Comparison between human supervisory control and hierarchical control system based on human's knowledge in petroleum plant

Presents a hierarchical control system which is designed based on a human's knowledge, which is a supervisor of conventional PID controllers. The hierarchical control system consists of two control function blocks, where we use (1) a statistical model using multi-regression analysis for the function block to estimate parameters of plant operation and (2) fuzzy logic for the control function block to compensate operating conditions. The hierarchical control system has been applied to hydrogen purity controls in a large scale petroleum refining plant and verified that the control system has shown better performance than human supervisory control. The human supervisor is necessary only when the system structure changes in the hierarchical system.