The model based method for energy saving of the separately excited DC motor drive system is proposed in the paper. The accurate power loss model is necessary for this method. Therefore, the adaptive tabu search algorithm is applied to identify the parameters in the power loss model. The field current values for minimum power losses at any load torques and speeds are calculated by the proposed method. The rule based controller is used to control the field current and speed of the motor. The experimental results confirm that the model based method can successfully provide the energy saving for separately excited DC motor drive. The maximum value of the energy saving is 48.61% compared with the conventional drive method.
1. Introduction
Nowadays, the separately excited DC motor is widely used in industries. This motor is suitable for the electric vehicle and conveyer
[1]
because it can provide the high torque to drive the load. Moreover, this motor is easy to control the speed for automatic control system. Presently, the energy saving in electrical system is a considered issue for industrial sectors, particularly in the electric motor drive. Therefore, the energy saving of the separately excited DC motor is proposed in the paper. In the previous works
[2

7]
, the conventional approach based on mathematics is applied to calculate the field current for energy saving. Unfortunately, the equation used in that approach is complicated and it is difficult to solve this problem. Therefore, the new method called the model based method (MBM) is used to calculate the field current for energy saving easily at any load torques and speed commands. The accurate power loss model is very important for the MBM method. Hence, the artificial intelligence technique called adaptive tabu search (ATS) is applied to identify the parameters in the power loss model. The ATS method is developed by Puangdownreong et al. in 2002
[8]
. In order to perform its effectiveness, the ATS has been tested against several wellknown benchmark functions, that is, Bohachevsky, Rastrigin, Shekel’s foxholds, Shubert and Schwefel functions
[9

13]
. Moreover, the convergence property of the ATS has been proved to assure that it can reach the optimal solution within finite search time
[9

14]
. Thus, the ATS algorithm can provide the global solutions for system identification. A single – phase bridge rectifier connected with two buck converters is a studied system in the paper as shown in
Fig. 1
. The expert controller called the rule based controller
[15]
is also presented for the field and speed controls. This controller is easy to implement and sufficient to control the field current and speed of the motor. In addition, the results of the electric energy consuming of the motor using the conventional drive method are compared with the MBM drive method. The results from the implementation in laboratory can confirm that the MBM method can provide the minimum losses at any load torques and speed commands. The maximum percent of the energy saving using the MBM drive method is equal to 48.61% compared with the conventional drive method. However, the MBM method cannot provide to save the energy in case of the high load torque.
The block diagram of a separately excited DC motor drive for energy saving
The reviews of power losses of the separately excited DC motor and the identification of the power loss model are presented in Section 2. The energy saving algorithm using the MBM method and the design of rule based controllers are fully described in Section 3 and 4, respectively. The experimental results of the energy saving using the MBM drive method compared with the conventional drive method and discussion are shown in Section 5. Finally, Section 6 concludes the overall advantages using the MBM method to save the energy in separately excited DC motor drive system.
2. Power Losses of Separately Excited DC Motor and Power Losses Identification
 2.1 Power losses of separately excited DC motor
The power losses can be classified into five types. These losses are both electrical and mechanical losses. The details of the losses are described in this section.
 2.1.1 Copper losses
The copper losses are divided in two cases. These losses consist of the armature and field copper losses that can be calculated by (1).
 2.1.2 Friction and windage losses
These losses occur from the friction of the bearings in the motor and the friction between the moving parts of the motor and the air in the motor case
[16]
. These losses are given in (2).
 2.1.3 Core loss
The core loss occurs in the iron core of the motor. The core loss consists of the hysteresis loss
[2
,
16]
and the eddy current loss
[17]
defined by (3).
 2.1.4 Brush loss
This loss that occur the voltage drop across the brushes in the motor can be calculated by (4).
 2.1.5 Stray loss
The stray loss depended on the armature current and the motor speed can be calculated by (5).
The friction and windage losses and eddy current loss are neglected because the motor using in the paper is small horsepower (0.5hp). Therefore, the total power losses can be defined by (6).
 2.2 Power Losses Identification
In the paper, the total power losses can be calculated by (6). The accurate parameters in (6) are necessary to achieve the precise total power losses value. Thus, the method to identify the power loss parameters is presented in this section. The armature current (
i_{a}
), field current (
i_{f}
), and motor speed (
ω
) values in (6) can be measured by sensors, while armature resistance (
R_{a}
) and field resistance (
R_{f}
) are also measured by an ohmmeter. In the paper,
R_{a}
and
R_{f}
are equal to 16 Ω and 735 Ω , respectively. Moreover, the coefficients of the stray losses (
K_{st}
) and hysteresis losses (
K_{h}
) in (6) are difficult to measure or calculate. In previous researches
[3]
, the equation to calculate the
K_{st}
and
K_{h}
values are not appeared. Therefore, the identification of two parameters are necessary. In the paper, the artificial intelligence technique called adaptive tabu search method (ATS) is applied to identify these parameters (
K_{st}
and
K_{h}
). The ATS method is confirmed by
[8]
to achieve the global solution. The back tracking and adaptive search radius are the main mechanisms of the ATS method to escape the local solution to global solution. The block diagram to identify the power loss parameters in the paper is illustrated in
Fig. 2
.
The block diagram to identify the parameters of power loss equation using ATS algorithm
In
Fig. 2
, the equation (7) is used as the objective function to calculate the cost value (
W
). This equation is the root mean square error between the power loss values from experiment (
P_{loss(experiment)}
) and computation (
P_{loss(computation)}
) as given in (8). The
n
value in (7) is the total data number. The experimental rig to measure the (
P_{loss(experiment)}
) is depicted in
Fig. 3
. In addition, the results from motor testing at 80%, 90%, 100%, 110% and 120% of rated speed are shown in
Table 1
. The data of 80%, 100% and 120% of rated speed are used to identify the power loss parameters (
K_{st}
and
K_{h}
), while the data of 90% and 110% of rated speed are used for validation.
The experimental rig for testing
The motor testing results
The motor testing results
From
Fig. 2
, the ATS is applied to search the
K_{st}
and
K_{h}
parameters. The new
K_{st}
and
K_{h}
values from ATS searching is used to calculate the power losses (
P_{loss(computation)}
). The new
P_{loss(computation)}
value calculated by using the new
K_{st}
and
K_{h}
parameters is subtracted with
P_{loss(experiment)}
again. The ATS process is used to tune the
K_{st}
and
K_{h}
values until the minimum
W
value can be obtained. The convergence of
W
value is shown in
Fig. 4
. In this paper, the initial solutions for ATS searching are determined by random. However, the authors define the suitable search space that can provide the initial solution as better as possible. As a result, the
W
value is equal to 4 at the beginning. If the search space is not appropriate, the initial
W
value will be higher than 4. As for this case, the count for ATS searching equal to 20 is not enough to achieve the best solution. The more details of the ATS searching process can be found in
[8]
. The results from the power loss parameters identification with ATS algorithm are shown in
Table 2
. In
Table 2
, the
K_{h}
value from the 1
^{st}
experimental is different from the others because of the random process of the initial solution in ATS searching. In the paper, the ATS searching process was conducted 5 trials with random initial solutions. Therefore, the average values of
K_{h}
and
K_{st}
are suitable for using in the paper. The average values of
K_{st}
and
K_{h}
are equal to 8.68×10
^{−7}
W·s
^{2}
/A
^{2}
and 4.77×10
^{−8}
W·s/A
^{2}
·rad, respectively. It can be seen from the results that the
K_{st}
and
K_{h}
values are very small because the motor using in the paper is a small size. In addition, the parameters validation is necessary to guarantee the identification results. The validation results show that the
K_{st}
and
K_{h}
values are appropriate to use at any operating points. Therefore, the equation to calculate the total power losses of the motor using in the paper is shown in (9). In the paper, the voltage drop across the brushes (
U_{e}
) is set to 2 V
[1]
.
The W convergence
The ATS searching results and the validation results of the power loss equation
The ATS searching results and the validation results of the power loss equation
3. Energy Saving using Model Based Method
 3.1 The conventional method
The conventional method to control the speed of the separately excited DC motor is divided into two cases. In the first case, the armature voltage control can be adjusted, while the field voltage is set to be constant at rated value. For this case, the speed command is less than the rated speed. In the second case, the speed command is more than the rated speed. The armature voltage is fixed to rated and the field voltage is adjusted to increase the motor speed. The input power (
P_{in}
) calculation of the motor for conventional control approach can be calculated by (10).
 3.2 The model based method
In the paper, the new method to minimize the power losses for separately excited DC motor is presented. This method is called the model based method (MBM). The MBM is applied to calculate the field current for energy saving at any operating points. The total power loss values calculated by (9) from section 2.2 at load torque (
T_{L}
) equal to 0.2, 0.4, 0.6, 0.8, 1.0, 1.2, 1.4 and 1.5 N·m are shown in
Fig. 5
. From this figure, the motor speed and field current are varied from 5003000 rpm and 0.10.3 A, respectively. The armature currents in
Fig. 5
are calculated by the basic equation as shown in (11) using
K
= 2.49 N·m /A
^{2}
.
The minimum power loss at any speeds and load torques
From
Fig. 5
, it can be seen that the minimum power loss value is depended on the field current at any load torques and speeds. Therefore, the field current for the minimum loss can be calculated. The optimal field current values for loss minimization at any load torques and speeds from
Fig. 5
are shown in
Fig. 6
. The curve in
Fig. 6
is called
i_{f}
−
N
curve for loss minimization. In case of
T_{L}
= 0.6 N·m and speed command
N
=1500 rpm, the field current (
i_{f}
) for minimum power loss is equal to 0.197 A. The
i_{f}
−
N
curve in
Fig. 6
can be approximated by linear regression as depicted in
Fig. 7
. The equation (12) is used to approximate the
i_{f}
value for loss minimization. The
a_{0}
and
a_{1}
coefficients at any load torques are shown in
Table 3
.
The i_{f}− N curves for loss minimization.
The linear approximation of i_{f}− N curve using linear regression method.
Thea0anda1coefficients from linear regression method
The a_{0} and a_{1} coefficients from linear regression method
In
Fig. 7
, there are eight
i_{f}
−
N
curves for
T_{L}
= 0.2, 0.4, 0.6, 0.8, 1.0, 1.2, 1.4 and 1.5 N·m. The
i_{f}
−
N
curves for other load torques can be determined by using the linear interpolation method. The
a
_{0}
and
a
_{1}
coefficients from linear interpolation method can be calculated by the equations in
Table 4
. For example, if
T_{L}
= 0.3 N·m, the equations to calculate the
a_{0}
and
a_{1}
coefficients can use the equations in 0.2 <
T_{L}
< 0.4 interval. The
i_{f}
−
N
curves for any load torques using the linear interpolate are shown in
Fig. 8
. In the paper, the equation (12) with the coefficients in
Table 4
is used to calculate the field current for loss minimization at any load torques and speeds.
Thea0anda1coefficients from linear interpolation method
The a_{0} and a_{1} coefficients from linear interpolation method
The i_{f}−N curves using linear interpolation method.
4. Rule Based Controller
The expert controller called the rule based controller is applied for speed and field current controls in the paper. The inference engine and knowledge based are the important parts of this controller. The rules of the controller are generated from the engineer experiences.
 4.1 Field current control
The MBM method described in Section 3 is used to calculate the field current for loss minimization. The buck converter in the field side of
Fig. 1
is used to adjust the field current to the command value. The duty cycle of
Q_{2}
is adjusted by the rule based controller. There are ten rules for field current control.
Rule 1
:
Rule 2
:
Rule 3
:
Rule 4
:
Rule 5
:
Rule 6
:
Rule 7
:
Rule 8
:
Rule 9
:
Rule10
:
The
err_f
in the rules is the error between the field current value for loss minimization calculated by MBM method (
i_{f,minimize}
) and the actual field current (
i_{f,act}
) as defined by (13). For example in
Rule 1
, the
i_{f,minimize}
is more than the
i_{f,act}
(more than 15 mA), the
err_f
is the positive value. In this case, the
i_{f,act}
should be increased by increasing the duty cycle of
Q_{2}
2.5%.
 4.2 Motor speed control
The rule based controller is also used to control the motor speed to the desired value. The buck converter connected in the armature side in
Fig. 1
is applied to adjust the armature voltage to control the motor speed. The duty cycle of
Q_{1}
is adjusted by the eight rules of the rule based controller.
Rule 1
:
Rule 2
:
Rule 3
:
Rule 4
:
Rule 5
:
Rule 6
:
Rule 7
:
Rule 8
:
The
err_s
is the error between the speed command (
N
^{*}
) and the actual speed (
N_{act}
) as shown in (14). For example in
Rule 5
case, the
N_{act}
is more than the
N
^{*}
(more than 200rpm), the
err_s
is the negative value. In this case, the duty cycle of
Q_{1}
is decreased 1.5% to reduce the armature voltage.
In the paper, the dsPIC30F2010 microcontroller is used to implement the rule based controllers for current and speed control loops. Moreover, the dsPIC30F2010 is used to calculate the
i_{f,minimize}
using the MBM method for energy saving. The overall procedures for energy saving and controls are explained in
Fig. 9
. From this figure, the speed command (
N
^{*}
) and load torque (
T_{L}
) are set by user ( ≤ rated of the motor). After setting
N
^{*}
and
T_{L}
values, the dsPIC30F2010 calculates the
i_{f,minimize}
for energy saving using MBM method. The actual field current (
i_{f,act}
) from current sensor is used to calculate the
err_f
. The
err_f
value is the input of the rule based controller to adjust the duty cycle of
Q_{2}
. After the duty cycle adjustment to adapt the field current to command value (
i_{f,minimize}
), the speed control is action. The actual speed (
N_{act}
) from speed sensor is detected to calculate the
err_s
. This error is used to check in eight rules for rule based controller. The controller adjusts the duty cycle
Q_{1}
to change the armature voltage for speed control. For this situation, the motor speed is equal to speed command and the field current is equal to
i_{f,minimize}
for energy saving. In case of the speed command or variation of load torque, the dsPIC30F2010 calculates the new
i_{f,minimize}
value again and the rule based controller starts to control the field current and motor speed to track the command value.
The block diagram of motor drive with the rule based controller
5. Experimental Results and Discussion
The experimental setup of the separately excited DC motor drive system is shown in
Fig. 10
. It can be seen in
Fig. 10
that the single phase bridge rectifier connected with two buck converters are the power converters of the system. The dsPIC30F2010 microcontroller board is used to calculate the command value of field current for energy saving based on MBM method. Moreover, it acts as the current and speed controllers using rule based control technique. The rating of the separately excited DC motor using in the paper is 0.5hp, 220 V, 2360rpm, 2.2 A. The conventional method and MBM described in Section 3 are applied to drive the motor. The experimental results using MBM method compared with the conventional method are depicted in
Table 5
. The load torque (
T_{L}
) values for testing are in the range 0.21.2 N·m. The speed commands (
N
^{*}
) for testing at any load torques are 1000, 1200, 1300, 1400 and 1500 rpm. In
Table 5
, the input powers using the conventional method are compared with the MBM method. In case of
T_{L}
= 0.2 N·m and
N
^{*}
=1000 rpm, the MBM method can save the energy to 48.61%. However, the MBM method cannot save the energy in high load torque cases. From
Table 5
, the percent of energy saving becomes lower for higher load torque. The MBM drive method can provide the small input power compared with the conventional drive. The small input powers in each situation indicate that the power losses of the motor are small. The power losses for MBM and conventional methods at any load torques and speeds are shown in
Fig. 11
. From this figure, it can be seen that the region of energy saving is a large area at slight load torque. At high load torque, the region of energy saving becomes small area. The experimental results confirm that the proposed MBM drive method can provide the good result in terms of energy saving for a separately excited DC motor drive system.
A driving system for separately excited DC motor
The energy saving of the motor using difference technique to control the motor speed
The comparison results of energy saving
The comparison results of energy saving
6. Conclusion
This paper presents the new method to drive the separately excited DC motor called the MBM method. The MBM method provides the good results in case of energy saving compared with the conventional drive method. The parameters of the power loss model are identified by using the adaptive tabu search (ATS) algorithm. This accurate power loss model is used in the MBM method to calculate the field current value for energy saving. The rule based controller is applied to control the field current and speed of the motor to track the command values. The experimental results show that the MBM drive method can provide the minimum losses at any load torques and speed commands. The maximum percent of the energy saving is equal to 48.61%. However, in case of the high load torque values, the MBM drive method cannot provide to save the energy
Nomenclature and Units
i_{a} , i_{f} are the armature and field currents. u_{a} , u_{f} are the armature and field voltages. u_{e} is the voltage drop across the brushes. R_{a} , R_{f} are the armature and field resistances. K_{m} is the coefficient of the friction and windage losses. K_{h} is the coefficient of the hysteresis loss. K_{e} is the coefficient of the eddy current loss. K_{st} is the coefficient of the stray loss. P_{a} , P_{f} are the armature and field copper losses. P_{m} is the friction and windage losses. P_{ai} is the core loss. P_{BD} is the brush loss. P_{s} is the stray loss. P_{loss} is the total power loss. P_{in} is the input power. P_{out} is the output power. N is the motor speed (rpm). w is the angular motor speed (rad/s). T_{L} is the torque load. W is the cost value. P_{loss(experiment)} is the power loss from experiment. P_{loss(computation)} is the power loss from computation. n is the total data numbers. a_{0} , a_{1} are the coefficients of the linear regression at any load torques. i_{f,minimize} is the field current value for the loss minimization. i_{f,act} is the actual field current. N^{*} is the speed command. N_{act} is the actual speed. err_ f is the error between i_{f,minimize} and i_{f,act}. err_ s is the error between N^{*} and N_{act}.
Acknowledgements
This work was supported by Suranaree University of Technology (SUT).
BIO
Sasiya Udomsuk was born in Nakhon Ratchasima, Thailand, in 1987. She received the B.Eng and M.Eng degrees in electrical engineering from Suranaree University of Technology (SUT), Nakhon Ratchasima, Thailand, in 2008 and 2011, respectively where she is currently studying toward the Ph.D. degree in electrical engineering. Her main research interests include motor drive, power electronics, AI application and control system.
Kongpol Areerak received the B.Eng, M.Eng, and Ph.D. degrees in electrical engineering from Suranaree University of Technology (SUT), Thailand, in 2000, 2003, and 2007, respectively. Since 2007, he has been a lecturer and Head of Power Quality Research Unit (PQRU) in the School of Electrical Engineering, SUT. He received the Assistant Professor in Electrical Engineering in 2009. His main research interests include active power filter, harmonic elimination, artificial intelligence applications, motor drive, and intelligence control systems.
Kongpan Areerak received the B.Eng. M.Eng degrees from Suranaree Uni versity of Technology (SUT), Thailand, in 2000 and 2001, respectively and the Ph.D. degree from the University of Nottingham, Nottingham, UK., in 2009, all in electrical engineering. In 2002, he was a lecturer in the Electrical and Electronic Department, Rangsit University, Thailand. Since 2003, he has been a Lecturer in the School of Electrical Engineering, SUT. His main research interests include system identifications, artificial intelligence applications, stability analysis of power systems with constant power loads, modeling and control of power electronic based systems, and control theory.
Chapman S. J.
1999
Electric Machinery Fundamentals
3rd ed.
McGraw
Boston
Zaki A. M.
,
ElSattar A. A.
,
Wahsh S.
,
Mashaly H. M.
,
Amer S. I.
1998
“EfficiencyOptimized Speed Control for a SeparatelyExcited DC Motor Using Adaptive Neuroncontroller,”
IEEE International Conference on Power Electronic Drives and Energy Systems for Industrial Growth
421 
426
Kusko A.
,
Galler D.
1983
“Control means for minimization of losses in ac and dc motor drives,”
IEEE Trans. Industrial Application
IA19
561 
570
DOI : 10.1109/TIA.1983.4504256
Angelo C. D.
,
Bossio G.
,
Garcia G.
1999
“Loss minimization in dc motor drives,”
International Conference Electric Machines and Drives (IEMD’99)
701 
703
Bernal F. F.
,
Cerrada A. G.
,
Faure R.
2000
“Modelbased loss minimization for dc and ac vectorcontrolled motors including core saturation,”
IEEE Trans. Industrial Application
36
(3)
755 
763
DOI : 10.1109/28.845050
Issa R. H.
2013
“Separately excited dc motor optimal efficiency controller,”
International Journal of Engineering and Innovative Technology (IJEIT)
3
533 
539
Sheta M. A.
,
Agarwal V.
,
Nataraj P. S. V.
2009
“A new energy optimal control scheme for a separately excited dc motor based incremental motion drive,”
International Journal of Automation and Computing
6
(3)
267 
276
DOI : 10.1007/s1163300902674
Puangdownreong D.
,
Areerak KN.
,
Srikaew A.
,
Sujijorn S.
,
Totarong P.
2002
“System Identification via Adaptive Tabu Search,”
IEEE International Conference on Industrial Technology (ICIT’02)
915 
920
Kluabwang J.
,
Puangdownreong D.
,
Sujitjorn S.
2012
“Multipath adaptive tabu search for a vehicle control problem,”
Journal of Applied Mathematics
Article ID 731623
2012
20 
Puangdownreong D.
,
Kulworawanichpong T.
,
Sujitjorn S.
,
Negoita M. G.
,
Howlett R. J.
,
Jain L. C.
2004
KnowledgeBased Intelligent Information and Engineering Systems, vol. 3215 of Lecture Notes in Computer Science
Springer
Heidelberg
“Finite convergence and performance evaluation of adaptive tabu search,”
710 
717
Kulworawanichpong T.
,
Puangdownreong D.
,
Sujitjorn S.
2004
“Finite convergence of adaptive tabu search,”
ASEAN Journal on Science and Technology for Development
21
(23)
103 
115
Sujitjorn S.
,
Kluabwang J.
,
Puangdownreong D.
,
Sarasiri N.
2009
“Adaptive tabu search and management agent,”
The ECTI Trans. on Electrical Engineering, Electronics, and Communications
7
(2)
1 
10
Sujitjorn S.
,
Kulworawanichpong T.
,
Puangdownreong D.
,
Areerak KN.
,
Zha X. F.
,
Howlett R. J.
2006
Integrated Intelligent Systems for Engineering Design
IOS Press
The Netherlands
“Adaptive Tabu Search and Applications in Engineering Design,”
233 
257
Sujitjorn S.
“Concepts of Design and Implementation of a RealTime Expert Controller,”
The 1991 engineering technology symposium and exhibition
Bangkok
Oct. 31Nov. 3, 1991
693 
703
Toro V. D.
1990
Basic Electric Machines
Prentice Hall
New Jersy
Nasar S. A.
1995
Electrical Machines and Power Systems
McGrawHill
New York