论文部分内容阅读
渗透率图像的预测模拟对油田的开发具有重要意义。充分利用条件数据可以提高渗透率模拟的精度,因此提出一种基于连续型多点地质统计法和软硬数据的渗透率图像模拟方法。首先,利用过滤器得分操作对渗透率图像降维,所有的过滤器得分形成了一个过滤器得分空间;其次,通过两步划分法将得分空间划分,得到每个非空得分类的特征,形成一个“特征库”;最后,通过自定义的距离函数从“特征库”提取已知模式的特征,以完成对节点的模拟。比较各情况下渗透率模拟图像可以看出,将软硬数据同时作为条件数据的模拟图像由于采用了较为丰富的条件数据信息,因此与真实情况下的渗透率分布在结构特征上最为接近。
The prediction of permeability images is of great importance to the development of oilfields. Making full use of the condition data can improve the accuracy of permeability simulation. Therefore, a method of permeability image simulation based on continuous multi-point geostatistics and hard and soft data is proposed. First, the filter operation is used to reduce the dimension of the permeability image. All the filter scores form a filter score space. Secondly, the score space is divided by two-step partition method to get the feature of each non-empty score class. A “signature library”; and finally, extract the known pattern features from the “signature library” through a custom distance function to complete the simulation of the nodes. Comparing the simulated images of permeability with different conditions, it can be seen that the simulated images with both hard and soft data as the condition data are the most similar to the real ones due to the abundant condition data.