ELM-based real-time prediction for hydrogenolysis degree of hydrofining reaction in hydrocracking process

Hydrogenolysis degree of hydrofining reaction in hydrocracking process is an important index which can reflect the running status of hydrocracking, but it's difficult to be detected directly. In this paper, hydrogenolysis degree is redefined with some relevant information which is easier to be detected online, such as hydrogen sulfide and ammonia, and then an ELM-based online prediction method is proposed to forecast the index. Considering that multiple working conditions exist in the long hydrocracking process, steady-state data is selected by double variances method, and a number of data sets are formed. The data sets include the information of catalyst activity, feed properties, operating parameters and the hydrogenolysis degree calculated using hydrogen sulfide and ammonia content by a given formula. Under specific catalyst activity and feed properties, the relationship between operating parameters and hydrogenolysis degree is established by ELM algorithm, so as to predict the hydrogenolysis degree of hydrofining process in real time. The simulation results demonstrate that the ELM-based real-time prediction method can effectively predict the hydrogenolysis degree of compounds containing S/N in hydrofining process, which is benefit to protect the activity of catalyst and ensure the quality of final products.