Enhancing interactive optimization with operating condition supervision for distillation units

Changing operating conditions is a common phenomenon in the distillation unit (DU), which poses difficulties to production operations. To fully cope with varying operating conditions and further improve production performance, an interactive operation optimization strategy (OOS) with operating condition supervision is proposed in this work. In this strategy, the operating condition supervision module perfectly cooperates with the interactive optimization strategy to reduce the two major performance losses incurred after changes in operating conditions. The bidirectional interaction between the optimization layer and the control layer during operation can eliminate losses and hazards caused by delayed response and inter-layer mismatches, ultimately achieving optimal closed-loop performance. Through detailed analysis of the specific industrial behavior, it provides professional support for the production operations under varying operating conditions. It is worth mentioning that a convolutional neural network (CNN) process model based on parameter transfer is established. It can be fine-tuned online. Experimental results show the proposed strategy can flexibly and effectively handle changes in operating conditions. The proposed OOS improves the product qualification rate and has broad application prospects in industrial processes.