Lithology identification method and system

Abstract

The invention provides a lithology identification method and system, and relates to the field of deep learning. The lithology identification method comprises the following steps: establishing a closed lithology section, and classifying and judging rock data samples according to different numbers of the rock data samples of a lithology identification data set formed by the lithology section; carrying out data preprocessing according to the rock data samples, removing true outliers from the rock data samples, carrying out oversampling on the samples with the rock data sample types marked as effective, and obtaining a balanced stratum rock data set. According to the method and system, a lithology identification data set can be formed through a lithology section, unified lithology identification is carried out, different parameters correspond to the characteristics of lithology, and after data mining is carried out on the lithology identification data set, a clear identification model is formed. In addition, the invention further provides a lithology identification system which comprises a classification and judgment module, a stripping module and an identification and classification module.

Description

technical field [0001] The present invention relates to the field of deep learning, in particular to a method and system for identifying lithology. Background technique [0002] Data mining is the process of extracting implicit, previously unknown, but potentially useful information and knowledge from a large number of incomplete, noisy, fuzzy, and random data. Data mining tasks are divided into two categories: description and prediction. The former derives a generalized pattern of potential relationships in the data, and the latter infers the current data to make predictions. The main methods are neural networks, support vector machines, Bayesian networks, and decision-making. Among them, BP neural network has been widely used in lithology identification, sedimentary facies division, permeability prediction, oil, gas and water layer identification, etc. Practice has proved that when dealing with complex geological problems affected by multiple factors, nonlinear neural netw...

AI Technical Summary

Problems solved by technology

However, one of the biggest disadvantages of BP neural network is that it is easy to "over-learn" the training samples. The method has too many variable parameters. If the algorithm is given enough time, it can "remember" almost anything, and the model established in this way will Without the geological background, there is no practical application value; in addition, neural networks, support vector machines and Bayesian networks have a common shortcoming that the predicted models are all "black boxes", and the relationship between sample data and attributes and relational fitting is not visible

[0003] Similarly, in the existing lithology identification, the KNN algorithm is mostly used to determine the rock sample type, but KNN may require a large amount of memory or space to store all the data, and the measurement method using distance or proximity may be difficult in crashes in very high dimensionality (with many input variables), which can negatively affect the performance of the algorithm on your problem

Images