You are on page 1of 14

A Control Engineer’s Guide t o Sliding Mode

Control
K. David Young, Vadim [. Utkin, Umit Ozgiiner

.4bstmct- This paper presents a guide to sliding mode torial [13] of design methods, or another more recent state
control for practicing control engineers. It offers an accurate of the art assessment [14], or yet another survey of sliding
assessment of the so-called chattering phenomenon, catalogs
implementable sliding mode control design solutions, and mode research [15],the purpose of this paper is to provide
provides a frame of reference for future sliding mode control a comprehensive guide to Sliding Mode Control for control
research. engineers. It is our goal to accomplish these objectives:
Keywds- Asymptotic Observers, Discontinuous Con- Provide an accurate assessment of the chattering phe-
trol, Discrete Time Systems, Disturbance Compensation, nomenon;
Disturbance Estimation, Disturbance Rejection, High Gain
Feedback Systems, Motion Separations, Multivariable Ser- offer a catalog of implementable robust sliding mode
vomechanisms, Parametric Uncertainties, Parasitic Dynam- control design solutions for real life engineering appli-
ics, Robust Control, Sampled Data Control Systems, Singu- cations;
lar Perturbations, Sliding Mode, Uncertain Systems, Vari-
able Structure Control. initiate a dialog with practicing control engineers on
sliding mode control by threading the many analyti-
cal underpinnings of sliding mode analysis through a
I. INTRODUCTION series of design exercises on a simple, yet illustrative
During the last two decades since the publication of the control problem; and
survey paper in the IEEE Transactions of Automatic Con- establish a frame of reference for future sliding mode
trol in 1977 [l],significant interest on Variable Structure control research.
Systems (VSS) and Sliding Mode Control (SMC) have been The flow of the following presentation conforms to the his-
generated in the control research community worldwide. torical development of VSS and SMC: First we introduce
One of the most intriguing aspects of sliding mode is the issues within Continuous Time Sliding Mode in Section 11,
discontinuous nature of the control action whose primary then in Section 111, we progress to Discrete time Sliding
function is to switch between two distinctively different Mode, followed with Sampled Data SMC Design in Section
system structures (or components) such that a new type IV.
of system motion, called Sliding Mode, exists in a mani-
fold. This peculiar system characteristic is claimed to re- 11. CONTINUOUS
TIME SLIDING
MODE
sult in superb system performance which includes insen- Sliding mode is originally conceived as system motion for
sitivity to parameter variations, and complete rejection of dynamic systems whose essential open loop behavior can be
disturbances. The reportedly superb system behavior of modeled adequately with ordinary differential equations.
VSS and SMC naturally invites criticisms and scepticisms The discontinuous control action, which is often referred
from within the research community, and from practicing to as Variable Structure Control (VSC), is also defined in
control engineers alike [a]. The sliding mode control re- the continuous time domain. The resulting feedback sys-
search community has risen to answer to some of these tem, the so-called VSS, is also defined in the continuous
critical challenges, while at the same time, contributed to time domain, and it is governed by ordinary differential
the confusion about the robustness of SMC with incom- equations with discontinuous right hand side. The mani-
plete analyses, and design fixes for the so-called chatter- fold of the state space of the system on which sliding mode
ing phenomenon [3]. Many analytical design methods were occurs is the Sliding Mode Manifold, or in brief, Sliding
proposed to reduce the effects of chattering [4], [5], [6], [7], Manifold. For control engineers, the simplest, but vividly
[8]- for it remains to be the only obstacle for sliding mode perceptible example is a double integrator plant, subject to
to become one of the most significant discoveries in modern time optimal control action. Due to imperfections in the
control theory; and its potential seemingly limited by the implementations of the switching curve, derivable using the
imaginations of the control researchers [9], [lo], [ll]. Pontrayagon Maximum Principle, sliding mode may occur.
In contrast to the previously works published since the Sliding mode was studied in conjunction with relay con-
1977 article [l],which serve as a status overview [l2], a tu- trol for double integrator plants, a problem motivated by
the design of attitude control systems of missiles with jet
K. D. Young is with YKK Systems, 2680 LaSalle Drive, Mountain thrusters in the 1950’s [16].
View, California 94040-4770, USA, Email: ykksystems@msn.com
V. I. Utkin is with the Departments of Electrical Engineering The so-called chattering phenomenon is generally per-
and Mechanical Engineering, Ohio State University, 2015 Neil Ave., ceived as motion which oscillates about the sliding man-
Columbus, Ohio 43210 USA, E m a i l utkin8ee.eng.ohic-state.edu ifold. There are two possible mechanisms which produce
U. Ozgiiner is with the Department of Electrical Engineering, Ohio
State University, 2015 Neil Ave., Columbus, Ohio 43210 USA, Email: such a motion. First, in the absence of switching nonide-
ozguner. l@osu.edu alities such as delays, i.e., when the switching device is

1996 IEEE Workshop on


Variable Structure Systems -1-
0-7803-3718-2/96 $5.00 01996 IEEE

Authorized licensed use limited to: Sejong University. Downloaded on November 11, 2009 at 06:45 from IEEE Xplore. Restrictions apply.
ideally switching at an infinite frequency. In this case, the s(t = 0 , t 2 t* where t* is the first time instant that
existence of parasitic dynamics in series with the plant. z(t*)= 0, a.e., once the state trajectory reaches the sliding
This may account for the otherwise negligible fast actua- manifold, it remains on it for good. However, even with
tor and sensor dynamics, and cause a small amplitude high such idealized switching device, unmodeled dynamics can
frequency oscillation to appear in the neighborhood of the induce oscillations about the sliding manifold. Suppose we
sliding manifold. In control engineering practice, such fast have ignored the existence of a second order “sensor” dy-
dynamics are often neglected in the open loop model for namics, and the true system dynamics are governed by,
control design if the associated poles are well damped, and
outside the desired bandwidth of the feedback control sys- x = -sgn(x,) , (3)
tem. Generally, the motion of the real system is close to +
p2x, 2p4, + xs = 5 , p << 1 , (4)
that of an ideal system in which the parasitic dynamics are
neglected, and the difference between the ideal and real mo- where z, and 5, are the states of the sensor dynamics.
tion, which is on the order of the neglected time constants, Clearly, sliding mode cannot occur on z = 0 since x is con-
decays rapidly. The mathematical basis for the analysis of tinuous, however, since 4sis bounded, x,(t) - x(t) = O(p)
dynamic systems with fast motion is the theory of singu- where p is the time constant of the sensor. Furthermore,
larly perturbed differential equations [17], and its exten- reaching an O(p) boundary layer of z ( t ) = 0 is guaranteed
sions to control theory have been developed and applied since:
in practice [18]. However, the theory is not applicable for P, = -sgn(x, O ( p ) ). + (5)
VSS since they are governed by differential equations with
The system behavior inside this O ( p ) boundary layer can
discontinuous right hand sides. The interactions between
be analyzed with the infinitely fast switching device re-
the parasitic dynamics and VSC generate a non-decaying
placed with a linear gain approximation whose gain tends
oscillatory component of finite amplitude and frequency,
to infinity asymptotically:
and this is generically referred to as chattering.
Second, the switching nonidealities alone can cause such U = -gxs, g+m. (6)
high frequency oscillations. We shall focus only on the de-
lay type of switching nonidealities since it is most relevant The root locus of this system, with g as the scalar gain
to any electronic implementation of the switching device, parameter, has third order asymptotes as g + CO. There-
including both analog and digital circuits, and micropro- fore, the high frequency oscillation in the boundary layer
cessor code executions. Since the cause of the resulting is unstable.
chattering phenomenon is due to time delays, discrete time With second order parasitic actuator dynamics in series
control design techniques such as the design of an extrap with the nominal plant, the closed loop dynamics are given
olator exist to mitigate the switching delays [19]. These bY
design approaches are perhaps more familiar to control en-
gineers.
Unfortunately, in practice, both the parasitic dynamics
and switching time delays exist. Since it is necessary to
compensate for the switching delays by using a discrete The characteristic equation of this system is identical to
time control design approach, a practical SMC design may that of the parasitic sensor case. This is not surprising
have to be unavoidably approached in discrete time. We since the forward transfer function is identical in both
shall return to the details of discrete time SMC after we cases. Thus, similar instability also occur with infinitely
illustrate our earlier points on continuous time SMC with fast switching.
a simple design example, and summarize the existing ap-
proaches to avoid chattering. B. Boundary Layer Control
The most commonly cited approach to reduce the effects
A . Chattering due to parasitic dynamics - a simple example
of “chattering” has been the so called piecewise linear or
The effects of unmodeled dynamics on sliding mode can smooth approximation of the switching element in a bound-
be illustrated with an extremely simple relay control system ary layer of the sliding manifold [20], [21], [22], [23]. Inside
example: Let the nominal plant be an integrator, the boundary layer, the switching function is approximated
by a linear gain. In order for the system behavior to be
?=U, x(to)=xo#O, (1) close to that of the ideal sliding mode, particularly when
a significant unknown disturbance is to be rejected, suffi-
and assume that a relay controller has been designed, ciently high gain is needed. Note that in the absence of
disturbance, it is possible to enlarge the boundary layer
U = -sgn(z) . (2) thickness, and at the same time reduce the effective lin-
The sliding “manifold is the origin of the state space ear gain such that the resulting system no longer exhibits
x = 0. Given any nonzero initial condition 50,the state any oscillatory behavior about the sliding manifold, how-
trajectory t ) is driven toward the sliding manifold. Ide- ever, this system no longer behaves as dictated by sliding
- z (~,
ally, if the relay controller can switch infinitely fast, then mode, i.e., simply put, in order to reduce chattering, the

-2-

Authorized licensed use limited to: Sejong University. Downloaded on November 11, 2009 at 06:45 from IEEE Xplore. Restrictions apply.
proposed method of piecewise linear approximation reduces Figure 2 shows the simulated error responses of the closed
the feedback system to a system with no sliding mode. This loop system for these two gain values which agree with the
proposed method has a wide acceptance by many sliding analysis. In this simulation, a unity reference command
mode researchers, but unfortunately it does not resolve the for the plant state and a constant disturbance d ( t ) = 0.5 is
core problem of the robustness of sliding mode as exhib- introduced. The tradeoff between chattering reduction and
ited in chattering. Many sliding mode researchers cited the disturbance rejection can be observed from the steady state
work in [3], [22]as the basis of their optimism that with value of -0.005 in x ( t ) whose magnitude (for the stable re-
boundary layer control, the implementation issues of con- sponse), or average value of -0.0025 (for the oscillatory
tinuous time sliding mode are solved. Unfortunately, the response) is proportional to the gain g. We note that even
optimism of these researchers was not shared by practicing with g > gc, the resulting responses are only oscillatory, but
engineers, and this may be rightly so. The effectiveness still bounded. This is because the linear analysis is valid
of boundary layer control is immediately challenged when only inside the boundary layer, and the VSC always forces
realistic parasitic dynamics are considered. An in-depth the state trajectory back into the boundary layer region.
analysis of the interactions of parasitic actuator and sensor However, as the gain increases, the frequency of oscillation,
dynamics with the boundary layer control [24]revealed the related to the magnitude of the imaginary part of the root
shortcomings of this approach where parasitics dynamics locus, also increases, hence the general description chatter-
must be carefully modeled and considered in the feedback ing applies as this frequency reaches the neglected resonant
design in order to avoid instability inside the boundary frequencies of the physical plant.
layer which leads to chattering. Without such information This example illustrates the advantages of boundary
of the parasitic dynamics, control engineers must opt for layer control which lie primarily in the availability of fa-
a worst case boundary layer control design in which the miliar linear control design tools to reduce the potentially
disturbance rejection properties of SMC are severely com- disastrous chattering. However, it should also be reminded
promised. that if the acceptable closed loop gain has to be reduced
sufficiently to avoid instability in the boundary layer, the
B.l A boundary layer controller resulting feedback system performance may be significantly
We shall continue with the simple relay control exam- inferior to the nominal system with ideal sliding mode. Fur-
ple, and consider the design of a boundary layer controller. thermore, the precise details of the parasitic dynamics must
We assume the same second order parasitic sensor dynam- be known and used properly in the linear design.
ics as before. The behavior inside the boundary layer is
describable by a linear closed loop system, C. Observer based Slzdzng Mode Control
Recognizing the essential triggering mechanism for chat-
x = -gx, + d ( t ) , (9) tering is due to the interactions of the switching action with
+
rzx, 27,xy x, + =2 , (10) the parasitic dynamics, an approach which utilizes asymp-
totic observers to construct a high frequency by pass loop
where d ( t ) represents a bounded, but unknown exogenous has been proposed[4]. This design exploits a localization
disturbance. Whereas discontinuous control action in VSC of the high frequency phenomenon in the feedback loop by
can reject bounded disturbances, by replacing the switch- introducing a discontinuous feedback control loop which is
ing control with a. boundary layer control, the additional closed through an asymptotic observer of the plant [25].
assumption that d be bounded is needed since according Since the model imperfections of the observer are suppos-
to singular perturbation analysis, the residue error is pro- edly smaller than that in the plant, and the control is
portional to ldI/g. Given a finite T,, we can compute the discontinuous only with respect to the observer variables,
root locus of this system with respect to the scalar pos- chattering is localized inside a high frequency loop which
itive gain g > 0. An upperbound gc exists which speci- bypasses the plant. However, this approach assumes that
fies the crossover point of the root locus on the imaginary an asymptotic observer can indeed be designed such that
axis. Thus, for 0 < g < gc, the behavior of this system is the observation error converges to zero asymptotically. We
asymptotically stable, i.e., for any initial point inside the shall discuss the various options available in observer based
boundary layer , sliding mode control in the following design example.
IXSI 5 1/91 (11)
C.l Design example of observer based SMC
the sliding manifold z = 0 is reached asymptotically as
t ---foc). The transient response and disturbance rejection
For the relay control example, we examine the utility
of this feedback system are two competing performance of the observed based SMC in localizing the high fre-
measures to be balanced by the choice of an optimum gain quency phenomenon. For the nominal plant, the follow-
value. If we assume p = .01,the associated root locus is ing asymptotic observer results from applying conventional
plotted in Figure 1 for 0.003 5 9-l 5 0.01 with a step state space linear control design,
size of 0.001, The critical gain is gc = 200. Thus from k = h(s, - 2) + U , (12)
the linear analysis, a boundary layer control with g = 100
results in a stable sliding mode,whereas with g = 200, os- where h > 0 is the observer feedback gain, and x, is the
cillatory behavior about the sliding manifold is predicted. output of the parasitic sensor dynamics. The SMC and the

-3-

Authorized licensed use limited to: Sejong University. Downloaded on November 11, 2009 at 06:45 from IEEE Xplore. Restrictions apply.
associated sliding manifold defined on the observer state if the high gain loop in the asymptotic observer is imple-
space is, mented with a switching function, it is referred to as a
U = -sgn(f) . (13) Sliding Mode Observer [28], [29], [30]. Since two sliding
manifolds are employed in the feedback loops, the closed
The behavior of the closed loop system can be deduced loop system robustness must be carefully examined when
from the following fourth order system, less than infinite switching frequencies are to be expected.
In such robustness analysis, the relative time scales of the
+
x = -sgn(z - e ) d(t), e = z - 2 ,
A
(14) various motions in the system can be managed with singu-
e = -he +
h(z -a,) d ( t ) , + (15) lar perturbation methods, similar to that applied to high
T;i?# +
2T,k, +
2, = 5 . (16)
gain observers.
The performance of the observer based SMC can be
First we consider the case when d(t)= 0. Using an infinite evaluated by simulation. We let the sensor dynamic time
gain linear function g(z - e ) to approximate the switching constant be T, = 0.01, and assume the same unity refer-
function sgn(z - e ) , and since p is finite, the above system ence command and constant disturbance as in the bound-
is a singular perturbed system with g-’ being the parasitic ary layer control example, d ( t ) = 0.5. A boundary layer
parameter. The slow dynamics which are of third order can of 0.002 is used to approximate the observer based SMC.
be extracted by formally setting 9-l = 0, and z - e = 0 , The closed loop eigenvalues are at {-5lO(due to bound-
ary layer), -12.92(from observer), -59.44, -127.62(shifted
1 = -hx,, (17) sensor poles). Figure 4 shows the error response between
+ +
r,”X, Z T , ~ , z, = e . (18) the reference and the observed state. The steady state er-
ror of -0.001 reflects the attenuation of the disturbance by
It is possible to further apply a singular perturbation anal- the high gain of 500. Note that sliding mode in the ob-
ysis to insure that given r8,there exists h > 0 such that server state space can be implemented with high gain with
the asymptotic observer dynamics are of first order, and no adverse interactions with the parasitic dynamics. How-
its eigenvalue is approximately -h. Clearly, the adverse ef- ever, as shown in Figure 5 , the plant state response has a
fects of the parasitic sensor dynamics are neutralized with steady state error of 0.05 which is due to the observation
an observer based SMC design. If a switching function is error caused by the relatively low feedback gain of the ob-
realized in the SMC design, the only remaining concerns server, h = 10. This error can be reduced by increasing the
will be switching time delays, and if the observer is to be value of gain h, provided that the time scales and stability
implemented in discrete time, the entire feedback design of the system are preserved. Also shown in this figure is the
including the compensation of switching time delays may superb rejection of the disturbance in the observer based
be best carried out in the discrete time domain. Figure 3 sliding mode.
is a block diagram of this design. Note that the switching
element is inside a feedback loop which passes through only
the observer, bypassing both blocks of the plant dynamics.
This is the so-called high frequency bypass effects of the D. Disturbance Compensation
observer based SMC [4], [26].
When d ( t ) # 0, its effects on the convergence of asymp- In SMC, the main purpose of sliding mode is to reject
totic observers are well known. If d ( t ) is an unknown con- disturbances and to desensitize unknown parametric per-
stant disturbance, a multivariable servomechanism formu- turbations. Building on the observer based SMC, a slid-
lation can be adopted to estimate both the state and exoge- ing mode disturbance estimator which uses sliding mode
nous disturbance in a composite asymptotic observer. The to estimate the unknown disturbances and parametric un-
resulting feedback system is the so-called Variable Struc- certainties has also been introduced [8]. In this approach,
ture (VS) Servomechanism design [25], [27]. the control law consists of a conventional continuous feed-
For bounded but unknown disturbances with bounded back control component, and a component derived from the
time derivative, the only known approach to solving the SM disturbance estimator for disturbance compensation.
robustness of the asymptotic observer is to introduce a If the disturbance is sufficiently compensated, there is no
high gain loop around the observer itself to reject the un- longer the need to evoke a discontinuous feedback control
known disturbance, e.g., by increasing the gain h in the to achieve sliding mode, thus, the remaining control design
observer such that the effects of d ( t ) are adequately atten- follows the conventional wisdom, and issues regarding un-
uated. However, the requirements for disturbance atten- modeled dynamics are no longer critical. Also chattering
uation and closed loop stability must be balanced in the becomes a non-issue since a conventional feedback control
design, and if sliding mode is to be preserved in the mani- instead of SMC is applied. The critical design issues are
fold 2 = 0, g must be sufficiently larger than h. A switch- transferred t o the SM disturbance estimator and its associ-
ing function implementation of the SMC would seem to ated sliding mode. While there are many engineering issues
ensure the necessary time scale separations, however, the to be dealt with in this approach, simulation studies and
condition g << 117, should also be imposed to avoid ad- experiment results [31] showing that desired objectives are
verse interactions with the parasitic dynamics. Note that indeed achievable.

-4-

Authorized licensed use limited to: Sejong University. Downloaded on November 11, 2009 at 06:45 from IEEE Xplore. Restrictions apply.
D.l An SM disturbance estimator E. Actuator Bandluidth Constraints
Once again we return to the simple relay example with Despite its desirable properties, VSC is mostly restricted
parasitic sensor dynamics for our design of a disturbance to control engineering problems where the control input of
estimator. The plant model is the plant is, by the nature of the control actuator, nec-
essarily discontinuous. Such problems include control of
j. = U + d ( t ), (19) electric drives where pulse-width-modulation is not the ex-
TZX, + 27,xs + x,3 = x . (20) ception, but the rule of the game. Space vehicle attitude
control is another example where reaction jets operated
We shall design a disturbance estimator with sliding mode
in an on-off mode are commonly used. The third exam-
as follows:
ple, which is closely related to the first one, is power con-
+
f = U sgn(x, - 2 ) . (21) verter and inverter feedback control design. For this class
From the error dynamics, of applications, the chattering phenomenon still need to be
a addressed, however, the arguments against using sliding
e = -sgn(z, - 2) + d ( t ) + S, -k , e = x, - 2, (22) mode in the feedback design are weakened. The issue in
if sliding mode occurs on e = 0, since u ( t ) is continuous this case is whether VSC should be utilized to improve sys-
tem performance, or standard PWM techniques are to be
and differentiable, /Ss- j.1 = O ( r S ) . By solving for the
equivalent control in d = 0 , applied, after a standard PID type design is completed, to
realize a low bandwidth servo loop. If VSC is to be used,
by adopting an observer based SMC, the high frequency
components of the discontinuous control can be bypassed,
Thus, within this estimator, there exists a signal which, and consequently, adverse interactions with the unmodeled
under the sliding mode condition, is O(7,) close to the dynamics which causes chattering are avoided.
unknown disturbance d ( t ) . This forms the basis of a feed- In plants where control actuators have limited band-
back control design which utilizes this signal to compensate width, e.g., hydraulic actuators, there are two possibilities:
the disturbance to O ( T ~ )The
. resulting control law has a First, the actuator bandwidth is outside the required closed
conventional linear feedback component, and a disturbance loop bandwidth. Thus the actuator dynamics become un-
compensating component, and for this system modeled dynamics, and our discussions in the previous sec-
U = -kf - [sgn(s, - ?)Ieq. tions are applicable. While in linear control design, it is
(24) possible to ignore the actuator dynamics, doing so in VSC
The extraction of the equivalent control from the sliding requires extreme care. By ignoring actuator dynamics in a
model control signal is by low pass filtering. While the- classical SMC design, chattering is likely to occur since the
oretically there exists such a low pass filter such that the switching frequency is limited by the actuator dynamics
equivalent control can be found, in practice, the bandwidth even in the absence of parasitic dynamics. Strictly speak-
of the desired closed loop system, the spectrum of the dis- ing, sliding mode cannot occur, since the control input to
turbance, are all important to the selection of the cutoff the plant is continuous.
frequency of this filter. Second, the desired closed loop bandwidth is beyond the
For evaluation by simulation, we let the sensor time con- actuator bandwidth. In this case, regardless of whether
stant be once again T, = 0.01, and assume the same unity SMC or other control designs are to be used, the actu-
reference command, and constant disturbance d ( t ) = 0.5. ator dynamics are lumped together with the plant, and
After canceling the disturbance, we desire a closed loop the control design model encompasses the actuator-plant
system with a time constant of one seconds which can be in series. With the actuator dynamics no longer negligi-
attained with k = 2. A boundary layer of 5 x replaces ble, often the matching conditions which are satisfied in
the switching function in the estimator. The closed loop the nominal plant model are violated. This results from
eigenvalues are { -2000. (from the boundary layer), -1 ,(the having dominant dynamics inserted between the physical
dominant closed loop pole), -96.75 i 3101.83(the shifted input, such as force, and the controller output, usually an
sensor poles). For low pass filtering, a third order butter- electrical signal. The design of SMC which incorporates
worth filter with a 3dB corner frequency of 50 rad./sec. is the actuator dynamics as a prefilter for the VSC was pro-
used to filter the equivalent control. Figure 6 shows the posed in [28]. This design utilizes an expansion of the orig-
error between the reference command and the plant state inal state space by including state derivatives, and formu-
which exhibits the desired one seconds time constant tran- late an SMC design such that the matching condition is
sient behavior, with the exception of initial minor distor- indeed satisfied in the extended space. Another alternative
tions which are due to the convergence of the disturbance approach is to utilize sliding mode to estimate the distur-
estimate shown in Figure 7. Despite the constant distur- bance for compensation as discussed earlier. Since sliding
bance, the steady state error is zero. While standard PID mode is not introduced primarily to reject disturbances,
controllers can achieve the same zero steady state error in the matching conditions are of no significance in this de-
the presence of unknown constant disturbance, the track- sign. Provided that a suitable sliding mode exists such that
ing error is regulated to zero even when d ( t ) is time varying the disturbance can be estimated from the corresponding
as it is reported before [8]. equivalent control, this approach resolves the limitations

-5-
imposed by actuator bandwidth constraints on the design the desired transient performance. The resulting feedback
of sliding mode based controllers. control law is given by

E.l An SMC Design with Pre-Filter U = -kl2 - k2xa - [sgn(x, - 2)leq. (33)
We shall use the example with a nominal integrator With 7 , = 0.01, and a = 0.2, the feedback gains IC1 =
plant, and actuator dynamics, 31.25, and k2 = 6.25 place the poles of third order system
dynamics, which is consists of the actuator dynamics and
x = 2, +d(t), (25) the integrator plant, at (-2.5, -2.5, -5}. Again, we use
cy2x, + 2aXa + 5 , = U , (26) the same third order butterworth low pass filter with a 50
Hz bandwidth as before to filter the equivalent control sig-
to illustrate this design. The actuator bandwidth limita- nal. Figure 8 shows the effects of the constant disturbance
tion is expressed in the time constant cy. Given a discontin- d ( t ) = 0.5 have been neutralized since the error between
uous input u(t),the rate of change of the actuator output the reference command and the plant state is reduced to
x a ( t ) is limited by the finite magnitude of cy. However, in zero in steady state. The disturbance estimate is shown in
order for the disturbance d ( t ) to be rejected, z a ( t )must be Figure 9 to reach its expected value in steady state.
an SMC. Also if x , can be designed as a control input, then
the matching condition is clearly satisfied. But since U is F. Frequency Shapzng
the actual input, the matching condition does not holds
An approach which has been advocated for attenuating
for finite a. The design begins with an assumption that the effects of unmodeled parasitic dynamics in sliding mode
d ( t ) has continuous first and second derivatives, and the involves the introduction of frequency shaping in the de-
definitions of new state variables,
sign of the sliding manifold [5]. In stead of treating the
sliding manifold as the intersection of hyperplanes defined
x 1 = x , x2=XC, x 3 = x . (27) in the state space of the plant, sliding manifolds which
are defined as linear operators are introduced to suppress
The control U is designed as an VSC with respect to the
sliding manifold, frequency components of the sliding mode response in a
designated frequency band. For unmodeled high frequency
+ +
4 x 1 , x2, x3) = ClXl c2x2 x3 = 0 . (28)
dynamics, this approach implants a low pass filter either
as a prefilter, similar to introducing an artificial actuator
With the equivalent control ueq computed from dynamics, or as a post-filter, functioning like sensor dynam-
ics. The premise of this so-called Frequency Shaped Slid-
ing Mode design, which was motivated by flexible robotic
manipulator control applications [32], is that the effects of
parasitic dynamics are as critical on the sliding manifold.
the resulting sliding mode dynamics are found to be com- However, robustness to chattering was not addressed in this
posed of two subsystems in series: design. By combining frequency shaping sliding mode and
the SMC designs introduced earlier, the effects of parasitic
+ +
x c2x c1x = 0 , (30) dynamics on switching induced oscillations, as well as their
+
x, = (c; - Cl)* c1c2x - di. (31) long term interactions with sliding mode dynamics can be
handled.
This design shows that although the embedded prefilter in
the plant model destroys the matching condition, an SMC F.l A frequency shaped SMC design
can still be designed to reject the unknown disturbance. For the nominal integrator plant with parasitic sensor
However, it is necessary to restrict the class of disturbances dynamics, we introduce a frequency shaping post-filter,
to those which have bounded derivatives. Furthermore,
derivatives of the state are also required in the design. +
xp 2w,xp w;x, = 2 , ,+ (34)
E.2 A,Disturbance Estimation Solution Yp = Plkp +p 2 z p (35)

For the nominal integrator plant with limited bandwidth The sliding manifold is defined as a linear operator, which
actuator dynamics given by Eqns.(25,26),we introduce the can be expressed as a linear transfer function,
same set of sensor dynamics as in Eqn.(20) and use a distur-
bance estimator similar to Eqn.(2l),with only x , replacing
U,
2 = 5, + sgn(x, - 2 ) . (32) Given an estimate of the lower bound of the bandwidth
With sliding mode occurs on x - x = 0, the disturbance d ( t ) of parasitic dynamics, the post-filter parameter wpcan be
is estimated with the equivalent control given by Eqn.(23) chosen to impose a frequency dependent weighting function
to O ( T ~ )With
. the disturbance compensated, the remain- in a linear quadratic optimal design whose solution provides
ing task is to design a linear feedback control to achieve an optimal sliding manifold. The optimal feedback gains

-6-
are implemented as p l , p 2 in Eqn.(36), and they ensures VSS and SMC, the notion of sliding mode subsumes a con-
that the sliding mode dynamic response has adequate roll tinuous time plant, and a continuous time feedback control,
off in the specified frequency band. albeit its discontinuous, or switching characteristics. How-
ever, Sliding Mode, with its conceptually continuous time
G. Robust Control Design based on Lyapunov Method characteristics, is more difficult to quantify when a discrete
For plants whose dynamic models are uncertain, robust time implementation is adopted. When control engineers
control design which utilizes Lyapunov functions of the approach discrete time control, the choice of sampling rate
nominal plant has been proposed. The origin of this ap- is an immediate, and extremely critical design decision,
proach can be traced to work published in the 1970's by unfortunately, in continuous time Sliding Mode, desired
Leitmann and Gutman [33]. The resulting feedback control closed loop bandwidth does not provide any useful guide-
law is of the form, lines for the selection of sampling rate. In the previous
section, we indicate that asymptotic observers or sliding
BTVV n dV mode observers can be constructed to eliminate chatter-
U = -p(z, t)
llBTVVll
,vv = -
dX
E Et", z E IR" (37)
ing. Observers are most likely constructed in discrete time
for any real life control design. However, in order for these
where p(., .) is a scalar feedback gain, z E Et" is the state observer-based design to work, sampling rate has to be rela-
vector, and B(z,t ) E Rnxm is the input matrix in an affine tively high since the notion of continuous time sliding mode
dynamic system, is still applied.
For Sliding Mode, the continuous time definition and its
k = f(x,t ) + B ( z ,t ) u + h ( z ,t ) , U E Rm, (38) associated design approaches for discrete time control im-
and V(z) > 0 is a Lyapunov function of the nominal open plementation have been redefined to cope with the finite
loop plant, i.e., Eqn.(38) with u = 0 and h ( . , . ) = 0. For time update limitations of discrete time controllers. Dis-
unity feedback gain, p(.;) = 1, the norm of the above crete time Sliding Mode (DSM) was introduced [34] for
feedback control is equal to unity for any (5, t ) ,thus it is discrete time plants. The most striking contrast between
also referred to as unit control. SM and DSM is that DSM may occur in discrete time sys-
It is critical to point out that in any uncertain plant given tems with continuous right hand side, thus discontinuous
by Eqn.(38) and unit control in Eqn.(37), the feedback con- control and Sliding Mode, are finally separable. In discrete
trol is in fact a sliding mode control which is discontinuous time, the notion of VSS is no longer a necessity in dealing
on the sliding mode manifold, with motion on a Sliding Manifold.

s(z) = BTVV = 0 , IV. SAMPLED


DATASLIDING
MODECONTROLDESIGN
(39)
We shall limit our discussions to plant dynamics which
provided that the unknown disturbance denoted by the can be adequately modeled by finite dimensional ordinary
term h ( z ,t ) can be rejected by the choice of the scalar feed- differential equations, and assume that an apriori band-
back functional p(., .). Under the matching condition [35] width of the closed loop system has been defined. The
that there exists a vector X(z, t ) E Rmsuch that feedback controller is assumed to be implemented in dis-
crete time form. The desired closed loop behavior includes
h ( s ,t ) = B ( z ,t)X(s, t ) , (40) insensitivity to significant parameter uncertainties and re-
sliding mode on s(z) = 0 is guaranteed with jection of exogenous disturbances. Without such a demand
on the closed loop performance, it is not worthwhile to
p(z,t) > Xo(zc,t) z Il4s,t)ll. (41) evoke DSM in the design. Using conventional design rule
of thumb for sampled data control systems, it is reason-
Since sliding mode is the principle mechanism with which able to assume that for the discretization of the contin-
uncertainties and disturbances are rejected in robust con- uous time plant, we include only the dominant modes of
trol of uncertain systems, the robustness of these feedback the plant whose corresponding corner frequencies are well
controllers with respect to unmodeled dynamics are identi- within the sampling frequency. This is always achievable
cal to continuous time SMC, and the respective engineering in practice by anti-aliasing filters which are inserted at the
design issues can be addressed as outlined in this section. plant outputs. Actuator dynamics are assumed to be of
higher frequencies than the sampling frequency. Otherwise,
111. DISCRETETIMESLIDINGMODE actuator dynamics will have to be handled as part of the
While it is an accepted practice for control engineers to dominant plant dynamics. Thus, all the undesirable para-
consider the design of feedback systems in the continuous sitic dynamics manifest only in the between sampling plant
time domain - a practice which is based on the notion behavior, which by default of using sampled data control,
that, with sufficiently fast sampling rate, the discrete time is essentially the open loop behavior of the plant. Clearly,
implementation of the feedback loops is merely a matter this removes any remote possibilities of chattering due to
of convenience due to the increasingly affordable micropre the interactions of sliding mode control with the parasitic
cessor. The essential conceptual framework of the feedback dynamics.
design remains to be in the continuous time domain. For We begin to summarize sampled data sliding mode con-

-7-

Authorized licensed use limited to: Sejong University. Downloaded on November 11, 2009 at 06:45 from IEEE Xplore. Restrictions apply.
trol designs with the well understood sample and hold pro- the sampling instants. Let the sliding manifold be defined
cess. This may seem to be elementary at first glance, it is by
however worthwhile since the matching conditions for the
continuous time plant are only satisfied in an approxima- Sk = c x k 0. k = 0 , 1 , .. . s(kT) = S k (52)
tion sense in the discretized models. We shall restrict our
discussions to linear time-invariant plants with uncertain- Two different definitions of discrete time sliding mode have
ties and exogenous disturbances, been proposed for discrete time systems. While these def-
initions share the common base of using the concept of
?=Ax+Bu+Ed, Z ,L E I R ~ ,E I R ' (42)
ZEIR~ equivalent control, the one proposed in [34] uses a defini-
tion of discrete time equivalent control U:' = u(kT) which
where A , B,E are constant matrices, and d ( t ) is the exoge- is the solution of
nous disturbance. For the plant (42), we assume that the
system matrices are decomposed into nominal and uncer- .$,++I =0, k = 0, 1,.. . (53)
tain components,
On the other hand, uiq is defined in [36] as the solution of
A=A+AA, B=B+AB (43)
A i = sk+l (54) - sk =0, k = 0,1,. . .
where A , B denote the nominal components. Let the
admissible parametric uncertainties satisfy the following Note that (53) implies (54), however, the converse is not
model matching conditions [35] true. Herein, the first definition given by Eqn.(53) shall be
used.

rank( [ B ! A A f AB f E ] )= rankB . (44) A . DSM Control Design for Nominal Plants


Given the nominal plant with no external disturbance,
The discrete time model is obtained by applying a sample
the DSM design becomes intuitively clear. In DSM, by
and hold process to the continuous time plant with Sam-
definition,
pling period T , which to O ( T 2 ) is
, given by:

Zk+l = F X k 4-G U k + Ddk , sk+l Cxkfl = c(FXk + GUk) = 0 , (55)


50 5(to), (45)
and provided that CG is invertible, the DSM control which
s(kT) = x k ' = uk ' d ( k T ) == dk ' (46) is also the equivalent control, is given by the linear contin-
where F, G and D result from integrating the solution of uous feedback control,
Eqn.(42) over the time interval t E [kT,( k 1)T] with +
uk = -[CG]-lCFxk. (56)
u ( t ) = u(kT), d ( t ) = d ( k T ) , (47)
The only other complication is that since l/Gll = O ( T - l ) ,
the required magnitude of this control may be large. If
F = exp(AT) , G = rB, (48) the bounds U. on uk are taken into account, the following

F = F i-
rAA, r= lT exp(AT)dT, (49)
feedback control has been shown [19] to force the system
into DSM:

G = G i-
F A B , D = FE (50) -[CG]-1CF5k, if 1uk/ < a
uk={ -iisgn(sk), if l t L k l 2 ii (57)
This discrete time model is an O ( T 2 ) approximation of
the exact model which is described by the same F and G A . l DSM Control of the integrator plant
matrices, but because the exogenous disturbance is a con- For the nominal integrator plant with parasitic sensor
tinuous time function, the sample and hold process yields dynamics, we design an DSM controller based on Eqn.(57).
a D matrix which renders the matching condition in the Let the sensor time constant T~ = 0.02, and the control
continuous time plant to be a necessary, but not sufficient magnitude E = 1. The desired closed loop bandwidth is
condition for the exact discrete time model. However, by given to be about one Hz. A good choice of the sampling
adopting the above O ( T 2 )approximated model, it follows frequency would be about 10 Hz (T=0.1) since the sen-
from (50) that, if the continuous time matching condition sor dynamics are of 50 Hz, and therefore can be neglected
(44) is satisfied, the following matching conditions for this initially in the design. The DSM control takes the form of
model hold :

From an engineering design perspective, the O ( T 2 )mod- where x i = z,(kT) is the sampled value of the sensor out-
els are adequate since the between sample behavior of the put z,(t). Note that due to the control bounds, a linear
continuous time plant is also O ( T 2 )close to the values at feedback control law is applied inside a boundary layer of

-8-

Authorized licensed use limited to: Sejong University. Downloaded on November 11, 2009 at 06:45 from IEEE Xplore. Restrictions apply.
thickness 2T about the sliding manifold xi = 0. With- with the sliding manifold given by
out sensor dynamics, the behavior inside the boundary
layer is that of a deadbeat controller. The sensor dynam- sk = c l x i + c 2 X i = 0, (64)
ics impose a third order discrete time system inside this
boundary layer, and its eigenvalues are inside the unit cir- and C2G is nonsingular. By eliminating xz, the reduced
cle at {-0.002,O.l -f 110.436). For reference, the discrete order sliding mode dynamics are O ( T 2 )approximated by
model of the open loop nominal plant and the sensor dy-
X k1+ l =( 4 1 - Fl2C$1)x:. (65)
namics has a pair of double real pole almost at the origin
(4.54x lop5), which result from sampling at a frequency B.l Discrete time disturbance compensation for the inte-
much lower than the sensor's 3dB corner frequency, and grator plant
a pole at unity which is due to the integrator plant. The
third order system response can be seen in Figure 10 where We continue with the DSM control design using the same
the sample values of the error between the constant unity sampling frequency and system parameter values. The con-
troller which takes into the one step delayed disturbance
reference command and the sensor output is plotted. Note
that only the behavior inside the boundary layer is shown, estimates is given by
and it agrees well with the predicted third order behav-
ior. The steady state error magnitude of 0.05 is due to the
constant disturbance d ( t ) = 0.5 as applied to this plant
before, and the effective loop gain being T-l = 10. Fig- Note the PID controller structure of this controller when
ure 11 displays the continuous time error of the plant state the system is inside the boundary layer. Figure 12 shows
and the discrete time error of the sensor output where the the sampled error between the reference command and the
time lag due to the sensor dynamics can be seen during the sensor output. The practically zero steady state error is
transient period. much better than our O ( T 2 )estimate due to the PID con-
B. DSM Control with Delayed Disturbance Compensation troller structure. The one step delayed disturbance esti-
mate is given in Figure 13, showing convergence to the ex-
The earlier DSM control design for nominal plants can pected value. Figure 14 displays the continuous time error
be modified to compensate for unknown disturbances in between the plant state and the reference, and its discrete
the system [37], [38]. From the discrete model in Eqn.(45), time measurements.
the one step delayed unknown disturbance
C. DSM Control with Parameter Uncertainties and Dzs-
a
di-l = D d k - 1 = X k - E X , - , - G U k - 1 , (59) turbances
can be computed, given the measurements x k , x k - 1 and With the presence of system parameter uncertainties, the
u k - 1 , and the nominal system matrices F , G . If we let
above approach using one step delayed disturbance esti-
mates can be still be applied. However the one step delayed
signal contains both delayed state and control values,
a
fk-1 = AFxk-l+AGUk-l+Ddk-1 = Xk-FXk-1-GUk-1 ,
The effectiveness of this controller is demonstrated by ex-
(67)
amining the behavior of the s k when the control signal is where A F = I'AA, AG = rAB. The DSM control is of
not saturated, the same form as Eqn.(60), with d i - l replaced by f k - 1 .
= CD(dk - d k - 1 ) .
The behavior of sk is prescribed by
Sk+l (61)
If the disturbance has bounded first derivatives, Le., 121 5 Sk+l =c(fk - fk-1) =CD(dk - dk-1) +
d < 03, d k - d k - 1 is of O ( T ) ,and from the definition given +CAF(Xk -xk-1) + C A G ( u k -Uk-1). (68)
in Eqn.(50), IlDll = O ( T ) , hence /ski = O ( T 2 ) ,implying
that the motion of the system remains within an O ( T 2 ) Since is bounded, xk - xk-1 is of O ( T ) , and since
neighborhood of the sliding manifold. This controller has llAFll = O ( T ) ,we have
also been shown [19] to force the system into DSM if the
control signal is initially saturated. ~ k + l= C A G ( U-
~u k - 1 ) +O(T2). (69)
On the sliding manifold, the system dynamics are, to
Due to the coupling between Sk and U k , it has been
O ( T 2 ) invariant
, with respect to the unknown disturbance. shown [38], [39] that the behavior outside the sliding man-
Since similar matching conditions exists for the O ( T 2 )dis- ifold is governed by the following second order difference
crete models we have adopted, it follows from continuous equation,
time sliding mode [28] that by using a change of state vari-
ables, the discrete model can be transformed into ~ k + l= - C A G ( C G ) - ' [ 2 s k - ~k-11 + O(T2), (70)
x k1+ l F1lxk + p 1 2 4 , (62) which has poles inside the unit circle for sufficiently small
xi+l = F 2 1 ~ : + F224 + G2Uk + Dzdk , (63) 11 AB 11. The permissible control matrix uncertainties are

-9-
dictated by the above stability condition which determines REFERENCES
the convergence on the sliding manifold. Note that pro- V. I. Utkin, ‘‘ Variable structure systems with sliding modes”,
vided that the parameter uncertainties on the system ma- IEEE R a n s . Automat. Contr., Vol.AC-22, No.2, pp. 212-222,
trix, they do not impact the convergence, nor they affect 1977.
Friedland, B. Advanced Control System Design, Prentice Hall,
the motion on the manifold. Englewood Cliffs, N.J., 1996.
Asada, H. and J-J. E. Slotine, Robot Analysis and Control,
C. 1 Compensation for gain uncertainties in integrator pp. 140-157, John Wiley and Sons, 1986.
Bondarev, A. G., S. A. Bondarev, N. E. Kostyleva, and V. I.
plant Utkin, “Sliding Modes in Systems with Asymptotic State Ob-
servers,’’ Automation and Remote Control, pp. 679-684, 1985.
We shall introduce gain uncertainties in the integrator Young K. D. and U. Ozgiiner, “Frequency Shaping Compensator
plant to examine their effects on the convergence of the Design for Sliding Mode,” Special Issue on Sliding Mode Control,
sliding manifold. The actual plant is given by International Journal of Control, pp. 1005-1019, 1993.
Young, K. D. and S. Drakunov, “Sliding mode control with chat-
tering reduction,” in Proceedings of the 1992 American Control
j. = (1 + 7)U +d(t) (71) Conference, Chicago, Illinois, pp. 1291-1292, June 1992.
Su, W. C., S. V. Drakunov, U. Ozgtinerand K. D. Young, “Sliding
mode with chattering reduction in sampled data systems ”, PTO-
where y represents the gain uncertainty in the integra- ceedangs of the 32nd IEEE Conference on Decision and Control,
tor. The DSM controller in Eqn.(66) can be used again San Antonio, Texas, pp. 2452-2457, December 1993.
because the right hand side of the one step delayed sig- Young, K. D., and S. V. Drakunov,“Discontinnous frequency
shaping compensation for uncertain dynamic systems,” Proceed-
nal is the same regardless of the parametric uncertainties. ings 12th IFAC World Congress, Sydney, Australia, pp. 39-42,
The root locus of the second order system governing the 1993.
motion outside the manifold is plotted in Figure 15 for Young, K. D. (editor), variable Structure Control f o r Robotics
-1 5 y 5 0.34. For y = 1, there is a pair of double poles _.
and Aerosvace Amlications. Elsevier Science Publishers. 1993.
[lo] Zinober, A . S. (editor), Vdriable Structure and Lyapunov Con-
at unity, and for y = 1/3, one of the poles becomes -1. trol, Springer Verlag, London, 1993.
The case for y = -0.5, corresponding to a pole of complex [ll] F.Garofalo and L.Glielmo (editors), Robust Control via Variable
Structure and Lyapunov Techniques, Lecture Notes in Control
pairs -0.5 +~0.5, is simulated with the same reference and and Information Sciences Series, Vol. 217, pp. 87-106, Springer-
disturbance as in the previous studies. Figure 16 shows Verlag, Berlin, Heidelberg, New York, 1996.
the sampled error between the reference command and the 1121 Utkin, V. I., “Variable Structure Systems: Present and Future,”
Avtomatika i Telemechanika, No. 9, pp. 5-25, 1983 (in Russian),
sensor output which converges to zero. Figure 17 displays English Translation, pp. 1105-1119.
the estimates of the exogenous disturbance and the residue [13] DeCarlo, R. A, S. H. Zak, and G. P. Matthews, “Variable struc-
control signal due to the gain uncertainty. The continuous ture control of nonlinear multivariable systems: A tutorial,” PTOC.
of IEEE, Vol. 76, NO. 3, pp. 212-232, 1988.
time error of the plant state and the discrete time error of
[14] Utkin, V. I., “Variable Structure Systems and Sliding Mode -
the sensor output are shown if Figure 18 for comparison. State of the Art Assessment,”, Young, K. D. (editor), Variable
Structure Control for Robotics and Aerospace Applications, pp. 9-
32, Elsevier Science Publishers, 1993.
V. CONCLUSIONS [15] Hung, J . Y., W. B. G a q and J. C Hung, “Variable structure
control: A survey,” IEEE n a n s . Ind. Electron., Vol. 40, No. 1,
We have examined systematically SMC designs which are pp. 2-22, 1993.
firmly anchored in sliding mode for the continuous time do- [16] Flugge-Lutz, I., Discontinuous Automatic Control, Princeton
main. Most of these designs are focused on guaranteeing University Press, 1953.
[17] Tikhonov, A. N.,“Systems of differential equations with a small
the robustness of sliding mode in the presence of practical parameter multiplying derivations,” Mathematicheskii Sbornik,
engineering constraints and realities, such as finite switch- Vol. 73, No. 31, pp. 575-586, 1952 (in Russian).
ing frequency, limited bandwidth actuators, and parasitic [18] Kokotovic, P. V., H. K. Khalil, and J. O’Reiley, Singular pertur-
bation methods in control : analysis and design, Academic Press,
dynamics. Introducing DSM, and restructuring the SMC 1986.
design in a sampled data system framework are both appro- [19] Utkin, V. I. , “Sliding Mode Control in Discrete-Time and
priate, and positive steps in sliding mode control research. Difference Systems,” Variable Stmcture and Lyapunov Control,
A.S.Zinober (editor), Springer Verlag, London, pp.83-103, 1993.
It directly addresses the pivotal microprocessor implemen- [20] Utkin, V. I., Slzding Modes and their applications tn Variable
tation issues; it moves the research in a direction which is Structure Systems, Moscow:MIR, 1978 (translated from Russian).
more sensitive to the concerns of practicing control engi- [21] Young, K-K. D., P. V. Kokotovic, and V. I. Utkin, “A Singular
Perturbation Analysis of High Gain Feedback Systems,”, IEEE
neers who are faced with the dilemma of whether to ignore 7 h n s . Auto. Contr., Vol. AC-22, No. 6, pp. 931-938, 1977.
this whole branch of advanced control methods for fear of [22] Slotine J-J. and S. S. Sastry, “Tracking Control of Nonlinear
the reported implementation difficulties, or to embrace it Systems using Sliding Surfaces with Application t o Robot Ma-
nipulator,” Int. J. Control, Vol. 38, No. 2, pp. 465-492, 1983.
with caution in order to achieve system performance oth- [23] Burton, J. A., A. S. I. Zinober, ‘Continuous approximation of
erwise unattainable. However, as compared with the ideal variable structure control,” Int. J . System Sci., Vol. 17, No. 6,
continuous time sliding mode, we should also be realistic pp. 875-885, 1986.
about the limitations of DSM control designs in reject- [24] Young, K-K. D. and P. V. Kokotovic, “Analysis of Feedback
Loop Interaction with Parasitic Actuators and Sensors,” Auto-
ing disturbances, and i n its a b i l i t y to w i t h s t a n d p a r a m e t e r matzca, Vol 18 , September 1982, pp. 577-582.
variations. The real test for the sliding mode research com- [25] Kwatny, H. G. , K. D.Young, “The Variable Structure Ser-
munity in the near future will be the willingness of control vomechanism,”, Systems and Control Letters, Vol. 1, No. 3,
pp. 184-191; 1981.
engineers to experiment with these SMC design approaches [26] Young, K. D., and V. I. Utkin, “Sliding Mode in Systems with
in their professional practice. Parallel Unmodeled High Frequency Oscillations,” Proceedings of

- 10 -
R m l Locus - Sensor dynamm.tau_s=O.Ol, ez=.w3-->
01
150
the Third IFAC Symposium on Nonlinear Control Systems De-
sign, Tahoe City, California, June 25-28, 1995.
[27] Young, K-K. D. and H. G. Kwatny, “Variable Structure Ser-
vomechanism Design and its Application to Overspeed Protection
Control,” Automatzca, Vo1.18, No. 4, pp. 385-400, 1982.
[28] Utkin, V. I., Sliding Modes in Control Optimizatzon, Springer-
Verlag 1992.
[29] Slotine, J-J. E., J. K. Hedricks, and E. A. Misawa, “On Sliding
Observers for Nonlinear Systems,” ASME J. Dynamic Systems,
Measurement and Control, Vol. 109, pp. 245-252, 1987.
[30] Xu, J-X., H. Hashimoto and F. Harashima, “On the design of a
VSS Observer for Nonlinear Systems,” ’ h n s . of the Society of In-
strument and Control Engineers (SICE), Vol. 25, No. 2, pp. 211-
217, 1989.
[31] Korondi, P., H. Hashimoto, K. D. Young, “Discretetime Sliding
t 1
Mode Based Feedback Compensation for Motion Control,” Pro-
ceedings of Power Electronics and Motion Control (PEMC’SG),
Budapest, Hungary, ,Sept. 2-4, 1996, Vo1.2, pp. 21244-2/248,1996.
[32] Young K . D., U. Ozgiiner, and J-X. Xu, “Variable Structure
Control of Flexible Manipulators,” Young, K. D. (editor), Vari- -’250
1 -2w
-150 -lw
Real
-50 50

able Structure Control for Robotzcs and Aerospace Applications,


pp. 247-277, Elsevier Science Publishers, 1993.
[33] Gutman, S. and Leitmann, G., “Stabilizing Feedback Control Figure 1: Root locus of boundary layer control, crossover
for Dynamic Systems with Bounded Uncertainties,” Proceedings gain g,=200.
of IEEE Conference on Decision and Contro1,pp. 94-99, 1976.
Sensor dynamics, tau-& 01, layer ez=O 01,O035, dis3urbbance;O 5
[34] Drakunov S. V. and V. I. Utkin, “Sliding mode in dynamic sys-
tems,” International Journal of Control, Vo1.55, pp. 1029-1037,
1990.
[35] Drazenovic, B., “The invariance conditions in variable structure
0 -008
- I -I
systems,” Automatica, Vo1.5, No. 3, pp. 287-295, 1969. 0006 i 1
[36] Furuta, K., “Sliding mode control of a discrete system,” Systems
and Control Letters,,,Vol.l4, pp. 145-152, 1990.
[37] R. G. Morgan and U. Ozgiiner, “A Decentralized Variable Struc-
ture Control Algorithm for Robotic Manipulators, IEEE Journal
of Robotics and Automation, 1, 1, pp. 57-65, 1985.
[38] W-C. Su, S. V. Drakunov and U. Ozgiiner, Sliding Mode Control
in Discrete Time Linear Systems, Prepnnts of IFAC 12th World
Ow
0 w2
l

0
I
4 002
Congress, Sydney, Australia, 1993.
[39] Su, W. C., S. V. Drakunov, U. Ozgiiner, “Implementation 4004
of Variable Structure Control for Sampled-Data Systems,” Ro-
bust Control via Variable Structure and Lyapunov Techniques,
F.Garofalo and L.Glielmo (editors), Lecture Notes in Control
and Information Sciences Series, Vol. 217, pp. 87-106, Springer- 4)w8 -
Verlag, Berlin, Heidelberg, New York, 1996.

Figure 3: Block diagram of Observer based Sliding Mode


Control.

- 11 -

Authorized licensed use limited to: Sejong University. Downloaded on November 11, 2009 at 06:45 from IEEE Xplore. Restrictions apply.
observer based SMC, ez-2a-3, tau_s.Ol. h-10. d=O 5
0 01 SM DisRlrbance estimator - tau-= 01 tau-fdlO2. d=O 5
08, I
0.008 - . . . . . ~. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

OW6-.. . . . . . . . . . . . . . . . . . ' . . .:... I . . . . . . . . . . . . .

o m- ........................ . . . . . . . . . . . .

o.w2-. . . . . . . . . . . . . . . . . . . :.. . . . . . . . . . . . . . .:. ..-

.: . . . . ..:..
g 0-..
b-
. . . . . . . . :. ...; ..... ...;.. .-

4.032 - ........................................ .-

o m - .............................. .: . . . . . . . .: .

4.006-. . . . . . . . . ~ . . . . . . . . . . . . . . . . . . . . . .
o@yj . . . . . . . . . . . . . :. . . . . . . . . . . . . . :. ... .-

1 I I
-001
0 0.5 1 1.5 2 2.5 3 35 4

Figure 4: Observer based SMC: error between reference


Figure 7: SM Disturbance estimator Control: distur-
command and observer state.
bance estimntp.
observer based SMC, ezSa-3, tau_* 01, h=10.d a . 5
11, Limited bandwidthactuatw with SM dlstukma, esm"ar

1.08 .. - 1 . . ..A.. .. L ........... L . . . > ........ .-

{
r
(n

Em
1.06 . . . . . . . . . . . . .

-m j w . . . . ... '

1.02 . . . . . .

1.
.1 . . . . . :
....;......

.....,......,.....

....
...........................

I .... I ..................

;. . . . . . . . . . . . . ;. . . . . I
. . . . . . . . .'. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
... :. . . . . .
08-...

0,6-.... \ i ! : : ! *
...........................

............. ..............................
L . ..: . . . . 1.. .

m
go.98. . . . . . . : . . . . . : . . . ...... ......
. L I .... 1 ....... :.

5
iiOW-.. :. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .....

0.94. ..... ., . . . . . ................................... .

o,g2-.... ......................... .:..... ...............

1
020 0.5 1 1.5 2 25 3 3.5 1
Qme,sBc

1,
Limitsd bandwlmh actuator wlth SM disturbance esbmator
o.g-. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . , .. 0.6

08 ................ ................................

0 4 - . .........................................................

EY
80,3-. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ........

2
3
*o,2- ................. .:. . . . . . .;. . . . . . .:.. . . . .:. . . . . . .:. . . . . . .

01 ...... .:. . . . -1.. ... -1. ..... .:.. . . . . .:. . . . . . . . . . . . . . . . .

O O B 4
bme,rec

oi 015 li5 d 215 A 3'5 4

-12-

Authorized licensed use limited to: Sejong University. Downloaded on November 11, 2009 at 06:45 from IEEE Xplore. Restrictions apply.
Discretetime SMC - tau_% 01,d=O 5

06-

l05-

i
EO4- U

4.4
1 :Ir;
I]
........................... ..,. .................

........
z
303-

OlCI
. . . ................................

0.5 1 15 ,:* 2.5 3 3.5


I
OO 5 10
l5
m
umesteps
a 3 0 3 5

tween reference command and sensor output. one step delaved disturbance estimate
02, ,, , Dlruete,!imeSMC;lau_s=Ol d=O5
Discreteb n m SMC wilh disturbanceemnmbon tau-* 01 d=O 5

O15t \.
go1t I 01

O1

4 15

"
0
t
I
05 1 15 bmesec
2 25 3 35 42
0 05 1
,

15
,

2
tlme sec
L

25
,

3 35
,I 4

Figure 11: DSM Control for nominal plant: continuous Figure 14: DSM Control with disturbance compensation:
time and discrete time error responses. continuous time and discrete time error resDonses.

:+ I
D"te lime SMC wlm distulrbanceBstimabon tau-- 01 d=O 5

-!
008

006

004

\ Iilji #
0.8

06

0.4
...........................................

....

...
. . . . . .

....... ....
!+i
. . . . . .

+
+ + + + +
......
. . . . . . . . . . .

I
..+ . . . . . . . . . .

1;
~

+ .

1
0 02

$ 0 g o + + + + + +*
4 02
P
E42

4 4 . ..... . , - + + + . . .+; - +. + ++ . + .,...

4 06 4 6 .. .... ........... ..... ...... -. ........

48 .... ., . . . . . , ... ..... . . . . . . . . . . .

1
1 4.5 0 0.5 1
Real part

Figure 12: DSM Control with disturbance compensation: Figure 15: DSM control with control parameter varia-
error between reference command and sensor output.
tions: root Locus for evaluating sliding manifold conver-
gence.

- 13 -

Authorized licensed use limited to: Sejong University. Downloaded on November 11, 2009 at 06:45 from IEEE Xplore. Restrictions apply.
Discretehme SMC with paramefetuncemnbes and dlslurbacne esbmabon
1

4.2
0 5 1 0 1 5 2 0 2 5 3 0 3 5
hmes step

Figure 16: DSM Control with control parameter varia-


tions: error between reference command and sensor output.

Discrere bme SMC wlm parametet unca"es and dishlrbacne emmaton

0.4 - In, --
L U
0.2 -

0
5 IO 15 x) 25 30 35 1
hmes step

Figure 17: DSM Control with disturbance compensation:


one step delayed parameter and disturbance estimate..
Disnete bme SMC with parmeter uncerfamnbes and disturbacneemmaton
1 , , I

Figure 18: DSM control with control parameter varia-


tions: continuous time and discrete time error responses.

-14-

Authorized licensed use limited to: Sejong University. Downloaded on November 11, 2009 at 06:45 from IEEE Xplore. Restrictions apply.

You might also like