Application of machine learning algorithms for refining processes in the framework of intelligent automation
V V Bukhtoyarov, Ivan Nekrasov, В С Тынченко, K A Bashmur, Ruslan Sergienko
Abstract
The oil refining industry is facing several challenges and issues in data handling. A large amount of data is generated by many different processes and equipment. This article is devoted to methods for efficient analysis of large amounts of data in an oil refinery. In particular, the effectiveness of machine learning methods for predicting failures of process equipment in the hydrocracking process is investigated. Machine learning, as an important element of digitalization, allows us to successfully solve many production problems. The article describes the application of some machine learning algorithms for solving problems of classifying and predicting failures of hydrocracking process equipment that occur during oil refining and diesel fuel production. The application of random forest methods, principal component analysis and hyperparameter tuning is considered. The effectiveness of these methods is compared on the basis of the Accuracy parameter. It is shown that the combination of these methods will improve the accuracy of the model by 2%. Keywords: automation; machine learning; hydrocracking; simulation; oil refinery.