You are on page 1of 14

ISA TRANSACTIONS

ISA Transactions 43 2004 257270

Smith predictor basedsliding mode controller for integrating processes with elevated deadtime
Oscar Camacho,a,* Francisco De la Cruzb
a

Postgrado en Automatizacion e Instrumentacion, Grupo en Nuevas Estrategias de Control Aplicado, Universidad de los Andes, Merida 5101, Venezuela b Departamento de Ingeniera Electronica, Vice Rectorado Barquisimeto Universidad Nac. Exp. Politecnica Antonio Jose de Sucre, Barquisimeto 3001, Venezuela

Received 23 October, 2002; accepted 25 August 2003

Abstract An approach to control integrating processes with elevated deadtime using a Smith predictor sliding mode controller is presented. A PID sliding surface and an integrating rst-order plus deadtime model have been used to synthesize the controller. Since the performance of existing controllers with a Smith predictor decrease in the presence of modeling errors, this paper presents a simple approach to combining the Smith predictor with the sliding mode concept, which is a proven, simple, and robust procedure. The proposed scheme has a set of tuning equations as a function of the characteristic parameters of the model. For implementation of our proposed approach, computer based industrial controllers that execute PID algorithms can be used. The performance and robustness of the proposed controller are compared with the Matausek-Micic scheme for linear systems using simulations. 2004 ISAThe Instrumentation, Systems, and Automation Society.
Keywords: Sliding mode control; Smith predictor; Integrating process; Deadtime

1. Introduction The presence of time delays in many industrial processes is a well-recognized problem. The achievable performance of typical feedback control systems can decline if a process has a relatively large time delay compared to the dominant time constant 1 . In the case of integral processes, deadtime compensation is necessary in order to enhance the performance of the control system. ODwyer 2,3 considered a wide variety of methods for the compensation of processes with time delay, in both the continuous time and dis-

*Corresponding author. Tel: 58-274-2402891; fax: 58-274-2402890. E-mail address: ocamacho@ula.ve

crete time domains. The compensators discussed are: 1. PID controllers and its variations. 2. Lead, lag or lead-lag controllers. 3. The Smith predictor 4 and its variations. 4. Direct synthesis methods, which are typically based on designing the controller to meet a required output specication; pole placement controllers are an example. 5. Optimal controller design methods, which may be based on a minimum variance or linear quadratic control strategy. 6. Predictive controllers. 7. Other compensation strategies for processes with time delays, including fuzzy implementation, neural networks, and expert systems. The wide spectrum of methods covered and the dependence of the choice of compensator method

0019-0578/2004/$ - see front matter 2004 ISAThe Instrumentation, Systems, and Automation Society.

258

Camacho, De La Cruz / ISA Transactions 43 (2004) 257270

on the application mean that an overall conclusion as to the best method to use is not appropriate. Viewing the compensation problem from a variety of perspectives, it appears that the Smith predictor is the optimal or a component of the optimal controller for dominant delay processes 2 . In those works, the compensation of integrating process with dead time is only achieved using PID controllers, Smith predictors, or its variations. Other approaches have not been reported for this kind of system. The Smith predictor SP , or deadtime compensator DTC , as it is also known, has many weak points, including possible instability and poor performance under modeling errors, and poor response to disturbances 1 . In addition, the original structure of the SP cannot reject constant load disturbance for processes with integration 5 . To overcome this obstacle many variations of SP have been proposed. Sliding mode control SMC is a robust and simple procedure to develop controllers for linear and nonlinear processes 6 . The design of a sliding mode controller SMCr depends on the process model, and the number of tuning parameters is proportional to the model order. Camacho and Smith 7 developed a simple and practical method for the design of a SMCr based on a simplied model of the actual process. The SP performs well for eliminating deadtime and the SMCr is a proven robust controller. It is desired to combine them into a single control structure that preserves the good qualities of both techniques and improves the bad qualities of the SP. A robust controller for integral processes with deadtime, the Smith predictor sliding mode controller SPSMCr , will be developed. This paper is organized as follows: Section 2 introduces the basic concepts of the sliding mode controller, the SP architecture and a review of the modied SP for integrating processes that have previously been proposed. Section 3 covers the development of the SPSMCr based on an integrating rst-order process model with deadtime. Section 4 shows the SPSMCr implementation using a PID algorithm of a computer based industrial controller. Section 5 provides simulation results to illustrate the approach and to compare it with the performance and robustness of a previous scheme. Finally, some conclusions are presented in Section 6.

Fig. 1. Graphical interpretation of SMC.

2. Basic concepts 2.1. Sliding mode control (SMC) Sliding mode control is a technique derived from variable structure VSC which was originally studied by Utkin 8 . A controller designed using the SMC method is particularly appealing due to its ability to deal with nonlinear systems and time-varying systems 9 . The robustness to the uncertainties becomes an important aspect in designing any control system. The idea behind SMC is to dene a surface along which the process can slide to its desired nal value. Fig. 1 depicts the SMC objective. The structure of the controller is intentionally altered as its state crosses the surface in accordance with a prescribed control law. Thus the rst step in SMC is to dene the sliding surface S ( t ) which represents a desired global behavior, like stability and tracking performance. The S ( t ) selected in this work, presented by Slotine and Li 9 , is an integral-differential equation acting on the tracking-error expression:

S t

f e t ,

e t dt,

de t , ,n , dt

where e ( t ) is the tracking error, that is, the difference between the reference value or set point r ( t ) and the output measurement x ( t ) , namely, e ( t ) r ( t ) x ( t ) . is a tuning parameter, which helps to dene S ( t ) . This term is selected by the designer, and determines the performance of the system on the sliding surface. n is the system order. The control objective is to ensure that the controlled variable be equal to its reference value at all times, meaning that e ( t ) and its derivatives

Camacho, De La Cruz / ISA Transactions 43 (2004) 257270

259

Fig. 2. Chattering reduction using a saturation function a:

0; b:

0.01; c:

0.1; d:

1.0 .

must be zero. Once the reference value is reached, Eq. 1 indicates that S ( t ) has reached a constant value, meaning that e ( t ) is zero at all times; it is desired to set

dS t dt

0.

Once the sliding surface has been selected, attention must be turned to design the control law that drives the controlled variable to its reference value and satises Eq. 2 . The SMC control law M ( t ) consists of two additive parts: a continuous part U C ( t ) and a discontinuous part U D ( t ) . That is,

M t

UC t

UD t .

The continuous part is given by

achieve the high switching control because of the presence of nite time delays for control computations or limitations of the physical actuators, thus causing chattering around of the sliding surface 8,9 . Chattering is a high-frequency oscillation around the desired equilibrium point. It is undesirable in practice, because it involves high control activity and can excite high-frequency dynamics ignored in the modeling of the system 9,10 . The aggressiveness to reach the sliding surface depends on the control gain i.e., , but if the controller is too aggressive it can collaborate with the chattering. To reduce the chattering, one approach is to replace the relaylike function by a saturation or sigma function, which can be written as follows:

UC t

f x t ,r t ,

4 UD t KD

where f x ( t ) ,r ( t ) is a function of the controlled variable and the reference value. The discontinuous part is nonlinear and represents the switching element of the control law. This part of the controller is discontinuous across the sliding surface. Mainly, U D ( t ) is designed based on a relaylike function i.e., U D ( t ) sgn( S(t) ), because it allows for changes between the structures with a hypothetical innitely fast speed. In practice, however, it is impossible to

S t S t

where K D is the tuning parameter responsible for the reaching mode. is a tuning parameter used to reduce the chattering problem. Fig. 2 shows the effect of variations in the shape of saturation function. Fig. 3 shows the effect of K D variations on the system trajectory on the phase plane from an initial state to a nal state.

260

Camacho, De La Cruz / ISA Transactions 43 (2004) 257270

Fig. 3. Graphical interpretation for K D variations. K1 K2 K3.

In summary, the control law usually results in a fast motion to bring the state onto the sliding surface, and a slower motion to proceed until a desired state is reached. 2.2. Smith predictor: Basic concepts and previous schemes for integrating processes As stated before, SP is a popular scheme for deadtime compensation. Fig. 4 shows the architecture of the SP. The process transfer function is G p ( s ) G ( s ) e t 0 s which is assumed to consist of a rational stable transfer function G ( s ) and a deadtime t 0 . A model of the process without deadtime, G m ( s ) , is used to predict the effect of the

control action on the process output and to increase the performance of the system. The difference between the output of the process and the model is fed back in order to correct modeling errors and load disturbances. If there is no process/model mismatch, G m ( s ) G ( s ) and t 0m t 0 , then the modeling error e m ( t ) y ( t ) y m ( t ) 0. Since the deadtime is separated from the model, and e m 0, the feedback only consists of the model without delay. Therefore the deadtime is isolated and compensated 1 , and thus, for controller design purposes, it can be ignored. However, the SP cannot be used with its original structure to control processes with integration since a constant load disturbance re-

Fig. 4. Smith predictor scheme.

Camacho, De La Cruz / ISA Transactions 43 (2004) 257270

261

Fig. 5. The modied SP proposed by Watanabe and Ito.

sults in a steady-state error 5 . To overcome this obstacle many variations of SP have been proposed. Watanabe and Ito 5 proposed a modication of the SP as shown in Fig. 5. The system can reject a load disturbance if the time delay of the process is exactly known. Otherwise, there will be a small steady-state error 12 . Simulations studies have shown that the setpoint and load disturbances are either very oscillatory or highly damped when the process has a large dead time 13 . Astrom et al. 13 proposed a new SP structure, as shown in Fig. 6, where the disturbance response is decoupled from the setpoint response. The controller has four adjustable parameters but a systematic tuning method was not given. The authors considered only the restricted integrating process G ( s ) e t 0 s /s. Zhang and Sun 12 improved the results of As trom et al. retaining the separation nature of load response from the setpoint response. The structure of the modied SP is shown in Fig. 7. M ( s ) was recommended as

M s

sM o s 1 sT
1

Mo s e

t0s

Mo s

T 2 T s T s 1 2 Ts

Eq. 6 contains a positive feedback loop that is a potential instability source, resulting in limited robustness 15 . Zhang and Sun rely on a general guideline to tune K r and rather than a systematic approach 16 . The proposed scheme does not pro vide much better performance than Astroms 12 . When the process is described by a high-order model, the controller is a derivative or a derivative with lag and a more complex M ( s) will result. Matausek and Micic 17,18 proposed a modied SP and gave a simple controller tuning method. Their scheme, given in Fig. 8, has a controller to remove the load disturbance although the setpoint and load responses cannot be decoupled. In a rst paper 17 , F ( s ) is a proportional controller K o , obtaining a simple structure with three adjustable parameters. However, for high values

Fig. 6. The SP structure proposed by Astrom et al.

262

Camacho, De La Cruz / ISA Transactions 43 (2004) 257270

Fig. 7. The structure of Zhan and Suns SP.

of the deadtime, the scheme has poor disturbance rejection 19 and becomes signicantly oscillatory with time delay deviation 15 . A second Matausek and Micics modied SP 18 , where F ( s ) is a lead-lag compensator, provides considerably faster load disturbance rejection but requires a trial-and-error procedure of the main controller gain in a tradeoff between closedloop system performance and stability/robustness. This scheme will be used to compare the proposed controller. The Tian and Gaos control scheme 15 has the same structure as Astroms SP and a parameters tuning as in Ref. 18 . The structure of this SP includes four controllers, as shown in Fig. 9. A local proportional feedback K o is introduced to prestabilize the integrator process. The introduction of G o ( s ) eliminates the effect of K o on setpoint tracking. To compensate the phase lag caused by integrator and time delay a proportional-derivative controller G c ( s ) is suggested, where

Go s

K oK p e s

t0s

Gc s

K c 1 T ds . 7

This structure behaves slightly better than the lat est Matausek and Micics scheme setpoint ITAE index reduction 0.5% 15 , despite the added structural complexity, and their tuning strategy can be applied to only pure integrator plus time delay models 16 .

3. Smith predictor based sliding mode controller for integrating processes The Smith predictor based sliding mode controller SPSMCr proposed in this paper uses the standard SP architecture while the controller is a sliding mode controller SMC . The block diagram of the proposed scheme is shown in Fig. 10. As mentioned, the original structure of SP is ineffective for integrator processes because it cannot reject a

Fig. 8. Modied SP proposed by Matausek and Micic.

Camacho, De La Cruz / ISA Transactions 43 (2004) 257270

263

Fig. 9. Tian and Gaos control scheme.

constant load disturbance 5 and proportionalderivative controller G d ( s ) is used for load disturbance rejection, where

d 2X 1 t
m

dt

dX 1 t dt

KmM t

11

Gd s

K o T ds 1 .

and

This controller will be discussed later. To develop SPSMCr, an integrating rst-order plus deadtime IFOPDT process model given in Eq. 9 is considered,

d 2X 1 t dt
2

1
m

KmM t

dX 1 t . dt

12

Gp s

G s e

t0s

K e s 1

t0s

In this case, we use an IFOPDT process and n 2. Then, from Eq. 1 , we selected as S ( t ) an integral-differential equation acting on the tracking error expression represented by

where K is the process gain, t 0 is the process deadtime, and is the process time constant. If there is no process/model mismatch, the model transfer function without deadtime is

S t

de t dt

t 1e

e t dt.
0

13

Gm s

X1 s M s

Km , ms 1

10

where K m is the model gain and m is the model time constant. Since the deadtime term has been isolated using a SP structure, we can ignore it in the SMC design. Then, transforming Eq. 10 into differential equation form,

Eq. 13 represents a PID surface. The parameters 1 and 0 can be chosen independently and, in our case, they were selected to obtain an overdamped response. From Eq. 2 ,

dS t dt

d 2e t dt 2
1

de t dt

0e

0 14

Fig. 10. Smith predictor based sliding mode controller.

264

Camacho, De La Cruz / ISA Transactions 43 (2004) 257270

but e ( t ) R ( t ) X 1 ( t ) and substituting into the above equation gives

d 2R t dt
2

d 2X 1 t dt
0e t 2 1

Furthermore, to assure that the sliding surface behaves as a critical or overdamped system, 0 should be
2 1

dR t dt

dX 1 t dt 15
0

0.

time

22

Camacho 11 has shown that the derivates of the reference value can be discarded without any effect on the control performance. Thus

d 2X 1 t dt
2 1

dX 1 t dt

0e

t .

16

Numerous simulations showed that the values of 1 and 0 are a function of the controllability relationship, CR t 0 / m , and the following values provide satisfactory system performance and robustness with time delay deviation:

Substituting Eq. 12 into Eq. 16 ,

dX 1 t 1 dt

0e t

1
m

KmM t

dX 1 t . dt 17

4
m 1

time time

if CR 4 if CR 4, 23

1.5
m

Thus the continuous part of the controller is

UC t

1 Km

m 1

dX 1 t dt

m 0e

t . 18
2 1 0

Then, the complete SPSMCr can be represented as

time

24

M t

1 Km KD

m 1

dX 1 t dt

m 0e

t 19

S t S t

with

The parameters and K D have a relationship with system speed, overshot, and chattering. Based on previous approaches 6,11,14 where the Nelder-Mead searching algorithm was used, the tuning parameters of the controller discontinuous part are

S t

sgn K
t 0

dX 1 t dt e t dt .
0

1e

t KD 20

0.75 t 0 Km m

0.76

fraction CO , 25

The function sgn(K) in Eq. 20 is included in the sliding surface equation to guarantee the appropriate action of the controller for the given system. Note that sgn(K) only depends on the static gain of the plant; therefore it never switches 6 . Eqs. 19 and 20 dene the controller equations to be used in the SPSMCr, which can be simplied by setting

2 0.68 0.12 K m K D fraction TO/time .

26

3.1. Disturbance rejection As mentioned, many modied SP for integrator processes with different structures have been proposed in the literature to remove the steady-state error produced by a constant load disturbance.

1
1 m

time

21

Camacho, De La Cruz / ISA Transactions 43 (2004) 257270

265

Tian and Gao 15 and Matausek and Micic 18 added derivative action to their proposed DTC to overcome this problem. A proportional-derivative controller G d ( s ) , given in Eq. 8 , is used in the proposed SPSMCr to enable the load disturbance rejection. The parameters of this controller, using 0.4, pm 64 1.117 rad, as recommended in Ref. 18 , are
pm 2

2 Ko Km
m

Fig. 11. Implementation of PID algorithm in an industrial controller.


pm 2

t0

0.7239 , Km m t0 Td
m

27 t0 0.4
m

t0 .

28

4. SPSMCr implementation using a PID algorithm Computer based industrial controllers that execute PID algorithms programmable logic controllers and remote terminal units can be used for SPSMCr implementation. The more common PID algorithm is based on the following equation:

is the controlled variable. Note that the derivative term is calculated using the measured controlled variable, not the error. Fig. 11 shows the block diagram that implements an algorithm for a PID. As can be observed, Eqs. 20 and 29 are similar. The term X 1 ( t ) in Eq. 20 is the same CV ( t ) in Eq. 29 . Therefore, to represent the sliding surface S ( t ) , the next step is to tune the PID based on the SPSMCr tuning equations as follows:

Kc
i D 1/

1, 0, 1.

30

1/

MV t

Kc

dCV t D dt 1
t i

e t

29

e t dt
0

where M V ( t ) is the manipulated variable, i is the integral time, D is the derivative term, and CV ( t )

With these equations the PID algorithm can be changed, representing S ( t ) . Then, to achieve the implementation, the sliding surface value S ( t ) is calculated from the PID output. Because the continuous and discontinuous parts of the controller are algebraic equations, they are easily programmable. The block diagram that implements an algorithm for the SPSMCr is shown in Fig. 12.

Fig. 12. Implementation of SPSMCr using an industrial controller.

266

Camacho, De La Cruz / ISA Transactions 43 (2004) 257270

Fig. 13. System responses to a setpoint change for different 0.33.

values.

1.33;

2.67;

Fig. 14. System responses to a setpoint change for different 0.056.

values.

0.22;

0.44;

Camacho, De La Cruz / ISA Transactions 43 (2004) 257270

267

5. Simulation results To illustrate the SPSMCr performance, two examples are given. In example 1, the system response to different parameter values is presented to show the relationship with the system performance. In example 2, the response of an integrating system with a long deadtime is shown and the results are compared to those obtained using the proposed controller in Ref. 18 . Example 1 Let us consider a process with IFOPDT transfer function G 1 ( s ) ,

G1 s

1 e s 3s 1

6s

31

with CR t 0 / 2. Using Eqs. 23 26 , the parameters of SPSMCr are

4
1 m

4 3

1.333,

2 1 0

8
0.76

0.222,

KD

0.75 t 0 K m

0.76

0.75 6 1 3

0.443,

2 * 0.68 0.12 K m K D

2 * 0.68 0.12 1 0.443 1.333 0.7239 Km m t0 T d 0.4


m

Fig. 15. System responses using SPSMCr for G 1 (s): a 1.5 and b 0.

1.5,

Ko

0.7239 1 3 6 t0 0.4 9

0.0809,

3.6.

Figs. 13 and 14 show the process output to a setpoint change when 1 and 0 variations were introduced. In these gures, the dotted lines correspond to responses using the values obtained with Eqs. 23 and 24 . The smaller values of 1 increase speed and overshoot. The smaller values of 0 increase speed response and decrease settling time. Numerous simulations showed that as speed increases, the system robustness decreases. Fig. 15 shows the system responses to a setpoint

value were introchange when reductions of duced. Fig. 15 a corresponds to the value obtained with Eq. 25 . Fig. 15 b corresponds to 0. The smaller increases speed and chattering. Fig. 16 shows the system responses to a setpoint change when reductions of K D value were introduced. In this gure, the solid lines correspond to responses to K D value using Eq. 26 . A larger K D increase speed and overshot system and the smaller K D values increase settling time. Example 2 Let us consider a process with a transfer function G 2 ( s ) , which has been investigated in Refs. 15,1719 ,

268

Camacho, De La Cruz / ISA Transactions 43 (2004) 257270

Fig. 16. System responses to a setpoint change for different K D values. K D 3.54.

: K D 0.44;

: K D 0.25;

G2 s 1 e s s 1 0.5s 1 0.2s 1 0.1s 1


The corresponding IFOPDT model is
20s

32

s introduced at time t 0 and a load disturbance d 0.1 is introduced at time t 70. When the model is exact, Fig. 17 a shows the response of both control schemes. Fig. 17 b shows the effect of a 20% deadtime modeling error on system performance. The MM99 scheme becomes unstable. 6. Conclusions

G 2m s

1 e s 1.28s 1

20.64s

33
In this paper, a robust control scheme for integrating systems with deadtime using a Smith predictor sliding mode controller SPSMCr has been presented. Sliding mode control improves robustness and stability to the scheme while the Smith predictor isolates and compensates the deadtime. This new method allows for a robust controller while compensating for the deadtime within the process. The robustness of the steady-state performance is improved by the sliding mode with 20% modeling errors. This controller combines the simplicity of the SP architecture implementation and the robustness of a sliding mode controller. Besides, the disturbance rejection for integrator processes of the original structure of SP has been improved, adding a proportional-derivative controller

with CR 16.125. In this example, the SPSMCr is compared to the controller proposed in Ref. 18 and denoted by MM99. For MM99, the equivalent time constant T e is set to 2.4 as in Ref. 18 to improve robustness. The rest of the parameters of MM99 and SPSMCr are given in Table 1. A unit step setpoint
Table 1 Controller tuning parameters for process G 2 (s). MM99 Tr Kr Ko Td
1 0

SPSMCr KD Ko Td

2.4 0.417 0.027 10.56 1.172 0.172 0.09 0.139 0.033 8.77

Camacho, De La Cruz / ISA Transactions 43 (2004) 257270

269

Fig. 17. Responses of the proposed SPSMCr and MM99 for G 2 (s). a Nominal case, b 10% error in t 0 .

originally proposed by Matausek and Micic 18 . The new proposed controller has a simple tuning procedure if an IFOPDT model is available. References
1 Tan, K. K., Lee, T. H., and Leu, F. M., Predictive PI versus Smith control for dead-time compensation. ISA Trans. 40, 1729 2001 . 2 ODwyer, A., The estimation and compensation of

processes with time delays. Ph.D. thesis, Dublin City University, Ireland, 1996. 3 ODwyer, A., A survey of techniques for the estimation and compensation of processes with time delay. Technical Report Number: AOD. 00.03, Dublin Institute of Technology, Ireland, 2000. 4 Smith, O. J. M., Close control loops with dead time. Chem. Eng. Prog. 53, 217219 1957 . 5 Watanabe, K. and Ito, M., A process-model control for linear systems with delay. IEEE Trans. Autom. Control 26, 12611269 1981 .

270

Camacho, De La Cruz / ISA Transactions 43 (2004) 257270

6 Camacho, O., Rojas, R., and Garca, W. M., Variable structure control applied to chemical processes with inverse response. ISA Trans. 38, 5572 1999 . 7 Camacho, O. and Smith, C. A., Sliding mode control: An approach to regulate nonlinear chemical process. ISA Trans. 39, 205218 2000 . 8 Utkin, V. I., Variable structure systems with sliding modes. IEEE Trans. Autom. Control AC-22, 212 222 1997 . 9 Slotine, J. J. and Li, W., Applied Nonlinear Control. Prentice-Hall, Englewood Cliffs, NJ, 1991. 10 Camacho, O. and Rojas, R., A general sliding mode controller for nonlinear chemical processes. Trans. ASME 122, 650 655 2000 . 11 Camacho, O., A New Approach to Design and Tune Sliding Mode Controller for Chemical Process. Ph.D. Dissertation, University of South Florida, 1996. 12 Zhang, W. D. and Sun, Y. X., Modied Smith predictor for controlling integrator/time delay processes. Ind. Eng. Chem. Res. 35, 27692772 1996 . 13 Astrom, K. J., Hang, C. C., and Lim, B. C., A new smith predictor for controlling a process with an integrator and long dead-time. IEEE Trans. Autom. Control 39, 343345 1994 . 14 Camacho, O., Rojas, R., Garca, W., and Alvarez, A., Sliding Mode Control: A Robust Approach to Integrating Systems with Dead Time. Proc. Second IEEE Int. Caracas Conf on Devices, Circuits and Systems, ICCDSC-98, Margarita, Venezuela, 1998. 15 Tian, Y. and Gao, F., Control of integrator processes with dominant time delay. Ind. Eng. Chem. Res. 38, 29792983 1999 . 16 Kwak, H. J., Sung, S. W., and Lee, I., Modied Smith predictors for integrating processes: Comparisons and proposition. Ind. Eng. Chem. Res. 40, 15001506 2001 . 17 Matausek, M. R. and Micic, A. D., A modied Smith predictor for controlling a process with an integrator and long dead-time. IEEE Trans. Autom. Control 41, 11991203 1996 . 18 Matausek, M. R. and Micic, A. D., On the modied Smith predictor for controlling a process with an inte-

grator and long dead-time. IEEE Trans. Autom. Control 44, 16031606 1999 . 19 Normay-Rico, J. E. and Camacho, E. F., Robust tuning of dead-time compensators for processes with an integrator and long dead-time. IEEE Trans. Autom. Control 44, 15971603 1996 .

Oscar Camacho received the Electrical Engineering, and M.S. in Control Engineering degrees from Universidad de Los Andes ULA , Merida, Venezuela in 1984 and 1992, respectively, and the M.E. and Ph.D. in Chemical Engineering at University of South Florida USF , Tampa-Florida, in 1994 and 1996, respectively. He has held teaching and research positions at ULA, CIEDPDVSA, and USF. His current research interests include sliding mode control, deadtime compensation, and hybrid systems. He is the author of more than 50 publications in journals and conference proceedings.

Francisco De la Cruz received the Electronic Engineering degree from the Insti tuto Universitario Politecnico IUP de Barquisimeto, Venezuela in 1980 and the Master degree in Electronic Engineering from the Universidad Simon Bolivar, Caracas, Venezuela in 1990. Since 1980, he has been at the Department of Electronic Engineering, Universidad Politecnica UNEXPO , Barquisimeto, Venezuela. He is currently working towards the Doctoral Degree in Applied Sciences, Universidad de Los Andes, Merida, Venezuela. His current research interests include systems with delays, sliding mode control, and predictive control.

You might also like