Heterogeneous rock digital rock core modeling method based on K-means clustering algorithm

Abstract

The invention discloses a heterogeneous rock digital rock core modeling method based on a K-means clustering algorithm, and belongs to the field of rock physical mechanics numerical simulation. The method comprises the following steps: firstly, carrying out XRD mineral composition testing, high-pressure mercury injection porosity testing, field emission environment SEM and CT scanning experiments on heterogeneous rocks; counting the mean value of the gray values of each mineral component of the BSE gray image to serve as the initial clustering center of the K-means clustering algorithm; preprocessing the heterogeneous rock CT scanning image; calculating the distance between the gray value of the CT image and the initial clustering center, and judging the membership degree of the gray value belonging to each category; and iteratively updating the clustering center and the gray value membership degree of the CT image to enable the target function to be minimum, completing the gray value classification and image segmentation of the CT image, and constructing the digital core of the heterogeneous rock. According to the method, the problems that the accuracy of a division result of randomly selecting the initial centroid and the k value cannot be determined and segmentation errors are caused by segmenting the image according to a threshold value can be solved, and the heterogeneous rock CT image data are optimally divided.

Description

Technical field [0001] The present invention relates to a non-homogeneous rock digital core modeling method, and is especially suitable for a non-homogeneous rock digital core modeling method based on K-Means clustering algorithm in the field of rock physical mechanics. Background technique [0002] The rock is a complex non-uniform geometry constituted by the hinders, structural surfaces, and weaknesses of different mineral particles units. These micro-component units determine the rupture characteristics of rocks and macroscopic mechanical properties, and inherent non-homogeneousness is an important reason for the complex rock mechanical properties. At present, the mechanical properties of the non-homogeneous rock microscopic scale are still facing huge challenges. The numerical methods of rock mechanics based on fine structures are usually used to assign the mechanical properties of rock materials. The authenticity of complex constituent rock reflections cannot be determined. ...

AI Technical Summary

Problems solved by technology

In view of the fuzzy edges of pores and minerals in CT images of heterogeneous rocks and the low contrast of gray values ​​between different minerals, the threshold segmentation algorithm and watershed algorithm lack the applicability to separate different mineral components, which may easily lead to under-segmentation or over-segmentation of the image.

K-means algorithm, as a classic algorithm in clustering algorithm, is fast and convenient to calculate, but its initial clustering center and the number of initial clustering centers need to be set manually, which has a great impact on the segmentation results

Images