Prediction Method for Hydraulic Fracturing Effect of Producing Well Based on Machine Learning Methods

Abstract Accurate prediction of producing fracturing oil enhancement is a prerequisite for decision-making on the implementation of production enhancement measures in producing wells. The traditional reservoir numerical simulation method requires accurate reservoir geological parameters and fracturing construction parameters, which makes the calculation process complicated and the on-site application effect unsatisfactory. Therefore, a method for predicting fracturing effect of producing well based on machine learning is proposed. A combination of correlation coefficient analysis and gray correlation analysis was used to determine the main controlling factors affecting fracturing effectiveness; Considering the nonlinear mapping relationship between the fracturing effect and the main control factors, a prediction model was established using a support vector machine. The model is applied with an actual block of an oil field as an example, and the evaluation of potential wells is completed with the analysis of economic benefits of fracturing measures. The results show that the model prediction results meet the accuracy requirements for engineering application, and the screened fracturing potential wells have achieved better oil enhancement results after the implementation of measures. The prediction model of oil enhancement effect established by this method is based on the actual production data of producing wells, with reliable prediction results and simple model solution, which has the value of popularization and application in the same type of blocks.