随机森林储层预测及关键参数探讨——以SC某工区储层预测为例

Random Forest Reservoir Prediction and Key Parameters Discussion—Taking Reservoir Prediction of a Work Area in SC as an Example

摘要: 随机森林是一种高度灵活的机器学习算法，可以解决回归和分类的问题。本文建立随机森林地震储层预测流程，探讨了随机森林储层预测过程中的关键参数，并利用SC某工区储层伽马值预测为例，分别对比分析了不同参数对预测结果的影响，并给出最优参数设置。同时对参与建模的地震属性重要性进行分析，优化属性样本集。最后利用最佳随机森林模型，以优化属性样本集为输入开展储层预测，取得良好效果。实践证明，随机森林算法能够有效进行储层预测，随机森林模型的参数调优很重要，影响到算法的效率和最终预测精度。

Abstract: Random forest is a highly flexible machine learning algorithm, which can solve the problems of regression and classification. According to the process of random forest seismic reservoir prediction, this paper discusses the key parameters in the process of random forest reservoir prediction. Taking the reservoir gamma prediction of a work area in SC as an example, the influence of different parameters on the prediction results is compared and analyzed, and the optimal parameter settings are given. At the same time, the importance of seismic attributes is analyzed, and the optimal attribute sample set is obtained. The optimal random forest model is used to carry out reservoir prediction with the optimal attribute as the input, and good results are achieved. It is proved that random forest algorithm can effectively predict reservoir. Parameter optimization of random forest model is very important, which affects the efficiency and final prediction accuracy of the algorithm.