TWI399527B - Interpretation method and system of outlier position of rock slope - Google Patents

Description

Rock slope outcrop position state interpretation method and system

The invention relates to a rock slope outcrop position state interpretation method and system, in particular to a rock slope outcrop position state interpretation method and system using a genetic algorithm.

In recent years, in the typhoon and torrential rains in Taiwan's mountainous areas, soil sand disasters have often been reported, which seriously threaten the safety of residents' lives and property. Therefore, the government has enacted laws and regulations to restrict and standardize the development of hillsides. The text contains the development of hillsides. Geological surveys must be carried out before, including the soil, rocks, geological structures, weak surfaces (discontinuous surfaces) and geological processes within the project area and its impact areas. Among them, the weak surface is often the main factor controlling the strength and stability of the rock mass.

In the geological survey project that must be carried out before the development of the hillside, the weak surface is formed by the external force or natural sedimentation process. The weak surface can be divided into the primary weak surface and the secondary weak surface depending on the occurrence of diagenesis before and after solidification. The former is the layer, the unconformity, and the latter is the fault, joint, cleavage, and leaf.

Generally, the local measurement of the weak surface state is measured by the geological compass instrument directly on the surface of the rock slope outcrop. After the body projection map is formed, the engineer is judged. This traditional method requires contact with the outcrop surface to measure. However, if the outcrop is blocked by the terrain or is a dangerous slope with potential collapse factors, the measurement is more difficult and has a higher risk. Therefore, it is necessary to seek a solution.

Accordingly, it is an object of the present invention to provide a rock slope outcrop positional interpretation method.

Therefore, the method for predicting the position of the rock slope outcrop of the present invention comprises the following steps: (A) obtaining a set of three-dimensional point cloud data corresponding to the surface of the outcrop site of a rock slope by three-dimensional field scanning; (B) utilizing A gridding and complementing technique, meshing and complementing the set of three-dimensional point cloud data to obtain a grid graphic file representing a grid pattern, wherein the grid pattern is composed of multiple grids (C) calculate the unit normal vector of all grids; and (D) divide the grid pattern into multiple clusters by specifying a group of poly numbers, and use a gene algorithm to derive unit normal vectors from the grids Calculate the unit normal vector of each clustered cluster center.

Another object of the present invention is to provide a rock slope outcrop positional interpretation system. The rock slope outcrop positional interpretation system is suitable for the user to determine the outcrop position of the rock slope outcrop site by a set of three-dimensional point cloud data corresponding to the surface of the one rock slope outcrop site. The system comprises a grid and patch module, a grid position calculation module and a gene algorithm module. The gridding and complementing module is used for meshing and complementing the set of three-dimensional point cloud data to obtain a grid graphic file representing one of the grid patterns, wherein the grid pattern is composed of multiple Grid composition. The grid position calculation module is used to calculate unit normal vectors of all grids. The gene algorithm module is configured to divide the grid pattern into a plurality of clusters according to a cluster number specified by the user, and calculate a unit normal vector from the grids by using a gene algorithm. The unit normal vector of each group of clusters.

The invention has the effect of avoiding the risk of falling from a high position when using the geological compass to measure the outcrop position in the field, and obtaining the result in a short time.

The above and other technical contents, features and advantages of the present invention will be apparent from the following detailed description of the preferred embodiments.

Before the present invention is described in detail, it is noted that in the following description, similar elements are denoted by the same reference numerals.

The rock slope outcrop positional interpretation method and system of the present invention utilizes a three-dimensional laser scanner to capture the point clouds of the outcrop surface, and indirectly through the use of the internal analysis of the Genetic Algorithm. The purpose of measuring the position of the outcrop is to avoid the danger of traditional direct measurement of the surface of the outcrop and the difficulty of implementation.

Referring to FIG. 1, a preferred embodiment of the rock slope outcrop positional interpretation system 1 of the present invention is suitable for a user to determine the rock slope by a set of three-dimensional point cloud data corresponding to the surface of a rock slope outcrop site. The outcrop position of the rock slope of the outcrop site. The rock slope outcrop positional interpretation system 1 comprises a three-dimensional laser scanner 11, a Global Positioning System (GPS) module 12, a gridded and complementary point module 13, and a grid position. The state computing module 14, a gene algorithm module 15 and a group of cluster computing modules 16. The detailed implementation steps of the present invention and the functions of the components 11-16 will be described below with reference to the flow chart of the rock slope outcrop positional interpretation method of the present invention.

Referring to FIG. 1 and FIG. 2, as shown in step 21, the present invention performs remote scanning using the three-dimensional laser scanner 11 to obtain point cloud data of a rock slope surface, wherein the three-dimensional laser scanner 11 is, for example, Available as Dibit LPM98 GeoScanner.

Then, as shown in step 22, in the preferred embodiment of the present invention, when the 3D laser scanning is performed in the field, the GPS module 12 performs GPS measurement to obtain the absolute coordinates of each data point, wherein the GPS module 12 includes a GPS receiver and a GPS data reader (not shown). In the preferred embodiment, the GPS receiver needs to receive the satellite transmission data at a fixed point for at least 1-2 hours to obtain more accurate data.

Then, as shown in step 23, the meshing and complementing module 13 performs meshing and complement processing on the set of three-dimensional point cloud data to obtain a grid graphic file representing one of the grid patterns, wherein A grid graphic consists of multiple meshes, such as a triangular mesh. In the preferred embodiment, the meshing and complementing module 13 utilizes the Fast Radial Basis Function Toolbox (Fast RBF Toolbox) processing technique in MATLAB to obtain the grid graphic file. The FastRBF Toolbox is based on the RBF function, and obtains a fast RBF function by specifying the precision, which simplifies the huge computational amount required to obtain the RBF. The RBF function is as shown in the following formula (1):

Where: s ( x ): radius function, p ( x ): low-order polynomial, one or quadratic polynomial, λ _i : RBF coefficient, Φ: basic function, x _i : center point cloud required for RBF.

The preferred embodiment of the present invention utilizes the MATLAB FastRBF Toolbox for data interpolation and meshing of point cloud data, wherein the RBF can represent a set of three-dimensional point data by a set of mathematical functions. Thereby, the present invention can perform interpolation and slope reconstruction for a rock slope that cannot be scanned due to the angle of view or an unevenly scattered point cloud generated by vegetation cover. Therefore, the data generated by interpolating and meshing the point cloud by the FastRBF Toolbox can be output for subsequent data analysis.

Then, as shown in step 24, the grid position calculation module 14 is used to calculate the unit normal vector and the inclination and inclination of all the grids. The grid position calculation module 14 calculates unit normal vectors, dips, and tendencies of all grids based on the following theoretical basis. Referring to Figure 3, the position of the weak face can be expressed by Dip and Dip Direction (DD). The dip angle is the acute angle at which the weak surface intersects the horizontal plane. The tendency is to describe the direction in which the weak face is tilted, with an angular range between 0 and 360 degrees, which is 90° out of phase with Strike. The definition and relationship of dip, tendency and direction are shown in Figure 3.

After the point cloud data is interpolated and meshed by the meshing and patching module 13, a new set of point cloud data and a set of grid matrices are generated to represent the grid pattern. The grid matrix is a 3×N matrix, and the point cloud numbers constituting each grid in the point cloud data are recorded, and the entire grid pattern contains N triangle grids. As shown in the following formula (2), using the characteristics of the three-point constituting plane, the plane equation of each triangular mesh can be calculated by formula (2), where A, B, and C represent the coefficients of X, Y, and Z, respectively, and D. Is a constant term:

AX+BY+CZ+D=0 (2)

Since the inclination angle is the angle between the weak surface and the horizontal plane, the angle range is from 0 to 90°, so the inclination angle is the normal vector of the triangular grid plane equation (A, B, C) and the normal vector of the XY plane (0, 0, 1). The angle of intersection is as shown in the following formula (3):

As for the calculation of the tendency, the following also refers to the weak surface geometric relationship of FIG. It is assumed that the plane P is a weak plane, the straight line L is the intersection of the weak plane P and the horizontal plane, and the vector n' is the horizontal projection vector of the normal vector n of the plane P. According to the definition of the direction and the direction of the inclination, θ is the direction of the weak surface, and φ is the tendency of the weak surface. When calculating the tendency, if the angle between the horizontal projection vector (A, B, 0) of the triangular mesh normal and the Y-axis vector (0, 1, 0) is directly calculated as the weak surface, an error will occur, because the two The cosine value (cos(κ)) obtained by the inner product of the vector, after the inverse trigonometric function (cos ^-1 (κ)), has a value between 0 and 180°, and the range of the tilt angle ranges from 0 to 360°. Therefore, the true dip direction must be calculated by discriminant, that is, the normal vector n and the cosine values of the Y and Z axes are used to judge the normal vector of the grid equation to point to the Octant. If α and γ are the intersection angles of the normal vector n and the X-axis and the Z-axis, respectively, the normal vector of the weak surface P of FIG. 3 is taken as an example. If the calculated normal vector is equal in size and direction is the same as the vector n, the calculation method is obtained. The vector points to the fourth limit (X>0, Y<0, Z>0), and its cos(α)>0, cos(γ)>0, then DD=cos ^-1 (κ). If the calculated normal vector is equal to the magnitude of the vector n and the direction is opposite, the calculated normal vector points to the sixth limit (X<0, Y>0, Z<0), and its cos(α)<0, cos(γ) ) < 0, the tendency is DD = 180 ° - cos ^-1 (κ). The correlation formula and discriminant formula of the tendency DD are as shown in the formulas (4) to (10).

Propensity = cos ^-1 (κ), if cos(α)≧0 and cos(γ)>0 (7)

Propensity = 180° + cos ^-1 (κ), if cos(α) > 0 and cos(γ) ≦ 0 (8)

Tendency = 360°-cos ^-1 (κ), if cos(α)<0 and cos(γ)≧0 (9)

Tendency = 180°-cos ^-1 (κ), if cos(α)≦0 and cos(γ)<0 (10)

Next, as shown in step 25, the gene algorithm module 15 is configured to divide the grid pattern into K clusters according to a cluster number K specified by the user, and use a gene algorithm to The unit normal vector of the grid calculates the unit normal vector of the cluster center of each cluster. The detailed flow of the gene algorithm used in the present invention will be described below with reference to FIG.

See Figure 4 for a flow chart of the gene algorithm. Gene algorithm is a superior random search technique. The basic principles underlying gene algorithms are natural selection (ie, Darwin's theory of evolution) and genetic recombination. That is, individuals who are more adaptable to the environment have a higher chance of survival and are more likely to be selected to participate in the work of breeding offspring. These individuals are selected in pairs as parents and produce offspring by means of genetic recombination. In addition, mutations may occur during the process of pairing the individual to produce an individual with special characteristics. This process of genetic recombination and mutation will occur from generation to generation.

Gene algorithms are widely used to solve multiple optimization problems in the real world, mainly because of their ease of programming and their ability to search for the best solutions in the universe. In the process of using the gene algorithm, the user first specifies the cluster number K , and then the gene algorithm module 15 (ie, the gene algorithm program module) automatically searches for the cluster center of many unit normal vectors. Then, the cluster state calculation module 16 obtains the inclination and inclination from the cluster centers. Therefore, the present invention can avoid the risk of falling from a high position in the conventional use of the geological compass to measure the outcrop position, and can obtain the result in a shorter time due to the use of the genetic algorithm. In fact, gene algorithms deal with this type of problem with a very fast convergence rate. In most cases, a satisfactory solution can be obtained as long as the calculation reaches 200 generations.

As shown in Figure 4, a typical gene algorithm program usually consists of the following elements:

(1) selecting parameters of the genetic algorithm;

(2) Define the fitness and fitness function;

(3) generating a random initial population;

(4) Calculating the fitness of individual chromosomes;

(5) choose a spouse;

(6) mating;

(7) Mutation.

The parameters of the genetic algorithm are selected by the user in advance and set in the gene algorithm module 15. Gene algorithm parameters include population size, selectivity, mutation rate, range, and stopping criteria. In a preferred embodiment of the present invention, the user can set the genetic algorithm parameters in the program of the gene algorithm module 15 as follows: the population size is 12, the selection rate is 0.5, and the mutation rate is 0.2. All values are between 1 and -1, with Population 12 being determined after multiple tests and applicable to the problems faced in embodiments of the present invention.

Similarly, the fitness and fitness functions are also predefined by the user and written in the gene algorithm module 15.

When using gene algorithms in different application areas, it is necessary to decide whether or not to encode a chromosome using a specific coding method depending on the actual situation, and to decode it when estimating the fitness level. However, in the preferred embodiment of the invention, since a unit normal vector is composed of three real numbers, the present invention can directly process the data without encoding. Such a gene algorithm that does not require an encoding and decoding process can be referred to as a continuous gene algorithm. Figure 4 is a flow chart of a continuous gene algorithm used in the present invention.

After the grid position calculation module 14 calculates the grid positions of all the grids as described above (as shown in step 24 of FIG. 2), the gene algorithm module 15 is generated as shown in step 251 of FIG. A random initial population in which each staining system consists of three real components of the unit normal vector of the cluster centers of all groups. The features (genes) selected in the present invention are K cluster centers, where K is the cluster number. Each cluster center is a unit normal vector, which is determined by the parameters of X, Y, and Z (between -1 and 1). Therefore, a chromosome consists of K cluster centers (or 3 x K real numbers). The number of chromosomes and the size of the population are important gene algorithm parameters and must be specified with caution. If a small population is used for calculation, the probability that the resulting candidate solution is close to the actual solution is lower. However, a larger population size may result in a decrease in the computational rate because the degree of fit of relatively more candidate solutions must be estimated. Therefore, in the present invention, it is necessary to perform multiple tests on different population sizes to determine the most suitable number of ethnic groups. As described above, the population size in the preferred embodiment of the present invention is 12 (i.e., the population size is 12 chromosomes), which is determined after a plurality of tests, and is applicable to the problems faced in the embodiments of the present invention.

then. As shown in step 252, the gene algorithm module 15 calculates the fitness value of each chromosome according to the fitness function of the following formula (11).

Where d _ik is the Euclidean Distance of a data point from the Kth cluster center.

Next, as shown in step 253, the gene algorithm module 15 performs a mating procedure based on the calculated fitness value and a predetermined selection rate. Since chromosomes with high fitness have a high probability of survival and generation of offspring, the chromosomes with low fitness levels will be eliminated, so the selection rate is the ratio of the chromosomes retained. Each chromosome organic rate is selected as a parent, and the probability of being selected is usually determined by the rank of its fitness. A preferred embodiment of the present invention uses Rank Weighting to determine the probability P _{i of a} chromosome being selected as a parent, as shown in the following equation (12):

Where m is the number of chromosomes retained and i is the order. Chromosomes with high ranks have a higher probability of being selected for pairing and produce ( n - m ) progeny, where n is the population size in order to maintain a fixed population size n . Thus, progeny can be generated via mating procedures to ensure that their genes are joined by the parent. Commonly used mating methods are single point mating, double point mating, and uniform mating, and a preferred embodiment of the present invention employs a single point mating method.

Next, as shown in step 254, the gene algorithm module 15 performs a mutation procedure based on a pre-specified mutation rate. That is, the preferred embodiment of the present invention also introduces a percentage of random mutations when performing a gene algorithm, that is, some genes are randomly selected and changed. This mechanism can potentially introduce new features that never appear on the original chromosome, and therefore have the opportunity to search in a new space. However, in order to avoid the most appropriate chromosomes being altered, in a preferred embodiment of the invention, the elite strategy is applied to preserve the optimal solution. That is, the Matlab program written in the preferred embodiment of the present invention keeps the most suitable chromosome intact and does not participate in the mutation, so that the optimal solution of the parent can be retained.

Then, as shown in step 255, the gene algorithm module 15 determines whether a stop criterion has been met. If the result of the determination is YES, it means that the unit normal vector of the cluster center of each cluster is calculated and the operation is stopped. On the other hand, if the determination result is no, the above steps 252 to 255 are repeatedly performed until the stop criterion is met and the operation is stopped. In a preferred embodiment of the invention, the gene algorithm module 15 is programmed to stop operations in the 200th generation. Alternatively, the program executor may also set a preset calculation number of the appropriate value (20 times in the preferred embodiment of the present invention), and when the value of the appropriateness is calculated, the cumulative 20 times are the same, that is, Stop the operation.

After the operation of the gene algorithm of FIG. 4 is completed (ie, step 25 of FIG. 2), then as shown in step 26 of FIG. 2, the cluster state calculation module 16 calculates the unit of the cluster center of each group according to the calculation. The normal vector, and the above formulas (3) to (10), calculate the inclination and inclination of the outcrop position of each group.

In summary, the rock slope outcrop position interpretation method and system of the present invention can avoid the danger that the traditional measuring personnel may face a high position when using the geological compass to measure the outcrop position by using the cooperative operation of the modules. Since the gene normal algorithm is used to obtain the unit normal vector of the cluster center of each cluster, the rock slope outcrop position can be obtained in a short time, so the object of the present invention can be achieved.

The above is only the preferred embodiment of the present invention, and the scope of the invention is not limited thereto, that is, the simple equivalent changes and modifications made by the scope of the invention and the description of the invention are All remain within the scope of the invention patent.

1. . . Rock slope outcrop positional interpretation system

11. . . 3D laser scanner

12. . . Global positioning system module

13. . . Grid and patch module

14. . . Grid position calculation module

15. . . Gene algorithm module

16. . . Clustering position calculation module

21~26. . . step

251~255. . . step

1 is a system block diagram showing a preferred embodiment of the system architecture of the rock slope outcrop positional interpretation system of the present invention;

2 is a flow chart showing a preferred embodiment of the rock slope outcrop positional interpretation method of the present invention;

Figure 3 is a schematic diagram showing the spatial relationship of the weak faces;

Figure 4 is a flow chart showing the method flow of the gene algorithm used in the present invention.

1. . . Rock slope outcrop positional interpretation system

11. . . 3D laser scanner

12. . . Global positioning system module

13. . . Grid and patch module

14. . . Grid position calculation module

15. . . Gene algorithm module

16. . . Clustering position calculation module

Claims (10)

A rock slope outcrop positional interpretation method comprises the following steps: (A) obtaining a set of three-dimensional point cloud data corresponding to the surface of a rock slope outcrop site by three-dimensional field scanning; (B) utilizing a network Grid and complement technique, meshing and complementing the set of 3D point cloud data to obtain a grid graphic file representing a grid pattern, wherein the grid pattern is composed of a plurality of grids; C) calculating the unit normal vector of all the grids; and (D) dividing the grid pattern into a plurality of clusters by specifying a group of poly numbers, and calculating the unit normal vectors from the grids by using a gene algorithm The unit normal vector of each cluster center. According to the rock slope outcrop positional interpretation method described in item 1 of the patent application scope, between the steps (A) and (B), the use of a global fixed-air system technology to obtain the absolute value of the three-dimensional point cloud data of the group is also included. coordinate. According to the rock slope outcrop position state interpretation method described in claim 1, wherein the meshing and complementing point technique is a fast radius function processing technique. The rock slope outcrop positional interpretation method according to claim 1, wherein the (D) step comprises the following substep: (D-1) generating a random initial population comprising a plurality of chromosomes, wherein each chromosome It consists of three real components of the unit normal vector of all clustered cluster centers; (D-2) calculates the fitness value of each chromosome according to a predefined fitness function; (D-3) According to the calculated fitness value and a selection rate, a mating procedure is performed on the chromosomes to obtain a mating population; (D-4) according to the mating population, and a mutation rate, The chromosome in the mating population undergoes a mutation procedure to obtain a mutant population; and (D-5) determines whether a stop criterion is met, and if the determination result is yes, it means that each cluster required is calculated. The unit normal vector of the cluster center, and if the determination result is no, the steps (C-3) to (C-6) are continued. According to the method for judging the rock slope outcrop position according to Item 1 of the patent application scope, after the step (D), calculating, according to the calculated unit normal vector of the cluster center of each cluster, The inclination and inclination of the outcrop position of a group of poly rock slopes. A rock slope outcrop positional interpretation system is suitable for a user to determine a rock slope outcrop position of a rock slope outcrop site by a set of three-dimensional point cloud data corresponding to a surface of an rock slope outcrop site. The system comprises: a gridding and patching module for meshing and complementing the set of three-dimensional point cloud data to obtain a grid graphic file representing a grid pattern, wherein the grid The graphic is composed of a plurality of meshes; a grid position calculation module for calculating a unit normal vector of all the grids; and a gene algorithm module for using one of the cluster numbers specified by the user The grid pattern is divided into a plurality of clusters, and a gene algorithm is used to calculate a unit normal vector of the cluster center of each cluster from the unit normal vectors of the grids. The rock slope outcrop positional interpretation system according to item 6 of the patent application scope also includes a global fixed system module for obtaining the absolute coordinates of the set of three-dimensional point cloud data. According to the sixth aspect of the patent application scope, the rock slope outcrop position state interpretation system, wherein the grid and the complement point module utilizes a fast radius function processing technique to obtain the grid pattern file. The rock slope outcrop positional interpretation system according to item 6 of the patent application scope, wherein the gene algorithm module is configured to perform the following steps: (E-1) generating a random initial group comprising a plurality of chromosomes, wherein each A dyeing system consists of three real components of the unit normal vector of all clustered cluster centers; (E-2) calculates the fitness value of each chromosome according to a predefined fitness function; -3) performing a mating procedure on the chromosomes according to the calculated fitness values and a selection rate to obtain a mating population; (E-4) according to the mating population, and a mutation rate a mutation program is performed on the chromosomes in the mating population to obtain a mutant population; and (E-5) determines whether a stop criterion is met, and if the determination result is yes, it indicates that each group required has been calculated If the unit normal vector of the cluster center is gathered, and the determination result is no, the steps (D-3) to (D-6) are continued. According to the sixth aspect of the patent application scope, the rock slope outcrop positional interpretation system further comprises a group of cluster state calculation modules, which are calculated according to the calculated unit normal vector of each clustered cluster center. Represents the inclination and inclination of the outcrop position of each group of rock slopes.