Professional Documents
Culture Documents
Dynamic Tracking Performance of Servo Mechanisms B
Dynamic Tracking Performance of Servo Mechanisms B
Abstract
This article provides systematic analysis and controller design methods for dynamic tracking performance of servo
mechanisms associated with practical systems. Discrete general composite nonlinear feedback, as a fundamental control-
ler, will be proposed to yield a good transient performance. Particularly, in the servo systems, there also exist unmo-
deled disturbances which may lead to tracking errors. A novel repetitive control scheme based on disturbance observer
configuration is incorporated into the controller to counteract this unexpected effect. Furthermore, to deal with any
periodic signal of variable frequency, a fractional-order repetitive control scheme based on disturbance observer strategy
is proposed. The stability of the overall closed-loop system is guaranteed via frequency domain analysis. Three control-
lers, that is, the proportional–derivative controller with zero-phase error tracking controller scheme and the conven-
tional disturbance observer, the integral backstepping controller, and the compound discrete general composite
nonlinear feedback controller with fractional-order repetitive control scheme based on disturbance observer are com-
pared. To demonstrate the dynamic tracking performance of the proposed control strategy, comparative experiments
are conducted.
Keywords
Servomechanisms, dynamic tracking performance, repetitive control, composite nonlinear feedback control, fractional
order
Creative Commons CC-BY: This article is distributed under the terms of the Creative Commons Attribution 3.0 License
(http://www.creativecommons.org/licenses/by/3.0/) 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 Advances in Mechanical Engineering
into the unit circle at low frequencies, and at the same However, providing that these disturbances, espe-
time, eliminates phase error induced by non-cancellable cially friction, have periodic and repetitive characteris-
zeros at all frequencies.6 To ensure the anticipated per- tics, then the compensation schemes become much
formance of ZPETC, it is indispensable to acquire the easier to design. Actually, all state variables as well as
precise model of the plant system. However, servo disturbances in direct current (DC) servo mechanisms
mechanisms may have a large extent of nonlinear have the same repetitive period as that of repetitive
uncertainties such as nonlinear friction and the large external command, and then, the stated assumption
change of payload. Therefore, the tracking perfor- becomes reasonable. As a consequence, repetitive con-
mance of ZPETC may be deteriorated by parameter trol (RC) is very effective to attenuate these periodic
perturbations and external disturbances. disturbances. Distinguished by its high performance as
Improving transient performance plays a crucial role well as the simple design and implementation criteria,
in tracking dynamic external signals. Compared to con- RC improves the periodic tracking performance12,13 by
ventional approaches, an intelligent nonlinear control implementing zN (N is the period of the disturbance/
technique, the so-called composite nonlinear feedback reference) into the feedback loop or the feedforward
(CNF) control, is capable of performing track following loop. Then, the errors in the previous repetition can be
with much better result. The CNF control, which is reused to improve the current control. However,
rooted in the article of Lin et al.,7 consists of a linear because of its strict stability condition and poor
feedback law and a nonlinear feedback law. The linear response to non-periodic disturbances, conventional
feedback part is designed to yield a closed-loop system RC fails to be applied in practice. Recently, a new
with a small damping ratio for a quick response. The structural configuration of disturbance of observer in
nonlinear feedback law, on the other hand, is supposed RC (RDOB), wherein designers have more flexibility in
to increase the damping ratio of the closed-loop system the repetitive loop-shaping design, is developed in Chen
as the system output approaches the target reference. and Tomizuka14 and Li et al.15 The proposed RDOB is
This transformation of system can reduce the overshoot an approach to extract the repetitive errors in feedback
caused by the linear part. Chen et al.8 developed the control, leading to a great decline in gain amplifications
CNF control technique for more general linear systems at the non-repetitive frequencies. An additional benefit
with input saturation. In the most previous forms of the of the scheme is that just like the design procedure of
CNF control technique, the target reference has always conventional DOB, the fundamental feedback control-
been assumed to be a step signal. In particular, Cheng ler and RDOB can be designed separately without any
et al.9 adopted a reference generator which produces interference.
more general signals to supplement the CNF control Unfortunately, the period of the disturbance or ref-
technique. Thus, it can yield a good performance for erence (N = fs =fr ), where fr is the fundamental fre-
tracking general non-step references. Moreover, in the quency of the reference signal and fs is the sampling
current digital control systems, a controller designed in rate, may be fractional in the case of a fixed sampling
continuous-time domain eventually has to be discre- rate fs . As it is known, only Z N with an integer N can
tized for implementation. It seems more efficient to be implemented in practice. Failing to compensate the
design controllers directly in the discrete-time domain exact fractional period signals, the conventional struc-
whenever possible. However, the existing studies hardly ture of RDOB14,15 is insufficient to track arbitrary fre-
focus on tracking more general reference signals via quency external signal or eliminate any periodic
CNF control strategy for a class of discrete-time disturbance.16,17 In other words, with the fixed sam-
systems. pling rate, the former RDOB would lead to significant
Friction, as the main source of torque disturbance, performance degradation in practice.
occurs unavoidably in direct-driving servo mechanical In this article, by borrowing ideas from general CNF
systems. Thus, there is a wide recognition that in order and RC based on DOB structure and integrating them
to achieve better tracking performance, it is essential to with some assumptions, a new compound controller is
design compensatory components which can inhibit the proposed. It can adapt to arbitrary frequency signals
effect of frictional disturbance. The conventional dis- for a high-accuracy flight simulator system driven by
turbance observer (DOB) becomes one of most effective DC motors. To verify the high performance of the pro-
ways to restrain the internal and external distur- posed controller, extensive comparative experimental
bance,10,11 which will enhance the rigidity of servo results have been obtained from the motion control of
mechanism eventually. However, due to the low-pass a DC motor. The distinctive characteristics of this work
Q-Filter, the compensations of DOB usually result in are as follows:
magnitude distortions and phase lags when confronted
with periodic inputs. Therefore, with regard to the 1. Unlike the general CNF designed in continuous-
dynamic tracking, the conventional DOB fails to com- time domain,9 the general CNF based on
pensate for specific disturbances. discrete-time domain which is more practical is
Wen et al. 3
Problem formulation
A flight simulator, as being a kind of servo mechanism
with high accuracy in position tracking, has been
widely used in modern engineering system. As shown
in Figure 1, a complex flight motion simulator is cur-
rently composed of an orthogonal outer axis frame, a Figure 2. Architecture of the inner-axis frame.
middle-axis frame which is horizontal to the outer axis,
and an inner-axis frame supported by the middle-axis J €u = Bp u_ + u + f (u, u,
_ t) ð1Þ
frame. The inner axis is driven by a DC motor that is
fixed onto the center of the middle frame. In practice, where u and J represent the angular displacement and
in order to achieve dynamic tracking performance, the inertia load, respectively. Bp represents the viscous
both the outer- and middle-axis frames have to be friction coefficient and f represents other unmodeled
designed largely enough to be consistent with more disturbances, such as nonlinear friction and external dis-
powerful motors. Whereas, the inner axis frames can turbances. u is the control input voltage. Equation (1)
not be manufactured large enough for the reason that can be rewritten into a state-space form as follows
the middle-axis frames fail to endure overweight load.
In consequence, the inner-axis frame, with the less x_ 1 = x2
powerful motor but the greater force of friction, has Bp 1 ð2Þ
trouble behaving well in frequency tracking. x_ 2 = x2 + u + d(x, t)
J J
Considerations should be given priority to the inner-
axis frame. where x = (x1 , x2 )T represents the state vector of the
The architecture of the inner-axis frame is described position and velocity and d represents the lumped
by Figure 2. It is a current-controlled permanent-mag- disturbance.
net DC motor with an analog amplified driver. At the same time, being ignored of unstructured
The equation of dynamic plant model in the inner- uncertainties, the nominal model of the plant can be
axis frame is described as19 expressed as follows
4 Advances in Mechanical Engineering
cult.21 To eliminate the effects of disturbance and where xe 2 R2 , r 2 R, and ue 2 R are the state, control
achieve highly precise motion control, significant efforts output, and control input, respectively. r is the refer-
have been devoted to compensation strategy. ence produced to be tracked. A, B, and C are the same
Given the approximate second-order plant, our aim as equation (4).
is to synthesize a control input u to make the state vec- Next, we design a linear control law for the auxiliary
tor x track the dynamic signal as precisely as possible system as follows
in spite of parameter uncertainties or unstructured non-
linear uncertainties. ue (k) = Fe xe (k) ð7Þ
Compound control scheme where Fe is the feedback gain matrix that can transform
the shape of equation (6).
Fundamental feedback controller design The auxiliary system (6) and the linear control law
In order to achieve the fundamental servo performance (7) are combined to form the final reference generator
and robustness, a feedback controller has to be as follows
designed. Instead of using conventional proportional– (
integral–derivative (PID) or Sliding Mode Control X xe (k + 1) = (A + BFe )xe (k) , xe (0) = xe0
: ue (k) = Fe xe (k) ð8Þ
scheme, a unique variable structure control strategy
ref r(k) = Cxe (k)
DGCNF is adopted to yield a good performance for
the second-order servo model mentioned above. By The reference generator (8) can generate an arbitrary
altering the damping ratio of the overall closed-loop type of sine signal r(k) = a sin (2pfr k) by setting the ini-
system corresponding to the error, the proposed con- tial value xe0 = ½0 2pf T and choosing appropriate Fe .
troller can perform well in quick response and succeed The proof of this is involved with matrix function calcu-
to reduce overshoot as well. It is a necessary character- lation method, and the similar design procedure can be
istic for tracking dynamic signal. The proposed found.9
DGCNF is consisted of an auxiliary reference genera- Then, let ^x(k) = x(k) xe (k), equation (9) can be
tor and a discrete CNF control scheme. obtained as
Specifically, a typical discrete second-order servo
model is characterized by ^x(k + 1) = A^x(k) + B(s½u(k) ue (k)) ð9Þ
X x(k + 1) = Ax(k) + Bs½u(k) This equation will be used in the design of the
: ð4Þ DGCNF control law. The following design procedure
y(k + 1) = Cx(k)
P
will be carried out in three steps similar to Chen et al.8
where x 2 R2 , y 2 R, and u 2 R are the state, control
output, and control input, respectively. A, B, and C are Step 1: Design a linear feedback law
appropriate dimensional constant matrices that can be
obtained by discretization of equation (2), and s repre-
sents the actuator saturation defined as UL (k) = F^x(k) + ue (k) ð10Þ
In this step, F is designed by two main principles.
s½u = sgn(u) minfumax , jujg ð5Þ
First, the stability of the overall system must be guaran-
with umax being the saturation level of the input (umax teed, and then, the closed-loop poles should be placed
usually equals 10 V in practical servo controller). In the to have a dominating pair with a small damping ratio.
same way, A, B, and C in the practical servo system can It will in turn yield a fast rise time in the closed-loop
be recognized that system response. This procedure corresponds with the
conventional proportional–derivative (PD) control
1. (A, B) is stabilizable. design.
2. (A, B, C) is invertible and has no invariant zeros
at z = 1. Step 2: Design a nonlinear feedback law
can alleviate the mismatch of substitution for the frac- implemented in real-time applications but also offer
tional N. Based on the principle of finite impulse high approximation accuracy in most cases.
response (FIR) filter, fractional delay can be well In this work, with the consideration of computa-
approximated by the given filter with integer orders. tional complexity and approximate precision, n is
Let Z N = Z Ni F with Ni being the maximum integer equaled to 3. Then, equation (18) is obtained as follows
but less than N and F (0 F\1) being the fractional
part of N. Z F can be approximated by a Lagrange zN ’ A0 zNi + A1 z(Ni + 1) + A2 z(Ni + 2) + A3 z(Ni + 3)
interpolation polynomial FIR filter as follows23 ð18Þ
X
n
Moreover, from equations (14) and (18), the final
Z F ’ Ak zk ð16Þ form of Q(z1 ) can be rewritten as
k =0
Figure 6. Magnitude responses of 1 zm Qf (z1 ) and Qf (z1 ) Figure 7. Magnitude responses of 1 zm Qf (z1 ) and Qf (z1 )
with a = 0:9131 m = 1, fs = 1000 Hz, fr = 11 Hz, and with a = 0:3669 m = 1, fs = 1000 Hz, fr = 5 Hz, and N = 200
N ’ 90:9091 are used as an illustrative example. are used as an illustrative example.
Wen et al. 7
System identification
Using the method of frequency domain identification,
the nominal model can be identified as
1
P(s) = ð31Þ
0:0005s2 + 0:006s
The distinctions of frequency characteristics between
the actual plant and the nominal model are shown in
Figure 10.
From Figure 10, there are extreme mismatches in
amplitude–frequency and phase–frequency characteris-
Figure 10. Frequency characteristics between the actual plant tics at both low frequency and high frequency. These
and the nominal model. uncertainties will definitely cause difficulties in achiev-
ing desired control performance.
1 1
T z Dp z () 1 ð30Þ
‘
Design of controllers
A factual example of stability and robust stability The following three controllers are compared:
analysis will be provided in Figures 11 and 12.
1. PD + DOB + ZPETC: This compound con-
trol strategy is commonly used in servo system.
Experimental validation It consists of three components, a PD feedback
controller, a DOB as the compensation of fric-
Experimental setup tion, and a ZPETC as the feedforward control-
To demonstrate the applicability of the proposed ler. The details of this control structure can be
method, an actual flight simulator system is employed found.26 At the end, the parameters are set as
as the test platform, which is shown in Figure 9. It con- Kp = 0:8 and Kd = 0:02, and t of the DOB is
sists of an inner-axis rotary actuator, a digital control chosen as 0.015.
system, and a DC pulse-width-modulated (PWM) 2. Integral backstepping: As a nonlinear control-
amplifier. ler, backstepping control with integral function
The rotary sensor of the device uses the encoder of can improve the system tracking response and
Heidenhain ECN125 whose accuracy is about 6 2000 . bandwidth in some degree. The design proce-
The proposed control algorithms are implemented via dure of this control scheme can be found,27 and
a C ++ language based on Windows-RTX real-time the parameters are set as c1 = 180:0, c2 = 180:0,
system on an industrial computer (E5300 2.8-GHz). and l1 = 2:0.
Wen et al. 9
Figure 11. (a) Stability and (b) robust stability when the external signal at 5 Hz.
Experimental results
Step Algorithm process
The three controllers are first tested for two specified
1. Initialization sinusoidal trajectories in 5 and 11 Hz, that is, the track-
2. Set constant matrixes A, B, F, and P of DGCNF ing command signals are employed as r = 0:5 sin
3. Confirm the frequency of the external dynamic signal (2p 5t) and r = 0:5 sin (2p 11t). Then, N is an integer
4. Compute the integer Ni and the fractional part F
5. According to equations (19), (21), and (22), and a fraction, respectively. The illustrations for stabi-
acquire the form of Qp lity and robust stability of the closed-loop system are
6. {main loop} analyzed as follows. With the parameter of DGCNF r
7. for i = 1 to m do varies in ½2(BT PB)1 , 0, the magnitude responses of
8. uQp =Pn (i) ( fe(i) !Qp (z1 )=Pn (z1 )g equations (28) and (30) present dramatic changes.
9. uQp =zm (i) ( fu1 (i) !Qp (z1 )zm g However, magnitudes are obviously lower than zero,
10. ufb (i) ( fe(i) ! Cfb (z1 )g and then, the stability and robust stability can be
guaranteed.
where Cfb (z1 ) = F + r(i)BT P(A + BF) (x(i) xe (i))
11. uff (i) ( fr(i) ! Cff (z1 )g, where Cff (z1 ) = Fe xe (i) The compared tracking errors of the three control-
12. u1 (i) ( uQp =Pn (i) + uQp =zm (i) + ufb (i) lers are shown in Figure 13. It shows that the proposed
13. uk (i) ( u1 (i) + uff (i) FRDOB + DGCNF controller performs extremely
14. End for better than PD + DOB + ZPETC controller and
15. Return uk (i) integral backstepping controller in terms of tracking
accuracy at both 5 and 11 Hz: the maximum tracking
errors decrease to 0.000568 with the help of
FRDOB + DGCNF. The precision raises two orders
3. FRDOB + DGCNF: The overall structure of of magnitude in the proposed schemes.
the proposed algorithm is depicted in Figure 8. To further show the robustness of the proposed
And according to Figure 8, the procedure of the strategy contrast to the previous methods, an unba-
algorithm is established as follows: lanced payload is installed on the inner axis. The condi-
tion is shown in Figure 14; the payload is about
where r = 0:2(BT PB)1 (2=p) arctan ( 0:2 jje(k)= 6.6 pounds.
e(k 1)j 1j). And Fe varies according to the com- The comparative performance is illustrated in
mand signal. In the filter of FRDOB Qp , a = 0:999N . Figure 15. In the case of extra payload, the tracking
10 Advances in Mechanical Engineering
Figure 12. (a) Stability and (b) robust stability when the external signal at 11 Hz.
Figure 13. Control performance of the three control strategies: (a) 5 Hz and (b) 11 Hz.
error of the proposed FRDOB + DGCNF controller remains almost unchanged in tracking dynamic signal
can be limited within a small level. It is also found in at 5 and 11 Hz. Particularly, with the increase in the
Figure 16(e) and (f), regardless of the additional pay- signal frequency, the errors of those two control
load, the proposed FRDOB + DGCNF scheme schemes increase obviously.
Wen et al. 11
Conclusion
In this article, a DGCNF combined with repetitive con-
troller based on DOB form is proposed for dynamic
tracking servo system. The developed controller has the
property of improving transient performance without
too much overshoot which mainly derives from CNF
Figure 14. Inner axis is installed with an unbalanced payload. technique. The FRDOB, with the fractional delay filter,
has an advanced improvement of rejecting any periodic
disturbance or tracking arbitrary repetitive external sig-
The strong robustness of the proposed scheme can nal. The stability as well as robustness of this controller
be explained by two primary factors: first, DGCNF is guaranteed via frequency domain analysis. The illus-
scheme is essentially a PID mode algorithm; therefore, trative experiments demonstrate the effectiveness of the
Figure 15. Control performance of the three control strategies with payload: (a) 5 Hz and (b) 11 Hz.
12 Advances in Mechanical Engineering
Figure 16. Comparison of tracking error with and without the payload: (a), (c), and (e) 5 Hz and (b), (d), and (f) 11 Hz.
Figure 17. Comparison of tracking performance of FRDOB and RDOB: (a and b) with payload and (c and d) without payload.
Wen et al. 13
proposed controller. The performance limitations of 11. Chen X, Komada S and Fukuda T. Design of a nonlinear
traditional PD + DOB + ZPETC and integral back- disturbance observer. IEEE T Ind Electron 2000; 47:
stepping in servo systems are experimentally investi- 429–437.
gated, while the proposed controller could overcome 12. Manayathara TJ, Tsao TC and Bentsman J. Rejection of
these limitations and improve the dynamic tracking per- unknown periodic load disturbances in continuous steel
formance effectively. casting process using learning repetitive control
approach. In: Proceedings of the 1993 IEEE international
symposium on intelligent control, Chicago, IL, 25–27
Declaration of conflicting interests August 1993, pp.36–41. New York: IEEE.
The author(s) declared no potential conflicts of interest with 13. Peery TE, OZbay H and Peery TE. HN optimal repetitive
respect to the research, authorship, and/or publication of this controller design for stable plants. J Dyn Sys Meas Con-
article. trol 1997; 119: 541–547.
14. Chen X and Tomizuka M. New repetitive control with
improved steady-state performance and accelerated tran-
Funding sient. IEEE T Contr Syst T 2014; 22: 664–675.
The author(s) disclosed receipt of the following financial sup- 15. Li CX, Gu GY, Yang MJ, et al. High-speed tracking of a
port for the research, authorship, and/or publication of this nanopositioning stage using modified repetitive control.
article: This work was partly supported by the grants of IEEE T Autom Sci Eng 2015; 1–11. DOI: 10.1109/
National Natural Science Foundation of China (No. TASE.2015.2428437.
61563041) and Inner Mongolia Natural Science Foundation 16. Zou Z, Zhou K, Wang Z, et al. Frequency-adaptive
(No. 2015MS0603). fractional-order repetitive control of shunt active power
filters. IEEE T Ind Electron 2015; 62: 1659–1668.
17. Nagahara M and Yamamoto Y. H-infinity-optimal frac-
References tional delay filters. IEEE T Signal Proces 2013; 61:
1. Kobayashi M and Horowitz R. Track seek control for 4473–4480.
hard disk dual-stage servo systems. IEEE T Magn 2001; 18. Na J, Gri~ nÓ R, Costa-Castelló R, et al. Repetitive con-
37: 949–954. troller for time-delay systems based on disturbance obser-
2. Devasia S, Eleftheriou E and Moheimani SOR. A survey ver. IET Control Theory A 2010; 4: 2391–2404.
of control issues in nanopositioning. IEEE T Contr Syst 19. Yao J, Jiao Z and Ma D. Adaptive robust control of DC
T 2007; 15: 802–823. motors with extended state observer. IEEE T Ind Electron
3. Liu ZZ, Luo FL and Rashid MH. Robust high speed 2014; 61: 3630–3637.
and high precision linear motor direct-drive XY-table 20. Peng J and Dubay R. Identification and adaptive neural
motion system. IEE P: Contr Theor Ap 2004; 151: network control of a DC motor system with dead-zone
166–173. characteristics. ISA T 2011; 50: 588–598.
4. Yao J, Yang G and Jiao Z. High dynamic feedback line- 21. Na J, Chen Q, Ren X, et al. Adaptive prescribed perfor-
arization control of hydraulic actuators with backstep- mance motion control of servo mechanisms with friction
ping. Proc IMechE, Part I: J Systems and Control compensation. IEEE T Ind Electron 2014; 61: 486–494.
Engineering 2015; 229: 728–737. 22. Kwan HK and Jiang A. Design of IIR variable fractional
5. Clayton GM, Tien S, Leang KK, et al. A review of feed- delay digital filters. In: ISCAS 2007. IEEE international
forward control approaches in nanopositioning for high- symposium on circuits and systems, New Orleans, LA,
speed SPM. J Dyn Sys Meas Control 2009; 131: 636–650. 27–30 May 2007, pp.2714–2717. New York: IEEE.
6. Park HS, Chang PH and Lee DY. Concurrent design of 23. Laakso TI. Splitting the unit delay. IEEE Signal Proc
continuous zero phase error tracking controller and sinu- Mag 1996; 13: 30–60.
soidal trajectory for improved tracking control. J Dyn 24. Venkataramanan V, Peng K, Chen BM, et al. Discrete-
Sys Meas Control 2001; 123: 127–129. time composite nonlinear feedback control with an appli-
7. Lin Z, Pachter M and Banda S. Toward improvement of cation in design of a hard disk drive servo system. IEEE
tracking performance nonlinear feedback for linear sys- T Contr Syst T 2003; 11: 16–23.
tems. Int J Control 1998; 70: 1–11. 25. Lee HS and Lee HS. Robust digital tracking controllers
8. Chen BM, Lee TH, Peng K, et al. Composite nonlinear for high-speed/high accuracy positioning systems. PhD
feedback control for linear systems with input saturation: Dissertation, Department of Mechanical Engineering,
theory and an application. IEEE T Automat Contr 2003; The University of California, Berkeley, CA, 1994.
48: 427–439. 26. Yunjie WU, Liu X and Tian D. Research of compound
9. Cheng G, Peng K, Chen BM, et al. Improving transient controller for flight simulator with disturbance observer.
performance in tracking general references using compo- Chinese J Aeronaut 2011; 24: 613–621.
site nonlinear feedback control and its application to 27. Tan Y, Chang J, Tan H, et al. Integral backstepping con-
high-speed-table positioning mechanism. IEEE T Ind trol and experimental implementation for motion system.
Electron 2007; 54: 1039–1051. In: IEEE international conference on control applications,
10. Lee HS and Tomizuka M. Robust motion controller Anchorage, AK, 27 September 2000, pp.367–372. New
design for high-accuracy positioning systems. IEEE T Ind York: IEEE.
Electron 1996; 43: 48–55.