Search

CN-121999148-A - Rock structural surface roughness high-precision mapping method and mapping system

CN121999148ACN 121999148 ACN121999148 ACN 121999148ACN-121999148-A

Abstract

The invention relates to the technical field of rock structure roughness mapping, in particular to a rock structure surface roughness high-precision mapping method and a rock structure surface roughness high-precision mapping system. The rock structural surface roughness analysis method comprises the steps of firstly, providing a rock classification model which classifies 3D point cloud data of a rock structural surface, adding a rock feature weighting layer to carry out weighting operation on geometric features, shape features and color features of a rock, secondly, providing a rock roughness mapping model which realizes mapping on local roughness and global roughness of the rock structural surface through a structural surface point cloud clustering algorithm, and finally, providing a rock structural surface roughness analysis method which establishes a three-dimensional rock structural surface model according to the obtained 3D point cloud data of the rock structural surface, carries out meshing division on the three-dimensional rock structural surface model, and carries out cause analysis on the rock structural surface roughness obtained through mapping by combining rock types and various related data.

Inventors

  • GAO HONGYI
  • Ma Ruilei

Assignees

  • 高弘毅

Dates

Publication Date
20260508
Application Date
20240420

Claims (10)

  1. 1. A rock structure surface roughness high-precision mapping method is characterized by comprising the steps of obtaining 3D point cloud data of a rock structure surface by using LiDAR technology, preprocessing the 3D point cloud data to obtain standard rock structure surface 3D point cloud data, carrying out rock classification by using a rock classification model to obtain a rock classification result, carrying out roughness mapping on the rock structure surface by using a rock roughness mapping model to obtain rock structure surface roughness, carrying out analysis on the rock structure surface roughness reasons to obtain rock structure surface roughness analysis vectors, wherein the obtaining process of the rock structure surface roughness analysis vectors comprises the steps of building a three-dimensional rock structure surface model according to the 3D point cloud data, dividing the three-dimensional rock structure surface model into K grids, wherein K is a natural number larger than 1, marking the grids according to the rock structure surface roughness, carrying out clustering on the grids according to the rock structure surface roughness to obtain a clustering grid set, constructing a rock roughness analysis function according to rock information and history related data, wherein the rock information comprises types and rock components, the related data comprises climate and hydrologic data, the roughness analysis data and the rock roughness analysis function is carried out correction on the rock roughness analysis according to the roughness analysis vectors.
  2. 2. The high-precision mapping method for the rock structural surface roughness is characterized in that the preprocessing process of the standard rock structural surface 3D point cloud data comprises the steps of denoising the 3D point cloud data by adopting a noise processing technology to obtain denoising 3D point cloud data, carrying out point cloud registration on the denoising 3D point cloud data to obtain registration 3D point cloud data, removing background points of the registration 3D point cloud data to obtain rock structural surface 3D point cloud data, carrying out point cloud density correction on the rock structural surface 3D point cloud data by using a density correction algorithm to obtain corrected rock structural surface 3D point cloud data, and converting a data format of the corrected rock structural surface 3D point cloud data to obtain the standard rock structural surface 3D point cloud data.
  3. 3. The method for high-precision mapping of rock structural surface roughness according to claim 1, wherein the classification process of the rock classification model comprises: inputting the standard rock structural surface 3D point cloud data to a feature extraction layer to obtain a rock feature map, wherein the rock feature comprises geometric features, shape features and color features; inputting the rock feature map into a rock feature weighting layer to obtain a rock attention feature map, and inputting the rock attention feature map into a feature fusion layer to obtain a rock feature fusion map; Inputting the rock feature fusion map to a full-connection layer, and obtaining a rock classification probability value by a normalization function; and obtaining the rock classification result according to the rock classification probability value.
  4. 4. The high-precision mapping method for the roughness of the rock structural surface according to claim 3 is characterized in that the rock attention feature map is obtained by respectively inputting three convolution kernels with the size of 1 multiplied by 1 and one with the size of 1 multiplied by n into the rock feature map to respectively obtain a first feature map of the rock, a second feature map of the rock, a third feature map of the rock and a fourth feature map of the rock, performing rank transformation on the second feature map of the rock to obtain a rank transformation feature map of the rock, performing multiplication on the rank transformation feature map of the rock and the first feature map of the rock and the second feature map of the rock to respectively obtain a first product feature map of the rock and a second product feature map of the rock, performing multiplication on the first product feature map of the rock, obtaining a third product feature map of the rock, performing normalization calculation on the third product feature map of the rock, performing product with the first feature map of the rock to obtain a fourth feature map of the rock, and performing stitching on the fourth product feature map of the rock and the fourth feature map of the rock to obtain the attention.
  5. 5. The rock structural surface roughness high-precision mapping method according to claim 1, wherein the rock roughness mapping model comprises the steps of selecting any point in the standard rock structural surface 3D point cloud data, clustering the standard rock structural surface 3D point cloud data according to the any point by adopting a structural surface point Yun Julei algorithm, repeating the structural surface point cloud clustering algorithm to obtain a structural surface 3D point cloud data clustering set s i (1≤i≤N), wherein s i is represented as an ith structural surface 3D point cloud data clustering set, N is represented as a clustering set total number, and calculating a normal vector of each point in the structural surface 3D point cloud data clustering set to obtain a point cloud normal vector set PCBN i is expressed as a point cloud normal vector set of the ith structural plane 3D point cloud data clustering set; The method comprises the steps of representing a point cloud normal vector of a jth point, representing M as the total number of points in a 3D point cloud data clustering set of an ith structural plane, calculating a point cloud normal vector difference matrix PCNVDM i ＝(NVDV i,j ) M×M , representing PCNVDM i as a point cloud normal vector difference matrix of a point cloud normal vector set i, and representing NVDV i,j as a normal vector difference value between the point i and the point j, wherein the calculation formula of the normal vector difference value is as follows: Calculating the local roughness of the structural surface according to the point cloud normal vector difference matrix, wherein the calculation formula of the local roughness of the structural surface is as follows: The method comprises the steps of expressing SSLR n as a structural surface local roughness of an nth structural surface 3D point cloud data clustering set, selecting a neighborhood structural surface 3D point cloud data clustering set, calculating height difference information between the neighborhood structural surface 3D point cloud data clustering set and each point in the neighborhood structural surface 3D point cloud data clustering set, calculating a structural surface neighborhood standard deviation of the neighborhood structural surface 3D point cloud data clustering set according to the height difference information to obtain a structural surface neighborhood standard deviation set SPNSD = { PSD n |1 +.n +.N +.A, expressing PSD n as an nth structural surface neighborhood standard deviation, and obtaining the rock structural surface roughness according to the structural surface local roughness and the structural surface neighborhood standard deviation set, wherein the rock structural surface roughness is calculated according to the formula: wherein RRSP is expressed as the roughness of the rock structural surface, eta 1 is expressed as the coefficient of the local roughness of the structural surface, and eta 2 is expressed as the coefficient of the neighborhood standard deviation of the structural surface.
  6. 6. The high-precision mapping method for the rock structural surface roughness is characterized by comprising the steps of initializing a connection matrix according to the standard rock structural surface 3D point cloud data, traversing according to the standard rock structural surface 3D point cloud data, and calculating the connection relation between any point and the rest points, wherein the connection relation comprises a step of 1 which indicates that the connection relation exists, a step of 0 which indicates that the connection relation does not exist, a step of updating the connection matrix and clustering according to the connection relation, and a step of repeating the steps from the first step to the third step, and stopping all point clustering.
  7. 7. The rock structural surface roughness high-precision mapping system is characterized by comprising a sensor unit, a data processing unit, a rock detection unit and an analysis unit, wherein the sensor unit is used for acquiring 3D point cloud data of a rock structural surface, the data processing unit is used for preprocessing the 3D point cloud data, the rock detection unit comprises a rock type detection module and a structural surface roughness mapping module and is used for detecting the type and the structural surface roughness of rock, the analysis unit is used for analyzing the roughness reason of the rock structural surface, and the report generation unit is used for generating a report according to the results of the rock detection unit and the analysis unit.
  8. 8. The high-precision mapping system for the rock structural surface roughness is characterized in that the preprocessing process of the data processing unit comprises the steps of denoising the 3D point cloud data by adopting a noise processing technology to obtain denoising 3D point cloud data, performing point cloud registration on the denoising 3D point cloud data to obtain registration 3D point cloud data, removing background points of the registration 3D point cloud data to obtain rock structural surface 3D point cloud data, performing point cloud density correction on the rock structural surface 3D point cloud data by using a density correction algorithm to obtain corrected rock structural surface 3D point cloud data, and converting a data format of the corrected rock structural surface 3D point cloud data to obtain standard rock structural surface 3D point cloud data.
  9. 9. The high-precision mapping system for rock structural surface roughness is characterized in that the specific process of the rock type detection module comprises the steps of inputting standard rock structural surface 3D point cloud data to a feature extraction layer to obtain a rock feature map, inputting the rock feature map into a rock feature weighting layer to obtain a rock attention feature map, inputting the rock attention feature map to a feature fusion layer to obtain a rock feature fusion map, inputting the rock feature fusion map to a full connection layer, obtaining a rock classification probability value by a normalization function, and obtaining a rock classification result according to the rock classification probability value.
  10. 10. The high-precision mapping system for rock structure surface roughness according to claim 7, wherein the analysis unit comprises a three-dimensional rock structure surface model built according to the 3D point cloud data, the three-dimensional rock structure surface model is divided into K grids, K is a natural number larger than 1, the grids are marked according to the rock structure surface roughness, the grids are clustered according to the rock structure surface roughness to obtain a clustered grid set, a rock roughness analysis function is constructed according to rock information and history related data, the rock information comprises rock types and rock components, the history related data comprises climate data, hydrogeology data and artificial activity data, the roughness influence of the clustered grids is calculated by the rock roughness analysis function, the roughness influence is corrected according to the rock structure surface roughness, and a rock structure surface roughness analysis vector is output.

Description

Rock structural surface roughness high-precision mapping method and mapping system Technical Field The invention relates to the technical field of rock structure roughness mapping, in particular to a rock structure surface roughness high-precision mapping method and a rock structure surface roughness high-precision mapping system. Background Rock is a solid substance consisting of one or more minerals, one of the main components of the crust. They are formed by geologic action and are classified according to their composition, structure and morphological characteristics. Rock is generally classified into three main types, igneous, sedimentary and metamorphic. The rock types have different causes, properties and purposes and have important significance for the earth science, engineering geology and engineering construction. Geologist have included many aspects in rock research such as rock classification, rock formation analysis, and rock property analysis, and are important in many studies for mapping rock structural planes. The rock structural plane refers to a fracture plane or a crack plane existing in the rock, and is an important component of the internal structure of the rock. Formation of rock face may be affected by a number of factors such as geologic formation movement, rock deformation, and earth's surface weathering attack. The mapping aspects of the rock structural surface, including joint mapping, layer mapping, fault mapping, roughness mapping, etc., relate to the detailed description and recording of the internal structural and structural features of the rock. Rock structural surface roughness mapping refers to the characteristic of surface irregularities, which has an important impact on engineering properties and geological properties of rock. With the continued update of geologic mapping techniques, researchers have proposed many effective methods for mapping rock structural planes. As in 2012, scholars Ge Yunfeng studied the influence of non-stationarity on roughness quantification, and proposed a new method for removing non-stationary features in rock joint point cloud data, which emphasizes that the inclination angle (non-stationarity) of the coarse features in the profile plays an important role in quantifying joint roughness, and suggests that the non-stationary features on the profile are primarily removed when evaluating the coarse features of the profile. At present, some defects still exist in a plurality of existing documents, for example, china patent with the prior art application number of CN201910162055.5 discloses a rock structural surface roughness coefficient refined characterization method, the method obtains morphology data of a rock structural surface, preprocessing three-dimensional point cloud data, conducting grid thinning processing on the point cloud data, conducting regular sequencing operation on seat data, calculating JRC values of all structural surface contour lines in the point cloud data, drawing a JRC numerical frequency distribution histogram, describing coarse conditions of the rock structural surface through the histogram, in addition, china patent with the prior art application number of CN202311447681.1 discloses a rock structural surface roughness intelligent extraction method and system. However, the above mentioned prior documents all have a defect that the roughness mapping of the rock structural surface is not comprehensive enough, so that the high-precision mapping result of the roughness of the rock structural surface cannot be obtained, and the specific analysis of the roughness cause of the rock structural surface in the later stage is not facilitated. In order to solve the problems, the invention provides a rock structural surface roughness high-precision mapping method and a mapping system. Disclosure of Invention The invention provides a rock structural surface roughness high-precision mapping method and system for realizing high-precision mapping of the roughness of a rock structural surface. The method and the system mainly achieve the technical effect of high-precision roughness mapping from the following aspects, firstly, the invention provides a rock classification model for rock classification of acquired 3D point cloud data of a rock structural surface, the rock is divided into a plurality of types, roughly divided into igneous rock, sedimentary rock and metamorphic rock, but different types of rock are small in distinguishing property and difficult to judge, therefore, the method specifically distinguishes geometrical characteristics, shape characteristics and color characteristics of the rock through learning in the model, weights distinguishing characteristics by adding a rock characteristic weighting layer in the model for improving the accuracy of rock classification, secondly, the invention provides a rock roughness mapping model for roughness mapping of the rock surface, the model clusters 3D point cloud data of the rock surf