Intelligent switching expert system for delayed coking unit based on iterative learning strategy

Delayed coking is the most effective process to decarbonize and demetallize heavy petroleum residues. However, it relies much on the field engineers’ experiences and expertise in practice for operating the controllers effectively and compatibly in delayed coking. This study establishes a knowledge database of intelligent switching expert system by analyzing the on-site data and operator’s experiences. A feed-forward control strategy based on iterative learning is introduced to erase disturbances arising from switching operation. The intelligent switching expert system (ISES) proposed here is guaranteed to be converged by introducing a convergence factor. The effectiveness and maneuverability of the ISES are proved by the simulation results on Shadow Plant Simulation Platform.