부산대학교 도서관

In this paper an adaptive control strategy for the 13 C isotope concentration process is proposed; the IMC structure based on the PD controller is compared to a classical strategy using the PI controller. The isotopic separation process is a distributed parameter one, depending on the position in the height of the column and on time. To obtain the enriched 13 C isotope, the carbamate and the carbon dioxide are subjects to a chemical exchange process. Since the process is a slow one, strongly nonlinear, with high time constants and long settling time, the PD controller is proposed. In order to improve the system performances, an adaptive form of the discretized PD controller operating with variable sampling time is implemented; all the coefficients of this conversion are learnt by means of artificial intelligence, using neural networks. The adaptive strategy successfully controls the sampling time and the disturbances effect occurring in the system are rejected.