It is well known that many of chemical process can be
described by first order system with integrator and time
delay [1-4]. Time delay may be caused by transmission or
measurement lag in chemical industry. General control
approaches will encounter great difficulties to stabilize
systems with time delay. Even if some of them can stabilize
these systems, it will take long settling time and oscillation
may occur in transient process. Therefore, such systems are
recognized as a challenging problem in process control [5, 6].
Smith predictor can eliminate system time delay shorten
settling time and reduce overshoot [7-9]. However,
conventional Smith predictor depends on the model heavily.
The performance may decrease in the presence of modeling
uncertainty, for example, great oscillation or even instability
may take place. Especially, conventional Smith predictor can
not reject constant external disturbance for processes with
integrator. Residual error may be caused by constant external
disturbance [10, 11].