Evaluation of Reservoir Non-Homogeneity Based on Random Forest

The evaluation of non-homogeneity of geological structure that stores oil and natural gas has the problem of non-uniform parameters and contradictory evaluation results of each parameter. In order to accurately evaluate reservoir non-homogeneity and improve reservoir prediction accuracy, this paper introduces a better effect random forest method based on logging data, which is designed not only to take into account the complexity of the reservoir non-homogeneity evaluation problem, but also presents the advantages of high evaluation accuracy and high tolerance to outliers and noisy data. From the results of reservoir non-homogeneity evaluation of the Chang 6 formation group in the YD oilfield. The accuracy of the random forest classifier is higher than 91%, which indicates that the algorithm has some application potential in reservoir non-homogeneity evaluation.