
nonlinearities in passive components such as inductors
and capacitors [6]. Controlled DC motors exhibit various
(4) types of complex behaviour such as bifurcations and
f  x t , t   A OFF x t   B OFF U chaotic operation. These phenomena are called the
nonlinear dynamics of the system [5-8]. In this section,
nonlinear phenomena in a controlled DC motor are
x   i L T  x1 x 2 T is the state vector, and studied using the time waveforms of state variables,
U  TL ET is the input vector. A and B are the system phase portraits and bifurcation diagrams. The input
voltage is used as bifurcation parameters to investigate
matrices that contain the system parameters, defined as: the changing behaviour of the system. Results are
validated theoretically, showing good agreement with
 B K t /J  simulation.
 J 
A ON  A OFF   A. Simulations results
 K /L  R/L 
 e  In order to study the dynamics of the controlled
system, Equation (3), which describes the dynamics of
DC motors, is solved using MATLAB/ SIMULINK. The
www.ijeit.misuratau.edu.ly ISSN 2410-4256 Paper ID: EN028
THE INTERNATIONAL JOURNAL OF ENGINEERING AND INFORMATION TECHNOLOGY (IJEIT), VOL. 3, NO. 1, Dec 2016 38

switching instants are determined by comparing the ramp 160

signal with the control signal. The input voltage was used
140
as the bifurcation parameter and was varied from 50V to
120
80V. The states were sampled at the start of each cycle of
the ramp; thus a sampled data map was obtained. The 100

Speed (rad/sec)
sampled values of the speed (neglecting the initial 80

transient) are plotted against the bifurcation parameter to 60

obtain the bifurcation diagram, shown in Figure 4. 40

106 20

105 -20
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Time (sec)

104
Speed (rad/sec)

Figure 5. Period-1 Waveforms of the DC Motor, E = 50V

103
4.5

102 4

Current (Amp)
101
3.5

3
100
50 55 60 65 70 75 80
values of Vin
2.5
Figure 4. Bifurcation Diagram of the System as the Input Voltage is
Changed 2

Normally, the controlled DC motor is designed to 1.5

101.4 101.5 101.6 101.7 101.8 101.9 102 102.1 102.2
operate in period-1. This is when the input voltage is less Speed (rad/sec)
than 55.6V for this specific system. However, increasing Figure 6. Phase Portrait of the System, E = 50V
the input voltage, a period-doubling bifurcation occurs at
55.6V and the stability of the period-1 is lost to another System waveforms as the input voltage increases to
periodic orbit, period-2. 60V are shown in Figures 7 and 8. It is obvious that the
This periodic solution continues until the input voltage system is working in period-2 operation i.e. the states
is near 67V then it loses its stability again and bifurcates repeat themselves every two switching cycles.
to period-4. As the input voltage increases further, a 20
cascade of period- doubling takes place and at some point
the system will enter into a chaotic region at an input 15
voltage of around 70V. Above this value the system
begins to operate in the chaotic region and exhibits some
Current (Amp)

10
complex behaviour.
The speed and the inductor current waveforms in the 5
time domain and the state space for an input voltage of
50V are shown in Figures 5 and 6, respectively. The 0
results indicate that the system is working in period-1
operation at this operating point.
-5
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
16 Time (sec)

14

12
160
10
140
Current (Amp)

8
120
6
100
Speed (rad/sec)

4
80
2
60
0
40
-2
20
-4
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0
Time (sec)
-20
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Time (sec)

Figure 7. Period-2 waveforms of the DC motor, E = 60V

www.ijeit.misuratau.edu.ly ISSN 2410-4256 Paper ID: EN028
A. Elbkosh and I. Farhat / Stability Analysis of Controlled DC Motor 39

5.5

5 B. Discrete maps and periodic solutions

4.5
In switching systems, the steady-state operation is a
periodic orbit not an equilibrium point. Furthermore, this
Current (Amp)

4
periodic orbit is non-smooth due to the switching action.
3.5 One way to check the existence of the periodic orbit and
3
find its location analytically is to derive a discrete map
that describes the system [9-12]. In this section, the data
2.5
sampled model in the form of a stroboscopic map is
2 derived, where the state variables are sampled at the
1.5
beginning of each cycle, to get a discrete model.
101 101.5 102 102.5 103 103.5 Assuming the controlled DC motor operates in the
Speed (rad/sec)
Figure 8. Phase portrait of the system, E = 60V nominal period-1 steady-state, in which there is only one
switching in one clock cycle occurring at the time instant
As the input voltage increases further, to 70V, the d′T as shown in Figure 11.
system operates in a chaotic state as shown in Figure 9.

Vramp & Vcon

The phase portrait at an input voltage of 70V shows a X(d′T)
bounded solution with non-periodic motion (Figure 10).
20
ФOFF(d′T,0)

15 X(0) X(T)

0 ФON(T,d′T)
,t)=
Current (Amp)

h(x
10

t=0 t=d′T t=T time

0

Figure 11. Typical Periodic Solution of the Controlled DC Motor

-5
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Time (sec) Since the controlled DC motor is a smooth piecewise
linear system, the dynamics of the system before and
160
after the switching can be described by a linear time
140 invariant (LTI) ordinary differential equation 3. Therefore
120 the solution of the system for each interval can be
Speed(rad/sec)

100 obtained directly by the exponential matrix method.

80 During the first interval, the ramp is crossed at t=d′T,
60 the switch is OFF, and the solution of the system will be:
40
20
xd ' T   e A OFF d 'T  x0   0d 'T e A OFF d 'T   B OFF U d
0
-20
0 0.2 0.4 0.6 0.8 1
Time(sec)  Φ OFF d T ,0 x0  Γ OFF d 
(7)
Figure 9. Chaotic waveforms of the DC motor, E = 70V

where Φ OFF d T ,0  , is the state transition matrix during

6
the first interval.
5
Likewise, during the second interval, the switch is ON
4 and the state vector is given by:
Current (Amp)

0
100 100.5 101 101.5 102 102.5 103 103.5 104
Speed (rad/sec)
Figure 10. Phase portrait of the system, E = 70V
www.ijeit.misuratau.edu.ly ISSN 2410-4256 Paper ID: EN028
THE INTERNATIONAL JOURNAL OF ENGINEERING AND INFORMATION TECHNOLOGY (IJEIT), VOL. 3, NO. 1, Dec 2016 40

xT   e AON T d 'T  xd T    d 'T e A ON T   B ON U d This transcendental nonlinear equation can be solved
T
numerically with a method such as the Newton-Raphson
(8) method to obtain the duty ratio. Figure 12 shows the
numerical values obtained for the duty cycle for different
values of the input voltage.
 Φ ON T , d T  xd T   Γ ON d  Once the duty cycle is calculated, the fixed point of the
sampled data map x0 can be obtained using equation
where Φ ON T , d T  , is the state transition matrix when (11) which corresponds to the location of the periodic
the switch is ON. orbit of the continuous system.
0.68
Since the vector field of the system is piecewise linear
0.66
(the solution of the system is continuous everywhere but
only piecewise differentiable), the final state just before a 0.64

switching instant can be taken as the initial state after the

Duty Cycle (d)

0.62
switching. This yields:
0.6

xT   Φ ON T , d T  Φ OFF d T ,0  x0   Γ OFF d    Γ ON d  0.58

(9) 0.56

This is a sampled data map of the system, also known

0.54
as the Poincaré map, and can be simplified to: 50 52 54 56 58 60 62
Input Voltage (V)
64 66 68 70

Figure 12. Evaluation of the Duty-Cycle for Period-1 Operation

xT   ΦT ,0  x0   Γd  (10)

where IV. STABILITY ANALYSIS OF THE DC MOTOR

USING POINCARE MAP METHOD
ΦT ,0   Φ ON T , d T   Φ OFF d T ,0  and The stability of any periodic system exhibiting
nonlinear behaviour such as the controlled DC motor is
Γd   Φ ON T , d T  Γ OFF d   Γ ON d  usually carried out as follows:
 Deriving the discrete time map (Poincaré map) of the
For a periodic solution xT   x0  must be satisfied thus system.
 Calculating the jacobian matrix of this map by
the following expression for x0 can be obtained: linearizing around its fixed point.
 Finally, finding the eigenvalues of the jacobian. The
x0   I  ΦT ,0 1  Γd  (11) natural response of the system will decay to zero (the
system will be stable) if and only if the eigenvalues
of the jacobian have a magnitude less than unity,
where I is the identity matrix of the same order as the
system matrix AON, and x0  is the fixed point of the
otherwise the system is unstable.

map. Since the nonlinear equation (11) is a function In a controlled DC motor this map can be established
of d  , the equation can be solved with the switching in a number of ways; one way is by sampling the state
equation hx(0), d   0 to obtain the duty cycle. variables at the beginning of each switching period T as
described in the previous section.
The switching equation hx(0), d   0 at the switching
instant is given by: A. Derivation of jacobian matrix of the DC motor
In section 3.b the Poincaré map of the DC motor has
hx(0), d   1 0Φ OFF d T ,0  x0   Γ OFF d   ref been derived (equation 10). For simplicity, this can be
written as:
VL  V d  (12)

KP xT   f x0 , d  (14)
Substituting equation (11) into equation (12), a
transcendental equation will be obtained which involves In order to check the stability of the system, we first
need to linearize this map around its fixed point x0  to
only one unknown, the switching instant:

xT 
1 0Φ OFF d T ,0 I  ΦT ,0 1  Γd   Γ OFF d  obtain the jacobian matrix
x0 
. This can be achieved
VL  Vd  (13)
 ref  0 by differentiating equation (14) with respect to x0  and
KP
using the series rule:
www.ijeit.misuratau.edu.ly ISSN 2410-4256 Paper ID: EN028
A. Elbkosh and I. Farhat / Stability Analysis of Controlled DC Motor 41

xT  f x0 , d  f x0 , d  d 

   (15)
x0  x0  d  x0  1

d 
To calculate , differentiation of the switching
x0 
E=56

Imaginary part of the Eigenvalues

0.5

manifold h x0 , d  with respect to x0  is needed; this

yields: 0

h x0 , d  h x0 , d  d 
0   (16) -0.5

x0  d  x0  E=55

-1

By substituting equation ( 16) into (15), the following

-1 -0.5 0 0.5 1
expression of the jacobian matrix is obtained: Real part of the Eigenvalues

xT   f x0 , d    f x0 , d    h x0 , d  

1
Figure 13. Evaluation of the Eigenvalues of the Iacobian Matrix as the
    
x0   x0    d    d  
Input Voltage Changes from 50V to 60V
(17)
 h x0 , d   C. Stability Analysis of the controlled DC motor
 
 x0  
Using the Monodromy Matrix Method
This is a general form of the jacobian matrix for any In this section an alternative approach for determining
switching systems. By differentiating the Poincaré map of the stability of any switching systems is proposed based
the controlled DC motor equation (10) and (12) with on linearizing the system around the whole periodic orbit.
respect to x0  and d  , the final expression for the
The new method requires the calculation of the
monodromy matrix of the controlled DC motor. The
jacobian of the voltage controlled DC motor is obtained. monodromy matrix of any switching system is the
product of the state transition matrices before, during and
xT   Φ ON B ON  B OFF U 1 0 Φ OFF 
 ΦT ,0 (18) after the switching for one complete cycle.
x0 1 0 A OFF xd T  B OFF U  V
K PT D. Derivation of the monodromy matrix of the DC
motor
B. Calculating the eigenvalues of the jacobian The controlled DC motor is a nonlinear system that
matrix represents different circuit topologies within one
In order to calculate the jacobian matrix and hence to switching cycle. For the continuous conduction mode,
check the stability of the system, one needs to find the there are two topologies. In each topology, the system can
location of the periodic orbit and the switching instant. be described by linear state equations. However, for a
This can be achieved numerically by solving the discrete complete switching cycle, the system becomes piecewise
map of the system with the switching equation to linear and the solution is not defined at the switching
generate a nonlinear equation (13) whose roots will instant. The solution of such a system can be defined on
define the switching instant. Once the switching instant the basis of differential inclusions using Filippov’s
has been identified, utilizing the fact that the system is concept [13-15].
LTI before and after the switching, it is possible to locate To define the solution of a system while it is on the
the limit cycle. Once these values are found, the jacobian switching manifold, Filippov suggested that the vector
matrix (18) can be expressed as a function of the input field at the switching instant will not be a single valued
voltage and its eigenvalues can be calculated. Figure 13 function but a set valued function whose limits are the
shows the evaluation of the eigenvalues of the jacobian values of the vector fields before and after the switching.
matrix for different values of the input voltage, clearly As a result the original piecewise equation (4) that
indicating the loss of stability through a smooth period- describes the dynamics of the DC motor has to be
doubling bifurcation around an input voltage of 56V as extended into a differential inclusion F (x, t) as:
demonstrated by the previous simulation and
experimental results. f  x, t  x V




x (t )  F (x, t )  co f  x, t , f  x, t  x  (19)



f  x, t 
 x V

www.ijeit.misuratau.edu.ly ISSN 2410-4256 Paper ID: EN028

THE INTERNATIONAL JOURNAL OF ENGINEERING AND INFORMATION TECHNOLOGY (IJEIT), VOL. 3, NO. 1, Dec 2016 42

where f  (x(t ), t ) and f  ( x(t ), t ) are the two smooth n   h x( t  ), t  

vector fields before and after the switching. They are
defined as: (23)
T
 K t x 2 t  B x1 t   TL   h h  1 
    
 J   x1 x 2  0 
 
f  x(t ), t  A ON xt   B ON U    (20)
 - K e x1 t  R x1 t   E 
  Therefore the projections of f  and f  on the hyper-
 L 
surface Σ are given by:
 K t x 2 t  B x1 t   TL 
   K t x 2 t   B x1 t   TL 
J  
 
f  x(t ), t  A OFF xt   B OFF U    (21) 
J

 - K e x1 t  R x1 t  n f   [1 0]  
T


 L

  - K e x1 t   R x1 t   (24)
 
 L 
It is obvious from equations (20) and (21) that there is K t x 2 t   B x1 t   TL
a discontinuity when the main switching element passes J
from ON state to OFF state, since f  ( x(t ), t )  f  ( x(t ), t ) .
The two dimensional state space is now divided into  K t x 2 t   B x1 t   TL 
three parts V-, V+ and Σ as shown in Figure 14 where V- is  
J
the time interval during which the switch is OFF, V+ is  
the time interval during which the switch is ON and Σ is n f   [1 0] 
T

the switching instant.  - K e x1 t   R x1 t   E  (25)
 
 L 
Switch ON h(x(d′T))=0 Switch OFF K t x 2 t   B x1 t   TL
J
x(0) The extension of a discontinuous system (4) into a
Inductor Current (A)

convex differential inclusion (14) is known as Filippov’s

convex method. The solution is unique for every initial
condition, if it crosses the hyper-surface transversally and
x(d′T)
spends almost zero time on the switching manifold.
f+
f-

nf+ n
A necessary condition for a transversal intersection at Σ is
V+ nf-
V- [15]:
Speed (rad/sec)
n T f  (x, t )  n T f  (x, t )  0 x(t )
Figure 14. Transversal Intersections in the Orbit of the Controlled DC
Motor
(26)
The smallest closed convex set is defined as:
K t x 2 t   B x1 t   TL K x t   B x1 t   TL
 t 2 0
 K t x 2 t   B x1 t   TL
J J

 J 
 
Therefore the solution is unique as shown in Figure 15.
 
co  f  , f    - K e x1 t   R x1 t   E The period-1 limit cycle of the system, given in Figure
 (1  q )  
 L 15, shows that the trajectory crosses the switching
 - K e x1 t   R x1 t  
manifold twice, at d′T and T. Therefore the fundamental
 q 
L solution matrix for one complete cycle, the monodromy
matrix, is given by:
q0,1 (22)
MT ,0 S 2 (T )  ΦON (T , d T )  S1 (d T )  ΦOFF (d T ,0) (27)

The normal to the switching manifold n is given by:

where Φ ON (T , dT ) , Φ OFF (dT ,0) are the state
transition matrices during the ON and OFF intervals,
respectively, and they are calculated by the exponential
matrix.
www.ijeit.misuratau.edu.ly ISSN 2410-4256 Paper ID: EN028
A. Elbkosh and I. Farhat / Stability Analysis of Controlled DC Motor 43

S1 and S2 are the state transition matrices during   0  

switching, also called the saltation matrices, and they are   0 
   
calculated by the following formula [1, 13-15]: A ON dT  
M T ,0   e  I   E
L


 0
f  x(t ), t   f  x(t ), t  nT   K t x 2 t   B x 1 t   TL  V  
 
SI  
hx(t  ), t   (28)  J K pT 
 
nT f  x(t  ), t    A OFF d T
t e
where t is the switching time (the time at which the (31)
The stability of the periodic orbit can be determined by
solution hits the switching manifold).
obtaining the Floquet multipliers which are the
eigenvalues of the monodromy matrix.
Switch ON

S2(T) E. Calculating the Floquet multipliers

The stability of the system can be determined by
Inductor Current (A)

ФOFF(d′T,0) calculating the Floquet multipliers which are the

h(x(T))=0
eigenvalues of the fundamental solution matrix MT ,0 .
Common problems that have to be addressed here are the
location of the limit cycle and the times at which the
h(x(d′T))=0
ФON(T,d′T)
switching take place. This can be achieved numerically
by deriving a nonlinear function equation 13 whose root
S1(d′T)
will define the switching instant. Once the switching
Switch OFF instants have been identified, utilizing the fact that the
system is LTI before and after the switching, it is possible
Speed (rad/sec) to locate the limit cycle. Once these values are found, the
Figure 15. Period-1 Limit Cycle of the DC Motor monodromy matrix can be expressed as a function of the
input voltage using equation 31. The computed loci of the
The switching manifold is defined by a scalar indicator eigenvalues with varying input voltage are shown in
function hx, t   0 , thus the derivative of h x, t  with Figure 16. The figure shows that the system loses its
stability through a smooth period-doubling bifurcation at
respect to t for period-1 operation t(0, T ) is:
an input voltage of around 55.6V. This result is in perfect
agreement with the previous analytical and simulation
h(x, t ) V results.
 (29)
t K pT 1.5

Substituting equations (23) – (29) into equation (28) at 1

the switching instant t  d T , the saltation matrix S1 can

Imaginary part of the Eigenvalues

be calculated: 0.5
E=56

 0  0

 0
 
E=55

S1  I   
-0.5
E (30)
 L 0
 K t x 2 t   B x1 t   TL  V  -1

 J K pT 
 
-1.5
-1.5 -1 -0.5 0 0.5 1 1.5
Real part of the Eigenvalues
To calculate S2, the time derivative of the switching
manifold at t  T is needed. Since the switching Figure 16. Loci of the Eigenvalues of the Monodromy Matrix for
Different Input Voltages
manifold is discontinuous (with respect to time) at this
point the time derivative will be infinite. Therefore the V. CONCLUSION
saltation matrix at this point is the identity matrix.
Knowing S1 and S2, it is possible to calculate the This paper focused on the stability analysis of the
eigenvalues of the monodromy matrix which, for a periodic orbit of the DC motor with speed mode control
period-1 response, must have amplitudes less than 1.The operating in continuous conduction mode. The existence
total fundamental solution matrix over one complete and location of the periodic solution of the system has
cycle of the buck converter is: been proven analytically in this paper. It was shown that
the system exhibits nonlinear phenomena, including
www.ijeit.misuratau.edu.ly ISSN 2410-4256 Paper ID: EN028
THE INTERNATIONAL JOURNAL OF ENGINEERING AND INFORMATION TECHNOLOGY (IJEIT), VOL. 3, NO. 1, Dec 2016 44

bifurcation and chaos as the input voltage or the control [7] V. Baushev, Zh. Zhusubaliev, Yu. Kolokolov, and I. Terekhin,
“Local stability of periodic solutions in sampled data control
parameter is changed. The nonlinearities have been
systems”, Automation and Remote Control, vol. 53, no. 2, pp.
shown analytically, by simulation validated. The stability 865-871, 1992.
analysis of the speed controlled DC motor has been [8] M. Di Bernardo, F. Garofalo, L. Glielmo, and F. Vasca,
considered using two approaches; the first is the Poincaré “Switchings, bifurcations and chaos in DC-DC converters,” IEEE
Transactions on Circuits and Systems-I, vol. 45, no. 2, pp. 133–
map method which is the conventional method for
141, 1998.
studying the stability of any periodic orbits. The second [9] M. Di Bernardo, and F. Vasca, “Discrete-Time Maps for the
approach is based on deriving the state transition matrices Analysis of Bifurcations and Chaos in DC/DC Converters” IEEE
before, during and after the switching takes place for one Transactions on Circuits and Systems-I, vol. 47, no. 2, pp. 130–
143, 2000.
switching cycle (the monodromy matrix). Unsurprisingly,
[10] E. Fossas and G. Olivar, “Study of chaos in the buck converter,”
both methods give the same results; however the stability IEEE Transactions on Circuits and Systems-I, vol. 43, no. 1, pp.
analysis using the monodromy matrix method was easier 13–25, 1996.
to implement when compared with the Poincaré map [11] S. Banerjee and G. C. Verghese, eds., Nonlinear Phenomena in
Power Electronics: Attractors, Bifurcations, Chaos, and Nonlinear
method. This represents a first step towards developing a
Control. New York, USA: IEEE Press, 2001.
technique for controlling bifurcations in controlled DC [12] C.-C. Fang and E. H. Abed, “Sampled-data modelling and analysis
motors which will be used in future work to stabilize the of the power stage of PWM dc-dc converters,” International
nonlinear behaviour in these systems Journal of Electronics, vol. 88, pp. 347 – 369, March 2001.
[13] R. I. Leine and H. Nijmeijer, Dynamics and Bifurcations of Non-
Smooth Mechanical Systems. Springer, 2004.
REFERENCES [14] R. I. Leine, D. H. V. Campen, and B. L. V. de Vrande,
“Bifurcations in nonlinear discontinuous systems,” Nonlinear
[1] A. Elbkosh, “MODELING OF SWITCHING DC-DC
Dynamics, vol. 23, pp. 105–164, 2000.
REGULATORS, REVIEW OF METHODS AND
[15] A. F. Filippov, Differential equations with discontinuous
APPLICATIONS ” Proceedings of the ICECE International
righthand sides. Dortrecht: Kluwer Academic Publishers, 1988.
Conference, (AsiaMIC 2013), Phuket, Thailand, April 2013.
[16] D. Giaouris, S. Banerjee, B. Zahawi, and V. Pickert, “Stability
[2] A. Elbkosh, “Stability Analysis of Switching Systems using the
Analysis of the Continuous Conduction Mode Buck Converter via
State Transition Matrix,” Proceedings of the ICECE International
Filippov's Method” IEEE Transactions on Circuits and Systems –
Conference, (ICECE 2013), Benghazi, Libya, March 2013.
I, 2008.
[3] M. A. Aizerman and F. R. Gantmakher, “On the stability of
periodic motions,” Journal of Applied Mathematics and
Mechanics (translated from Russian), pp. 1065–1078, 1958.
[4] G. C. Verghese, M. E. Elbuluk, and J. G. Kassakian, “A general
approach to sampled-data modeling for power electronic circuits,”
IEEE Transactions on Power Electronics, vol. PE-1, no. 2, pp. 76–
89, 1986.
[5] C. K. Tse, Complex Behavior of Switching Power Converters.
Boca Raton, USA: CRC Press, 2003.
[6] C.K. Tse and M. Di Bernardo, "Complex Behavior in Switching
Power Converters," Proceedings of IEEE, Special Issue on
Applications of Nonlinear Dynamics to Electronic and
Information Engineering, vol. 90, no. 5, pp. 768-781, May 2002.

www.ijeit.misuratau.edu.ly ISSN 2410-4256 Paper ID: EN028