论文部分内容阅读
针对常规岩性识别方法在复杂储层的非均衡测井数据准确率不高的问题,本文提出了一种面向复杂储层非均衡测井数据的多层BP神经网络的岩性识别算法.根据测井数据的特点及测井数据间的非线性关系,对测井数据进行均衡化处理,利用多层BP神经网络进行逐层岩性识别.在13个含油气盆地的7793个地层元素测井数据中测试表明,本文提出的方法在数据非均衡比达到1∶6的情况下,岩性识别的符合率仍能达到92%以上,在复杂储层的岩性识别方面具有很好的应用前景.
Aiming at the problem that the conventional lithology identification method is not accurate in the non-equilibrium well logging data of complex reservoirs, a lithological identification algorithm of multi-layer BP neural network for complex reservoir imbalance logging data is proposed in this paper. According to Logging data characteristics and the non-linear relationship between well logging data, the well logging data are equalized and multi-layer BP neural network is used to identify lithology by layer.Among 7793 stratigraphic well logs in 13 petroliferous basins The test results show that the proposed method can still achieve over 92% agreement with lithology identification under the condition of non-equilibrium data ratio of 1: 6, which has a good application prospect in the lithology identification of complex reservoirs .