R. Srinivasan et R. Rengaswamy, Use of inverse repeat sequence (IRS) for identification in chemical process systems, IND ENG RES, 38(9), 1999, pp. 3420-3429
Use of a pseudo random binary sequence as an input signal for identificatio
n of finite impulse response models of chemical process systems has been qu
ite popular. In this paper, we show the utility of another binary signal, t
he inverse repeat sequence (IBS) signal in identification of chemical proce
sses. The attractiveness of IRS lies in the fact that, by its design, even-
order kernels in the process response are canceled out. This reduces the ef
fect of nonlinearity in identification. The improvements that can be derive
d by using IRS are shown through the aid of continuous stirred tank reactor
and fluidized catalytic cracking unit; case studies. Fm ther, the utility
of the identified models in the model predictive control framework is also
demonstrated.