Вы находитесь на странице: 1из 6

Speed control of a Motor-Generator system using

Internal Model Control techniques


L. Angel, J. Viola and M. Paez
Universidad Pontificia Bolivariana
Bucaramanga, Colombia
(luis.angel, jairo.viola, miller.paez)@upb.edu.co

Abstract This paper presents the speed control for a motor- a function of a single parameter Q [5] - [8]. For single input-
generator system using an internal model control based controller. output systems (SISO) represented by transfer functions, when
Initially, the internal model control technique is presented, which applying the IMC methodology, it is possible to obtain a set of
is compared with a pole placement based PID controller. Then, a tuning rules based on a parameter that fit into the model of a
linear model of the motor-generator system is identified employing PID controller. The selection of specifies the closed loop
the captured data from a data acquisition card and the matlab response speed and has a direct relationship with the robustness
identification toolbox. Subsequently, the internal model control of the system [9].
based controller and the pole placement PID controller are
designed based on the identified linear model. Controllers This paper presents the identification, design, and
implementation are performed using Matlab Stateflow toolbox in implementation of a speed control for a motor-generator system
conjunction with a data acquisition card. The performance of the using an IMC controller. Initially, the response of the motor-
controllers is evaluated in presence of external disturbances for generator system against a stepped reference signal is captured.
tracking tasks and is quantified through performance indices. From the captured information, a linear model of the system is
Results show that employing the internal model control based obtained using the Matlab identification toolbox. Then, an IMC
controller, the motor-generator system has a better performance controller based on the PID structure is tuned. After that, a PID
for tracking tasks than the pole placement PID controller. controller is designed using the pole assignment methodology to
contrast with the performance of the IMC controller.
KeywordsInternal model control, PID controller by pole
placement, motor-generator system, Matlab Stateflow Toolbox,
Subsequently, controllers are implemented using the Matlab
performance indices. Stateflow toolbox in conjunction with a data acquisition card.
Then, the IMC and PID controllers performance for tracking
task is evaluated in presence of external disturbances as well as
I. INTRODUCTION the presence of random noise in the feedback loop. To quantify
The dynamic behavior of a system is influenced by external the performance of the controllers, performance indices for the
and internal factors that impact the system performance. To systems temporal response and the control action are employed.
reduce the influence of these factors, control system is designed
The main contributions of this paper are the practical
to ensure the system operation under desired operating
implementation of a IMC controller based on the PID structure
conditions.
and the quantitative performance analysis of the IMC and PID
In the industry, most of the control systems use PID by pole placement controllers.
controllers due to its high reliability and cost/benefit ratio, as
This article is structured as follows. Initially the concept of
well as its simplicity for design and implementation [1].
IMC control and the design methodology of an IMC controller
For the PID controllers tuning, there are different based on the PID structure are presented. Then, the pole
methodologies that allow to find the constants of the controller. placement methodology used in the design of a PID controller is
Some of the design methodologies are empirical and are based presented. Subsequently, the study case is presented which
on the graphical analysis of the system response to different corresponds to the design of a speed control for a motor-
input signals [1] - [3]. The main limitation for the application of generator system using an IMC controller and a PID controller
these methodologies is that it depends on a stable open loop by pole placement, as well as the performance analysis of the
response of the system to obtain the controller constants. controllers for tracking tasks in the presence of external
disturbances. Finally, conclusions are presented.
Other methodologies employ an analytical model of the
process, which can be obtained through identification
techniques or by applying the physical laws that govern the II. IMC CONTROL
behavior of the system. These methodologies ensure close-loop Internal model control structure given by [1], [9] is based on
compliance of a set of desired operating specifications of the the internal model principle which proposes that a precise
control system [2] - [4]. control of a system can be reached only if the control system
Internal model control (IMC) is a methodology for the consider the model of the process to be controlled. A closed
controller design based on the Q parameterization technique, loop control system is given by Fig.1, where is the system
which describes all possible controllers that stabilize a system as output, is the desired input of the system, is the error

978-1-5386-0398-7/17/$31.00 2017 IEEE


signal, is the control action of the controller , is an
external disturbance and is the model of the system to be
controlled.

Fig. 2. IMC control scheme

However, to implement an IMC controller must be considered


Fig. 1. Feedback system that the plant estimated model approximates . Also, as
shown in (5) is the inverse of , forming an improper
From Fig.1, the system output is given by (1).
fraction which implementation is not possible. For this, a filter
( ) given by (8) is established.
= + (1)
1
If = ( )= (8)
( + 1)
= + (2)
where is an adjustable parameter and is the filter order
which eliminate the improper fraction. Thus, the transfer
Solving in (2), the control action is defined in (3). function for the Controller is given by (9).
1
= ( ) (3) (9)
( )=
( + 1)
From Fig.1, control action is defined as: Fig.3 shows the block representation of the IMC controller
including the filter defined in (9).
= ( ) (4)

If = , comparing (3) and (4) is:

1
= (5)

This means that to achieve a perfect control action which Fig. 3. Generalized IMC control scheme
ensure that = , the controller model must be the inverse
From Fig.3, the control action for the IMC controller is
of the system , considering ideal. However, in most cases
defined by (10).
is not ideal, so that an approximated model of the system
( ) is employed. For the disturbance , it is possible to find ( )
an estimated model given by (6), subtracting and the ( )= ( ) (10)
1 ( )
estimate model of the process.

(6) If is a second order with real poles system given by (11).


=

Replacing (6) on (3), the control action of the system depends ( )= (11)
of the reference signal and the disturbance as shown in (7). ( + 1)( + 1)

1 and replacing (11) in (10), the IMC controller transfer function


= ( ) (7) is defined in (12).

( + 1)( + 1) (12)
From (6), Fig.2 shows the block representation of a IMC ( )=
controller. As can be observed, the plant effect is eliminated
considering only the disturbance effect for the desired operating
point.
Adjusting (12), the IMC controller transfer function (13) 1 1 1
+ + + + + + =0 (18)
resembles a PID controller which is given by (14).

+ 1 From (18), the closed loop characteristic equation of the


( )= 1+ + (13) system is a third order polynomial. So that the desired closed
( + ) +
loop characteristic equation of the system (17), is modified as
1 show in (19).
( )= 1+ + (14)
( + )( +2 + )=0 (19)
where is the proportional constant, is the integral time and
where is a non-dominant pole of the system located five or
is the derivative time. Comparing (13) with (14), the terms
ten times farther than the dominant poles of the closed loop to
of IMC Controller based on the PID structure are defined by
not influence the dynamics of the closed loop system. Equating
(15).
(18) and (19), the constants for the PID controller for a second
+ order system with real poles are given by (20).
=
(1 + 2 )1
=
1
= (15)
( + )
(1 + 2 )1
= (20)
=
( + )
( +2 )
=
As shows in (15), and depends only of the model of the (1 + 2 ) 1
system, while the proportional constant depends on the model
of the system and the parameter which defines the closed loop
speed response of the system. IV. STUDY CASE: SPEED CONTROL OF A
MOTOR-GENERATOR SYSTEM
III. PID CONTROLLER BY POLE PLACEMENT Control techniques presented in Section II and Section II are
According to [1], the pole placement methodology aims to employed for the speed control of a motor-generator system
obtain a closed loop system with a set of desired poles. It shown in Fig.4. This system is a motor-generator conformed by
requires a complete model of the process usually restricted to a two DC motors Bodine brand with a maximum speed of 2500
first or second order system. RPM and 130v DC nominal voltage. As can be shown, the
To apply this design methodology, initially the closed loop motors M1 and M2 are mechanically coupled by its axis where
desired operating conditions of the system are defined. These M1 works as motor and M2 as generator. M1 has a power driver
conditions establish the behavior of the system in the transient M4 with an 10v input. A tachogenerator M3 with 7v output
state and the steady state and are given in terms of the damping voltage is coupled to M2 for the speed feedback of the system.
ratio , the natural frequency of the system , the setting time
and the steady state error. Then, a PI or PID controller is
selected and the controller characteristic equation plus plant in
closed loop is calculated which is given by (16).

1+ ( ) ( )=0 (16)

Then, the desired characteristic equation of the system (17) is


calculated using the operating conditions defined above.

( +2 + )=0 (17)
Fig. 4. Motor-Generator System
Finally, comparing the coefficients of (16) and (17), results
an equations system whose solution allows to find the constants A. System Identification
of the controller. For the motor-generator system identification, the system is
Applying the pole placement methodology for a second order excited with a stepped speed profile shown in Fig.5. This profile
system with real poles and a PID controller given by (11) and allows to observe the dynamical behavior of the system for
(14) respectively, the closed loop characteristic equation of the different operating points in both turn senses. As output signal
system applying (16) is:
for the identification process the tachogenerator voltage M3 is data acquisition card capture the tachogenerator signal to
employed. generate a new control action by the controller implemented in
Matlab Stateflow.
10
V. PERFORMANCE TEST
Output (v)

5
0
To analyze the performance of the IMC controller and the
-5 PID controller by pole placement, four tests for tracking task
-10
0 1000 2000 3000 4000 5000 are performed. The first test evaluates the system performance
Time (s) in nominal operating conditions, it means without consider any
10 external disturbance. In the second test, the system is submitted
5 to an external disturbance in the control action. The third test
Input (v)

0 analyze the system behavior in presence of random noise in the


-5
feedback loop. For the fourth test, a resistive load is connected
to the generator output of the motor-generator system.
-10
0 1000 2000 3000 4000 5000 The time response, the control action, and the error signal of
Time (s) the motor-generator system for the IMC controller and the PID
Fig. 5. Identificacion trajectory and system response
controller by pole placement during the first test for nominal
operating conditions is shown in Fig.6. As can be observed, the
Data acquisition of the system is performed using a National motor-generator system reaches the desired operating speed
Instruments data acquisition card NI DAQ 6008 M5 altogether employing the IMC controller and the PID controller by pole
with Matlab Simulink M6, where the steeped speed profile is placement for tracking tasks. On the other hand, the
generated and the system output is captured. From the acquired performance of the control action is similar for the IMC and
data, the model of the system is identified using the Matlab PID pole placement controllers. For the error signal, IMC
Identification Toolbox, resulting in a second order system with controller is closer to zero than PID controller by pole
real poles given by (21) with a fit of 93%. placement showing a better behavior in tracking tasks.

3000 8
0.62 PID controller 7 PID controller
( )= (21)
2000 IMC controller IMC controller
(21.725 + 1)(5.12 + 1)
6
Speed (RPM)

1000 5

Output (v)
4
0
3
-1000 2
B. Controllers tuning -2000
1
0
From (21), the IMC controller and the PID controller by -3000 -1
0 1000 2000 3000 4000 5000 0 1000 2000 3000 4000 5000
pole placement are tuned. For the PID controller by pole Time (s) Time (s)

placement, the desired operating conditions are an overshoot of 3000


a) b)
10% and a setting time of 150s. For this, a damping ratio of PID controller
2000 IMC controller
= 0.6 and a natural frequency of the system =
Speed error (RPM)

0.04 rad / s are required. For the IMC controller, is the only 1000

tuning parameter, which is adjusted to obtain a system response 0

like the obtained by the desired PID controller by pole -1000

placement. Using (15) and (20), the constants of the IMC and -2000

PID by pole assignment controllers are presented in Table I. -3000


0 1000 2000 3000 4000 5000
Time (s)

TABLE I. CONTROLLERS CONSTANTS c)


Fig. 6. Tracking response for the motor-generator system for nominal
Controller
operating conditions a) time response b) control action and c) error signal.
IMC 8.65 26.85 4.14
PID by pole placement 2.98 19.02 15.35
Fig.7 shows the time response, the control action, and the
error signal of the motor-generator system for the IMC
C. Controllers practical implementation controller and the PID controller by pole placement for the
For the practical implementation of the IMC controller and second test in presence of external disturbance in the control
the PID controller by pole placement, controllers are discretized action introduced in = 3200s . Controllers performance
by the backward difference method and programmed in Matlab against the external disturbance is highlighted into a red square
Stateflow. The practical implementation includes the anti- and is similar for the tracking tasks and the control action. In
wind-up scheme design using the back-calculation presence of external disturbance, the error signal of the PID
methodology and the actuator saturation range [1]. controller by pole placement shows a better performance
The control action is transmitted through the data acquisition against the IMC controller.
card NI DAQ 6008 to the motor-generator system. At once, the
3000 8 the IMC and PID by pole placement controllers in presence of
PID controller PID controller
2000 IMC controller
7
6
IMC controller resistive load in the generator output. As can be shown, the
presence of resistive load in the generator output affects the
Speed (RPM)

1000 5

Output (v)
0
4 controllers performance, impeding reach the maximum speed
3
-1000 2 of the system. The error signal exhibit a better performance for
-2000
1 the IMC controller than PID controller by pole placement with
-3000
0
-1
a lower amplitude.
0 1000 2000 3000 4000 5000 0 1000 2000 3000 4000 5000
Time (s) Time (s)
3000 8
a) b) PID controller 7 PID controller
3000 2000 IMC controller IMC controller
PID controller 6

Speed (RPM)
2000 IMC controller 1000 5

Output (v)
Speed error (RPM)

4
1000 0
3
-1000 2
0
-2000 1
-1000 0
-3000 -1
-2000 0 1000 2000 3000 4000 5000 0 1000 2000 3000 4000 5000
Time (s) Time (s)
-3000
0 1000 2000 3000 4000 5000 a) b)
Time (s) 3000
c) PID controller
2000 IMC controller
Fig. 7. Tracking response for the motor-generator system against external

Speed error (RPM)


disturbance in the control action a) time response b) control action and c) error 1000

signal. 0

-1000
The time response, the control action, and the error signal for
-2000
the IMC and PID by pole placement controllers for the third test
against the presence of random noise in the feedback loop is -3000
0 1000 2000 3000
Time (s)
4000 5000

presented in Fig.8. This noise is produced by the motor- c)


generator system which is eliminated in the other tests using a Fig. 9. Tracking response for the motor-generator system with resistive load
low pass filter with a capacitor to the tachogenerator output. As at generator output a) time response b) control action and c) error signal.
shows Fig.8, the IMC and PID by pole placement controllers
are similarly affected in the time response and the control action To establish a quantitative performance comparison between
for tracking tasks. For the error signal, the IMC controller is the IMC and PID by pole placement controllers, The RMSE
less affected by the random noise than the PID controller by value (Root Mean Square Error) for the time response and the
pole placement. RMS value (Root Mean Square) for the control action are
calculated. These indices measure the tracking tasks
3000 8 performance and the mean control action. The RMSE value is
2000
PID controller
IMC controller
7 PID controller
IMC controller defined by (22).
6
Speed (RPM)

1000 5
Output (v)

4
0
1
3
-1000 2
1
= ( ( )) (22)
-2000
0
-3000 -1
0 1000 2000 3000 4000 5000 0 1000 2000 3000 4000 5000
Time (s) Time (s)
a) b) where n is the total number of samples and ( ) is the error
3000
PID controller signal. The RMS value for the control action is defined by (23).
2000 IMC controller
Speed error (RPM)

1000

1
(23)
0
= ()
-1000

-2000
where u( ) is the control action of the IMC and PID by pole
-3000
0 1000 2000 3000
Time (s)
4000 5000
placement controllers.
c) Table II shows the performance indices for the four tests in
Fig. 8. Tracking response for the motor-generator system against uniform tracking tasks. As can be observed, the RMSE value is less
random noise in the feedback loop a) time response b) control action and c)error when using the IMC controller with respect to the PID by pole
signal. placement controller for all the tests. Also, as the improvement
percentage shows that the IMC controller has a better
For the fourth test, Fig.9 shows the time response, the control
performance especially against the presence of random noise in
action, and the error signal of the motor-generator system for
the feedback loop. For the control action, there is no exist a
significate difference between the IMC and the PID by pole control action, there is not a significate difference between the
placement controllers. IMC and PID by pole placement controllers.

TABLE II. PERFORMANCE INDICES FOR POSITION TASK OF THE REFERENCES


MOTOR-GENERATOR SYSTEM
[1] K. J. Astrm and T. Hgglund, Control PID avanzado. Madrid,
Performance Improvement Espaa: Pearson Prentice Hall, 2009.
Test Controller Indices percentage (%) [2] N. K. Sinha and G. P. Rao, Identification of Continuous-Time
RMSE RMS RMSE RMS Systems: Methodology and Computer Implementation. Springer
Nominal IMC 0,85 3
17,65 1 Netherlands, 1991.
operation PID 1 3,03
External IMC 0,95 3,02 [3] L. Ljung, System Identification: Theory for the User. Prentice Hall
13,68 0,33
disturbance PID 1,08 3,03 PTR, 1999.
Random IMC 0,86 3,01 [4] S. A. Billings, Nonlinear System Identification: NARMAX Methods
26,74 1,32
noise PID 1,09 3,05 in the Time, Frequency, and Spatio-Temporal Domains. Wiley, 2013.
Resistive IMC 1,48 3,09
11,48 0,32 [5] N. Pachauri, A. Rani, and V. Singh, Bioreactor temperature control
load PID 1,65 3,1
using modified fractional order IMC-PID for ethanol production,
VI. CONCLUSIONS Chem. Eng. Res. Des., vol. 122, pp. 97112, 2017.
This paper presents the design and practical implementation [6] M. Yazdanian and A. Mehrizi-Sani, Internal Model-Based Current
of a speed control for a motor-generator system. The motor- Control of the RL Filter-Based Voltage-Sourced Converter, IEEE
generator system was identified employing the Matlab Trans. Energy Convers., vol. 29, no. 4, pp. 873881, 2014.
identification toolbox for a steeped signal, obtaining a second [7] A. Ramdani and S. Grouni, Dynamic matrix control and generalized
order model with real poles for the system. An IMC controller predictive control, comparison study with IMC-PID, Int. J.
for the speed control is designed. To contrast the IMC controller Hydrogen Energy, pp. 110, 2017.
performance, a PID by pole placement controller was designed. [8] W. K. Ho, T. H. Lee, H. P. Han, and Y. Hong, Self-tuning IMC-PID
The performance of IMC and PID by pole placement control with interval gain and phase margins assignment, IEEE
controllers was analyzed against the presence of external Trans. Control Syst. Technol., vol. 9, no. 3, pp. 535541, May 2001.
disturbances. Obtained results showed that employing the IMC [9] D. E. Rivera Flores, Una metodologa para la identificacin
controller the motor-generator system has a better performance integrada con el diseo de controladores imc-pid, Rev. Iberoam.
for tracking tasks regarding the PID by pole placement Automtica e Informtica Ind. RIAI, vol. 4, no. 4, pp. 518, 2007.
controller. This can be noticed in a lower RMSE value for the
IMC controller for all the performed tests. Regarding to the

Вам также может понравиться