用自组织特征映射神经网络识别岩相

为了解决测井岩性识别问题,引入有较强的聚类和容错能力的自组织特征映射(SOFM)神经网络。文中说明了SOFM网络的模型,并结合实际探井资料,用MATLAB语言建立SOFM网络岩性识别模型,并进行了具体的应用研究。通过与已知资料的对比,证明该方法是一种行之有效的岩相识别方法,且具有良好的应用前景。

The Logging Lithological Identification by Using Self-organizing Feature Map Neural Networks

The self organizing feature map(SOFM) neural network has superior clustering and fault-tolerant ability.In this paper SOFM was used to solve the well-logging lithological identification.We described firstly SOFM network modeling and then,by combined it with a set of actual well-log data,a lithological identification model was set up by Matlab programming and the specific research on this field was carried on.By compared the SOFM-based results with some known well-logging information,it was shown that the SOFM modeling is very effective and efficiency for the well-logging lithological identification.It is also shown that SOFM modeling has a good prospect for the petroleum reservoir engineering.