This action might not be possible to undo. Are you sure you want to continue?

# 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.

it will be left undisturbed. 4. By mathematician.2010 at 17:46:05 UTC from IEEE Xplore.-se And by aclassicalcontrol engineer: = 1 (1 + s0/n)" where n is a large number. Block diagram for a typical time-delay system. I and for 8 By the Routh-Hurwitz criterion this transfer function is stable if all denominator the coefficients are positive. and Y.bother us.so) -( . r .1 < K < 1 + 2718. would be the same as the signal B after a delay of 8. may 1983 Authorized licensed use limited to: The University of Arizona. = Kp/(rps 1) and T. as shown in Fig.st?.I ) I -1 < K < V' + (2r/eI2. they do allow us to make the following generalization. I f the ficticiousvariable B could be measured.57~/e)'. The following development of the Smith predictor algorithm is based on Despande andAsh [I]. For simplicity. Bya digital controlengineer. represents the plant time delay. G.R(s) KO's' 2KePse f 2(7. This implies . L represents potential load disturbances. 1. Y(s) -. 33 of [3]. The approximation methods shown here are not? in general. Downloaded on July 09. T. 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 . TheSmith predictor algorithm [2] avoids this reduction of gain and consequent poorer performance. The Pad$: approximation could indicate stability for some unstable systems.W S ) Ke-Se ( . a series expansion: the Taylor Fig. System performance be could improved the if B could be fed back instead of the output Y. signal Y. in equation (1). and nh = 8. 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. n is an integer. 2.-so z- -2 - -nh fictitious variable where h is the system sampling period. If 8 is small so that se << 1 for frequencies of interest. tion techniques have been suggested. the gain would have to be reduced to maintain stability.R(s) 0 r s 2 + ( 2 r + 0 .as shown on p.K0)s + 2(K 1) All denominator coefficients will be positive and the system will be stable if 0 < K < r/8. 1. These constraints depend on the relationship of 8 to 7: for 8 < < r stability requires -1 < K < v' 1 + (1. The block diagram for conventional control of a time-delay system is shown in Fig. As shown in Fig.(s).K0)s 7 R(s) -E (K -.K 8 ) s + 2 ( K 1 ) + 3. Using the Pade approximation in equation (1) produces a third transfer function. However. The first technique implies that the original system has an infinite number of poles that can be reduced by using an approximation.L(s).(s). For a simple first-order plant with a pure time delay G. Pade approximation: the lL Fig. represents plant dynamics. 17 I / . These three approximations yield different stability limits. (2) 2. this cannot be done in a physical system. C represents the controller. For the sake of comparison. by using e-@ = 1 .Sometimes they yield bizzareresults. Byprocess a control engineer. we can use the Nyquist criterion [ 111 to derive stability constraints for equation (1). I will use theshorthand notation of Marshall [3]: R represents the system input. for 8 = r stabilityrequires -. 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. This would move the time delay outside the control The loop. Restrictions apply. G. The Taylor series approximations err on the safeside. Of course. the process can be conceptually split into delay free system dynamics and a pure time delay.the transform equivalent: .(s).The following four approxima. therefore. The introduction of a time delay makes it moredifficult to access the stability of a system. T. the response of the system would be improved. This should be expected because they are approximations. we could connect it to the controller. 8 : issmall compared to the plant time constant. Y. The Smith Predictor If a time delay were introduced into an optimally tuned system. 1. . R(s).. Y S) K(2 . Large gains can only be used in timedelay systems if the plant time delay. 2.1 < K < 710. = e -S*P. C(s). 1. then we can derive a single pole solution. good methods for assessingthe stability of asystem. The exponential in the denominator will be approximated by an algebraic expression. represents the plant output.

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

o * l s - J = 0. C = 4(0. changes 1 9 Authorized licensed use limited to: The University of Arizona. = T. conjugate gradient.S’ -I.5 4. Suppose that model the of the plant dynamics isin error. e. = 1/(3s + l).(s) 2Tp(s3 -I. = G. approximation was used here for pedagogical reasons. A rearrangement predictor the of Smith model and plant mismatches.or square error. so that AG is non zero. Alternatively.5s + I). equation (4).e . [5].errors in estimatingtheplant gain or timeconstants could cause instabilities. with G . 6. = l/(s + I). and assume its initial value. 5 as Fig. If there is an error in modeling only the time delay so that A G = 0. - ‘ I 1A T l?G Fig.-s . Typical function minimization techniques include gradient. illustrating effects of Y.(2s + 4)(Tp .isin parallel with awelldefined model.Rogers andLatham [4] or Palmor [5] are much more comprehensive. It is not agoodgeneral technique for analyzing stability of Smith predictor systems.200 R(s) The negative coefficient in the denominator shows thatthis is unstable.5 This transfer function appears to be stable by use of the Pad$: approximation and the Routh-Hurwitzcriterion.. although 8. monetary cost: error.24s2 + 84s f 80) __3s3 .5s + I ) .disease. energy. wear. temperature. we get may 1983 Themathematicsbecome complicated if we allow both OP and 8. and plant and model time constants of1 sec. Based on [3].=Gp+AGp=G+AG T . The techniques of Ioannides. let us minimize the square error between the model output and plant output. G . Let us presume that the plant time delay changes. We are leftwith a simple closed-loop feedback control system with feedback of (1 -AT). to change continuously. 4. so that T.1 sec. 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). -36s 1 s2 This produces + 22s 4. = I/(s + l ) . The gradient method tells AO. This description is close enough to that of the Smith predictor to makethis type of adaptive control applicable. but the model of the plant time delay is correct.ATp = TAT Thissimplified notation gets rid of the subscripts and allows us to redraw Fig.86s 4.. = .~ ( O ~ + A B . Let the magnitude of the plant time delay be represented by 0 . 6.Let us try to do this for the Smith predictor of Fig. 6 becomes zero.R(s) + 14s 4. 6. The time delay in this feedback loop could cause instability. Therefore. aplant time delay of 1 sec. The performance function is + 4)Tp 3s 4.4)Tp s2 + 18s 4.to see the effects of a mismatch between model and plant parameters. has been changed by a small amount called AO.2010 at 17:46:05 UTC from IEEE Xplore.etc. However. The magnitudeof the time delay affects the controller via the A G term in Fig. . An Adaptive Smith Predictor Control System If is oftenadvantageous to change the controller to compensate forchanges in plant parameters caused by age. Its use indicates stability for manyunstable systems. a PD controller. C = 4(0. . and design a system that will automatically change the model time delay. For example.In one form of this scheme the controlled system. using the gradient method of minimization. we desire a generalized on-line method of adapting the model time delay. The Pad.y.fatigue. Typical performance criteria are minimization of time. although it has been shown tobe unstable by the techniques of [4]. The gain k must be selectedforeachapplication. G. ) = e-sepe-sAep = T. oceancurrents. With only one parameter the gradient is simplythe partial derivative. then the inner positive feedback loop of Fig.40 Y.5 e2 dt. [ 141. Therefore. and AT = 1. for our Smith predictor we wish to modify the model time delay so that it tracks the plant timedelay thus minimizingtheperformance us function. Let us use a simple plant. = -kVJ. From this wecould compute the plant time delay. The adaptive control scheme Combining examples by Landau [6] and Marshall (31. which is composed of a time-varying plant with adjustable parameters. and Fletcher-Powell. if we use the Pade approximation on equation (4). Restrictions apply. Therefore. This is a positive feedback loop T of AG. In the model reference approach we would now change-gains in the controlled system to minimize the performance function. There are many different types of adaptive control schemes [ 121. One easily understoodscheme is model the reference adaptive control scheme discussed by Landau [ 6 ] . . and a model time delay of 0. The easiest way to do this would be to apply a step input and measure the output. Downloaded on July 09.Theparameters ofthecontrolled system are varied to make the controlled system behave like the parallel model [ 131.G.(s) (6s’ -.). TWO major items must be chosen before an adaptive controller can be designed: a performance criterion a and method of function minimization..T m ) (2s Using the Pade approximation for the exponentials yields T -T P m = . e = (y. we can study the transfer function.

First we identify changes in 6. + CG. = -k The rate at which plant time delay char.)R (1 + CG. dt dJ a YP ae P - The error. . in the lower Smith predictor are supposed to contain the plant values. s . Method on 131.5 e2dt Using (14) twice yields e. ae.? To a first approximation y. Therefore. 7. We will let the identification and control phases alternate. e.continuously.) Therefore.ges is Weare pleased with this result because the 6y. onlyat discrete intervals. By the chain rule Figure 7 shows one method for producing dy. with a digital computer it could be' approximatedover any givenfrequency range./dO. dt =-k I ::a e- dt But now how do we get aelae..lae. dt Using Equation (9) Forsimplicitythetransferfunction has been labeled M(s). Now. It has been assumed the that Laplace transformofay.CG.T.. is a function of theplant time delay. =-kg P 5_ a ae.=- - ae2 2 ae.T. (For each specific application this should be verified by simulation as was done by Marshall [3]. called a sensitivity function. I Now.(s)/ae. l2 Substituting this and (12) into (15) yields =-k-k ae dt 0. = -k I . of generating the sensitivity function ay. = /6.2010 at 17:46:05 UTC from IEEE Xplore. . we get . Restrictions apply. ae. and T. AO./dB. A 0 can be removed from the integral and Finally. Downloaded on July 09. can easily be computed. The boxes labeled G.Tp)( 1 4. Based Authorized licensed use limited to: The University of Arizona. we calculate the partial derivative If theadaptation is slow. For a small change inthe time delay. From equation (4) we have The pure derivative function shown in this figure would not be physically realizable analog with components. but they are not available. Substituting (12) into (11) we get AO.CG. and update them as often as possible. will now be omitted. we will have to use G .. is not a function of e.(t)/a8. However. and T. we can compute the desired change in the model time delay as AO. we will change 8. I I 1 Fig..: e -dt aJ Ad. . we stillhave to evaluate the constant k in (13).(-sCG. = AO.. -cG.'e can interchange the order of differentiation . T.. and the complex frequency argument. From equations ( 1 1 ) and (13) we have = -AO. Computing the gain k d NOW.. usingthedefinition of the differential and assuming the initial value of this error is zero (so that A e = e). is equal to au.

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

A . BahillandM. C.10. Navy teaching mathematics and electrical engineering for four years to the students of the Navy Nuclear Power School. New York: Academic Press. a human can track a predictable target. 1981. 227-250. Control of Time-Dela). pp. Bahilland J . autopilotsfortankers. “An adaptive control model for human and movements head eye while walking.Thetargetmoved f5 degrees. Thorell. References [I] P. N. resulting The system still tracks targets with a time delay. in 1975. the eye continues to follow the predicted target for about 150 msec. Gilbart and G. SMC-13.“The smooth pursuit eye movement system uses anadaptivecontrollertotrackpredictable targets. pp. McDonald.E. The human oculomotor systemhas a 150 msec time delay.Thisresearchwaspartiallysupported National by Science Foundation grant ECS-8121259. McDonald.21. Narendra R. “Stability properties of [5] Z . for an optical “Adaptive compensations trackingtelescope.“ lnt J Control. [2] 0 . Watanabe M. and the Ph. 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. The model performed as well as the human [7]. Marshall. J .S. “Stabilitycriterion of linearcontrolsystems with time delay.” in Proc Int Conf Cxbern Soc. Acknowledgements I thank Jeffery S. Rogers and V . [I31 J. Downloaded on July 09. He received the B. Elements of Computer Process Control With Advanced Control Applications. 1141 . Restrictions apply. L. such as one the shown in Fig. vol. and V . Human zero-latency tracking of a cubic target waveform.1979. kT. 753762. plant time delay. Lee and F. California. modeling physiological systems. T. the and bottom trace shows target velocity (solid) and eye velocity(dotted).” IEEE Trans Svst Man Cybern. pp. then ra(t) can be computed. AI-Tikriti. 1957. 1970. J .TheSmithpredictor algorithm larger allows gains. [7]A. A. Berkeley. [4] A.S. Mare Island. p p . J. Kallstrom. vol. Conclusion To maintain stability of a control system after a time delay is introduced. W . Agin.pp. AC2 6 .” in 9th International Symposium on Industrial Robots. and hT are known. “Real time control of a robot with a mobile camera. Jamshidi.Thiscontrolscheme has onlybeen studied for single-input single-output systemswithscalarKandonlyonetime delay. [3] J . in 1967. March 1979: Washington D. vol. pp. ”Adaptive Automatica. ‘without latency or phase lag. Palmor. Kallman and two perceptivereviewersforconstructivecriticisms. B.(t) can be estimated.29.” 15. [ 101 G . vol. An adaptive control loop added to the Smith predictor can automatically adjust model parameters match time to the varying plant parameters. in electrical engineering and computer science fromthetiniversity of California. Smith.C. Latham. in 1970. J. When a target starts moving there is a 150 msec delay before the eye starts moving. M. Applications of Adaptive Control.” Int J Control. [I ] F. 1261-1269. 125-131. Ioannides.” Meas Control. certain In circumstances controllers canbe designed to track targets with no latency in spite of plant time delays.pp. 1975. and if r. D. [I61 K.. N-Nagy and M.’‘ Inr J Control. Adaptive Control: The New York: Model Reference Approach. vol. J . He served as a Lieutenant in the U. Winston. 1979.” dead Progress. 1979. it requires an exact matchingof model and plant parameters.S.. 3. The time axis is labeled in seconds. Mendel. Eriksson and L. G . 1974. The output can be made equal to the input. pp. compensating for both the time delay and the plant dynamics. The model required knowledge about plant dynamics. E. G .46.” Automatica. However.“ IEEE Trans Aurom Control. Pennsylvania. vol. vol.2010 at 17:46:05 UTC from IEEE Xplore. 1981). the matrix b . Ash. 57-70. 217-219. the gainmustbereduced. h o m . vol. 563.pp. 10. Habibi-Ashrafi. 1981. pp. Terry Bahill (S’66-M’68-Sh4’81) i n o r ns b wa Washington. processmodelcontrolforlinearsystemswithdelay. J .pp. San Jose. The top trace shows the target position (dotted) andtheeye position (solid). model A was built to help explain how the human couldovercomesuchatimedelayand track with no latency.If the time delay 0. head and eye coordination. Sten. S . [9]M. “A novel approach to seismic signal processing andmodeling. D. C . three-stage non-linear systemstimecontrol with delay. 10. and computer text and data processing.New York: IEEE. Sysrems. K . [8]. andthevectors Fig. However. 1983. J. “Closer control of loops Chemical Engineering with time. [I21 K. vol. 22 control systems magazine Authorized licensed use limited to: The University of Arizona. the b1. . the gain K. and Clinical Engineering (Prentice-Hall. One reason studying zerofor such latencytracking is thatthehumaneye movementcontrolsystemseemsto use target-selective adaptive control. N. Tucson. The target position waveform was that of a repeated cubic segment. Monopoli. 1398-1414. Smith dead-time compensator controllers. California. “A design of [I51 M. A . Aminzadeh. 937-949.Hisresearchinterests include control theory.“ Geophvsics. When the target stops.C. Medical. Landau. Research Triangle Park NC: Instrument Society of America. He is the author of Bioengineering: Biomedical. F. Friedman. 241-254. E.1981. 1980. 1980. S.E: degree in electrical engineering from San Jose State Univesity. 53(5). on January 31. degree in electrical engineering from the University of Arizona. Marcel Dekker Inc. Stevenage United Kingdom: Peter Peregrinus Ltd.E. 1946.targetfrequencyandtargetwaveform.“Stabilitylimits of aSmithcontroller in simple systems containing a time 557delay. Deshpande and R. 86-87. in press. T . D. J.E. 32. [6] I. H.. target amplitude. pp. [SI J . 269-278. Kormylo.1981.D. B. 1979. “A and Ito. 1979. vol.