Slope stability evaluation system based on geometric characteristic parameters of rock mass structural plane

Abstract

The present application belongs to the technical field of geotechnical engineering investigation and slope stability evaluation, and is used to solve the problem that the prior art cannot use the inherent local geometric boundary conditions of the slope to dynamically check the rationality of the grouping assumption, thereby blocking the propagation path of statistical false convergence to local stability evaluation, and specifically is a slope stability evaluation system based on rock mass structural plane geometric characteristic parameters, comprising a data acquisition module, a candidate scheme generation module, a local block identification module, a back projection verification module and a grouping result output module; the present application generates multiple candidate grouping schemes and introduces free face geometric constraints for back projection verification, converts the inherent local boundary conditions of the slope into independent criteria for the grouping assumption, so that any false convergence grouping assumption that does not match the local block shear-out condition will be automatically rejected, thereby cutting off the propagation path of global statistical false convergence to local stability evaluation from the mechanism.

Description

Technical Field

[0001] This invention belongs to the field of geotechnical engineering investigation and slope stability assessment technology, specifically a slope stability assessment system based on the geometric characteristic parameters of rock mass structural surfaces. Background Technology

[0002] In current rock slope engineering, the mainstream assessment method generally involves obtaining the geometric parameters of exposed structural surfaces of the slope through on-site measurements, grouping the structural surfaces based on the statistical characteristics of these parameters, assuming that structural surfaces with similar attitudes belong to the same group, and assigning uniform mechanical parameters to the group; identifying potentially unstable blocks based on the grouping results and calculating their safety factors; this process can achieve good results under simple geological conditions where the attitude distribution of structural surfaces is clear and the boundaries between groups are obvious.

[0003] In actual slopes, structural surfaces of different origins and scales often overlap in the attitude space. When using global measurement data for one-time grouping, statistical errors inevitably arise due to overlapping areas. For example, two groups of structural surfaces with different mechanical behaviors may be incorrectly classified into the same mixed group. More importantly, the true controlling factors of slope stability are often reflected in local areas, such as wedge-shaped blocks enclosed by a few structural surfaces and free faces. Local blocks are far more sensitive to mechanical parameters than global statistics. When global grouping results are forcibly applied to local block analysis, the differences between groups that are statistically "averaged" will be amplified in local calculations. This parameter mismatch caused by statistical false convergence stems from the lack of a mechanism in the conventional process to verify whether the global grouping assumption is consistent with local geometric constraints.

[0004] Therefore, the urgent technical problem to be solved is: how to use the inherent local geometric boundary conditions of the slope (such as the spatial constraint relationship between the free face and the structural face) to dynamically verify the rationality of the grouping assumption, thereby blocking the propagation path of statistical false convergence to local stability assessment. Summary of the Invention

[0005] In current rock slope engineering, the mainstream assessment method generally involves obtaining the geometric parameters of exposed structural surfaces of the slope through on-site measurements, grouping the structural surfaces based on the statistical characteristics of these parameters, assuming that structural surfaces with similar attitudes belong to the same group, and assigning uniform mechanical parameters to the group; identifying potentially unstable blocks based on the grouping results and calculating their safety factors; this process can achieve good results under simple geological conditions where the attitude distribution of structural surfaces is clear and the boundaries between groups are obvious.

[0006] In actual slopes, structural surfaces of different origins and scales often overlap in the attitude space. When using global measurement data for one-time grouping, statistical errors inevitably arise due to overlapping areas. For example, two groups of structural surfaces with different mechanical behaviors may be incorrectly classified into the same mixed group. More importantly, the true controlling factors of slope stability are often reflected in local areas, such as wedge-shaped blocks enclosed by a few structural surfaces and free faces. Local blocks are far more sensitive to mechanical parameters than global statistics. When global grouping results are forcibly applied to local block analysis, the differences between groups that are statistically "averaged" will be amplified in local calculations. This parameter mismatch caused by statistical false convergence stems from the lack of a mechanism in the conventional process to verify whether the global grouping assumption is consistent with local geometric constraints.

[0007] Therefore, the urgent technical problem to be solved is: how to use the inherent local geometric boundary conditions of the slope (such as the spatial constraint relationship between the free face and the structural face) to dynamically verify the rationality of the grouping assumption, thereby blocking the propagation path of statistical false convergence to local stability assessment.

[0008] 1. By introducing geometric constraints on the free surface for back projection verification, the inherent local boundary conditions of the slope are transformed into independent criteria for judging the grouping assumptions. This ensures that any false convergence grouping assumptions that do not conform to the local block shearing conditions are automatically rejected. From a mechanism perspective, this can cut off the path of global statistical false convergence to local stability assessment and reduce statistical bias caused by mismatch of structural mechanical parameters. 2. By extracting and labeling trend indicators and fluctuation indicators, cross-data type cross-validation of attitude grouping and spatial arrangement patterns is achieved; when the spacing gradient characteristics of the structural surfaces within a local block deviate, the abnormal situation that the corresponding structural surfaces may belong to different physical groups can be automatically identified, thereby triggering a rejection flag and ensuring the rationality of grouping. 3. Calculate the local self-consistency of each candidate grouping scheme and perform hierarchical output through its numerical range; quantify the uncertainty in the grouping process into a visualized decision indicator, so that the stability calculation can be carried out in a targeted manner according to the certainty of the data itself, reducing the false accuracy caused by forcibly outputting a unique group. 4. The orientation distribution of the rejected structural surfaces is reverse-mapped to the stereographic projection map to identify high-density regions and locally tighten the clustering tolerance only in these regions, thereby establishing a two-way self-consistent closed loop between global grouping and local geometric constraints. The asymmetric feedback strategy can accurately suppress the erroneous merging of false convergence regions while protecting the reasonable grouping granularity of other regions. After iterative optimization, the self-consistency of the grouping scheme can be further improved. Attached Figure Description

[0009] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0010] Figure 1 This is a system structure block diagram of Embodiment 1 of the present invention; Figure 2 This is a flowchart of the system operation method according to Embodiment 1 of the present invention. Detailed Implementation

[0011] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0012] In rock slope engineering investigation and stability assessment, structural plane grouping is a fundamental task. Existing technologies generally adopt a linear processing flow of "one-time grouping across the entire domain and subsequent application throughout the process." However, existing technologies lack a verification mechanism for the self-consistency between the global statistical grouping results and local geometric constraints. This leads to the inability to identify the "false convergence of dominant orientation" phenomenon when there are statistically overlapping areas in the orientation of structural planes. It is understandable that two groups of structural planes with different origins and mechanical behaviors (such as early shear joints and late tension joints) may overlap in the orientation space in the large-scale statistical domain due to the mixing of data points or the random projection of regional structures, and be mistakenly "falsely converged" into a mixed group. Existing technologies force the application of the unified mechanical parameters of the "false convergence group" to local block analysis, causing structural planes that should belong to high-strength gently dipping joints to be assigned the parameters of low-strength steeply dipping joints.

[0013] If the above problems are not addressed, the slope stability assessment system will continue to lose its ability to objectively distinguish the allocation of mechanical parameters of local blocks. In particular, the failure to identify the false convergence of dominant orientations will lead to the forced merging of structural surfaces with different mechanical behaviors, resulting in the systematic underestimation or overestimation of the anti-sliding force of local blocks, which will lead to two extreme consequences: first, misjudging stable slopes as unstable, resulting in unnecessary engineering reinforcement expenditures; second, misjudging unstable slopes as stable, leading to sudden landslide disasters.

[0014] Example 1: As Figure 1-2 As shown, the slope stability assessment system based on the geometric characteristic parameters of rock mass structural surfaces includes: Data acquisition module: Acquires geometric feature data of multiple structural surfaces within the slope area and a model of the free face of the slope; wherein, the geometric feature data includes at least the attitude data, spacing data and spatial location data of each structural surface; This article provides a clear and hierarchical explanation of the multiple data sources acquired in the data acquisition module: The data acquisition module is responsible for collecting geometric feature data of multiple structural surfaces within the slope area and the slope free face model. The geometric feature data includes at least the attitude data, spacing data, and spatial location data of each structural surface.

[0015] For acquiring attitude data, this embodiment uses a geological compass in conjunction with a 3D laser scanner or a UAV photogrammetry system. On each exposed structural surface, at least three representative measuring points are selected, and their dip values ​​(the angle between the projection of the structural surface normal onto the horizontal plane and true north, ranging from 0° to 360°) and dip angle values ​​(the maximum angle between the structural surface and the horizontal plane, ranging from 0° to 90°) are measured respectively. Specifically, to eliminate measurement errors, the attitude data for each measuring point is measured three times consecutively, and the arithmetic mean is taken. A set of dip values ​​is obtained for each structural surface. and tilt angle value subscript The unique identifier for the structural surface; the formula for converting dip and dip angle into the unit normal vector of the structural surface is: ; in The direction pointing towards the upper plate of the structural plane (i.e., the positive direction is defined as the normal pointing outwards from the slope); when the dip angle When the angle is less than 1°, it is considered a near-horizontal structural plane, and its inclination can be arbitrarily set to 0° without affecting subsequent calculations.

[0016] For obtaining spacing data, this embodiment measures the slope surface along a pre-set survey line. The survey line layout principle is to prioritize directions perpendicular to the dominant orientation of the structural surface, and the survey line length should cover at least 10 structural surface exposure points. Starting from one end of the survey line, the horizontal distance between adjacent structural surface exposure points is recorded sequentially to form a spacing sequence. ; where each spacing value The first corresponding measurement line The exposed structural surface and the first The straight-line distance between exposed structural surfaces is expressed in meters. It is important to note that, in order to reduce measurement deviations caused by local weathering or vegetation cover, each survey line should be measured at least three times, and the median of the spacing values ​​at the corresponding locations should be taken as the final recorded value. If the same structural surface is exposed multiple times on the survey line (for example, due to the alternating exposure of structural surfaces caused by slope undulations), the spacing should be recorded in the order of exposure, and the structural surface number to which each exposure point belongs should also be recorded.

[0017] For acquiring spatial location data, this embodiment uses a 3D laser scanner or differential GPS to collect coordinates of at least three outcrop points on each structural surface; the 3D spatial coordinates of each outcrop point are defined as follows: ,in Number the structural surfaces. Dew point number The coordinate system adopts the geodetic coordinate system of the project location and projects it onto the Gauss-Kruger plane coordinate system; the elevation adopts the normal height system; the three outcrop points should not be collinear and should be distributed as much as possible in the upper, middle, and lower parts of the structural surface to accurately represent the spatial position and attitude change trend of the structural surface; for example, for a shear joint exposed on a slope, its number is F001, and the coordinates of the three outcrop points are respectively , , All units are meters.

[0018] Regarding the acquisition of the slope free face model, this embodiment utilizes a digital elevation model (DEM) or dense point cloud data generated through 3D laser scanning. The grid resolution of the DEM is typically set to 0.5 meters to 1.0 meters, which can reflect the macroscopic undulation of the slope surface. Based on the DEM data, the normal vector of each grid cell is calculated, thereby obtaining the local free face attitude of that grid cell: dip value. and tilt angle value In the specific calculation, the elevation values ​​of each grid cell and its eight adjacent grid cells are taken, and the normal vector of the cell is obtained by fitting a plane or using a difference method. This normal vector is then converted into dip and dip angle. Subsequently, the entire slope is divided into several analysis sectors according to natural valleys or ridgelines. The arithmetic mean of the free face attitude of all grid cells within each sector is taken as the representative free face attitude of that sector. The unit normal vector outside the free face is obtained by converting this representative attitude. ; in Pointing outwards from the slope (i.e., the direction of the outward normal to the slope surface); for example, if a sector contains 200 grid cells with an average dip of 185.5° and an average dip angle of 42.3°, its outward normal unit vector can be directly calculated using the above formula.

[0019] It is worth noting that all the above raw data must undergo data cleaning and coordinate alignment before entering the subsequent processing module: the system automatically removes records in the attitude data where the dip exceeds 0°–360° or the dip angle exceeds 0°–90°, as well as outliers with duplicate spatial coordinates or significant deviations from the main body of the slope; the system uniformly transforms the coordinates of the outcrop points of all structural surfaces to the same geodetic coordinate system to ensure spatial consistency of data obtained from different measurement batches and different measurement methods; after the above preprocessing, The data acquisition module will assign the number and orientation of each structural plane. ,inclination Spacing sequence The structure surface dataset consists of the three-dimensional coordinates of three or more outcrop points. The sector number and its representative free surface orientation are assigned. Composition of slope free face partition model ; Structure Surface Dataset Slope free face partition model Once generated, it is stored in the context of this request for subsequent modules to call and calculate; Any person skilled in the art, following the above description and using conventional geological surveying instruments and mapping methods, can reliably obtain the required structural surface geometric feature data and slope free face model.

[0020] Candidate scheme generation module: Based on attitude data, a clustering algorithm is used to generate multiple candidate grouping schemes; each candidate grouping scheme contains the mapping relationship between each structural surface and group label; Since the candidate scheme generation module needs to reflect the mapping relationship between each structural surface and group label in the candidate grouping scheme, this embodiment naturally provides a specific implementation process for generating multiple candidate grouping schemes based on attitude data: receiving the structural surface attitude data (i.e., the tendency of each structural surface) output by the data acquisition module. and tilt angle ,in , After the total number of structural surfaces in the slope area; to overcome the statistical bias that may be introduced by a single clustering scheme, this module does not output a unique grouping result, but generates a set containing multiple candidate grouping schemes for the subsequent back-projection verification module to filter.

[0021] This embodiment defines a similarity metric between structural surfaces; for any two structural surfaces and Each of them tends to ,inclination as well as , Substituting into the formula for calculating the unit normal vector, we get: , ; The spherical angle between the normal vectors of the two structural surfaces Calculation by dot product ; in The dot product of two unit vectors, with values ​​ranging from 1 to 2. Take the absolute value and then find the inverse cosine, such that... Always fall Within the interval; and The smaller the value, the closer the orientation of the two structural planes are, and the more likely they are to be classified into the same group.

[0022] This embodiment uses a hierarchical agglomerative clustering algorithm for grouping. Initially, each structural surface forms its own class. During iteration, the algorithm calculates the minimum spherical angle distance between all current classes and merges the two classes with the smallest distance that is less than the preset clustering tolerance into a new class. The distance between classes is calculated using the complete-linkage method, which takes the maximum value of the spherical angle between any two structural surfaces in two classes as the inter-class distance. This distance measurement method ensures that the angle between any two structural surfaces in the merged class does not exceed this maximum value, thus forming a more compact cluster. The merging process is repeated until the minimum distance between any two classes is greater than or equal to the preset clustering tolerance, or all structural surfaces are merged into a single class.

[0023] To generate diverse candidate grouping schemes, this embodiment sets an initial clustering tolerance. This is denoted as the first preset angle threshold; based on this, this module independently runs the above-mentioned hierarchical agglomerative clustering process multiple times, executing a total of [number missing] times. Each time; before each run, the tendency of each structural surface. and tilt angle Add random errors that follow a uniform distribution: , ,in and Independent and all follow a uniform distribution The purpose of adding random errors is to simulate measurement uncertainty and the subjective differences among different engineers in interpreting the attitude, thereby obtaining different grouping schemes that may be reasonable. It should be noted that random errors are only used for distance calculation in this clustering and do not modify the original measurement data. After adding errors, the spherical angle matrix between all structural faces is recalculated, and then hierarchical agglomerative clustering is performed.

[0024] Note that in this embodiment, if the tilt value exceeds the range of [0°, 360°) after adding random error, it is returned to the range by modulo 360° calculation; if the tilt angle value exceeds [0°, 90°], it is truncated to the boundary value.

[0025] After each clustering run, a grouping scheme is output; each scheme contains the following information: total number of groups. Each structural plane is assigned a group label. (Values) And the average attitude of each group; wherein, the average attitude of the group is obtained by the vector average of the original measured attitudes (without random error) of all structural surfaces within the group: for the th Group, let the first The set of structural surface numbers of the group is The number of structural surfaces it contains is denoted as Then the average normal vector of this group is The average normal vector is converted back to dip and tilt angle. If two runs produce completely identical group label mappings (i.e., each structural surface is assigned to the same group in both schemes, and the group order can be rearranged), it is considered a duplicate scheme, and only one copy is retained. To determine whether two grouping schemes are the same, this embodiment uses a label standardization method: the group labels in each scheme are rearranged in ascending order according to the smallest number of the structural surface within the group to obtain a standardized label mapping. If the two standardized mappings are completely identical, they are considered duplicate schemes.

[0026] This embodiment collects all The set of candidate solutions comprises all unique grouping schemes in each run. ,in Usually not exceeding Each candidate solution This corresponds to a mapping relationship from structural surface number to group label.

[0027] It is worth noting that the specific parameters of hierarchical agglomerative clustering using the fully connected method, the fixed initial tolerance of 20°, the repetition 20 times, and the addition of ±2° random error in this embodiment are determined based on a large amount of engineering statistics and sensitivity analysis: the tolerance of 20° can cover the statistical variation range of most reasonable groupings, while the error range of ±2° matches the typical accuracy of field compass measurement (±2° to ±5°); those skilled in the art can adjust the above parameters according to the actual slope complexity, but the core processing logic (generating a set of candidate grouping schemes through multiple perturbation clustering) remains unchanged.

[0028] The candidate solution generation module ultimately outputs a set of candidate solutions. The output data, along with the average attitude of each group corresponding to each scheme (for quick lookup of representative attitudes), will be passed to the back-projection verification module as the basis for subsequent calculation of local block sliding direction and rejection decision analysis.

[0029] Local block identification module: Based on the slope free face model and the attitude data and spatial location data of multiple structural surfaces, it identifies at least one potential critical block in the slope area and determines the set of structural surfaces contained in each potential critical block and the attitude of its local free face. In a specific implementation, for the local block identification module, this embodiment provides a full-space stereographic projection method based on Goodman-Shi block theory, used to identify potential key blocks from the slope free face model and the attitude data of all structural surfaces, and to determine the set of structural surfaces contained in each block and its local free face attitude; this module receives the structural surface dataset output by the data acquisition module. (Including the tendency of each structural plane) ,inclination (and spatial location) and slope free face zoning model (Including representative free surface orientations of each sector) ).

[0030] This embodiment converts the attitude data of the free face of the slope and all structural surfaces into a great circle representation on a stereographic projection; for a single structural surface, its normal vector is... In stereographic projection, the corresponding great circle is the circle formed by the intersection of the structural surface and the reference sphere with its center as the origin. In practice, the great circle is not drawn directly; instead, the dip of each structural surface is calculated. and tilt angle The corresponding pole (i.e., the intersection of the normal vector and the sphere), or the great circle arc automatically generated using equal-angle stereographic projection software; in order to identify the block enclosed by the structural plane and the free surface, the free surface of the slope needs to be regarded as a virtual "structural plane", the attitude of which is normal vector Calculate using the same formula, but the outer normal is taken as pointing outwards from the slope.

[0031] This embodiment employs the full-space stereographic projection analysis method in the block theory proposed by Goodman-Shi. The specific process is as follows: all structural surface great circles and free surface great circles are projected onto the same stereographic diagram. For each group of three or four structural surfaces (including or excluding free surfaces), they will intersect on the stereographic diagram to form several closed arc-shaped triangles or quadrilaterals. Each closed region corresponds to a cone enclosed by these structural surfaces. If the sliding direction of the cone points outward from the free surface and its vertex points towards the slope direction, it constitutes a potential unstable block. Classical block theory states that a movable block must satisfy the condition that "on the stereographic diagram, the closed region corresponding to the block is located on one side of all structural surface great circles, and that side is consistent with the outward slope direction indicated by the free surface great circle." The specific criterion is: for a group of structural surfaces (including free surfaces), if the intersection of the half-space (i.e., the inside of the structural surface) pointed to by the normal vectors of these structural surfaces is non-empty, then the intersection corresponds to a possible block.

[0032] In practical engineering implementation, manual drawing is unnecessary; this embodiment uses a block recognition algorithm based on vector analysis, the specific process of which is as follows: Let the outward normal unit vector of the free face of the slope be... The expression is obtained by substituting the sector's representative attitude into the previous formula; for any set consisting of r structural planes (r=3 or r=4), let the unit normal vectors of these r structural planes be respectively... Note that the normal vectors here all point inward into the rock blocks separated by the structural planes, usually pointing outward from the free surface, or according to the convention of "block normal" in block theory. The classic formula of block theory states that the necessary and sufficient condition for the existence of a block enclosed by these structural planes is the existence of a direction vector. Make For all Established simultaneously; this direction This represents the sliding direction of the block; the set of directions that satisfy this set of inequalities constitutes the cone region where the block may slide.

[0033] This embodiment uses a linear inequality solution method; a matrix is ​​constructed. Its row vectors are the normal vectors of each structural surface. Determine if there exists a non-zero vector. satisfy (All components are greater than zero); This is a classic linear feasibility problem, which can be solved by linear programming or by sequentially calculating the directions of the intersection lines of each pair of structural surfaces and checking whether all inequalities are satisfied; in specific engineering projects, commonly used block identification software (such as Unwedge) automatically completes this task based on this principle; this embodiment utilizes the output of this software, namely the set of structural surface numbers contained in each identified potential critical block. , and the average free surface orientation of the exposed area of ​​the block.

[0034] Considering that the slope may be divided into multiple sectors, each with a different free face attitude, this embodiment determines the sector to which each identified block belongs based on the spatial coordinates of its outcrop point. The representative free face attitude of that sector is then used as the local free face attitude of the block, denoted as... If the block spans multiple sectors, the weighted average of the attitude of each sector is taken (with the exposed area of ​​the block in each sector as the weight).

[0035] For example, suppose the free face of a certain slope sector dips to the surface. ,inclination Within this region, the structural plane F001 dips in the direction of dipping. ,inclination The orientation of structural plane F002 is dipping. ,inclination The orientation of structural plane F003 is dipping. ,inclination The block identification software calculates and outputs a potential critical block B1, which is formed by three structural planes (F001, F002, and F003) and the free surface. The set of structural plane numbers is as follows: The local free surface attitude is as follows At the same time, the system also records the approximate location of the block in space (e.g., coordinate range) so that it can be located when outputting alarm information later.

[0036] For each identified potential critical block, this module outputs its unique number and the set of structural surface numbers it contains. and the attitude of the local free surface If no potential critical blocks are identified on the slope, the local block identification module outputs an empty list. In this case, the subsequent back-projection verification module can be skipped, and the solution with the highest self-consistency in the candidate solution set (considered as having no local constraints) can be directly adopted. Otherwise, all identified blocks will form a block list. , as input to the back-projection verification module; If no potentially critical blocks are identified on the slope (i.e. If the local block identification module outputs an empty list and passes a special flag to the grouping result output module, the subsequent back projection verification module and self-consistency calculation are skipped. After receiving the special flag, the grouping result output module directly selects the first scheme in the candidate scheme set (or uses the conventional attitude clustering result) as the final global grouping scheme, and adds the output prompt message: "No potential movable blocks pointing out of the slope were found in the current slope. The grouping result is only based on attitude statistics and has not been verified by local geometric constraints."

[0037] Back-projection verification module: For each candidate grouping scheme, determine the representative attitude of the structural surface contained in each potential critical block under the candidate grouping scheme; based on the representative attitude and the attitude of the local free surface of the potential critical block, calculate the sliding direction of the potential critical block under the candidate grouping scheme; and perform rejection decision analysis on the potential critical block based on the sliding direction. For the back-projection verification module, this embodiment provides a complete rejection decision analysis process based on candidate grouping schemes and potential key blocks: receiving the candidate scheme set output by the candidate scheme generation module. and the list of potential key blocks output by the local block recognition module. ,in The total number of potential critical blocks; for each candidate solution and every potential key block This module executes the following sub-steps in sequence: Determine the current block The included structural surfaces in the current candidate scheme The following represents the occurrence; let the mass be... The set of structural surface numbers is According to the candidate solutions Group label mapping in each structural surface Assigned to a group label If a group contains multiple structural surfaces, the attitude is represented by the vector average of the original measured attitudes (without random error correction) of all structural surfaces in that group, with the average normal vector... The calculation formula is as described above, then converted back to dip and dip angle; if a certain structural plane is in If a group forms its own category (i.e., the group only contains the structural plane itself), it means that the attitude directly adopts the original measured attitude of the structural plane; thus, the block In the plan Below Each structural surface yields a representative attitude, which is then converted into the corresponding unit normal vector. .

[0038] Calculate the block size in the scheme The sliding direction downwards; according to block theory, by The possible sliding directions of a potential critical block enclosed by several structural surfaces are the intersection directions of any two non-parallel planes among these structural surfaces; this embodiment uses the following steps to calculate the sliding direction vector. : The first step is for the block. Any two structural surfaces in and Calculate the cross product of its normal vectors. And calculate its modulus. ;like If the two structural planes are nearly parallel and do not produce a valid intersection line, then the combination is skipped directly; if Then the unit vector of the intersection direction is .

[0039] in This represents the cross product of three-dimensional vectors, with the denominator being the magnitude of the cross product. It's important to note that the intersection line has two opposite directions. and These correspond to the two directions of the intersection line of the structural surfaces; since the block can only slide along one of these directions, the direction will be determined later through direction filtering. The second step is to assign a unit vector to each intersection direction. Calculate the two pointing directions and the unit vector normal to the local free plane respectively. The dot product; the out-of-plane normal of the local free surface is derived from the block. Local free surface attitude The same formula is used for conversion; the direction in which the dot product is positive is taken as the candidate direction of the intersection line pointing outwards from the slope; specifically, if Then keep ;like Then take If the dot product is zero, the intersection line is parallel to the free surface, and theoretically, the block cannot be cut out; therefore, this intersection line is excluded. After the above processing, each valid intersection line combination corresponds to a unit vector pointing outwards from the slope, denoted as... ; The third step is to select the direction closest to the direction of gravity as the potential sliding direction from all candidate intersection directions pointing outwards from the slope; let the unit vector of the direction of gravity be... (Vertical downward); Calculate each candidate direction and The direction of the sliding direction is chosen as the direction of the dot product. ; Sliding direction In a physical sense, this corresponds to the direction of the intersection where the sliding force component is largest, i.e., the direction in which sliding is most likely to occur.

[0040] Assuming there are no intersection directions pointing outwards from the slope (i.e., the dot product of all intersections and the outward normal of the free plane is not greater than zero), then the block In the plan If a sliding direction pointing outwards from the slope cannot be formed, then the sliding direction is defined. It is a zero vector and is marked as "pointing inwards or parallel".

[0041] The rejection decision analysis is based on the sliding direction; this embodiment uses the following criteria: Scenario 1: When zero vector or When the sliding direction points inwards from the slope or is parallel to the slope surface, the potential critical block is directly determined to be a candidate component. The decision was not rejected. Furthermore, the block does not generate any rejection information; Design rationale: The block pointing into the slope is geometrically non-sliding and belongs to a stable block. Regardless of the grouping assumption, it is impossible to reject the assumption from a kinematic perspective.

[0042] Scenario 2: When At this point, the sliding direction clearly points outward from the slope; further consistency comparison is needed by utilizing the local gradient features of the spacing; this embodiment first obtains the block The local gradient characteristics of the spacing of each structural surface included; the local gradient characteristics of the spacing include trend indicators and fluctuation indicators, the definitions of which are found in the data acquisition module regarding the processing of the spacing sequence; specifically, for the structural surface spacing sequence acquired along the preset survey line... Calculate the difference between adjacent spacings ,in Define symbolic functions :when When +1 is taken, When -1 is taken, Take 0 at that time.

[0043] like The number of alternations between positive and negative is... To meet the following conditions Number of: From 1 to ,and And both are not zero; at this point, the volatility indicator is defined as Its value range is: When the signs of all adjacent differences change alternately In this embodiment, in order to make the fluctuation index equal to 1 when the denominator is completely alternating, the denominator is adopted. Those skilled in the art will understand that, in other implementations, the following methods are adopted: A similar effect can be achieved by using it as the denominator, only requiring adjustment of the subsequent consistency threshold.

[0044] like If there is only one difference, it is impossible to determine the fluctuation. (or marked as invalid); like If so, the survey line will not participate in the calculation of the spacing gradient feature.

[0045] Trend Indicators The unit is meters, and its positive or negative sign indicates the overall monotonic change direction of the spacing along the survey line (positive for increasing, negative for decreasing), and the absolute value indicates the rate of change; fluctuation index ,in for Number of times positive and negative signs alternate (traversal) to ,like (If neither of them is zero, then it counts once). The range of values ​​is A larger value indicates that the spacing fluctuates more frequently.

[0046] For the current candidate solutions and the current block Each structural surface within the block Spacing local gradient features With the plan Other structural surfaces in the same group as this structural surface (i.e., those with the same group label but not in the block) The consistency of the characteristics of the internal structural surfaces is compared. The specific process is as follows: For the plan Each group in Collect all elements that belong to this group but are not in the current block. The internal structural surfaces constitute a set. Let the number of its elements be . ;like Then calculate the mean of these structural trend indicators respectively. and standard deviation and the mean of volatility indicators and standard deviation : , ; , ; For blocks Each structural surface within (This structural plane belongs to the group) ): like and If the consistency comparison result of the structural surface is consistent, then the consistency comparison result is determined to be consistent. Otherwise (i.e., any indicator falls outside the corresponding range), it is judged as inconsistent; Once the block If the consistency comparison result of memory on at least one structural plane is inconsistent, then the candidate solution is deemed invalid. In the block The proposal was rejected. ; when When the sample size is insufficient for reliable statistical inference, the system does not reject the structural surface within the block based on the local gradient features of the spacing. Instead, it directly determines that the consistency comparison result of the structural surface is passed (i.e., it does not reject it), and the system records a warning message: "Insufficient samples, spacing gradient verification skipped".

[0047] If the sliding direction of all potential critical blocks points inward or parallel to the slope surface, the system outputs the prompt message: "There are no obvious movable blocks on the current slope. The grouping results are not sensitive to stability assessment. It is recommended to use conventional empirical grouping." Then, the most stable clustering scheme in the candidate scheme concentration (i.e., the scheme corresponding to the median number of groups in multiple clusters) is directly used as the final output.

[0048] For each candidate solution and each potential key block This module outputs a Boolean value. And the additional reasons for rejection when rejected (such as "the sliding direction points out of the slope but the spacing gradient is mismatched"); ultimately, all Each verification result will be passed to the grouping result output module for self-consistency calculation.

[0049] Grouping Result Output Module: For each candidate grouping scheme, calculate the local self-consistency of the candidate grouping scheme based on the number of blocks that are rejected in all potential key blocks; output the grouping result based on the local self-consistency.

[0050] In a specific implementation, this embodiment provides a complete output process based on local self-consistency calculation and hierarchical decision-making for the grouping result output module; the module receives the verification results output by the back-projection verification module, that is, for the candidate scheme set... and a list of potential key blocks For each combination, the obtained Boolean veto flag (in Indicate candidate solutions In the block The proposal was rejected. (This indicates that the proposal was not rejected).

[0051] like If the self-consistency calculation and threshold judgment are skipped, the first solution in the candidate solution set is directly output with the "no potential key block" comment; otherwise, this module sequentially performs self-consistency calculation, hierarchical decision output, and optional feedback modulation. For each candidate solution Count the number of blocks that were rejected among all potential key blocks. ,in The indicator function takes a value of 1 when the condition is true and 0 otherwise; then the local self-consistency of the candidate solution is calculated. Defined as the verified block proportion: ; in This represents the total number of potential critical blocks; it is obvious and direct that... The range of values ​​is A larger value indicates that the candidate solution is more consistent with the local geometric constraints; when When the value is 1, it indicates that the scheme does not exhibit spacing gradient mismatch on any blocks pointing outwards from the slope; while when the value is 1, it indicates that the scheme does not exhibit spacing gradient mismatch on any blocks pointing outwards from the slope. If the value is zero, it means that the solution is rejected on every block that has a tendency to slide outwards from the slope.

[0052] This embodiment presets three decision thresholds: the first self-consistency threshold... Second self-consistency threshold and preset membership threshold These thresholds are empirical values ​​obtained through offline testing and statistics on multiple typical slope projects, achieving a balance between conservatism and sensitivity; the maximum local self-consistency among all candidate solutions is determined. ; And let the scheme that obtains the maximum value be denoted as (If multiple options are available, choose one or retain all of them); according to Different output strategies are applied to different values: First scenario: At this point, it is assumed that there exists a highly consistent candidate grouping scheme, sufficient to serve as the final global grouping scheme; this module selects... As the final grouping result, the final group label for each structural surface and the average attitude of that group are output; specifically, for each structural surface... Output its in Group tags in For each group Output the average tendency of this group. and mean dip angle The value is calculated by the vector average of the original attitudes of all structural surfaces within the group, as described above. This embodiment also outputs the local self-consistency of the grouping scheme. As a confidence reference, this module may optionally trigger a feedback modulation step (see below) before output to further optimize the grouping results.

[0053] The second scenario: (Right now The presence of a '0' indicates that all candidate schemes have very low local self-consistency, meaning that no grouping assumption is compatible with the geometric constraints of most local blocks. This suggests that the distribution of structural surfaces on the slope may exhibit strong non-uniformity or multi-phase tectonic superposition, making global unified grouping inapplicable in this region. This module outputs a forced splitting alarm message, including: a reliable global grouping cannot be established, and it is recommended to perform independent local block analysis on each block. Simultaneously, it lists the spatial locations and constituent structural surface numbers of all rejected potential critical blocks. Specifically, for each block rejected by at least one candidate scheme... Output its block number, center coordinates or range coordinates, and the set of structural surface numbers that make up the block. Engineers can use this list to perform limit equilibrium calculations on each block individually using a 3D block program (such as Unwedge), employing the original measured attitude of the structural surfaces involved in that block, without relying on global grouping.

[0054] The third scenario: (Right now At this point, there are candidate solutions with a certain degree of self-consistency, but the confidence level is insufficient to uniquely determine the solution; this module enters the probability modulation mode; for each structural surface... Statistics on all candidate solutions The frequency at which it is assigned to each group; let the total number of different group labels appearing in all candidate schemes be... (Group label numbers may differ for different schemes; alignment is required according to the structural surfaces within the group); for structural surfaces Its group membership vector ,in That is, the structural plane is assigned to a group. The proportion of candidate solutions to the total number of solutions; obviously, Output the group membership probability vector for each structural surface for subsequent stability calculations. In subsequent calculations, the mechanical parameters of different groups are proportionally weighted and averaged according to this membership degree to obtain equivalent mechanical parameters. Simultaneously, structural surfaces with frequently changing group labels in different schemes are labeled, i.e., those with the highest membership degree. The structural surfaces are designated as sensitive areas for grouping. For these structural surfaces, this module outputs their numbers and recommendations: direct shear tests or high-precision re-measurement of the occurrence are recommended to reduce grouping uncertainty.

[0055] In the first case ( After selecting the final global grouping scheme, this embodiment can further execute a feedback modulation module to optimize the clustering tolerance using local veto information. The feedback modulation module is communicatively connected to the grouping result output module, and its specific implementation is as follows: Extract the selected global grouping scheme The structural surfaces involved in the potentially critical blocks that were rejected; (Note: The original text appears to be incomplete and contains several grammatical errors. A more accurate translation would require the full context.) The set of veto blocks below is Collect all the structural surface numbers that appear from these blocks to form a set of veto structural surfaces. The number of its elements is denoted as This embodiment sets a preset number of feedback triggers. ;when If the condition is met, feedback modulation is performed; otherwise, it is skipped.

[0056] When the triggering condition is met, the set of structural surfaces will be rejected. The orientation of the structural surfaces in the diagram is mapped to the stereographic projection; specifically, each structural surface... Corresponding to a pole on the stereographic diagram, its coordinates are determined by the dip direction. and tilt angle The transformation is obtained; in polar coordinate stereographic projection, the commonly used transformation formula is: radial distance polar angle It should be noted that the above conversion tends to... By mapping the coordinates to the unit circle using sine and cosine functions, the Euclidean distance between two measurements of different inclinations on the plane naturally corresponds to their minimum included angle (i.e., the shortest arc length considering the 360° periodicity), without the need for additional circumferential kernel density processing; for example, inclinations 5 and 355 differ numerically by 350°, but on the transformed plane, the distance between their corresponding points is approximately 0.174°, precisely corresponding to a central angle of 10°, consistent with the physical meaning of the similarity of inclinations.

[0057] This embodiment employs a two-dimensional kernel density estimation method to identify high-density regions in the attitude space where structural planes are densely distributed; the kernel density estimation formula is as follows: To reduce boundary effects, this embodiment employs mirror boundary processing: the projected points are reflected relative to the unit circle boundary to generate virtual image points, which participate in density estimation; the actual high-density points are still selected only from the original points.

[0058] in For Gaussian kernel function, bandwidth (Dimensionless, representing the Euclidean distance on the stereographic projection plane. This value, calibrated experimentally, can effectively identify local clustering regions in the attitude space); Calculate the density distribution across the entire stereographic map, marking regions with density values ​​exceeding twice the average density of all points as high-density regions; Each high-density region corresponds to an attitude range, which can be described by dip intervals and dip angle intervals.

[0059] For the clustering distance between structural surfaces in these high-density regions, the clustering tolerance is adjusted from the initial first preset angle threshold. Reduce to the second preset angle threshold However, the clustering distance between structural surfaces outside the high-density region remains unchanged. This local tightening strategy is achieved by modifying the distance threshold in hierarchical agglomerative clustering: during the clustering process, when determining whether two structural planes can be merged, if both structural planes are located in high-density regions, then the distance threshold is adjusted. Otherwise use Based on the adjusted local tightening tolerance, the candidate solution generation module is invoked again to generate a new set of candidate solutions. Then, the back-projection verification and self-consistency calculation were re-executed to obtain the modified scheme with the highest self-consistency. ;Compare self-consistency Compared with the original plan self-consistency :like If so, then adopt This will be the final global grouping scheme; otherwise, the original scheme will be retained. .

[0060] To prevent the iteration process from getting stuck in an infinite loop and to avoid invalid calculations, this embodiment sets the following iteration termination condition: First, the maximum number of iteration rounds is set to 2 rounds (i.e., a maximum of two feedback modulations are performed).

[0061] Second, set a convergence threshold. After each feedback modulation cycle, calculate the highest degree of self-consistency obtained in this round. The highest degree of self-consistency in the previous round The difference between ;like If the self-consistency has converged, the iteration will be terminated early even if the maximum number of rounds has not been reached, and no further feedback modulation will be performed.

[0062] Third, if after a certain round of feedback modulation If the self-consistency decreases, then the results of this round of adjustments are abandoned, the previous round's solution is retained, and subsequent iterations are terminated immediately.

[0063] By following the above rules, the feedback modulation module can stop in time when the improvement effect is not significant or degradation occurs, thus avoiding invalid calculations.

[0064] Convergence threshold This is an empirical value set based on engineering experience. Those skilled in the art can adjust this threshold according to actual engineering needs without deviating from the essence of this invention.

[0065] After completing the entire process of data acquisition, candidate scheme generation, local block identification, back projection verification, grouping result output and feedback modulation, this embodiment achieves substantial improvement over the prior art: by using the inherent free surface geometric constraints of the slope as an independent criterion, a dual back projection verification of the kinematic shearing condition and spacing gradient consistency is performed on each candidate scheme on the local block, so that any false convergence grouping assumption that does not conform to the local geometric self-consistency will be automatically rejected and accumulated as a rejection count.

[0066] The above description is merely an example and illustration of the structure of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the structure of the invention or exceed the scope defined in the claims, all of which should fall within the protection scope of the present invention.

[0067] In the description of this specification, references to terms such as "an embodiment," "example," "specific example," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the invention. In this specification, illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples.

[0068] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A slope stability assessment system based on geometric characteristic parameters of rock mass structural surfaces, characterized in that, include: Data acquisition module: Acquires geometric feature data of multiple structural surfaces within the slope area and a model of the free face of the slope; wherein, the geometric feature data includes at least the attitude data, spacing data and spatial location data of each structural surface; Candidate scheme generation module: Based on attitude data, a clustering algorithm is used to generate multiple candidate grouping schemes; each candidate grouping scheme contains the mapping relationship between each structural surface and group label; Local block identification module: Based on the slope free face model and the attitude data and spatial location data of multiple structural surfaces, it identifies at least one potential critical block in the slope area and determines the set of structural surfaces contained in each potential critical block and the attitude of its local free face. Back-projection verification module: For each candidate grouping scheme, determine the representative attitude of the structural surface contained in each potential critical block under the candidate grouping scheme; based on the representative attitude and the attitude of the local free surface of the potential critical block, calculate the sliding direction of the potential critical block under the candidate grouping scheme; and perform rejection decision analysis on the potential critical block based on the sliding direction. Grouping Result Output Module: For each candidate grouping scheme, calculate the local self-consistency of the candidate grouping scheme based on the number of blocks that are rejected in all potential key blocks; output the grouping results based on the local self-consistency. The specific process of rejection decision analysis for potential critical blocks includes: when the sliding direction points inward or parallel to the slope surface, it is directly determined that the potential critical block will not be rejected under the candidate grouping scheme; when the sliding direction points outward, the local gradient characteristics of the spacing of each structural surface contained in the potential critical block are obtained, and the local gradient characteristics of the spacing of these structural surfaces are compared with the local gradient characteristics of the spacing of other structural surfaces in the same group in the candidate grouping scheme. If the consistency comparison result of at least one structural surface in the potential critical block is inconsistent, it is determined that the potential critical block will be rejected under the candidate grouping scheme; otherwise, it will not be rejected.

2. The slope stability assessment system based on the geometric characteristic parameters of rock mass structural surfaces according to claim 1, characterized in that, In the geometric feature data: attitude data includes the dip and dip angle values ​​of each structural facet; spacing data includes the sequence of horizontal distances between the outcrop points of adjacent structural faces recorded sequentially along the preset survey line; spatial location data includes the three-dimensional spatial coordinates of each structural facet at at least three outcrop points on the slope surface.

3. The slope stability assessment system based on the geometric characteristic parameters of rock mass structural surfaces according to claim 1, characterized in that, The specific process of generating candidate schemes includes: using a hierarchical agglomerative clustering algorithm, taking the spherical angle between the normal vectors of the structural surfaces as the distance metric, setting the initial clustering tolerance as a first preset angle threshold, and clustering structural surfaces with spherical angles smaller than the first preset angle threshold into the same group; repeating hierarchical agglomerative clustering multiple times, adding random errors that follow a uniform distribution within a preset range to the tendency value and tilt angle value of each structural surface before each clustering, and collecting the non-repeating grouping schemes generated by each clustering to form a candidate grouping scheme set.

4. The slope stability assessment system based on the geometric characteristic parameters of rock mass structural surfaces according to claim 1, characterized in that, The specific process of local block identification includes: using the full-space stereographic projection method in Goodman-Shi block theory, projecting the attitude data of the free face and all structural faces of the slope onto the stereographic map, finding the closed cone formed by the intersection of three or four great circles of structural faces and the great circle of the free face, identifying each closed cone as a potential critical block, and recording the set of structural face numbers contained in each potential critical block and the average free face attitude of the exposed area of ​​the block.

5. The slope stability assessment system based on the geometric characteristic parameters of rock mass structural surfaces according to claim 1, characterized in that, The specific calculation process for the sliding direction of a potential critical block under the candidate grouping scheme includes: determining the representative attitude of each structural surface contained in the potential critical block based on the group label of each structural surface in the candidate grouping scheme, and converting the unit normal vector of each structural surface; calculating and normalizing the cross product of the normal vectors of any two structural surfaces contained in the potential critical block to obtain the unit vector of each intersection direction; calculating the dot product of each intersection direction unit vector with the outward normal unit vector of the local free plane of the potential critical block, and selecting the intersection directions with a dot product greater than zero as candidate directions pointing outward from the slope; among the candidate directions pointing outward from the slope, selecting the direction with the largest dot product with the unit vector of gravity direction as the sliding direction of the potential critical block under the candidate grouping scheme.

6. The slope stability assessment system based on the geometric characteristic parameters of rock mass structural surfaces according to claim 1, characterized in that, The methods for obtaining the local gradient characteristics of the spacing include: calculating the difference between adjacent spacings in the structural surface spacing sequence obtained along the preset survey line; obtaining a trend index by dividing the algebraic sum of the differences by the number of spacings minus one, which is used to characterize the overall monotonic change direction and degree of the spacing along the survey line; and obtaining a fluctuation index by dividing the number of positive and negative alternations of the differences by the number of spacings minus two, which is used to characterize the frequency of fluctuations in the spacing value.

7. The slope stability assessment system based on the geometric characteristic parameters of rock mass structural surfaces according to claim 6, characterized in that, The specific process of consistency comparison includes: for each group in the current candidate grouping scheme, collecting the trend index and volatility index values ​​of all structural surfaces belonging to that group but not within the current potential key block; when the number of collected structural surfaces reaches the preset sample size limit, calculating the mean and standard deviation of the trend index and the mean and standard deviation of the volatility index respectively; for each structural surface within the current potential key block, determining whether its trend index falls within the range of the first mean plus or minus twice the first standard deviation, and whether its volatility index falls within the range of the second mean plus or minus twice the second standard deviation; if either index falls outside the corresponding range, the consistency comparison result of that structural surface is determined to be inconsistent.

8. The slope stability assessment system based on the geometric characteristic parameters of rock mass structural surfaces according to claim 1, characterized in that, The specific process for outputting grouping results based on local self-consistency includes: determining the maximum local self-consistency among all candidate grouping schemes; when the maximum local self-consistency is greater than or equal to a first self-consistency threshold, selecting the candidate grouping scheme corresponding to the maximum local self-consistency as the final global grouping scheme, and outputting the final group label and average attitude of each structural surface; when the maximum local self-consistency is less than a second self-consistency threshold, outputting a forced splitting alarm message, listing the spatial location and constituent structural surface numbers of all rejected potential key blocks, and indicating that independent local block analysis should be performed on these blocks; when the maximum local self-consistency is between the second self-consistency threshold and the first self-consistency threshold, calculating the frequency of each structural surface belonging to each group in all candidate grouping schemes, outputting the grouping membership probability vector of each structural surface, and marking structural surfaces with a maximum membership degree less than a preset membership threshold as grouping sensitive areas.

9. The slope stability assessment system based on the geometric characteristic parameters of rock mass structural surfaces according to claim 8, characterized in that, The grouping result output module is also communicatively connected to a feedback modulation module: extracting the structural surfaces involved in the potentially critical blocks rejected under the selected global grouping scheme, forming a set of rejected structural surfaces; when the number of structural surfaces in the set of rejected structural surfaces reaches a preset feedback trigger number, mapping the attitude of the structural surfaces in the set of rejected structural surfaces to a stereographic projection map, and identifying high-density regions in the attitude space through kernel density estimation; for the clustering distance between structural surfaces in the high-density region, reducing the clustering tolerance from a first preset angle threshold to a second preset angle threshold, while maintaining the first preset angle threshold for the clustering distance between structural surfaces outside the high-density region; regenerating the candidate grouping scheme set based on the adjusted local tightening tolerance, recalculating the local self-consistency, and selecting the new scheme with the highest self-consistency as the final global grouping scheme.

Images