Development of Mathematical Model for Forecasting the Production Rate

This article is devoted to the issue of digitalization of the field. The authors reviewed the existing methods of building digital systems and methods of production level estimation. On the basis of the method of expert evaluations, the conceptual and then mathematical model of the field is constructed. After that, the behavior of the field depending on the operating conditions was simulated with the help of a digital system design. Forecasting models for production levels are constructed. A distinctive feature of this study is the application of a digital system design to the oil and gas industry. The results demonstrated in this study have found their reflection in practical application in fields complicated by paraffin content. The authors described the idea of Random Forest which is to combine multiple trees to obtain an accurate result, each trained on a random dataset and selecting random subsets of parameters. The Random Forest model was imported from the sklearn library. The authors determined that the model mean square error and mean absolute error remained large for three wells: 15/9-F-11, 15/9-F-12, and 15/9-F-15 D. The presented studies are the final stage in the framework of the work carried out as a dissertation research at Empress Catherine II Saint Petersburg Mining University.