Beruflich Dokumente
Kultur Dokumente
Abstract
Positive position feedback control is the most common resonant control technique that has been studied for last three
decades. As a low-pass filter, positive position feedback is very sensitive to low-frequency disturbances. To overcome this
shortcoming of positive position feedback controller, negative derivative feedback controller, which acts as a bandpass
filter and can effectively control the lower and higher frequency disturbances, has been developed recently. So far, there is
no comparison work between positive position feedback and negative derivative feedback on flexible manipulator system.
Consequently, to fill this gap, in this article, both positive position feedback and negative derivative feedback controllers
are applied experimentally and analysed in terms of settling time and vibration attenuation at different damping ratios on a
single link flexible manipulator featuring piezoelectric actuator. Moreover, robustness with respect to natural frequency
variation is studied for the first time on flexible manipulator system. Based on experimental study conducted on the
particular system developed in this article, it has been observed that negative derivative feedback controller is more
effective than positive position feedback controller based on evaluated performance measures.
Keywords
Active vibration control, flexible manipulator, piezoelectric, modal control technique, PPF controller, NDF controller
Introduction to separate the two dynamics, that is, rigid joint motion and
elastic link motion and then designing different control
Flexible manipulator offers a large number of advantages
laws for both dynamics. H1 control has also been applied,
over rigid manipulators. These include faster response time,
which describes the control problem as a mathematical
lower power requirement, lower cost, high payload-to-weight
optimization problem and then finds appropriate controller
ratio and so on.1 These advantages have urged the use of
flexible manipulators in industry and space application. How-
ever, despite all these advantages, they are susceptible to Mechanical Engineering Department, DHA Suffa University, Karachi,
severe tip vibrations. These vibrations should be eliminated Sindh, Pakistan
or at least reduced for accurate and fast execution of task.
Several vibration control techniques have been applied to Corresponding author:
Hassaan Hussain Syed, Mechanical Engineering Department, DHA Suffa
flexible manipulator system during the last two decades.1–3
University, DG-78, Off Khayaban-e-Tufail, Ph-VII (Ext.), DHA, Karachi,
Siciliano and Book applied single perturbation approach Sindh 75500, Pakistan.
to control flexible manipulator system.4 The main idea was Email: hassaanhussain@gmail.com
Creative Commons CC BY: This article is distributed under the terms of the Creative Commons Attribution 4.0 License
(http://www.creativecommons.org/licenses/by/4.0/) which permits any use, reproduction and distribution of the work without
further permission provided the original work is attributed as specified on the SAGE and Open Access pages (https://us.sagepub.com/en-us/nam/
open-access-at-sage).
2 International Journal of Advanced Robotic Systems
for its solution.5 Intelligent controllers such as neural net- via implementation on a flexible manipulator featuring
work6 and fuzzy logic7 have also been considered for vibra- smart piezoelectric actuator, where vibration is induced due
tion control of flexible manipulator system. The other to slewing manoeuvre of the flexible arm. Furthermore, in
control techniques used to control flexible manipulator sys- order to enhance the practicality and feasibility of these
tem include input shaping technique,8 output redefinition,9 techniques in real-world environment, robustness (with
stable inversion in time domain,10 optimizing independent respect to natural frequency variation) of the controllers
motion of flexible arms11 and so on. has been experimentally performed, studied and compared.
In many flexible structures, only a few vibration modes Moreover, in the previous research regarding NDF con-
are significant, so vibration can be controlled by controlling trol,17 compensator damping ratio was kept equal to the
only certain number of modes. In these situations, resonant system damping ratio, while in this research, compensator
control techniques are appropriate solution to vibration damping ratio is considered as a controlling parameter
problem. In these techniques, a first- or second-order which ensures more flexibility of design under different
dynamic system (compensator) is used to apply control operating conditions. In this process, the effect of compen-
force on the system. Some state-of-the-art resonant control sator’s damping ratio on controller’s performance has been
techniques include active modal tuned mass damper evaluated. Moreover, both controllers are compared in
(AMTMD), positive position feedback (PPF) control, inte- terms of settling time and vibration attenuation. Hence, a
gral resonant control (IRC) and negative derivative feed- comprehensive performance evaluation of controllers is
back (NDF) control. IRC is one of the novel techniques provided.
developed recently,12 and it takes advantage of pole-zero As far as the application fields of these controllers are
interlacing property of collocated sensor–actuator pair to concerned, they mostly have been applied to suppress
damp vibration modes. This property ensures that the phase vibration of flexible beam. Author believes that these con-
of a collocated transfer function lies between 0 and 180 . trol techniques can be further applied to suppress vibra-
However, to satisfy this condition, only collocated sensor– tion in aircraft wings and other flexible structures. Large
actuator pair should be used, this design constraint is not flexible space structures and space inflatable structures
practically feasible in certain applications. Recently, are also one of the key applications of such techniques,
AMTMD has been introduced,13 which uses passive TMD where PPF and NDF controllers can be used to control
formulation to define a resonant controller and then structural vibration. Since these controllers work on reso-
finds the control force from its equation. One of the nant control, therefore, they can be applied to control
oldest resonant control methods is PPF control and it was sound waves as well.
introduced to control vibration of large flexible space This article is organized as follow. In ‘Resonant control
structures.14,15 In PPF, control structural position coordi- technique’ section, PPF and NDF control schemes are dis-
nate is sent as an input to the compensator, and the product cussed in detail providing physical insight especially in the
of the compensator’s output and gain is fed back to the case of NDF controller. Experimental set-up and results are
structure in a positive sense to act as a control force. Shan discussed in ‘Experimental set-up’ section along with fur-
et al. applied PPF algorithm to control the vibration of a ther investigation of the controllers. Finally, conclusion is
single link flexible manipulator.16 Lately, NDF control included in the last section.
has been introduced.17 It works on feeding the velocity
of structure to compensator and then feeding back the
compensator velocity multiplied with some gain to the Resonant control technique
structure in a negative sense.
The working principle of resonant control technique is very
There are numerous advantages associated with PPF
simple, and it produces a damping effect on particular
controller, such as it is insensitive to spillover, its magni-
vibration modes by applying a control force that is opposite
tude rolls off quickly at high frequencies, it is relatively
in phase to the velocity of structure at the resonance fre-
easier to apply in real-world situations and its stability can
quencies under consideration. Figure 1 shows the block
be ensured by adjusting only the stiffness properties of the
diagram of resonant controller. Here, ‘q’ is the response
structure.16 Despite these advantages, PPF is very sensitive
of the structure, that is, sensor output and ‘’ is the output
to low-frequency disturbances as it works as a low-pass
of the compensator.
filter and hence it causes a worsening of system behaviour
below the natural frequency. On the other hand, NDF
works as a bandpass filter and eliminates both the higher PPF control
and lower uncontrolled mode effect; therefore, it avoids PPF is one of the widely used controllers in active vibration
low-frequency worsening and it is more robust to spillover control area. To explain its working principle, consider a
effects. So far, there is no direct comparison work between single degree of freedom system which is basically a single
PPF and NDF on a single link flexible manipulator system. decoupled mode as shown in the following equation
Consequently, the main contribution of this work is to
compare vibration control performances of PPF and NDF _ þ on2 qðtÞ ¼ gon2
q€ðtÞ þ 2on qðtÞ (1)
Syed 3
Figure 3. Bode plot of PPF controller. (a) c ¼ 0.2. (b) c ¼ 0.6. PPF: positive position feedback.
NDF control
NDF includes a negative feedback of the velocity of the
Figure 4. Control scheme of NDF controller. NDF: negative
compensator output as a control force rather than the posi- derivative feedback.
tion as used in PPF control. Let us consider a single degree
of freedom system which is basically a single decoupled
mode In order to find the stability of NDF, controller combin-
ing equations (5) and (6), placing them in their second-
q€ðtÞ þ 2on qðtÞ
_ þ on2 qðtÞ ¼ go
_ n (5)
order form and assuming no external forces give
" #
where and on are the damping ratio and natural frequency q€ 2on gon q_ on2 0 q 0
of the structure, respectively. g is the plant gain, _ is the þ þ ¼
€ koc 2 c oc _ 0 oc2 0
derivative of the compensator output, that is, velocity of the
compensator and q is the position of the structure. The NDF (7)
controller is implemented using a second-order dynamic
Again, using the same formulation as used in the previ-
system defined as
ous section, we can conclude that the NDF control stability
€ _ þ oc2 ðtÞ ¼ k qo
ðtÞ þ 2 c oc ðtÞ _ c (6) depends on the elements in the damping matrix only.
Again, using the same formulation as used in the previous
where c and oc are the damping ratio and frequency of the section to determine the stability condition of PPF control,
compensator, respectively, and k is the controller gain. Fig- the NDF control stability can be found as
ure 4 shows the control scheme of NDF method. In single
degree of freedom form, NDF introduces a second-order ð2on Þð2 c oc Þ þ ðkoc Þðgon Þ > 0
compensator which is forced by velocity response of the oc on ð4 c þ kgÞ > 0
structure. The velocity response of the compensator multi- kg > 4 c
plied with some gain and natural frequency is then fed back
in a negative sense to give the force input to the structure. So, the NDF controller will remain stable if the above
The transfer function of the compensator rolls off quickly equation is satisfied. The transfer function of NDF is
both at high and low frequencies, hence, making it suitable ðsÞ koc s
for controlling both the lower and higher frequency modes ¼ 2 (8)
qðsÞ s þ 2 c oc s þ oc 2
of the structure with well-separated modes.
Syed 5
Figure 5. Bode plot of NDF controller. (a) c ¼ 0.2. (b) c ¼ 0.6. NDF: negative derivative feedback.
Table 1. Properties of piezoelectric sensor and actuator. Table 2. Comparison of natural frequencies.
Figure 8. Response at manipulator’s tip. (a) Uncontrolled time response. (b) Uncontrolled frequency response.
Table 3. Tuned gain values at different c. value the better is the robustness of the controllers and vice
versa. In order to experimentally validate this concept, two
Gain ‘g’ experiments were conducted using 0.8 and 0.2 as ‘ c’ value.
c PPF controller NDF controller These experiments were performed to study the effect of
compensator damping ratio on compensator’s robustness
0.2 0.00030 0.000045 when the difference between the natural frequency of the
0.5 0.00035 0.00009
system and the controller centre frequency is increased.
0.8 0.0005 0.0001
The discrepancies in natural frequency estimation were
PPF: positive position feedback; NDF: negative derivative feedback. visualized by changing the compensator centre frequency
from 0.75*w to 4*w in regular steps and then obtaining the
time response of the manipulator at each centre frequency;
of c value on controller’s performance will be further
here, ‘w’ represents the original natural frequency of the
discussed in robustness study. Finally, settling time results
system. A total of six experiment trials were performed at
show that NDF controller has performed better at all c
each compensator centre frequency, and these trails were
values as compared to PPF controller.
further used to obtain the vibration attenuation in terms of
In many automated industries, robotic arms are working
auto-power spectrum of manipulator. Figure 10(a) and (b)
continuously and repeatedly; in this regard, the controller
shows the vibration attenuation using different controller
should be able to provide constant performance during
centre frequency at c ¼ 0.2 and c ¼ 0.8, respectively. In
repetitive manoeuvrings. To check this, we conducted
general, it can be observed from the graphs that when the
seven experimental trials at each ‘ c’ value under same
difference between the original natural frequency of the
conditions. We further calculated the average settling time
system and the controller centre frequency is increased, the
and standard deviation of all the experimental trials which
magnitude of vibration attenuation degrades significantly.
have been shown in Table 4. One can easily conclude from
This degradation of performance is quite severe at c ¼ 0.2,
Table 3 that settling time readings at each ‘ c’ value are
where there is large drop in vibration attenuation as we
almost constant and there is not much deviation in the
move away from ‘w’. While at c ¼ 0.8, the drop in vibra-
results. Hence, it is shown that the controllers are able to
tion attenuation is not that extreme and our controller is still
give constant performance during different experimental
able to provide 35 dB of vibration attenuation even if there
trials and overall NDF is a better choice than PPF.
is þ300% (if controller centre frequency ¼ 4w) error in
natural frequency estimation.
As explained in ‘Resonant control technique’ section
Robustness study of each control law regarding robustness of the controllers, the same can be
In this section, we will evaluate the performance of the deduced from the experimental analysis presented in this
controllers when there is a difference between the natural section, that is, the robustness of the controllers can be
frequency of the system and the controller centre fre- increased by increasing the c value. Furthermore,
quency. This study is useful in situations where the natural through these experiments, the effect of c values on
frequency of the system is difficult to obtain or have large controller’s performance has been highlighted; hence, it
errors in it. In ‘PPF control’ and ‘NDF control’ sections, it is shown that the optimal value of c depends on whether
was concluded from Bode plot analysis that robustness of the natural frequency of the system is known accurately
controllers depends on the ‘ c’ value, the higher the ‘ c’ or not and how much it is expected to change. If the
8 International Journal of Advanced Robotic Systems
Figure 9. Time response of controllers. (a) c ¼ 0.2. (b) c ¼ 0.5. (c) c ¼ 0.8.
PPF controller (settling time ‘s’) Average settling time Standard deviation
Trials 1 2 3 4 5 6 7
c ¼ 0.2 5.522 5.530 5.518 5.548 5.534 5.530 5.519 5.529 0.010
c ¼ 0.5 6.345 6.359 6.349 6.340 6.352 6.359 6.349 6.350 0.007
c ¼ 0.8 7.369 7.367 7.360 7.357 7.359 7.360 7.375 7.364 0.007
NDF controller (settling time ‘s’)
Trials 1 2 3 4 5 6 7
c ¼ 0.2 3.572 3.578 3.572 3.570 3.565 3.578 3.585 3.574 0.007
c ¼ 0.5 4.325 4.327 4.316 4.328 4.329 4.325 4.329 4.326 0.005
c ¼ 0.8 6.589 6.595 6.581 6.585 6.581 6.588 6.594 6.588 0.006
PPF: positive position feedback; NDF: negative derivative feedback.
system natural frequency is known accurately, lower c experimentally on a single link flexible manipulator featur-
value is a better choice. ing piezoelectric actuator and were further analysed in terms
of settling time and robustness at different controllers damp-
ing ratios. After thorough study of time domain results and
Conclusion robustness, it has been concluded that NDF controller is
In this article, vibration suppression performances of two overall more effective in suppressing vibration than PPF
resonant control techniques namely PPF and NDF have been controller based on particular system developed in this study.
compared. For this purpose, these techniques were applied Author believes that these control techniques can be further
Syed 9
Figure 10. Vibration attenuation as compared to the uncontrolled case at different compensator frequencies. (a) c ¼ 0.2. (b) c ¼ 0.8.
applied to control structural vibration in flexible space 8. Chang PH and Park HS. Time-varying input shaping tech-
robots, multi degree of freedom manipulators, lightweight nique applied to vibration reduction of an industrial robot.
wind turbine blades and other structures where only few Control Eng Pract 2005; 13(1): 121–130.
modes are contributing to the vibration of the system. The 9. Yang H, Krishnan H and Ang MH. Tip-trajectory tracking
future work will be focused on applying these control control of single-link flexible robots via output redefinition.
techniques to multi degree of freedom system. In: Robotics and Automation, 1999. Proceedings. 1999 IEEE
international conference on 1999, Detroit, MI, USA, Vol. 2, pp.
Declaration of conflicting interests 1102–1107. IEEE.
The author(s) declared no potential conflicts of interest with respect 10. Kwon DS and Book WJ. A time-domain inverse dynamic
to the research, authorship, and/or publication of this article. tracking control of a single-link flexible manipulator. J Dyn
Syst Meas Control 1994; 116(2): 193–200.
Funding 11. Yushu B, Zhihui G and Chao Y. Vibration reduction of open-
The author(s) received no financial support for the research, chain flexible manipulators by optimizing independent
authorship, and/or publication of this article. motions of branch links. Chin J Aeronaut 2008; 21(1): 79–85.
12. Aphale SS, Fleming AJ and Moheimani SR. Integral resonant
References control of collocated smart structures. Smart Mater Struct
1. Tokhi MO and Azad AK. Flexible robot manipulators: mod- 2007; 16(2): 439.
elling, simulation and control. London, UK: Iet, 2008. 13. Cazzulani G, Resta F and Ripamonti F. Active modal tuned
2. Benosman M and Le Vey G. Control of flexible manipula- mass damper for smart structures. Eng Lett 2011; 19(4): 297.
tors: a survey. Robotica 2004; 22(05): 533–545. 14. Goh CJ and Caughey TK. On the stability problem caused by
3. Dwivedy SK and Eberhard P. Dynamic analysis of flexible finite actuator dynamics in the collocated control of large
manipulators, a literature review. Mech Mach Theory 2006; space structures. Int J Control 1985; 41(3): 787–802.
41(7): 749–777. 15. Fanson JL and Caughey TK. Positive position feedback con-
4. Siciliano B and Book WJ. A singular perturbation approach trol for large space structures. AIAA J 1990; 28(4): 717–724.
to control of lightweight flexible manipulators. Int J Robotics 16. Shan J, Liu HT and Sun D. Slewing and vibration control of a
Res 1988; 7(4): 79–90. single-link flexible manipulator by positive position feedback
5. Yim JG, Yeon JS, Lee J, et al. Robust control of flexible robot (PPF). Mechatronics 2005; 15(4): 487–503.
manipulators. In: 2006 SICE-ICASE international joint confer- 17. Cazzulani G, Resta F, Ripamonti F, et al. Negative derivative
ence 2006 Oct 18, Bexco, Busan, Korea, pp. 3963–3968. IEEE. feedback for vibration control of flexible structures. Smart
6. Isogai M, Arai F and Fukuda T. Modeling and vibration control Mater Struct 2012; 21(7): 075024.
with neural network for flexible multi-link structures. In: 18. Tarazaga PA, Inman DJ and Wilkie WK. Control of a space
Robotics and Automation, 1999. Proceedings. 1999 IEEE inter- rigidizable-inflatable boom using embedded piezoelectric com-
national conference on 1999, Detroit, Michigan, Vol. 2, pp. posite actuators. In: AIAA Paper. 2006 May, Newport, RI, United
1096–1101. IEEE. States, (2006–1976).
7. Lin J and Lewis FL. Two-time scale fuzzy logic controller of 19. Caughey TK. Dynamic response of structures constructed
flexible link robot arm. Fuzzy Sets Systems 2003; 139(1): from smart materials. Smart Mater Struct 1995; 4(1A):
125–149. A101.