Prediction Model for Minimum Miscibility Pressure in CO₂-Oil Based on Automated Machine Learning

As a key means of improving oil recovery, the application effect of CO<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</inf> flooding technology is restricted by many factors, the effectiveness of CO<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</inf> flooding technology is restricted by many factors, with the minimum miscibility pressure (MMP) required to achieve the miscible phase of CO<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</inf>-crude oil being the most critical parameter for determining oil displacement efficiency. In this study, the CO<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</inf>-oil MMP prediction model was established using automatic machine learning, and the MMP was predicted and the results were analyzed. The results demonstrate that automatic machine learning can simplify the model development process through automatic model selection, hyperparameter optimization, and model fusion, significantly improving the prediction performance and accuracy of the model while accurately predicting the CO<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</inf>-oil MMP. By comparing the predicted values of the model with the results of the slim tube experiment, the error was found to be within 5 %, indicating that the automatic machine learning model can effectively predict the MMP in Bayan Oilfield, providing valuable guidance for the development of CO2 flooding in oilfields.