Abstract
The article presents the approach of machine training to the modeling of reactor - regenerative unit of a catalytic cracking unit of an oil refinery. In order to increase the efficiency of the process, the influence of the basic parameters of the cracking process on the quantity and quality of products produced - catalysis - has been studied. Based on the results of research and processing of the collected quantitative and qualitative information with the help of a controlled model of machine learning, regression algorithms have been constructed, describing: the influence of the temperature of the cracking reactor on the output of catalysis and the stability of the catalyst; The influence of the feed rate on the quantity and quality (octane number) of catalysis. This article is an attempt to provide an overview of SVM machine learning applications for catalytic cracking control. The purpose of the study was to create an optimal control system for the distillation column of a catalytic cracking unit using a mathematical model of the process.