BP神经网络技术在识别西湖凹陷平湖组煤层中的应用

The Application of BP Neural Network Technology in Identifying Coal Seam of Pinghu Formation in Xihu Sag

摘要: 西湖凹陷是东海陆架盆地内最大的含油气凹陷，主力烃源岩平湖组为含煤的煤系地层，煤层具有厚度薄、层数多、生烃能力强的特征，而煤层的准确厚度常难以计算。本文利用BP神经网络技术，以实际岩性为学习预测样本，以对煤层响应敏感的测井参数为输入变量，通过对建立的三层BP神经网络模型的训练测试来实现对煤层的识别。结果表明，声波时差、中子和密度对煤层的测井响应最明显，利用BP神经网络技术识别出的煤层数量比原测井解释的煤层数有所增加，能识别出薄煤层，且与实际岩性的吻合度更高。BP神经网络技术可更好地识别钻井剖面中的煤层，尤其是薄煤层，从而更准确地计算出钻井剖面中的煤层厚度，具有良好的准确性和实用性。

Abstract: Xihu sag is the largest oil and gas-bearing depression in the East China Sea Continental Shelf Basin. Pinghu Formation, the main source rock, is a coal-bearing strata, and the coal seam is characterized by thin thickness, multiple layers and strong hydrocarbon generation ability. However, the accurate thickness of coal seam is often difficult to calculate. This paper uses BP (Back Propagation) neural network technology, takes actual lithology as the learning and prediction sample, takes the logging parameters sensitive to coal seam as the input variable, and realizes the identification of coal seam through the training and testing of the established three-layer BP neural network model. The results show that acoustic time difference, neutron and density have the most obvious logging response to coal seam, and the number of coal seams identified by BP neural network technology is increased compared with the number of coal seams interpreted by original logging, and the thin coal seams can be identified, and the coincidence degree with the actual lithology is higher. BP neural network technology can better identify the coal seam in the drilling section, especially the thin coal seam, so as to calculate the coal seam thickness in the drilling section more accurately, which has good accuracy and practicability.