The automatic recognition of petrologic features based on gene expression programming in well logging

摘要

The geologic interpretation of log information is important in well logging. Many problems exist in the recognition of petrologic features, such as few core materials, massive log information and fuzzy parameter distribution. Neural Network is a commonly used method in the automatic recognition of petrologic features, but its convergence rate is slow. Gene Expression Programming (GEP) is a new genetic algorithm based on the genome group and the phenotype group, which works with greatly strengthened function discovery ability and high efficiency. This paper introduces GEP into the automatic disposal of log information. We did experiment on 4 wells in one oil field area, and the result proved GEP's validity in automatic modeling. Compared with the existing methods, GEP shows its advantages in automation, diversity of model structure and higher precision in data fitting and predication. GEP provides a new method in the automatic recognition of petrologic features.