Professional Documents
Culture Documents
net/publication/3919676
CITATIONS READS
52 6,321
2 authors, including:
Samer Saab
Lebanese American University
76 PUBLICATIONS 2,037 CITATIONS
SEE PROFILE
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Samer Saab on 03 June 2014.
Samer S. Saab, Senior Member, IEEE, and Raed Abi Kaed-Bey, Member, IEEE
02001 IEEE.
0-7803-7057-O/Ol/$lO.OO 98 1
mouse click. In order to collect the input voltage and 1
the dynamometer outputs simultaneously at a desired
sampling rate, particular software is coupled with the
dynamometer software to collect different
measurements automatically at a desired fixed time
interval.
The parameter identification procedure also requires
This system is composed of two linear algebraic
measurements 'of angular acceleration and the
equations with ~inknownsL , R, K T,J , and,f: The
derivative of the armature current. A Kalnian filter
system coefficients are obtained from the state
[3]-[4] is designed as a state observer to estimate
these variables. A similar Kalman filter is applied to dw di
variable measurements 8, U, -, i, -, v,
estimate angular rate and acceleration using wheel dt dt
pulse generator of a freight train [ 5 ] . If the model of the motor is precisely given by
Finally, to improve the performance of the overall ( l ) , along with nominal values of i(0, v(0, Q(t),
system, the DC parameters were first estimated, di dw
- - , U ( / ) , and -, thcn equation ( 2 ) must hold
employing least-squares algorithm, by decoupling the dr dt
AC parameters using a DC input signal. These for all time /. Let o ( n )denote the speed at time
estimated DC parameters are then used to identify the
AC parameters. nT, i(n) denote the current at time nT, etc., define
This novel experimental approach is the main 4,
contribution of this paper. -(U) i(n) w(n) 0 R',
This paper is arranged as follows. Section 2 includes dlt'
the dynamics of a DC motor with notation and the 0 0 -i(n) -(/I) it.(/?)
clt
corresponding least-squares algorithm. Section 3 -f -
presents the proposed system showing armature
- - -
voltage generation committing to this problem, data
acquisition, and implementation of a Kalnian filter.
Section 4 includes experimental results and the
and J'( n ) = La'"'] E 912.Equation (2) above may be
performance of the proposed system. Finally, Section written in the form M ( n ) K = y ( n ) (3)
5 concludes the findings of this work. A4 is referred to as the regressor matrix. Multiply both
2 LEAST-SQUARES ALGORITHM sides of (3) by M ( n ) . we have
M T ( n ) M ( n ) K= M ( n ) y ( / ?.)Consequently,
The system dynamics of a DC motor is given by
L dgt - R i ( t ) - K -b w ( t )+ v ( t )
=
J dwht
Equation (4) can be rewritten as follows
K , i ( t ) bw ( t )
= - (1)
R,w K = R,bl!. (5)
v
where R,,, E M ( n ) M ( n )E 3'"'and
In order to design a controller based on these lI=I
measurements of v(i), i(t) and e ( d , then K = R y; R,\/, The key to this method is ensuring that
,
which are then used to estimate the
parameters. This can be accomplished R ,, is invertible. This is completely dependent on
by considering the first two equations choosing the correct input to the motor, that is any arbitrary
above which can be written as follows input generally does not guarantee the nonsingularity of
4 1 '
982
DC decoupling
In the “DC decoupling” methodology R, and k are first Synchronous
estimated using DC input. The corresponding state Generator
*
I
983
dynamometer software to collect different automatically at a desired fixed time interval; 3-
measurements automatically at a desired fixed time implementing a steady-state Kalman filter, as a state
interval. observer, to estimate the angular acceleration and the
Table 1. Experimental results before using Kalman derivative of the armature current. Experimental
results illustrated the efficiency of the proposed
Paramet. La R, K J system.
0.5
AC
Input to
6.3e-3 32.78 0.251 2e-4
Pi
DC mtr
DC -0.53 27.73 1.592 -0.014
Decoupl
10
[Henr [Ohm
Input to
the DC
motor
DC
Decoupl
0.658
1.198
32.91
27.73
rt- 0.556
1.592
1.k-3
0.004
le-4
2e-4
0.145
0.730
814.2
8.438
0 1 2 3
S
4
ing
I I I I I I I I
1
0 1 2 3 4 5 6 7
Figures 3 and 4 show the performance of Kalman
filter as an observer and smoother operator. The top
of Figures 3 and 4 show the measurement of the
armature current and the motor angular rate and the
superimposed smooth plots are the Kalman filter -20
outputs. Similarly, bottom of Figures 3 and 4 show 0 1 2 3
5
4 5 6 7
the derivatives of the armature current and the angular Figure 4. Angular speed, w(top), and d d d t (bottom)
rate. The “non-differentiable’’signals are obtained by
explicit differentiation, where the smooth plots are the 6 REFERENCES
output of the Kalman filter. Tables 1 and 2 include a
summary of all the considered methodologies. By S. Sastry, and M. Bodson, Adaptive Control
examining the condition of the matrix RM,it is Stability, Convergence, and Robustness,
obvious that the DC decoupling method with a steady- Prentice-Hall Adv. Reference Series, 1989.
state Kalman filter implementation resulted in the best K. J. Astrom, and B. Wittenmark, Adaptive
results. In addition, the parameters estimated are Control, Addison- Wesley Series in Electrical
consistent with the physical parameters of the Engineering: Control Engineering, 1989.
employed DC motor. R. E. Kalman, and R. S. Bucy, “New results in
linear filtering and prediction theory,” ASME, J.
5 CONCLUSION Basic Eng., vol. 83D, pp. 95-108, 1961.
R. E. Kalman, “A new approach to linear
This paper showed that parameter identification of a filtering and prediction problems,” ASME, J.
DC motor, without the use of a D/A converter and a Basic Eng., vol. 82D, pp. 34-45, 1960.
power amplifier, is made possible. This was basically S. Saab, “A Map Matching Approach for Train
achieved by: 1- employing a synchronous generator Positioning. Part 11: Application and
coupled to a prime mover in order to generate the Experimentation,” IEEE Transactions on
“required” DC motor input voltage; 2- integrating a Vehicular Tech.n, vol. 49, No. 2, March, 2000.
special software to the existing data acquisition
system resulting in collecting different measurements
984