Journal of Science and Technology - Engineering and Computer Sciences, Vol. 14, No.

2, December/ 2013

Optimal Speed Control for Direct Current Motors Using Linear

Quadratic Regulator

Aamir Hashim Obeid Ahmed

Control Engineering Department
School of Electrical & Nuclear Engineering
Sudan University of Science and Technology
Khartoum, Sudan
aamirahmed@sustech.edu; Aamir_elec@yahoo.com

Abstract: Direct Current (DC) motors have been extensively used in many industrial
applications. Therefore, the control of the speed of a DC motor is an important issue and has
been studied since the early decades in the last century. This paper presents a comparison of time
response specification between conventional Proportional-Integral-Derivatives (PID) controller
and Linear Quadratic Regulator (LQR) for a speed control of a separately excited DC motor.
The goal is to determine which control strategy delivers better performance with respect to DC
motor’s speed. Performance of these controllers has been verified through simulation using
MATLAB/SIMULINK software package. According to the simulation results, liner quadratic
regulator method gives the better performance, such as settling time, steady state error and
overshoot compared to conventional PID controller. This shows the superiority of liner quadratic
regulator method over conventional PID controller.

Keywords: Optimal Control, Linear Quadratic Regulator, Proportional-Integral-Derivative

Controller, Direct Current Motors, Speed Control.

‫ ﻟﺬﻟﻚ ﻓﺈن اﻟﺘﺤﻜﻢ ﻓﻰ‬.‫( ﺗﺴﺘﺨﺪم ﻋﻠﻰ ﻧﻄﺎق واﺳﻊ ﻓﻰ اﻟﻌﺪﯾﺪ ﻣﻦ اﻟﺘﻄﺒﯿﻘﺎت اﻟﺼﻨﺎﻋﯿﺔ‬DC) ‫ ﻣﺤﺮﻛﺎت اﻟﺘﯿﺎر اﻟﻤﺴﺘﻤﺮ‬:‫اﻟﻤﺴﺘﺨﻠﺺ‬
‫ ھﺬه اﻟﻮرﻗﺔ ﺗﻘﺪم ﻣﻘﺎرﻧﺔ ﺑﯿﻦ ﻣﻮاﺻﻔﺎت‬.‫ﺳﺮﻋﺔ ﻣﺤﺮﻛﺎت اﻟﺘﯿﺎر اﻟﻤﺴﺘﻤﺮ ﻗﻀﯿﺔ ﻣﮭﻤﺔ ﺗﻤﺖ دراﺳﺘﮭﺎ ﻣﻨﺬ اﻟﻌﻘﻮد اﻷوﻟﻰ ﻟﻠﻘﺮن اﻟﻤﺎﺿﻰ‬
‫( ﻟﻠﺘﺤﻜﻢ ﻓﻰ ﺳﺮﻋﺔ ﻣﺤﺮك‬LQR) ‫( و اﻟﻤﻨﻈﻢ اﻟﺘﺮﺑﯿﻌﻰ اﻟﺨﻄﻰ‬PID) ‫اﻟﺘﻔﺎﺿﻠﯿﺔ‬-‫اﻟﺘﻜﺎﻣﻠﯿﺔ‬-‫اﻹﺳﺘﺠﺎﺑﺔ اﻟﺰﻣﻨﯿﺔ ﺑﯿﻦ اﻟﺤﺎﻛﻤﺔ اﻟﺘﻨﺎﺳﺒﯿﺔ‬
‫ اﻟﮭﺪف ﻣﻦ ذﻟﻚ ﺗﺤﺪﯾﺪ إﺳﺘﺮاﺗﯿﺠﯿﺔ اﻟﺘﺤﻜﻢ اﻟﺘﻰ ﺗﻌﻄﻰ أداء أﻓﻀﻞ ﻓﯿﻤﺎ ﯾﺘﻌﻠﻖ ﺑﺎﻟﺘﺤﻜﻢ ﻓﻰ ﺳﺮﻋﺔ‬.‫اﻟﺘﯿﺎر اﻟﻤﺴﺘﻤﺮ ذو اﻟﺘﻐﺬﯾﺔ اﻟﻤﻨﻔﺼﻠﺔ‬
‫ وﻗﺪ ﺗﻢ اﻟﺘﺤﻘﻖ ﻣﻦ أداء ھﺬه اﻟﻤﺘﺤﻜﻤﺎت ﻣﻦ ﺧﻼل اﻟﻤﺤﺎﻛﺎة ﺑﺈﺳﺘﺨﺪام ﺣﺰﻣﺔ ﺑﺮﻧﺎﻣﺞ‬.‫ﻣﺤﺮك اﻟﺘﯿﺎر اﻟﻤﺴﺘﻤﺮ‬
،‫ وﻓﻘﺎ ً ﻟﻨﺘﺎﺋﺞ اﻟﻤﺤﺎﻛﺎة ﻧﺠﺪ أن طﺮﯾﻘﺔ اﻟﻤﻨﻈﻢ اﻟﺘﺮﺑﯿﻌﻰ اﻟﺨﻄﻰ ﺗﻌﻄﻰ أداء أﻓﻀﻞ ﻣﺜﻞ زﻣﻦ اﻟﺘﺮﺳﯿﺦ‬.MATLAB/SIMULINK
‫ وھﺬا ﯾﺪل ﻋﻠﻰ ﺗﻔﻮق طﺮﯾﻘﺔ اﻟﻤﻨﻈﻢ اﻟﺘﺮﺑﯿﻌﻰ اﻟﺨﻄﻰ‬.PID ‫اﻟﺨﻄﺄ ﻋﻨﺪ ﺣﺎﻟﺔ اﻹﺳﺘﻘﺮار وﺗﺠﺎوز اﻟﮭﺪف ﻣﻘﺎرﻧﺔ ﻣﻊ اﻟﺤﺎﻛﻤﺔ اﻟﺘﻘﻠﯿﺪﯾﺔ‬
.PID ‫ﻋﻠﻰ اﻟﺤﺎﻛﻤﺔ اﻟﺘﻘﻠﯿﺪﯾﺔ‬

48
 Journal of Science and Technology - Engineering and Computer Sciences, Vol. 14, No. 2, December/ 2013

ntroduction:
Electrical derives involving various types everybody is able to imagine what is
of DC motors turn the wheel of industry. happening inside the
The main reason for their popularity is controller during the control process.
the ability to control their torque and flux Furthermore, it has the ability to eliminate
easily and independently. Therefore, DC steady state offset trough integral action
motors are comprehensively used in and it can anticipate the changes through
various industrial applications such as derivative action. In addition to this,
electrical equipment, computer traditional PID controllers have very
peripherals, robotic manipulators, simple control structure and inexpensive
actuators, steel rolling mills, electrical cost. In spite of the major features of the
vehicles, and home appliances. Its fixed PID controller, it has some
applications spread from low horse power disadvantages such as the high starting
to the multi-mega watt due to its wide overshoot in speed, the sensitivity to
power, torque, speed ranges, high controller gains and the sluggish response
efficiency, fast response, and simple and due to sudden change in load torque
continuous control characteristics [1-4]. disturbance. Therefore, a great deal of
Controlling the speed of a DC motor is a attention has been focused on adaptive or
pivotal issue. The speed of DC motor can self-tuning of conventional PID controller
be changed by controlling the armature gains. Tuning PID controller parameters is
and field voltages. In this paper, the very difficult, poor robustness; therefore,
controller is designed to control the it's difficult to achieve the optimal state
armature voltage while the field voltage under field conditions in the actual
is fixed as a constant. Over the past production. In order to overcome some
decades, many techniques have been problems that faced by conventional PID
developed for the DC motor control. controller and achieve accurate control
Some of these methods were based on performance of speed control of a DC
classical and also intelligent approaches motor, the other type of control methods
[5-10]. For DC motors, factors such as can be developed such as linear quadratic
unknown load characteristic and regulator [11-16].
parameter variation influence seriously Linear quadratic regulator design
the controlling effect of speed controller. technique is well known in modern
The most commonly used controller for optimal control theory and has been
the speed control of DC motors is widely used in many applications. It has a
conventional PID controller. Traditional very nice robustness property. This
PID controllers have been successfully attractive property appeals to the
used in control applications since 1940s practicing engineers. Thus, the linear
and are the most often used industrial quadratic regulator theory has received
controller today. Conventional PID considerable attention since 1950s. The
controllers have several important liner quadratic regulator technique seeks
features. The reason is that the to find the optimal controller that
conventional PID controller is easy to minimizes a given cost function
implement either by hardware or by (performance index). This cost function is
software. No deep mathematical theory is parameterized by two matrices, Q and R,
necessary to understand how the that weight the state vector and the
conventional PID controller works, so system input respectively. These

49
 Journal of Science and Technology - Engineering and Computer Sciences, Vol. 14, No. 2, December/ 2013

weighting matrices regulate the penalties disadvantages are a torque that decreases
on the excursion of state variables and at high speeds. In this paper, the
control signal. One practical method is to separately excited DC motor model is
Q and R to be diagonal matrix. The value chosen according to its good electrical
of the elements in Q and R is related to and mechanical performances more than
its contribution to the cost function. To other DC motor models. The electric
find the control law, Algebraic Riccati circuit of the separately excited DC motor
Equation (ARE) is first solved, and an is shown in figure 1. Objective is to
optimal feedback gain matrix, which will control the speed of the separately excited
lead to optimal results evaluating from DC motor by armature voltage control [1-
the defined cost function is obtained [17- 4].
20].
In this paper, to achieve accurate control
performance of speed control of DC motor,
optimal linear quadratic regulator
technique is presented. The remainder of
the paper is organized as follows: at first
the dynamic model of the separately
excited DC motor is briefly reviewed for
the purpose of speed control. The next
section the basic concept and design of
linear quadratic regulator controller is
briefly reviewed. Then the simulation
results are presented. Finally, the last
section states the main conclusion. Figure 1: A separately excited DC motor
Dynamic Model of DC Motor: model
Direct current motors are widely used for
various industrial and domestic Assuming constant field excitation the
applications. Examples are as robotic and armature circuit electrical equation is
actuator for automation process, written as:
mechanical motion, and others. Accurate di a
speed control of the DC motor is the Va  R a ia  L a  Eb
basic requirement in such applications. dt
di a (1)
There are two main ways of controlling a
Va  R a ia  L a  K b
DC motor: The first one named armature dt
control consists of maintaining the stator Where Va is the input terminal voltage
magnetic flux constant, and varying the (armature voltage) in volt, Eb is the back
armature current. Its main advantage is a emf in volt, Ra is the armature resistance
good torque at high speeds and its in ohm, La is the armature inductance in
disadvantage is high energy losses. The H. Kb is the back emf constant in Vs/rad,
second way is called field control, and  is represents angular speed in rad/s, and
has a constant voltage to set up the
ia is the armature current in A. The
armature current, while a variable voltage
dynamics of the mechanical system is
applied to the stator induces a variable
given by the following torque balance
magnetic flux. Its advantages are energy
equation:
efficiency, inexpensive controllers and its

50
 Journal of Science and Technology - Engineering and Computer Sciences, Vol. 14, No. 2, December/ 2013

d Armature Resistance, Ra 1
T  K T ia = J  B
dt
(2) Armature Inductance, La 0.05H
Moment of Inertia, J 0.01kgm2/s2
Where J is the moment of inertia of the Viscous Friction Coefficient, B 0.0000 3Nms
motor in kgm2/s2, T is the motor torque in
The Back EMF Constant, Kb 0.023Vs/rad
Nm, B is the viscous friction coefficient in
Nms, and KT is the torque factor constant The Torque Factor Constant, KT 0.023Nm/A
in Nm/A. Equation (1) and equation (2) are
rearranged to obtain:

d ia Ra Kb V
  ia    a (3)
dt La La La Design of the LQR Controller:
Linear quadratic regulator design technique
d KT B
 ia   (4) is well known in modern optimal control
dt J J theory and has been widely used in many
applications. The standard theory of the
To design a desired controller using the optimal control is presented in [17-20].
linear quadratic regulator technique, the Under the assumption that all state variables
system must first be expressed in the state are available for feedback, the LQR
space form. In the state space model of a controller design method starts with a
separately excited DC motor, the equation defined set of states which are to be
(3) and equation (4) can be expressed by controlled. In general, the system model can
choosing the angular speed () and be written in state space equation as
armature current (ia) as state variables and follows:
the armature voltage (Va) as an input. The
output is chosen to be the angular speed [1- x  Ax  Bu (6)
4].
Where x  R n and u  R m denote the state
R Kb  variable, and control input vector,
dia   a  0 respectively. A is the state matrix of order
 dt    La La  ia   1 V nn; B is the control matrix of order nm.
 d   K   L  a Also, the pair (A, B) is assumed to be such
 dt   T B   a  (5)
 J J  that the system is controllable. The linear
quadratic regulator controller design is a
ia method of reducing the performance index
y  0 1  
 to a minimize value. The minimization of it
is just the means to the end of achieving
The physical and functional parameters of acceptable performance of the system. For
the separately excited DC motor used for the design of a linear quadratic regulator
simulation testing are given in Table1. controller, the performance index (J) is
given by:
Table 1: Parameters of the separately
 T
excited DC motor
Parameters Values
 T

J   x Qx  u Ru dt
0
(7)

51
 Journal of Science and Technology - Engineering and Computer Sciences, Vol. 14, No. 2, December/ 2013

Where Q is symmetric positive semi- closed loop optimal control law is defined
definite (  0 ) state weighting matrix of as:
order nn, and R is symmetric positive *
u   Kx (8)
definite (  0 ) control weighting matrix of Where K is the optimal feedback gain
order mm. The choice of the element Q matrix, and determines the proper
and R allows the relative weighting of placement of closed loop poles to
individual state variables and individual minimize the performance index in
control inputs as well as relative equation (7). The feedback gain matrix K
weighting state vector and control vector depends on the matrices A, B, Q, and R.
against each other. The weighting There are two main equations which have
matrices Q and R are important to be calculated to achieve the feedback
components of an LQR optimization gain matrix K. Where P is a symmetric
process. The compositions of Q and R and positive definite matrix obtained by
elements have great influences of system solution of the ARE is defined as:
performance. The designer is free to T 1 T
choose the matrices Q and R, but the A P  PA  PBR B P  Q  0 (9)
selection of matrices Q and R is normally Then the feedback gain matrix K is given
based on an iterative procedure using by:
experience and physical understanding of 1 T
KR B P (10)
the problems involved. Commonly, a trial Substituting the above equation (8) into
and error method has been used to equation (6) gives:
construct the matrices Q and R elements. x  Ax  BKx  ( A  BK)x (11)
This method is very simple and very
If the eigenvalues of the matrix (A-BK)
familiar in linear quadratic regulator
have negative real parts, such a positive
application. However, it takes long time
definite solution P always exits.
to choose the best values for matrices Q
Simulation Results:
and R. The number of matrices Q and R
In order to verify the validity of the linear
elements are dependent on the number of
quadratic regulator controller, several
state variable (n) and the number of input
simulation tests are carried out using
variable (m), respectively. The diagonal-
MATLAB/SIMULINK software package.
off elements of these matrices are zero for
The performance of linear quadratic
simplicity. If diagonal matrices are
regulator controller has been investigated
selected, the quadratic performance index
and compared with the conventional PID
is simply a weighted integral of the
controller. Simulation tests are based on
squared error of the states and inputs. The
the facts that whether the linear quadratic
term in the brackets in equation (7) above
regulator controller is better and more
are called quadratic forms and are quite
robust than the traditional PID controller
common in matrix algebra. Also, the
or not. For the comparison, simulation
performance index will always be a scalar
tests of the speed response were
quantity, whatever the size of Q and R
performed according to the nominal
matrices [21-25]. The conventional linear
condition, moment of inertia variation,
quadratic regulator problem is to find the
and armature inductance variation of the
optimal control input law u * that separately excited DC motor. To
minimizes the performance index under determine the feedback gain matrix K, the
the constraints of Q and R matrices. The elements of the weighting matrices Q and

52
 Journal of Science and Technology - Engineering and Computer Sciences, Vol. 14, No. 2, December/ 2013

R are chosen as: Q = [0.2 0;0 0.028] and there is no steady state error using linear
R = 0.2, respectively. After solving the quadratic regulator controller. However,
ARE and substituting into equation (10), the rise time for traditional PID controller
the optimal values of control feedback is smallest value than for linear quadratic
gain matrix K are obtained as K= [0.9742 regulator controller.
1.3709]. Figure 2 shows the step
responses of speed control of the Table 2: Performances metrics for LQR
separately excited DC motor at nominal and PID controllers
condition by two controllers. According Time Response LQR PID
to the simulation results, linear quadratic Specifications
regulator method give the better Settling Time (Ts) 2s 4.5s
performance compared to traditional PID Rise Time (Tr) 2.5s 1.1s
controller.
Overshoot % 0 17
1.4
LQR Steady State Error (ess) 0 0.03
PID
1.2

1
For high performance applications the
proposed linear quadratic regulator
Speed (rad/sec)

0.8

0.6 scheme should be robust to parameter

0.4 variations. Changes in the moment of
0.2
inertia and the armature inductance are
investigated through simulations. The
simulation studies are undertaken by
0
0 2 4 6 8 10 12 14 16 18 20
Time (sec)

Figure 2: Comparison of output speed changing one parameter at a time while

responses among LQR and conventional keeping other parameters unchanged. The
PID controllers separately excited DC motor is
commanded to accelerate from rest to
The time response specifications of the reference speed under no torque load.
conventional PID controller and linear Figure 3 shows the separately excited DC
quadratic regulator technique obtained motor responses of optimal linear
from the simulation of the separately quadratic regulator approach and
excited DC motor speed control is shown conventional PID controller when the
in Table 2. Based on the Table 2, linear moment of inertia is increased by 100%
quadratic regulator technique has the of its original value, whilst figure 4
fastest settling time of 2s while traditional depicts the speed response when the
PID controller has the slowest settling armature inductance increased by 100%
time of 4.5s. For the percent overshoot, of its original value.
linear quadratic regulator technique does

not have overshoot and conventional PID

controller has the greatest value of
percent overshoot of 17%. Furthermore,

53
 Journal of Science and Technology - Engineering and Computer Sciences, Vol. 14, No. 2, December/ 2013

Settling Time (Ts) 3.6s 7.8s

1.4
LQR
Rise Time (Tr) 5.1s 2.1s
PID
1.2

Overshoot % 0 4
1

Steady State Error (ess) 0 0.03

Speed (rad/sec)

0.8

0.6
It is very much clear from figure 4 that the
0.4
proposed linear quadratic regulator
0.2
controller is less sensitive to parametric
variations and a robust tracking
0
0 2 4 6 8 10 12 14 16 18 20 performance is achieved in presence of the
uncertain parameters. Furthermore, it can
Time (sec)

Figure 3: Responses of the DC motor be noted that the increase in armature

using two controllers with variation in the inductance causes greatest value of percent
moment of inertia overshoot, settling time, and steady state
error in classical PID controller than
1.4
optimal linear quadratic regulator
LQR
PID
controller which is affected only by
slowest rise time. The time response
1.2

1 parameters percent overshoot; settling

time, rise time, and steady state error for
Speed (rad/sec)

0.8

LQR and PID controller are presented in

0.6
Table 4.
0.4

Table 4: Performances of two controllers

under increased La
0.2

0
0 2 4 6 8 10 12 14 16 18 20 Time Response LQR PID
Time (sec) Specifications
Figure 4: Responses of the DC motor 3.3s 10.3s
Settling Time (Ts)
using two controllers with variation in the
armature inductance Rise Time (Tr) 2.6s 1.4s

From figure 3, it can be seen that the Overshoot % 2 31

increment of the moment of inertia does
not impose any significant effect on the Steady State Error (ess) 0 0.03
performance of the linear quadratic
regulator technique but only affects the rise
time. A comparison is illustrated in Table 3 Conclusions:
between LQR and PID controller Optimal LQR strategy and conventional
quantitatively. PID controller have been considered in
this paper for controlling the speed of a
Table 3: Performances of two controllers separately excited DC motor. The
under increased J performance of the two controllers is
Time Response LQR PID validated through simulations. A number
Specifications of simulation results are presented for
comparison. Based on the comparative

54
 Journal of Science and Technology - Engineering and Computer Sciences, Vol. 14, No. 2, December/ 2013

simulation results, one can conclude that Proportional Integral Control for Brushless
the linear quadratic regulator controller DC Motor”, ELECTRONICS’ Sozopol,
realises a good dynamic behaviour of the BULGARIA, September, pp. 21-23, 2005.
separately excited DC motor with a rapid 7. Uma Maheshwararao.Ch,
settling time, no overshoot, and zero Y.S.kishoreBabu and K. Amaresh, “Sliding
steady state error compared to Mode Speed Control of DC Motor”,
conventional PID controller under nominal International Conference on
condition. But the comparison between the Communication Systems and Network
speed control of the separately excited DC Technologies, 2011.
motor by linear quadratic regulator 8. Rahul Malhotra, Tejbeer Kaur, “DC
technique and conventional PID controller Motor Control using Fuzzy Logic
shows clearly that the linear quadratic Controller”, International Journal of
regulator technique gives better Advanced Engineering Sciences and
performances than conventional PID Technologies, Vol.8, No.2, pp. 291-296,
controller against parameter variations. 2011.
Furthermore, the simulation results so 9. Arpit Goel, Ankit Uniyal, Anurag
obtained show that the conventional PID Bahuguna, “Performance Comparison of
controller gives greatest value of percent PID and Fuzzy Logic Controller using
overshoot and longer settling time. Different Defuzzification Techniques for
References: Positioning Control of DC Motors”,
1. Weiyao Lan and Qi Zhou, “Speed Journal of Information Systems and
Control of DC Motor using Composite Communication, Vol.3, No.1, pp. 235-238,
Nonlinear Feedback Control”, 2009 IEEE 2012.
International Conference on Control and 10. Basil Hamed, Moayed Almobaied,
Automation Christchurch, New Zealand, “Fuzzy PID Controllers using FPGA for
December 2009. Real Time DC Motor Speed Control”,
2. Moleykutty George, “Speed Control of Intelligent Control and Automation, Vol. 2,
Separately Excited DC Motor”, American pp. 233-240, 2011.
Journal of Applied Sciences, Vol. 5, No. 11. K. Ang, G. Chong, Y. Li, “PID control
3, pp. 227-233, 2008. system analysis, design, and technology”,
3. Y. J. Huang, T. C. Kuo, “Robust IEEE Trans. Control System Technology,
position control of DC servomechanism Vol. 13, pp. 559 – 576, 2005.
with output measurement noise”, Electr. 12. O. Yaniv, M. Nagurka, “Robust, PI
Eng., Vol. 88, pp. 223-238, 2006. controller design satisfying sensitivity and
4. S.J.Chapman, “Electric Machinery uncertainty specifications”, IEEE Trans.
Fundamentals”, The McGraw-Hill Automation Control, Vol. 48, pp.2069-
Companies, 1999. 2072, 2003.
5. Tossaporn Chamsai, Piyoros 13. H. Kim, I. Maruta, T. Sugie, “Robust
Jirawattana and Thana Radpukdee, PID controller tuning based on the
“Sliding Mode Control with PID Tuning constrained particle swarm optimization”,
Technique: An Application to a DC Automatica, Vol. 44, Iss. 4, pp. 1104 -
Servo Motor Position Tracking Control”, 1110, 2008.
Energy Research Journal 1 (2), pp. 55- 14. P. M. Meshram and Rohit G.
61, 2010. Kanojiya, “Tuning of PID Controller
6. Igor Karaskakovski1, Goce Shutinoski2, using Ziegler-Nichols Method for Speed
“Comparison of Sliding Mode and Control of DC Motor”, IEEE-

55
 Journal of Science and Technology - Engineering and Computer Sciences, Vol. 14, No. 2, December/ 2013

International Conference On Advances In multimachine power system’, Electrical

Engineering, Science And Management Power and Energy System 23, pp. 263-
(ICAESM -2012), pp. 117-122, March 271, 2001.
2012. 21. A.N.K.Nasir, M.A. Ahmad and
15. J. C. Basilio and S. R. Matos, “Design M.F.Rahmat, “Performance Comparison
of PI and PID Controllers With Transient between LQR and PID Controller for an
Performance Specification“, IEEE Trans. Inverted Pendulum System”, International
Education, vol. 45, Issue No. 4, pp. 364- Conference on Power Control and
370,2002. Optimization, Chiang May, Thailand,
16. J. G. Juang, M. T. Huang and W. K. July 2008.
Liu, “PID control using prescribed 22. P.O.M. Scokaert and J. B. Rawlings,
genetic algorithms for MIMO system”, “Constrained linear quadratic regulation”,
IEEE Trans. Systems, Man and IEEE Trans. Automatic Control, Vol. 43,
Cybernetics, vol. 38, no.5, pp. 716–727, No. 8, pp.1163-1169, 1998.
2008. 23. R. Yu, R. Hwang, ‘Optimal PID
17. T.L.Chem, Y.C.Wu, ‘An optimal Speed Control of Brushless DC Motors
variable structure control with integral using LQR Approach’, In Proc. IEEE
compensation for electrohydraulic International Conference on Man and
position servo control systems’, IEEE Cybernetics, Hague, Netherlands,
Transactions Industrial Electronics, October 2004.
Vol.39, pp.460-463, October 1992. 24. B. Chu, S. Kim, D. Hong, J. Park, and
18. M.Ghribi, H.Le-Huy, ‘Optimal T. Choy, “Optimal Control for
control and variable structure Synchronizing Positions of Multi-Axis
combination using a permanent magnet Driving System with Cross-Coupled
synchronous motor’, In Conf. Rec. Structure”, 2001 Proc. of the KSPE
IEEE-IAS Annul Meeting, Vol.1, Spring Annual Meeting, pp. 271-274,
Denver, Co, pp.408-415, October 1994. 2001.
19. Osama S. Ebrahim, Mohamed M. 25. G. Marro, D. Prattichizzo, and E.
Negm and Mohamed Z. Youssef, “An Zattoni, “Geometric Insight into Discrete-
Improved DTC scheme for The AC Time Cheap and Singular Linear
Drives Based on Optimal Preview Quadratic Riccati (LQR) Problems”,
Control Technique”, IEEE ISIE 2006, IEEE Trans. Automatic Control, Vol. 47,
Montréal, Canada, pp. 2194-2199, 2006. No. 1, pp. 102- 107, 2002.
20 I.Robadi, K.Nishimori, R.Nishimura,
N.Ishihara, ‘Optimal feedback control
design using genetic algorithm in

56