Method for evaluating three-dimensional roughness of rock mass structural surface

Abstract

The invention belongs to the field of data processing, and particularly relates to a three-dimensional roughness evaluation method for a rock mass structural plane, which comprises the following steps of: performing multi-scale decomposition on point cloud data by utilizing two-dimensional continuous wavelet transform, and introducing scale confidence to correct the unreliability of small-scale analysis; the method comprises the following steps: automatically identifying valley point scales in a double logarithm correction scale roughness spectrum, and separating macroscopic undulations and microcosmic rough bodies representing roughness types; according to the method, the total roughness and the wave roughness energy ratio are combined, a double-threshold classification system is established, rock mass structural surfaces are divided into four types, and therefore more accurate and more physical mechanical response prediction and parameter bases are provided for stability analysis and support design of geotechnical engineering.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of data processing. More specifically, the present application relates to a method for evaluating three-dimensional roughness of rock mass structure surface. BACKGROUND

[0002] In the field of mining and geotechnical engineering, the roughness of rock mass structure surface is a key physical property that determines its shear strength and stability. In order to evaluate this property, the existing technology usually obtains three-dimensional point cloud data of the rock mass structure surface by means of three-dimensional laser scanning, and calculates a global statistical parameter to represent the overall roughness based on the data. However, this single parameter evaluation method has the inherent defect of scale confusion: the true topography of the rock mass structure surface is a superposition of features of different scales, including large-scale gentle undulations and small-scale sharp rough bodies. The undulations and the rough bodies dominate different mechanical responses such as dilatancy effect and initial engagement strength in the process of rock shear. The traditional global parameter confuses these two physically different features, resulting in a serious disconnection between the evaluation result and the true mechanical behavior of the rock mass structure surface, and thus cannot provide accurate and reliable basis for engineering design.

[0003] In order to solve the above-mentioned scale confusion problem, the academic and engineering circles have begun to try to introduce multi-scale analysis methods. Among them, two-dimensional continuous wavelet transform, as an effective signal processing tool, can decompose the topography of the rock mass structure surface into a series of different physical scales for analysis. However, when applying wavelet transform to the discrete point cloud data obtained by three-dimensional scanning, due to the limited sampling density of data acquisition, interpolation processing is needed when generating the two-dimensional height field for analysis. This process inevitably introduces false, high-frequency interpolation noise, which mainly affects the accuracy of small-scale analysis results. Therefore, the small-scale components obtained by directly applying wavelet transform have low credibility and cannot fully and truly reflect the actual morphology of the micro rough bodies, constituting a technical obstacle for accurate evaluation. SUMMARY

[0004] In order to solve the above-mentioned problems of scale confusion and inaccurate evaluation of the roughness of rock mass structure surface caused by interpolation noise, the present application provides a method for evaluating the three-dimensional roughness of rock mass structure surface, comprising: The three-dimensional point cloud data of the rock mass structure surface is acquired, and a two-dimensional height field matrix is generated; the two-dimensional height field matrix is subjected to two-dimensional continuous wavelet transform, and a series of wavelet coefficient matrices at different scales are obtained; the energy sum of the wavelet coefficient matrices at different scales is calculated, and scale roughness at different scales is obtained; scale confidence at different scales is obtained based on the ratio of the size of each scale to the average point spacing of the two-dimensional height field matrix; the scale roughness at different scales is corrected based on the scale confidence at different scales, and the corrected scale roughness at different scales is obtained, and the corrected scale roughness at all scales is used to form a corrected scale roughness spectrum; in a double logarithmic coordinate system, the size of the maximum point of the negative value of the second derivative of the corrected scale roughness spectrum is taken as the valley point scale; the corrected scale roughness spectrum is divided into a micro-roughness energy band and a macro-wavy energy band according to the valley point scale; the ratio of the integral of the corrected scale roughness in the macro-wavy energy band to the integral of the corrected scale roughness in the micro-roughness energy band is taken as the wave-roughness energy ratio; the sum of the corrected scale roughness at all scales is taken as the total roughness, and the rock mass structure surface is classified and evaluated according to the total roughness and the wave-roughness energy ratio.

[0005] Firstly, the three-dimensional point cloud data is converted into a two-dimensional height field matrix, and multi-scale decomposition is performed by using two-dimensional continuous wavelet transform, so that the morphological features at different scales are separated in the corresponding scale channels, and the problem of multi-scale feature aliasing in the traditional method is avoided. Subsequently, considering that the sampling density of the point cloud data is limited, false high-frequency noise is easily introduced by interpolation in small-scale analysis, and the scale confidence is introduced to correct the roughness at each scale. The confidence is constructed based on the ratio of the analysis scale to the average point spacing of the point cloud, which can adaptively reduce the weight of unreliable small-scale components, thereby improving the authenticity of the full-scale roughness spectrum, especially the credibility of the micro-roughness feature. On this basis, the second derivative of the corrected scale roughness spectrum is analyzed in the double logarithmic coordinate system to automatically identify the valley point scale. The valley point scale objectively reflects the transition position between the micro-roughness and the macro-wavy, and provides a stable and repeatable basis for the energy division of the two types of morphological features. Further, the wave-roughness energy ratio is calculated by integrating the corrected roughness in the macro- and micro-energy bands, to quantify the relative contribution of large-scale undulation and small-scale concave-convex. At the same time, the total roughness is obtained by summing the corrected roughness at all scales, forming two evaluation indexes with physical meaning. Finally, the total roughness and the wave-roughness energy ratio are combined, and an adaptive threshold method is used to divide the classification boundary in the two-dimensional feature space, so as to divide the rock mass structure surface into four types. The classification not only reflects the roughness level, but also embodies the nature of the dominant morphology, which is helpful to more accurately predict the mechanical response of the rock mass in the shearing process, such as the initial bite strength or the shear dilation tendency.

[0006] Preferably, the scale confidence satisfies the expression: ; wherein, is the scale confidence of the rock mass structural plane at the scale ; is the scale size, is the average point distance of the original three-dimensional point cloud on the reference surface, is the confidence adjustment coefficient, is a natural exponential function.

[0007] The present application improves the reliability of each scale component in multi-scale roughness analysis by introducing scale confidence. The scale confidence is based on the ratio of the analysis scale to the average point distance of the original three-dimensional point cloud on the reference surface, adopts a Gaussian-type attenuation form, and makes the confidence decrease smoothly as the scale decreases, avoiding abrupt truncation of small-scale information. When the analysis scale is much larger than the average point distance, the confidence tends to the maximum value, indicating that the roughness calculation result at this scale has high reliability. When the scale is close to or smaller than the average point distance, the confidence decreases rapidly, effectively suppressing the interference of high-frequency noise introduced by interpolation on small-scale roughness. The confidence adjustment coefficient of the scale confidence provides the implementer with the freedom to adjust the confidence decay rate, enabling a reasonable balance between preserving true microscopic features and suppressing false noise, improving the robustness of small-scale analysis, and providing a more reliable data basis.

[0008] Preferably, the valley point scale satisfies the expression: ; wherein, is the scale size, is the identified valley point scale, is an operator that finds the parameter that makes the function value in the parentheses maximum, is a natural logarithm operation, is the modified scale roughness of the rock mass structural plane at the scale .

[0009] The present application realizes the objective division of the energy boundary between macro-waviness and micro-roughness by adopting a valley point scale identification method based on the curvature characteristics of the modified scale roughness spectrum. The second derivative of the modified scale roughness spectrum is calculated in the double logarithmic coordinate system, and the scale corresponding to the maximum point of the negative value of the derivative is taken as the valley point scale, thereby automatically positioning the transition position between the micro and macro components in the energy spectrum. Since the double logarithmic coordinate can amplify the details of the spectrum, and the second derivative can accurately describe the concave and convex changes of the curve, this strategy reduces the subjectivity and inconsistency caused by manually setting the segmentation scale, improves the scientificity of the energy band division, and provides a unified and physically meaningful scale reference for subsequent wave roughness energy ratio calculation and structural plane classification, enhancing the automation level and engineering applicability of the entire evaluation process.

[0010] Preferably, the generating the two-dimensional height field matrix comprises: performing plane fitting on the three-dimensional point cloud data by using a least square method to obtain a best fitting reference surface representing a macroscopic average trend of the rock mass structural surface; projecting all the three-dimensional point cloud data vertically onto the reference surface to obtain height data of the projected points; constructing a two-dimensional grid on the reference surface, and performing interpolation on the two-dimensional grid based on the height data of the projected points to generate a two-dimensional height field matrix.

[0011] Preferably, the two-dimensional continuous wavelet transform adopts a two-dimensional Mexican hat mother wavelet function.

[0012] The present application adopts a two-dimensional Mexican hat mother wavelet function, because the rough and undulating morphological characteristics on the rock mass structural surface generally do not have obvious directionality, and are isotropic, the two-dimensional Mexican hat wavelet is sensitive to the isotropic characteristic response, and can effectively identify irregularly distributed protrusions and depressions; at the same time, the two-dimensional Mexican hat wavelet has good localization characteristics, which ensures that not only the morphological characteristics of different scales can be identified in the decomposition process, but also the spatial position and undulation strength on the structural surface can be accurately corresponded, so that the wavelet coefficient matrix under each scale can more truly reflect the real morphology of the rock mass structural surface under each scale, and provide a reliable data basis for subsequent calculation.

[0013] Preferably, the wave roughness energy ratio satisfies the expression: ; wherein, is the wave roughness energy ratio of the rock mass structural surface, is the scale size, is the modified scale roughness, is the valley point scale, is the total scale range of analysis, is the minimum scale of analysis, is the maximum scale of analysis.

[0014] The present application realizes the evaluation of the relative development degrees of the macroscopic undulation and the microscopic roughness of the rock mass structural surface by introducing the wave roughness energy ratio, the ratio takes the valley point scale as the boundary, respectively integrates the modified scale roughness of the macroscopic scale segment and the microscopic scale segment, and takes the ratio, so as to objectively reflect the energy proportion of the large-scale undulation and the small-scale concave-convex in the overall morphology. The index can effectively distinguish different types of structural surfaces dominated by undulating or fine roughness, and provides a key criterion for subsequent classification and evaluation. When used in combination with the total roughness, the wave roughness energy ratio further enriches the description dimension of the geometric characteristics of the rock mass structural surface, so that the evaluation result is closer to the actual situation, and helps to improve the pertinence and reliability of the rock mass stability analysis.

[0015] Preferably, the classification evaluation of the rock mass structural plane according to the total roughness and the wave roughness energy ratio comprises: taking the total roughness and the wave roughness energy ratio of each sample as a point in a two-dimensional feature space, segmenting the two-dimensional feature space by using a two-dimensional adaptive threshold algorithm to obtain a total roughness threshold and a wave roughness energy ratio threshold ; and classifying the rock mass structural plane according to the comparison result of the total roughness, the wave roughness energy ratio and the threshold , .

[0016] Preferably, the classification of the rock mass structural plane comprises: determining a low roughness, rough body dominant type in response to the total roughness being less than or equal to and the wave roughness energy ratio being less than or equal to ; determining a low roughness, waviness dominant type in response to the total roughness being less than or equal to and the wave roughness energy ratio being greater than ; determining a high roughness, rough body dominant type in response to the total roughness being greater than and the wave roughness energy ratio being less than or equal to ; and determining a high roughness, waviness dominant type in response to the total roughness being greater than and the wave roughness energy ratio being greater than .

[0017] Preferably, the confidence adjustment coefficient is set to 2.

[0018] Preferably, the three-dimensional point cloud data of the rock mass structural plane is obtained by using a three-dimensional laser scanner.

[0019] The present application has the beneficial effects that: the present application effectively separates the contribution of macroscopic waviness and microscopic roughness in the rock mass structural plane by two-dimensional continuous wavelet transform and scale confidence correction, overcomes the problem of ambiguous mechanical meaning caused by scale confusion of the traditional single roughness parameter; the present application introduces the scale confidence related to the average point distance of the point cloud to adaptively suppress the high-frequency noise introduced by interpolation, so that the corrected scale roughness spectrum is closer to the real topography, and the reliability of small-scale features is improved; the present application constructs a two-parameter classification framework based on the total roughness and the wave roughness energy ratio, and uses adaptive threshold segmentation, which avoids subjective experience setting, so that the classification of the rock mass structural plane has more physical basis and engineering interpretability; the classification result of the present application can provide more targeted basis for stability evaluation, support design and the like in geotechnical engineering. BRIEF DESCRIPTION OF DRAWINGS

[0020] Figure 1 is a flowchart schematically showing a three-dimensional roughness evaluation method of a rock mass structural plane in the present application; Figure 2is a flow chart schematically showing S2 in the present application. DETAILED DESCRIPTION

[0021] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are some but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative effort are within the protection scope of the present application.

[0022] The specific embodiments of the present application will be described in detail below with reference to the drawings.

[0023] The embodiments of the present application disclose a three-dimensional roughness evaluation method of a rock mass structural surface, referring to Figure 1 , comprising steps S1 to S5: S1, acquiring three-dimensional point cloud data of the rock mass structural surface, and generating a two-dimensional height field matrix, performing two-dimensional continuous wavelet transform on the two-dimensional height field matrix to obtain a series of wavelet coefficient matrices at different scales.

[0024] It should be noted that, in order to evaluate the three-dimensional roughness of the rock mass structural surface, the present application converts the acquired original three-dimensional point cloud data into a two-dimensional height field matrix for multi-scale analysis, and decomposes the topographic features of the rock mass structural surface at different scales through two-dimensional continuous wavelet transform.

[0025] Specifically, a three-dimensional laser scanner or a high-precision digital photogrammetry system is used to acquire high-density three-dimensional point cloud data of the rock mass structural surface to be measured. In order to establish a unified analysis reference, the least square method is used to perform plane fitting on the point cloud data to obtain a best-fitted reference surface representing the macroscopic average trend of the rock mass structural surface. The three-dimensional point cloud data is vertically projected onto the reference surface to obtain height data of the projected points; based on the height data of the projected points, interpolation is performed on a two-dimensional grid whose relative size of a grid cell size to the average point spacing of the original three-dimensional point cloud data is within a certain range to generate a two-dimensional height field matrix , wherein the value of each element in the two-dimensional height field matrix represents the deviation of the actual height of the rock mass structural surface from the best-fitted reference surface at the grid coordinates , a positive value indicates a protrusion, and a negative value indicates a depression. In this embodiment, the bilinear interpolation method is used to interpolate the two-dimensional grid, and in other embodiments, the implementer can select an interpolation algorithm according to actual conditions.

[0026] More specifically, the grid cell size of the two-dimensional grid should ensure that the two-dimensional height field can both sufficiently preserve the spatial details of the original point cloud and avoid introducing too much interpolation noise due to over-dense grid or losing valid information due to under-dense grid. The grid cell size is set to 0.8-1.2 times of the average point spacing of the original three-dimensional point cloud on the reference surface: when the grid size is less than 0.8 times of the average point spacing, most grid cells lack direct observation points, and the interpolation result is easily affected by algorithm assumptions, resulting in false high-frequency components; when the grid size is greater than 1.2 times of the average point spacing, it may not be able to effectively distinguish the real existing small undulations, resulting in the loss of topographic details. Therefore, by controlling the grid cell size to be between 0.8-1.2 times of the average point spacing, a reasonable balance can be achieved between preserving the real geometric features of the structural surface and suppressing the uncertainty of interpolation, providing a reliable height field input for subsequent multi-scale analysis. In this embodiment, the grid cell size is set to 1.1 times of the average point spacing of the original three-dimensional point cloud on the reference surface, and in other embodiments, the implementer can set the grid cell size according to the actual situation.

[0027] Further, a series of preset physical scale sizes, such as, from to are taken in an exponential sequence . The two-dimensional continuous wavelet transform (2D-CWT) is used to process the two-dimensional height field matrix: for each scale size, the mother wavelet function of the two-dimensional continuous wavelet transform is convolved with the two-dimensional height field matrix to obtain the wavelet coefficient matrix at each scale. The numerical distribution in the wavelet coefficient matrix reflects the strength and position of the undulation features in the original rock mass structural surface topography that match the current scale.

[0028] It should be noted that the two-dimensional Mexican Hat wavelet is selected as the mother wavelet function in this embodiment because the undulation features of the rock mass structural surface are usually isotropic, i.e., they have similar statistical properties in different directions, and the two-dimensional Mexican Hat wavelet has good localization properties and is sensitive to isotropic morphological features, which can effectively capture these irregularly distributed but not strongly directional protrusions and depressions, thereby accurately identifying and quantifying rough bodies and undulations of different sizes. In other embodiments, the implementer can select the mother wavelet function of the two-dimensional continuous wavelet transform according to the actual implementation situation.

[0029] S2, calculate the scale roughness of each scale, obtain the scale confidence of each scale, modify the scale roughness of each scale based on the scale confidence of each scale, and construct the modified scale roughness spectrum by all the modified scale roughnesses of all scales.

[0030] It should be noted that after obtaining the wavelet coefficient matrix of each scale, in order to quantitatively characterize the roughness of the rock mass structure surface at different scales from the multi-scale decomposition result, and overcome the distortion problem of small-scale analysis caused by the limited sampling density of the point cloud, the present application evaluates the contribution of each scale to the total roughness, and adaptively corrects the calculation error introduced at the small-scale end due to insufficient data sampling density.

[0031] Specifically, the flowchart of step S2 refers to Figure 2 , including steps S201 to S203: S201, calculate the energy sum of the wavelet coefficient matrix of each scale to obtain the scale roughness of each scale.

[0032] It should be noted that since the energy of the wavelet coefficient can directly reflect the fluctuation intensity of the rock mass structure surface at the corresponding scale, in order to evaluate the fluctuation intensity of the rock mass structure surface at a specific scale, the present application obtains the scale roughness of the rock mass structure surface at each scale according to the wavelet coefficient energy at each scale.

[0033] Specifically, the scale roughness of the rock mass structure surface at each scale satisfies the expression:

[0034] wherein, is the scale roughness of the rock mass structure surface at the scale size is the scale size; is the grid coordinate of the measurement plane area; is the wavelet coefficient value at the scale size and the coordinate , and the integral region of covers the entire measurement plane area. The greater the value of , the more significant the corresponding topographic feature of the rock mass structure surface at the scale , and the more intense the fluctuation.

[0035] S202, obtain the scale confidence of each scale based on the ratio of the size of each scale to the average point distance of the two-dimensional height field matrix.

[0036] It should be noted that the scale roughness can evaluate the roughness of the rock mass structure surface at each scale from a macroscopic perspective, but the average point distance of the original point cloud data limits the minimum topographic size that can be reliably analyzed. When the scale size is close to or smaller than the average point distance, false high-frequency noise will be introduced during interpolation, resulting in overestimation of the roughness and reduction of the reliability. Therefore, the present application corrects the roughness of each scale based on the scale roughness combined with the scale confidence to obtain the corrected scale roughness.

[0037] Specifically, the scale confidence satisfies the expression:

[0038] wherein, is the scale confidence of the rock mass structural plane at the scale ; is the scale size; is the average point distance of the original three-dimensional point cloud on the reference surface, which can be directly calculated by projecting the three-dimensional point cloud data vertically onto the reference surface; is the coefficient for adjusting the influence of the average point distance, is a natural exponential function.

[0039] In the present embodiment, the preferred range of the coefficient for adjusting the influence of the average point distance is to When the value of is less than , the confidence function has too strong attenuation effect, which may excessively suppress part of the effective small-scale information; when the value of is greater than , the suppression effect on the interpolation noise is not obvious. In the present embodiment, the value of is set to , which can achieve a good balance between effectively suppressing noise and retaining true features. In other embodiments, the implementer can select the value of according to actual conditions.

[0040] When the scale size is much greater than the average point distance, the ratio is very large, the exponential term tends to , so that tends to , indicating that the analysis result of the large-scale feature has high confidence. Conversely, when is close to or less than , the ratio is close to , the exponential term tends to , so that tends to , indicating that the analysis result at this time is mainly dominated by interpolation noise, and the confidence is very low.

[0041] S203, based on the scale confidence of each scale, the scale roughness of each scale is corrected to obtain the corrected scale roughness of each scale, and the corrected scale roughness of all scales is composed of the corrected scale roughness spectrum.

[0042] The scale roughness of each scale is multiplied by the scale confidence of each scale, and the corrected scale roughness of the rock mass structural plane at each scale satisfies the expression:

[0043] wherein, is a scale size, is a modified scale roughness of the rock mass structural plane at the scale size . is a scale roughness of the rock mass structural plane at the scale size . is a scale confidence of the rock mass structural plane at the scale size .

[0044] Through this modification, the unreliable energy value at the small scale end is effectively suppressed, while the reliable energy value at the large scale end is retained, and the modified scale roughness at each scale is formed into a modified scale roughness spectrum that is more in line with physical reality.

[0045] S3, in a double logarithmic coordinate system, taking the scale size corresponding to the maximum point of the negative value of the second derivative of the modified scale roughness spectrum as the valley point scale; according to the valley point scale, the modified scale roughness spectrum is divided into a micro-roughness energy band and a macro-waviness energy band.

[0046] It should be noted that for the three-dimensional roughness evaluation of the rock mass structural plane, not only the roughness at each scale needs to be evaluated, but also the roughness type of the rock mass structural plane needs to be evaluated, because the roughness contains characteristics at different scales: such as macro-waviness, which refers to the undulating morphology of the rock mass structural plane in a larger range, affecting the overall sliding trend and stability; such as micro-roughness, which refers to the small bumps and pits on the surface of the rock mass structural plane, mainly affecting the local contact and friction performance.

[0047] It should be further noted that the modified scale roughness spectrum of the natural rock mass structural plane usually presents a double-peak morphology, with energy concentrated in the small scale region representing micro-roughness and the large scale region representing macro-waviness, and there is a valley point scale with relatively low energy between the two, which can be used as an objective basis for dividing the macro-waviness energy band and the micro-roughness energy band. The valley point scale of the modified scale roughness spectrum corresponds to the position where the curve bends most sharply in mathematics, i.e., the point with the maximum curvature. The present application can realize automatic identification of the valley point scale by finding the extreme value of the second derivative.

[0048] Specifically, to accurately identify the valley point scale, the modified scale roughness spectrum is plotted in a double logarithmic coordinate system, with the horizontal axis as and the vertical axis as .

[0049] Further, the valley point scale satisfies the expression:

[0050] wherein, It refers to scale or size. It is the scale of the identified valley points; It is a parameter that maximizes the value of the function within the parentheses. Operators; For natural logarithm operations; It is the rock mass structural plane at the scale The corrected scale roughness.

[0051] When the second derivative is positive, the curve is concave, meaning it bends upwards; when it is negative, it is convex, meaning it bends downwards. The valley point is precisely the point with the strongest concavity. By calculating the negative value of the second derivative, the problem of finding the maximum concavity is transformed into a standard problem of finding the maximum value of a function. The operator automatically locates the scale corresponding to the maximum value point. This is the valley point scale we are looking for. .

[0052] Therefore, the entire scale range [ It is objectively divided into micro-roughness energy bands. and macroscopic undulation energy band .

[0053] S4. The ratio of the integral of the modified scale roughness within the macroscopic waviness energy band to the integral of the modified scale roughness within the microscopic roughness energy band is taken as the waviness-roughness energy ratio.

[0054] It should be noted that after completing the energy band segmentation, this invention extracts a parameter for evaluating the relative development of macroscopic waviness and microscopic roughness, denoted as the waviness-roughness energy ratio.

[0055] Specifically, the wave roughness energy ratio of the rock mass structural surface satisfies the following expression:

[0056] in, The wave roughness energy ratio of the rock mass structural surface. For scale, To correct dimensional roughness, For valley point scale, [ [This represents the total scale range of the analysis.] As the smallest scale for analysis, This represents the maximum scale of the analysis.

[0057] The wave roughness energy ratio reflects the energy contrast between two states: when When the macroscopic undulation energy is greater than the microscopic roughness energy, it indicates that the morphology of the rock mass's structural surfaces is dominated by large-scale, gently undulating undulations; when When the wave-roughness energy ratio is small, it indicates that the surface morphology is mainly composed of small-scale, sharp roughness. The size of the wave-roughness energy ratio evaluates the relative dominance of macro-wave and micro-roughness energy.

[0058] S5, the sum of the modified scale roughnesses of all scales is taken as the total roughness, and the rock mass structural plane is classified according to the total roughness and the wave-roughness energy ratio.

[0059] It should be noted that in geotechnical engineering practice, the mechanical behavior of rock mass structural plane is not only affected by its overall roughness, but also closely related to the dominant type of its surface morphology: the structural plane mainly composed of large-scale wave is prone to produce shear dilation effect in the shearing process, while the structural plane mainly composed of small-scale roughness mainly provides initial biting resistance. If only a single index is used to describe the roughness, it is difficult to distinguish the two morphology characteristics with different engineering responses. Therefore, the present application constructs a two-dimensional classification system composed of total roughness and wave-roughness energy ratio, and automatically determines the classification boundary according to the distribution characteristics of the actual sample data by using the self-adaptive threshold method. Since the rock mass of different engineering sites has different genesis, weathering state and tectonic history, the distribution range and concentration trend of its roughness and wave-roughness energy ratio are different, and it is difficult to be universal with fixed threshold. The present application dynamically determines the segmentation threshold, so that the classification result can not only accurately reflect the high and low of the roughness of the rock mass structural plane, but also effectively identify the dominant type of its morphology, so as to realize the fine description of the mechanical properties of the rock mass structural plane.

[0060] Specifically, the sum of the modified scale roughnesses of all scales of the rock mass structural plane is taken as the total roughness of the rock mass structural plane.

[0061] Further, the total roughness of each rock mass structural plane sample is taken as the horizontal axis, and the wave-roughness energy ratio of each rock mass structural plane sample is taken as the vertical axis, a two-dimensional feature space is constructed, and each rock mass structural plane sample is represented as a point in the two-dimensional feature space, whose coordinates are (total roughness, wave-roughness energy ratio). A large number of sample points are analyzed, and two optimal segmentation thresholds are automatically determined by using a two-dimensional self-adaptive threshold algorithm: a total roughness threshold and a wave-roughness energy ratio threshold The two thresholds divide the two-dimensional feature space into four quadrants, corresponding to four typical types of rock mass structural plane. The maximum inter-class variance method is used as the two-dimensional self-adaptive threshold algorithm in this embodiment, and the two-dimensional self-adaptive threshold algorithm can be selected according to actual needs.

[0062] In response to the total roughness being less than or equal to and the wave-roughness energy ratio being less than or equal to , the rock mass structural plane is judged to be of low roughness and micro-dominant type, and the surface of the rock mass structural plane is nearly smooth, only with weak and fine rough texture; in response to the total roughness being less than or equal to and the wave-roughness energy ratio being greater than​ , the rock mass structure surface is determined as low roughness, wave type, the rock mass structure surface is determined as low roughness, macro type, the rock mass structure surface is macro smooth, but the amplitude of the fluctuation is not large, and the whole is relatively smooth; in response to total roughness greater than and wave roughness energy ratio less than or equal to , the rock mass structure surface is determined as high roughness, micro type, the rock mass structure surface is like coarse sandpaper, densely covered with a large number of sharp small-size protrusions; in response to total roughness greater than and wave roughness energy ratio greater than , the rock mass structure surface is determined as high roughness, macro type, the rock mass structure surface has significant large-scale wave fluctuations, and may also superimpose moderate roughness.

[0063] Thus, three-dimensional roughness evaluation of the rock mass structure surface is realized.

Claims

1. A three-dimensional roughness evaluation method for rock mass structural surfaces, characterized in that, include: Acquire three-dimensional point cloud data of rock mass structural surfaces and generate a two-dimensional height field matrix; Perform a two-dimensional continuous wavelet transform on the two-dimensional height field matrix to obtain a series of wavelet coefficient matrices at different scales; The energy sum of the wavelet coefficient matrices at each scale is calculated to obtain the scale roughness at each scale; the scale confidence is obtained based on the ratio of the size of each scale to the average point spacing of the two-dimensional height field matrix. The scale roughness of each scale is corrected based on the scale confidence of each scale to obtain the corrected scale roughness of each scale, and the corrected scale roughness of all scales constitutes the corrected scale roughness spectrum. In a double logarithmic coordinate system, the scale corresponding to the maximum value of the negative value of the second derivative of the modified scale roughness spectrum is taken as the valley scale. Based on the valley point scale, the modified scale roughness spectrum is divided into a micro-roughness energy band and a macro-waviness energy band. The ratio of the integral of the modified scale roughness within the macroscopic waviness energy band to the integral of the modified scale roughness within the microscopic roughness energy band is taken as the waviness-roughness energy ratio. The sum of the modified scale roughness of all scales is taken as the total roughness, and the rock mass structural surfaces are classified and evaluated based on the total roughness and the wave roughness energy ratio.

2. The method for evaluating the three-dimensional roughness of rock mass structural surfaces according to claim 1, characterized in that, The scale confidence level satisfies the expression: ； in, It is the rock mass structural plane at the scale The scale confidence level is as follows; For scale size; The average point spacing of the original 3D point cloud on the reference plane; This is the confidence level adjustment coefficient; It is a natural exponential function.

3. The method for evaluating the three-dimensional roughness of rock mass structural surfaces according to claim 1, characterized in that, The valley point scale satisfies the expression: ； in, It refers to scale or size. It is the scale of the identified valley points; It is a parameter that maximizes the value of the function within the parentheses. Operators; For natural logarithm operations; It is the rock mass structural plane at the scale The corrected scale roughness.

4. The method for evaluating the three-dimensional roughness of rock mass structural surfaces according to claim 1, characterized in that, The generation of the two-dimensional height field matrix includes: The least squares method is used to fit the three-dimensional point cloud data to a plane to obtain a reference surface that represents the macroscopic average orientation of the rock mass structure. All three-dimensional point cloud data are vertically projected onto the reference surface to obtain the height data of the projected points. A two-dimensional grid is constructed on the reference surface, and interpolation is performed on the two-dimensional grid based on the height data of the existing projected points to generate a two-dimensional height field matrix.

5. The method for evaluating the three-dimensional roughness of rock mass structural surfaces according to claim 1, characterized in that, The two-dimensional continuous wavelet transform uses the two-dimensional Mexican hat wavelet function.

6. The method for evaluating the three-dimensional roughness of rock mass structural surfaces according to claim 1, characterized in that, The wave roughness energy ratio satisfies the expression: ； in, The wave roughness energy ratio of the rock mass structural surface. For scale, To correct dimensional roughness, For valley point scale, [ [This represents the total scale range of the analysis.] As the smallest scale for analysis, This represents the maximum scale of the analysis.

7. The method for evaluating the three-dimensional roughness of rock mass structural surfaces according to claim 1, characterized in that, The rock mass structural surfaces are classified and evaluated based on the total roughness and the wave roughness energy ratio, including: The total roughness and wave roughness energy ratio of each sample are used as points in a two-dimensional feature space. A two-dimensional adaptive thresholding algorithm is then used to segment the two-dimensional feature space to obtain a threshold for the total roughness. Threshold of wave roughness energy ratio ; through total roughness, wave roughness energy ratio and threshold , The comparison results are used to classify and evaluate the rock mass structural planes.

8. The method for evaluating the three-dimensional roughness of rock mass structural surfaces according to claim 7, characterized in that, The classification and evaluation of rock mass structural planes includes: In response to a total roughness less than or equal to And the wave roughness energy ratio is less than or equal to It is determined to be low roughness, rough body-dominated type; in response to total roughness less than or equal to And the wave roughness energy ratio is greater than It is determined to be of low roughness and wavy texture dominant type; in response to a total roughness greater than And the wave roughness energy ratio is less than or equal to It is determined to be of high roughness, with roughness-dominated type; in response to a total roughness greater than And the wave roughness energy ratio is greater than It was determined to be a high-roughness, wavy-dominant type.

9. The method for evaluating the three-dimensional roughness of rock mass structural surfaces according to claim 2, characterized in that, The confidence level adjustment factor is set to 2.

10. The method for evaluating the three-dimensional roughness of rock mass structural surfaces according to claim 1, characterized in that, The acquisition of three-dimensional point cloud data of rock mass structural surfaces includes: A 3D laser scanner was used to acquire 3D point cloud data of the rock mass structural surfaces.

Images