Firstly, two kinds of soft sensor are developed and compared for alcoholysis degree in the process of PVA on the basis of mechanistic analysis, which are based on techniques of the least square regression and artificial neural network (ANN). Compared with the former, the soft sensor ANN-based is more reliable in performance due to the more input information available. It can provide operating guide for industrial use. Secondly, the optimal control strategies are presented for the third distillation column in Beijing Organic Chemical Plant. The effectiveness of the method has been proved through industrial application. Through system studies, improved suggestions for further optimization control are put forward.
|