论文部分内容阅读
支持向量机是识别水淹层的有效方法,但其预测性能受多种因素的影响。研究提出一种水淹层识别新方法,采用Relief-F算法进行自动化特征选择,通过遗传算法优化模型参数以及使用加权支持向量机改善样本类数据分布不平衡对分类准确率的影响。将该方法应用于克拉玛依油田六中区克下组砾岩油藏水淹级别划分中,结果表明效果良好,增强了支持向量机的预测能力,进一步提高了水淹层解释的精度。
Support vector machines (SVMs) are effective methods for identifying flooded layers, but their predictive performance is affected by many factors. This paper presents a new method of water flooded layer identification, using Relief-F algorithm to select automatic features, optimizing the model parameters through genetic algorithm and using weighted support vector machine to improve the classification accuracy of sample data. The method is applied to the subdivision of conglomerate reservoir in Kezhaka Formation, Karamay Oilfield. The results show that the method is effective and the prediction ability of SVM is enhanced and the accuracy of interpretation of water flooded layer is further improved.