3146)/1-38/(1977). Wieslander, J . (1979a): Interaction in computer sysaidedanalysis and design control of tems. PhD thesis. DeptofAutomaticControl.

Lund Institute of Technology. Lund, Sweden. Report CODEN: LUTFDZ/(TFRT1019)/1-222/(1979). Weislander. J . (1979b): Designprinciplesfor computer aided design software. Preprints, IFAC Symposium on CAD of Control Systems. Zurich. 393. Wieslander. J . (1980a): Interactive programGeneral guide. Dept of Automatic Control. Lund Institute of Technology, Lund. Sweden. Report CODEN: LUTFD?/(TFRT3156)/1-30/(1980). Wieslander. J. (1980b): IDPAC commandsUser's guide. Dept of AutomaticControl. Lund Institute of Technology. Lund. Swe-

den, Report CODEN: LUTFD?/(TFRT3157)/1-108/(1980). conman&Wieslander, J . (1980~):MODPAC User'sguide. Dept Automatic of Control, Lund Institute of Technology. Lund. Sweden. Report LUTFDZRTFRTCODEN: 3158)/1-81/(1980). Wieslander. J . (1980d):SYNPACcommandsUser'sguide. Dept Automatic of Control. Lund Institute of Technology. Lund, Sweden. CODEN: Report LUTFD?/(TFRT3159)/1-130/(1980). Wieslander. J. and Elmqvist, H . (1978): INTRAC, A communicationmodule for interactiveprograms.Language manual. Dept of Automatic Control. Lund Institute of Technology, Lund. Sweden.ReportCODEN: LUTFD2/(TFRT-3149)/1-60/(1978) Wilkinson. J . H. and Reinsch. C. (1971):Linear Algebra. Springer-Verlag. Berlin.

Winston. P. H . and Horn, B. K. P. (1981): LISP. Addison-Wesley, Reading, Mass.
Karl Johen Astrom was born in Ostersund, Saeden on August 5 , 1934. He was educated at the Royal Institute ofTechnology (KTH) in Stockholm, Sweden. Heheld has various teaching positions at KTH,and he has worked for the Research Institute of Swedish Defense, and for IBM. In 1965hewas appointed to the chair of Automatic Control at Lund Institute of Technology (LTH). Hismainresearchinterests are stochastic control theory, identification, adaptive control, and computer aided design of control systems.Astromhasalso worked with industrial applications of of inertial automatic control in the fields guidance, paper mills, flight control, and ship steering. Apart from his professional work he and his family enjoy skiing and sailing.

A Simple Adaptive Smith-Predictor for Controlling Time-Delay Systems
A Tutorial, by A. Terry Bahill
Biomedical Engineering Program, Department Electrical Engineering, of Carnegie-MellonUniversity, Pittsburgh, PA 15213
ABSTRACT: This heuristic paper presents several simple techniques for analyzing the f stability o time-delay systems. It explains the Smith predictor control scheme for time-delay systems and shows how errors in modeling the plant parameters can causeinstability. Then anadaptiveconto the Smith predictor troller is added system; this pedagogical example offers a complete derivation of a simple adaptive control system. Finally, a new control scheme is discussed that allows zerolatency tracking of predictable targets by a time-delay system.
new system has good performance characteristics, but it tracks input signals with a timedelay. In some circumstances is it possible to design time-delay systems that trackpredictabletargets with nolatency 171, 181. The examples of this paper treat timedelay systems, the Smith predictor, and an adaptive control system. The examplesare complete and the derivations are explicit; no steps are omitted. Many research papers discuss adaptive control systems, but most of them are too complicated for the novice to understand; fewtextbookshaveincorporated simple examples of adaptive control systems. One purpose of this paper is tofillthisgap.Thispapershowssome simple techniques that can be used to gain insight about time-delay systems, explains theSmithpredictorcontrolscheme, and presents a complete, but simple, example of an adaptive control system. times (such as those required for making a chemical composition analysis, cortical processing of a visual image, analyzing a TV picture by a robot, or evaluating the output of a digital control algorithm) [I], [3], [7-lo]. Most elementary control theory textbooks slighttime-delay systems, because they are more difficult to analyze and design. For example, in time-delay systems initial conditions must be specified for the whole interval from -0 to 0, where 0 is the timedelay.For simplicity, in this paper I only discuss steady-state behavior, or equivalently I assume the initial conditions are zero. A unity-feedback, closed-loop control system with KGH = K/(Ts+ 1) has a transfer function of

Introduction
If a time delay is introduced into a well

tuned system, the gain must be reduced to
maintain stability [l]. The Smith predictor control scheme can help overcome this limitation and allow larger gains [2], but it is critical thatthemodelparameters exactly match the plant parameters [3-51. An adaptive control system [6] can be added to theSmithpredictor to change themodel parameters, so that they continually match thechanging plant parameters [3]. This ReceivedJune 22. 1982:revisedOctober 25, 1982:revisedJanuary 18, 1983. Accepted in revised form by Technical Associate Editor
F. Aminzadeh.

Y(S) -- -

K
7 s -I- 1

R(s)

fK

Why time-delay are plicated?

sysrems more com-

Time delays occur frequently in chemical, biological, mechanical, and electronic systems.Theyareassociated with travel times (as of fluids in a chemical process, hormones in the blood stream, shock waves in the earth, electromagnetic or radiation in space), or with computation
0272-1 70818310500-0016$1.0001983IEEE

This is stable for - 1 < K.If a time delay of the form e-@ is introduced in the forward path, stability is no longer guaranteed. The transfer function of such a system is Y(S> -Ke-Se (1)

R(s) -(TS f 1 f KeCSe)

The stability limits are not obvious. The does not exponential in the numerator

1 6

control systems magazine
Authorized licensed use limited to: The University of Arizona. Downloaded on July 09,2010 at 17:46:05 UTC from IEEE Xplore. Restrictions apply.

and Y. Y(s) -.K0)s 7 R(s) -E (K -. The Taylor series approximations err on the safeside. r .so) -( . this cannot be done in a physical system.I ) I -1 < K < V' + (2r/eI2.R(s) 0 r s 2 + ( 2 r + 0 . they do allow us to make the following generalization. because the time delay is probably distributed-not lumped-and there is no a priori reason to place the timedelay after the plant dynamics rather than before it. 8 : issmall compared to the plant time constant. The Pad$: approximation could indicate stability for some unstable systems. = Kp/(rps 1) and T.L(s). then we can derive a single pole solution. I and for 8 By the Routh-Hurwitz criterion this transfer function is stable if all denominator the coefficients are positive. Bya digital controlengineer. Large gains can only be used in timedelay systems if the plant time delay. Using the Pade approximation in equation (1) produces a third transfer function. Y. C represents the controller. good methods for assessingthe stability of asystem.The following four approxima.1 < K < 1 + 2718.bother us. (2) 2. The approximation methods shown here are not? in general. we can use the Nyquist criterion [ 111 to derive stability constraints for equation (1). TheSmith predictor algorithm [2] avoids this reduction of gain and consequent poorer performance.(s). Since there would be no delay in the feedback signal. Thetransfer function becomes All denominator coefficients will be positive and the system will be stable if . may 1983 Authorized licensed use limited to: The University of Arizona.st?. Of course. 33 of [3]. This should be expected because they are approximations. These constraints depend on the relationship of 8 to 7: for 8 < < r stability requires -1 < K < v' 1 + (1. we could connect it to the controller. 1. would be the same as the signal B after a delay of 8. However. If 8 is small so that se << 1 for frequencies of interest. For a simple first-order plant with a pure time delay G. The following development of the Smith predictor algorithm is based on Despande andAsh [I].-se And by aclassicalcontrol engineer: = 1 (1 + s0/n)" where n is a large number. Block diagram for a typical time-delay system. 1. represents the plant time delay. For the sake of comparison. As shown in Fig.it will be left undisturbed. T. and nh = 8.K0)s + 2(K 1) All denominator coefficients will be positive and the system will be stable if 0 < K < r/8. represents plant dynamics. The following two-pole solution can be derived by substituting the first three terms of ( 2 ) into (1): >> r stability requires + -1 <K < + (nT/e)2. 2.2010 at 17:46:05 UTC from IEEE Xplore. This implies . in equation (1). the process can be conceptually split into delay free system dynamics and a pure time delay. By mathematician.57~/e)'. Restrictions apply. I will use theshorthand notation of Marshall [3]: R represents the system input. 1. Y S) K(2 . C(s). The introduction of a time delay makes it moredifficult to access the stability of a system. T. This would move the time delay outside the control The loop.(s). I f the ficticiousvariable B could be measured. Pade approximation: the lL Fig. by using e-@ = 1 .-so z- -2 - -nh fictitious variable where h is the system sampling period. n is an integer. The first technique implies that the original system has an infinite number of poles that can be reduced by using an approximation. signal Y. for 8 = r stabilityrequires -. G. 4.K 8 ) s + 2 ( K 1 ) + 3. 1. 17 I / . Downloaded on July 09. The block diagram for conventional control of a time-delay system is shown in Fig. a series expansion: the Taylor Fig. These three approximations yield different stability limits. the gain would have to be reduced to maintain stability. The exponential in the denominator will be approximated by an algebraic expression.Sometimes they yield bizzareresults.the transform equivalent: . therefore. = e -S*P. For simplicity. represents the plant output.(s). R(s). tion techniques have been suggested. . L represents potential load disturbances.1 < K < 710.W S ) Ke-Se ( .as shown on p.R(s) KO's' 2KePse f 2(7. The Smith Predictor If a time delay were introduced into an optimally tuned system. 2. as shown in Fig. G. System performance be could improved the if B could be fed back instead of the output Y. the response of the system would be improved. Byprocess a control engineer..

s G . 4. represents the model of the plant dynamics. To compensate for these errors a second feedback loop is implemented using E. = e Althoughthefictitious variable B is unavailable. Preliminary form of the Smith predictor.. The control of the system output.For The dashed box labeled G. as shown in Fig. This arrangement controls the model well. and E represents the error between the output of the model and the output of the plant. = G . but not the overall system. in Fig. -CG./(T. G . PD (Proportional plus Derivative). can be used as the feedback signal. the system tracks disturbances with a time delay. is open loop: it will not accommodate either load disturbances or inaccurate models. 4. Therefore. Sometimes the Smith predictor is drawn as shown in Fig. = -1. where 0.GpTp [ 1 + GmC(11 +CG. ) ] (6) Once again there is no time delay in the denominator. we will not consider load disturbances again in this paper. (5) The effects of thetime delay have been removed from the denominator of the transfer function. IL I I I Fig. and the system performance has been improved. and T. T. = T. for L = 0.To improve the design let us model the plant as shown in Fig.However. isstable. or . However. the transfer function between load disturbance and system output becomes ~~ simplicity. this reduces to CGpTp Yp(s) 1 R(s) + CG. The controller C is a conventionru PI (Proportional plus Integral). Hz. or PID (Proportional plus Integral plus Derivative) controller. is the magnitude of the planttimedelayin seconds. 5. Downloaded on July 09. T. The closed-loop transfer function of this system. is Yp(s) -.Tp (4) If G .T. Thesystem has dynamic poor response unless load disturbancesare restricted to frequencies below 2/0. it tracks input variations with a time delay. R(s) + CG. calIed the Smith predictor controller. is a stable closed-loop feedbackcontrol system with a transfer function of ” The condition for instability is CG.2010 at 17:46:05 UTC from IEEE Xplore. 4. represents the model of the plant timedelay. What happens if the model does not match the plant exactly? Following the development of Marshall [3]. let us set control systems 1 8 magazine Authorized licensed use limited to: The University of Arizona. the result may be unstable. I iI ____~ ~ The complete Smith predictor control scheme.. + CG. Stability of the Smith Predictor Control System T ..CGPTP 1 lL I I Fig. Restrictions apply. 3. . = K.s + 1) and T. when this controller is used in a closed-loop system. as shown in Fig. 3. B.- Clearly the magnitude the of righthand side is never one. 3. CG. Assuming perfect model matching. For the previous exampleof a f i t order pcg. Y. the controller G.which can be tuned more closely because the effect of the time delay in the feedback loop has been minimized. which is equivalent to Fig. However. This is the Smithpredictorcontrol strategy. 5.

The magnitudeof the time delay affects the controller via the A G term in Fig. and plant and model time constants of1 sec. Alternatively.5s + I ) . It is not agoodgeneral technique for analyzing stability of Smith predictor systems.G. Normally we cannot do this because applying a step input to an operating plant woulddisturb the process we wanted to control (this would become annoying if we did it every second). However.ATp = TAT Thissimplified notation gets rid of the subscripts and allows us to redraw Fig. This description is close enough to that of the Smith predictor to makethis type of adaptive control applicable.(s) (6s’ -. = I/(s + l ) . G. The gain k must be selectedforeachapplication. - ‘ I 1A T l?G Fig. = .~ ( O ~ + A B . oceancurrents. which is composed of a time-varying plant with adjustable parameters. 6 becomes zero. we desire a generalized on-line method of adapting the model time delay.o * l s - J = 0. An Adaptive Smith Predictor Control System If is oftenadvantageous to change the controller to compensate forchanges in plant parameters caused by age. although it has been shown tobe unstable by the techniques of [4]. One easily understoodscheme is model the reference adaptive control scheme discussed by Landau [ 6 ] . Suppose that model the of the plant dynamics isin error. The performance function is + 4)Tp 3s 4. and Fletcher-Powell. 6. .5 4. Let the magnitude of the plant time delay be represented by 0 . = T.errors in estimatingtheplant gain or timeconstants could cause instabilities. e = (y. temperature.fatigue. Based on [3].R(s) + 14s 4. 4. 5 as Fig. The Pad. and design a system that will automatically change the model time delay. .. then the inner positive feedback loop of Fig. Restrictions apply. 6. = l/(s + I). Its use indicates stability for manyunstable systems. and a model time delay of 0. The time delay in this feedback loop could cause instability. if we use the Pade approximation on equation (4). e.2010 at 17:46:05 UTC from IEEE Xplore. = -kVJ. we can study the transfer function.86s 4. -36s 1 s2 This produces + 22s 4.to see the effects of a mismatch between model and plant parameters. TWO major items must be chosen before an adaptive controller can be designed: a performance criterion a and method of function minimization.=Gp+AGp=G+AG T . monetary cost: error. but the model of the plant time delay is correct. conjugate gradient. The techniques of Ioannides.40 Y. Downloaded on July 09. . Typical function minimization techniques include gradient. and AT = 1.e . The adaptive control scheme Combining examples by Landau [6] and Marshall (31. for our Smith predictor we wish to modify the model time delay so that it tracks the plant timedelay thus minimizingtheperformance us function.5s + I). In the model reference approach we would now change-gains in the controlled system to minimize the performance function. changes 1 9 Authorized licensed use limited to: The University of Arizona. although 8. If there is an error in modeling only the time delay so that A G = 0. to change continuously. = 1/(3s + l). Let us use a simple plant. There are many different types of adaptive control schemes [ 121. so that T.In one form of this scheme the controlled system.S’ -I.1 sec. This is a positive feedback loop T of AG.Theparameters ofthecontrolled system are varied to make the controlled system behave like the parallel model [ 131. [ 141. using the gradient method of minimization. C = 4(0. Typical performance criteria are minimization of time. and assume its initial value. [5]..-s . let us minimize the square error between the model output and plant output. approximation was used here for pedagogical reasons. Therefore. so that AG is non zero.isin parallel with awelldefined model. ) = e-sepe-sAep = T. has been changed by a small amount called AO. Let us presume that the plant time delay changes.5 This transfer function appears to be stable by use of the Pad$: approximation and the Routh-Hurwitzcriterion. The gradient method tells AO. A rearrangement predictor the of Smith model and plant mismatches.24s2 + 84s f 80) __3s3 .or square error. a PD controller. equation (4). With only one parameter the gradient is simplythe partial derivative. energy.4)Tp s2 + 18s 4. wear. We are leftwith a simple closed-loop feedback control system with feedback of (1 -AT).y.).Let us try to do this for the Smith predictor of Fig.etc. Therefore.200 R(s) The negative coefficient in the denominator shows thatthis is unstable. From this wecould compute the plant time delay..disease. illustrating effects of Y. = G. we get may 1983 Themathematicsbecome complicated if we allow both OP and 8. aplant time delay of 1 sec.(s) 2Tp(s3 -I. Therefore. The easiest way to do this would be to apply a step input and measure the output. For example.5 e2 dt. C = 4(0.T m ) (2s Using the Pade approximation for the exponentials yields T -T P m = .(2s + 4)(Tp . 6. G . with G .Rogers andLatham [4] or Palmor [5] are much more comprehensive.

we stillhave to evaluate the constant k in (13). We will let the identification and control phases alternate. -cG. It has been assumed the that Laplace transformofay. = AO. =-kg P 5_ a ae. of generating the sensitivity function ay. will now be omitted. . 7.)R (1 + CG. Method on 131./dO. and the complex frequency argument.. Now. we will have to use G . can easily be computed.=- - ae2 2 ae.CG. Downloaded on July 09./dB. in the lower Smith predictor are supposed to contain the plant values.T. . called a sensitivity function.(s)/ae.: e -dt aJ Ad. we can compute the desired change in the model time delay as AO. AO.5 e2dt Using (14) twice yields e. e.. and T.CG. For a small change inthe time delay.(t)/a8. but they are not available. we calculate the partial derivative If theadaptation is slow. l2 Substituting this and (12) into (15) yields =-k-k ae dt 0.. and T. I Now. . s . = -k I . onlyat discrete intervals.T.) Therefore.. Based Authorized licensed use limited to: The University of Arizona. with a digital computer it could be' approximatedover any givenfrequency range. The boxes labeled G. Therefore. dt dJ a YP ae P - The error. A 0 can be removed from the integral and Finally. I I 1 Fig. Substituting (12) into (11) we get AO.Tp)( 1 4. dt =-k I ::a e- dt But now how do we get aelae. we will change 8.. T.ges is Weare pleased with this result because the 6y.? To a first approximation y. First we identify changes in 6.. Restrictions apply.lae.continuously.(-sCG. ae. (For each specific application this should be verified by simulation as was done by Marshall [3]. However. and update them as often as possible. From equation (4) we have The pure derivative function shown in this figure would not be physically realizable analog with components.2010 at 17:46:05 UTC from IEEE Xplore. usingthedefinition of the differential and assuming the initial value of this error is zero (so that A e = e). is equal to au.'e can interchange the order of differentiation . is a function of theplant time delay. By the chain rule Figure 7 shows one method for producing dy. + CG. is not a function of e.. Computing the gain k d NOW. = -k The rate at which plant time delay char. dt Using Equation (9) Forsimplicitythetransferfunction has been labeled M(s). we get . ae. From equations ( 1 1 ) and (13) we have = -AO. = /6.

(t) and the closed-loop transfer function becomes VS) - Zero-latency Tracking In Time-delay Systems In some tracking systems target the position can be predicted. Fig. the usual case. 181. indicates the vector transpose operation. kT (the 1 x n feedbackcontrolvector).A).(s) to equation (18) produces: . Downloaded on July 09. the reference source.A)+ hT(sI.aswell as on the timedelayanddynamics of theplant.Theothersymbols represent Y(s) (thescalaroutput). INVERT Fig. + R. It was stable and had satisfactory dynamicswhen the initial model time delay was wrong by as much as 8 0 8 .bKe-. For most linear systems there is a maximum value of gain that will ensure stability.(t) is turned off. there are ways of applying optimal control techniques to the Smith predictor scheme [3]. There are ways of dealing with nonlinear timedelay systems [15]. The feedmust be back vector kT and the gain K selected to achieve stability.A)-'bK may 7 9 8 3 Authorized licensed use limited to: The University of Arizona. composed is of two parts. rs(t). The system input. Ri(s) (thescalarsysteminput). Applying the requirement Y(s) = R. system with a time delay in the forward path. 8.1 1 Rs [ hT(sI . K (the scalar gain). ri(t). [16]. the a then gain must be reduced. and there are ways of making certain systems track targets with zero-latency [ 7 ] .However. [8]. and b (the n x 1 input coefficient vector). to be yield If the ay. I SELECTIVE CONTROLLER y : : E : : " 1I 7 r b I rig. Marshall [3] has simulated the system of Fig. Thisadaptationsignalisappliedtothe time-delay system. Instead. I (then x n identity matrix). it would have to be calculated on line. Based on [3]. one computes an adaptation signal that depends on thetargetmovement. is known. A (the n x n system matrix). then this constant could be precomputed.) e'' 1 (19) Solving for R. r. An adaptive controller on a Smith predictor. LI . yields: Ra = kT(d . 9.' The e-'* term in the numerator is a pure timedelaythatremains in spite of the feedback. The target-selective adaptive control scheme. and the adaptive signal. ri(t) then equals r. For zero-latency tracking output the must be identically equal to the reference input: y(t) = rs(t). Otherwise. Thedimensions of thevectors and matrices are such that the numerator and the denominator of(1 8) are scalars. When rs(t) is not a known target waveform. Superscript T (R. Figure 9 shows this scheme applied a to simple state-variable feedback control - hT(sI . If a time delay is introduced into suchsystem.' %=[ 1 kT(sI -A)- + ' 'bKe-. The e-" term in the denominator produces the phase lag that reduces theallowablegain. I will discuss only the last of these techniques. Thesystemadjusted its model time delay to within 10% c f the correct value in one time delay period. 7) on both a hybrid computer and on a digital computer. r.A)-'b'b . do this one does not simplypredictthe futuretargetpositionandfeedthisinto the time-delay system.2010 at 17:46:05 UTC from IEEE Xplore. and settled at the correct value in two periods.then both sides can divided by kA0. 8 (which encompasses the system of Fig. and it is posTo sible to produce zero-latency tracking. the system still tracks targets with a time delay. and it allows zerolatency tracking [7]. hT (the 1 x n output coefficient vector). Restrictions apply. 8 shows this implementation./&3.A)-l bKe-S' 1 kT(sI--A)-' + bKe-Se (18) hT(sI.(t). Use of a Smith predictor controller will allowthe gain to be restored to itsoriginalvalue. Recently many embellishments timefor delay systems have been reported. This now gives complete the algorithmforchangingthemodeltime delay.

A. 1981.S. SMC-13. California. Human zero-latency tracking of a cubic target waveform. pp. The human oculomotor systemhas a 150 msec time delay.E: degree in electrical engineering from San Jose State Univesity. J . The top trace shows the target position (dotted) andtheeye position (solid). Stevenage United Kingdom: Peter Peregrinus Ltd. D.1981. it requires an exact matchingof model and plant parameters.1981. vol..D. C . 269-278. vol. head and eye coordination. a human can track a predictable target. Mendel. 563. Ash.. [8]. target amplitude. 1981). C. model A was built to help explain how the human couldovercomesuchatimedelayand track with no latency. pp. K . vol. E. and the Ph. AC2 6 .” in Proc Int Conf Cxbern Soc.2010 at 17:46:05 UTC from IEEE Xplore. pp.Hisresearchinterests include control theory. “A design of [I51 M. J . [I ] F.“ IEEE Trans Aurom Control. 10. such as one the shown in Fig.“The smooth pursuit eye movement system uses anadaptivecontrollertotrackpredictable targets. When a target starts moving there is a 150 msec delay before the eye starts moving. Elements of Computer Process Control With Advanced Control Applications. h o m . “Stabilitycriterion of linearcontrolsystems with time delay.pp. 32. the b1. [6] I. AI-Tikriti. McDonald. .” Meas Control. p p . He is an Associate Professor of Electrical and Biomedical Engineering at Carnegie-Mellon University and also an Adjunct Associate Professor of Neurology at the Universitv of PittsburghMedicalSchool.S. D. “A and Ito.’‘ Inr J Control. [I61 K. modeling physiological systems. The time axis is labeled in seconds. 241-254. Ioannides. plant time delay.targetfrequencyandtargetwaveform.E. then ra(t) can be computed. N. However. 753762. degree in electrical engineering from the University of Arizona. 1975. [3] J . 1979. McDonald.C. Research Triangle Park NC: Instrument Society of America. and computer text and data processing. and Clinical Engineering (Prentice-Hall. J. California. N-Nagy and M. 1979. 1980. “Closer control of loops Chemical Engineering with time. BahillandM. He served as a Lieutenant in the U. 86-87.Thiscontrolscheme has onlybeen studied for single-input single-output systemswithscalarKandonlyonetime delay. 1141 . An adaptive control loop added to the Smith predictor can automatically adjust model parameters match time to the varying plant parameters. the eye continues to follow the predicted target for about 150 msec. kT. 1957. Latham. ‘without latency or phase lag. andthevectors Fig. Marshall. “A novel approach to seismic signal processing andmodeling. Deshpande and R.1979.21. [I31 J. 1981. Tucson. Palmor. Berkeley. for an optical “Adaptive compensations trackingtelescope. in press. B. autopilotsfortankers. Sysrems. T. Gilbart and G. and hT are known. W .New York: IEEE. He is the author of Bioengineering: Biomedical. pp. A . Thorell. “Stability properties of [5] Z . the matrix b . G . The model required knowledge about plant dynamics. A .pp. G . [4] A. 1398-1414. E.C. Smith dead-time compensator controllers. Conclusion To maintain stability of a control system after a time delay is introduced. 1980. vol. When the target stops. 1974. One reason studying zerofor such latencytracking is thatthehumaneye movementcontrolsystemseemsto use target-selective adaptive control. vol. certain In circumstances controllers canbe designed to track targets with no latency in spite of plant time delays.E. 1979. D.If the time delay 0. Adaptive Control: The New York: Model Reference Approach. 217-219. the and bottom trace shows target velocity (solid) and eye velocity(dotted). New York: Academic Press. Bahilland J . vol. The target position waveform was that of a repeated cubic segment. 125-131. Pennsylvania. Rogers and V . F. M. 10.Thetargetmoved f5 degrees. 22 control systems magazine Authorized licensed use limited to: The University of Arizona. in 1970. [9]M.S. Applications of Adaptive Control. “Real time control of a robot with a mobile camera.” Int J Control. Restrictions apply. J.pp. vol.46. [SI J . Control of Time-Dela). Smith. S..“ Geophvsics. J .” IEEE Trans Svst Man Cybern. Kallstrom. Eriksson and L. J. T .10. on January 31. vol. Sten. three-stage non-linear systemstimecontrol with delay.Thisresearchwaspartiallysupported National by Science Foundation grant ECS-8121259. Jamshidi.TheSmithpredictor algorithm larger allows gains. ”Adaptive Automatica. Narendra R. Landau. 227-250. vol. compensating for both the time delay and the plant dynamics. Habibi-Ashrafi. 1946. March 1979: Washington D. Kallman and two perceptivereviewersforconstructivecriticisms. Medical.” 15. Winston. pp. Friedman. Kormylo. vol. N.” in 9th International Symposium on Industrial Robots. processmodelcontrolforlinearsystemswithdelay. [7]A. and V . resulting The system still tracks targets with a time delay. Marcel Dekker Inc. 57-70.E. Mare Island. 1983. J. in 1967. 937-949. Navy teaching mathematics and electrical engineering for four years to the students of the Navy Nuclear Power School.” dead Progress.pp.29. Downloaded on July 09.” Automatica. Acknowledgements I thank Jeffery S. “An adaptive control model for human and movements head eye while walking.“ lnt J Control. pp. J . Terry Bahill (S’66-M’68-Sh4’81) i n o r ns b wa Washington. Aminzadeh. Agin. 3. in 1975. pp. San Jose. However. 1979. References [I] P. The model performed as well as the human [7]. [ 101 G .“Stabilitylimits of aSmithcontroller in simple systems containing a time 557delay. the gain K. and if r. B. 1970. the gainmustbereduced. in electrical engineering and computer science fromthetiniversity of California. 1261-1269. [2] 0 . H. L. Monopoli. He received the B.(t) can be estimated. Lee and F. S . 53(5). Watanabe M. The output can be made equal to the input. [I21 K.

Sign up to vote on this title
UsefulNot useful