PREDICTION OF DRILL STRING VIBRATIONS USING MACHINE LEARNING TOOLS

Abstract

The objective of this work is to apply machine learning algorithms for the analysis of dynamic vibrations of drill strings. As part of the research, a mathematical model describing the vibrations of the drill string was developed; a finite difference scheme was implemented, and numerical modeling was carried out using the three-point sweep method. Based on the data obtained from the numerical solution, a machine learning model was created. The numerical modeling was implemented using C++, while data collection and the construction of the machine learning model were performed using the Python programming language. As a result, a predictive model was obtained, capable of accurately forecasting the dynamic vibrations of the drill string and determining optimal parameters, thereby improving the efficiency and safety of drilling operations.