An automatic lithology naming method based on logistic regression

Abstract

The invention discloses an automatic lithology naming method based on logistic regression, which comprises the following steps: 1) Obtain the mineral composition percentage data of rock samples to be named, and obtain the structural characteristics of grain size, roundness and cement composition by visual observation ; 2) Collect the rock sample data with existing thin section identification and naming conclusions as the training set; 3) Take the layer to be named as the test set, take the thin section identification conclusion as the dependent variable, take the particle size, roundness, and cement as the factors, and take the mineral Constituent percentage content is a covariate for unordered multi-category logistic regression, and a logistic regression model of various lithologies is established; 4) The mineral composition percentage, grain size, roundness, and cement composition data of the layer to be named are substituted into the logistic regression model; 5) According to the principle of maximum subordination, the lithology of the layer to be named is named. The method of the invention makes the naming result more scientific and objective, solves the problem that qualitative data such as structure cannot participate in quantitative calculation, and can meet the demand for quick naming while drilling on site.

Description

technical field [0001] The invention relates to the field of petroleum exploration and development, in particular to a method for naming lithology in the field of mud logging technology in the process of petroleum exploration and development, in particular to an automatic lithology naming method based on logistic regression. Background technique [0002] Lithology naming is the basic work of mud logging, which plays an important role in later reservoir evaluation. At present, the methods of mud logging lithology naming mainly include ① naked eye observation; ② thin section identification; ③ X-diffraction whole-rock analysis. The current lithology naming method has the following deficiencies: [0003] 1. Inner eye observation is the first-hand information on site, but it has high requirements for naming personnel, and the method is simple, which cannot meet the needs of fine naming; [0004] 2. Observation and naming under the microscope are accurate, the process is complex,...

AI Technical Summary

Problems solved by technology

[0003] 1. Inner eye observation is the first-hand information on site, but it has high requirements for naming personnel, and the method is simple, which cannot meet the needs of fine naming;

[0004] 2. Observation and naming under the microscope are accurate, the process is complicated, the requirements for personnel and equipment are high, and the identification period is long, which is not suitable for the field and cannot meet the needs of the field naming while drilling

[0005] 3. Through X-diffraction whole-rock analysis, the mineral composition of the rock can be obtained for naming. Currently, methods for naming mineral composition include curve intersection method, plate analysis method, etc. These methods use fewer mineral types and low dimensional utilization. Not suitable for high-dimensional mineral analysis, and no reference to factors such as mineral structure when naming

Images