Lithology classification is a very important basic work in petroleum exploration and development, and accurate lithology recognition is critical to evaluate the reservoir properties in geophysical logging data. In view of the heterogeneity of the actual reservoir, this paper mainly based on the K clustering algorithm and DBSCAN clustering algorithm for a basin with 10-dimensional characteristics of the logging data set up a Lithologic classification identification model. According to the limitation of Lithology classification technology and the advantages of clustering algorithm, this paper designs Lithologic classification system, which can be used to identify and label the mass well logging data in the case of no excessive accuracy compared with machine learning classification algorithm. In this system, we first data processing of the original logging curve, get 10-D characteristic data as the experimental data set, then use K-means, MeanShift and DBSCAN cluster model to cluster the experimental data, and then through the manual tagging method, Finally, the logging data with Lithology label is obtained, and the classification of lithology is realized. Compared with the traditional logging data processing method based on linear mathematics, clustering analysis algorithm improves the accuracy of classification recognition, and the recognition result is closer to the real characteristics of reservoir, and has the advantages of good classification ability and adaptability and easy deployment. The IPPTC- 20181934 Clustering Analysis algorithm also lays a good foundation for identifying the possible unknown lithology.
本作品采用
知识共享署名
4.0 国际许可协议进行许可。
CSTR:11738.14.NCDC.XDA14.PP6143.2024