BIM (Building Information Modeling) and VR (Virtual Reality)-based visual modeling restoration method and equipment for underground powerhouse of hydropower station

Abstract

The invention relates to the technical field of visual modeling, in particular to a BIM (Building Information Modeling) and VR (Virtual Reality)-based visual modeling reduction method and equipment for an underground powerhouse of a hydropower station. Determining a curvature incongruous characteristic value according to the difference between the polar angle of each data point in each grid unit and the median of the polar angle interval corresponding to the grid unit; and evaluating the spatial distribution regularity of the curvature anisotropic characteristic values of all the grid units in the region where each cluster is located and the energy distribution in the frequency domain, and determining a comprehensive characteristic value so as to optimize a voxel filtering algorithm, filter the three-dimensional point cloud, and perform visual modeling on the hydropower station underground powerhouse based on the filtered three-dimensional point cloud. According to the method, the problems of detail loss and insufficient simplification are solved through self-adaptive recognition of a fine structure and adjustment of filtering, and the modeling precision and efficiency of the underground powerhouse of the hydropower station are improved.

Description

Technical Field

[0001] This invention relates to the field of visualization modeling technology, specifically to a method and equipment for visualization modeling and restoration of underground powerhouses in hydropower stations based on BIM and VR. Background Technology

[0002] In the collaborative design review of underground powerhouses in hydropower stations, integrating BIM and VR technologies is crucial for a detailed review of equipment layout and spatial relationships. BIM technology is responsible for constructing a 3D model that integrates information throughout the entire lifecycle, while VR technology provides an immersive experience, allowing users to intuitively perceive the interior of the powerhouse. By importing the information-rich BIM model into the VR environment, the design team and review experts can conduct an intuitive and interactive comprehensive review of the powerhouse structure, equipment, and systems in a virtual setting.

[0003] To achieve a 1:1 realistic VR recreation of the underground powerhouse of a hydropower station and allow users to walk into the internal structures such as the spiral shell, it is essential to ensure a high degree of preservation of key details in the original point cloud. However, traditional filtering methods, such as voxel filtering, often use a uniform simplification intensity when dealing with such complex structures and significant scale differences. This results in severe loss of fine structural details such as the spiral shell, while flat areas such as the ground are not sufficiently simplified, reducing the accuracy and efficiency of modeling the underground powerhouse of the hydropower station. Summary of the Invention

[0004] To achieve the above objectives, this invention provides a method and equipment for visual modeling and restoration of underground powerhouses in hydropower stations based on BIM and VR. By adaptively identifying fine structures and adjusting filtering, it solves the problems of lost details and insufficient simplification, thereby improving the accuracy and efficiency of modeling underground powerhouses in hydropower stations.

[0005] To address at least one of the aforementioned technical problems, according to one aspect of the present invention, a method for visual modeling and reconstruction of underground powerhouses in hydropower stations based on BIM and VR is provided, comprising the following steps:

[0006] Obtain the 3D point cloud of the underground powerhouse of the hydropower station;

[0007] All data points in the 3D point cloud are clustered. Within each cluster, a surface is fitted to each data point and all data points in its neighborhood. Based on the difference between the two principal curvatures of each data point on the fitted surface, the relative curvature difference value of each data point is determined. The region where each cluster is located is divided into polar radius-polar angle grids. Based on the difference between the polar angle of each data point in each grid cell and the median of the polar angle interval corresponding to the grid cell, the angle weight of each data point in each grid cell is determined. Combined with the relative curvature difference, the curvature anisotropic feature value of each grid cell is determined.

[0008] The spatial distribution regularity of the curvature anisotropic eigenvalues ​​of all grid cells within the region of each cluster is evaluated to determine the structural eigenvalues ​​of each cluster. Based on the energy distribution of the curvature anisotropic eigenvalues ​​of all grid cells within the region of each cluster in the frequency domain, the frequency domain eigenvalues ​​of each cluster are determined, and combined with the structural eigenvalues, the comprehensive eigenvalues ​​of each cluster are determined.

[0009] Based on the comprehensive feature values, the voxel filtering algorithm is optimized to filter the 3D point cloud of the underground powerhouse of the hydropower station, and the underground powerhouse of the hydropower station is visualized and modeled based on the filtered 3D point cloud.

[0010] Preferably, the distance metric used in the clustering process is the Euclidean distance between the three-dimensional coordinates of the data points.

[0011] Preferably, the relative curvature difference value of each data point is the result of the absolute difference between the two principal curvatures of each data point on the fitted surface divided by the sum of the absolute values ​​of the two principal curvatures.

[0012] Preferably, the expression for the angle weight of each data point within each grid cell is: In the formula, This represents the angle weight of data point j within grid cell i; Indicates the polar angle of data point j within grid cell i; This represents the median of the polar angle interval corresponding to grid cell i; The value represents the length of the preset polar angle interval; exp() represents an exponential function with the natural constant as the base.

[0013] Preferably, the method for determining the curvature anisotropic eigenvalues ​​of each grid cell is as follows:

[0014] Calculate the product of the normalized relative curvature difference value and the normalized angle weight value for each data point in each grid cell. Take the mean of the product of all data points in each grid cell as the curvature anisotropic characteristic value of each grid cell.

[0015] Preferably, the method for determining the structural feature values ​​of each cluster is as follows:

[0016] Within each cluster region, a curvature anisotropy matrix is ​​constructed based on the curvature anisotropy eigenvalues ​​of all grid cells, where the elements in the curvature anisotropy matrix are curvature anisotropy eigenvalues.

[0017] The curvature anisotropic matrix is ​​used as the input to the singular value decomposition algorithm. All singular values ​​are output, and the sum of the first preset number of singular values ​​in the descending order of all singular values ​​is used as the structural feature value of each cluster.

[0018] Preferably, the method for determining the frequency domain feature values ​​of each cluster is as follows:

[0019] Calculate the information entropy of all amplitudes of all elements in the frequency domain of the curvature anisotropic matrix of each column in the region where each cluster is located, and denote it as the first eigenvalue of each column.

[0020] Based on the preset volute encirclement angle, calculate the harmonic order in the frequency domain signal corresponding to each column element of the curvature anisotropic matrix, calculate the absolute value of the amplitude at the harmonic order corresponding to the preset volute encirclement angle, and record it as the second characteristic value of each column.

[0021] The third characteristic value of each column is determined based on the difference between the second characteristic value of each column and its adjacent preceding column;

[0022] The first, second, and third eigenvalues ​​of each column are used to form an eigenvector. All eigenvectors of the curvature anisotropic matrix are used as inputs to the Bayesian algorithm, and the output probability values ​​are used as the frequency domain eigenvalues ​​of each cluster.

[0023] Preferably, the comprehensive feature value of each cluster is the result of the positive fusion of the structural feature value and the frequency domain feature value of each cluster.

[0024] Preferably, the optimized voxel filtering algorithm includes:

[0025] In the voxel filtering algorithm, the voxel grid size is used when filtering cluster m. The expression is: ; This represents the normalized value of the comprehensive eigenvalues ​​of cluster m; , These represent the upper and lower limits of the preset voxel mesh size, respectively.

[0026] Preferably, the visualization modeling of the underground powerhouse of the hydropower station based on the filtered 3D point cloud includes:

[0027] The filtered 3D point cloud of the underground powerhouse of the hydropower station is used in 3D modeling and visualization software to obtain a visualized model of the underground powerhouse of the hydropower station.

[0028] According to one aspect of the present invention, a BIM and VR-based visualization modeling and restoration device for underground powerhouses of hydropower stations is provided, which is used to execute the BIM and VR-based visualization modeling and restoration method flow of the present invention in a hardware system.

[0029] According to another aspect of the present invention, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the BIM and VR-based visualization modeling and reconstruction method for underground powerhouses of hydropower stations of the present invention.

[0030] According to another aspect of the present invention, a computer device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps in the BIM and VR-based visualization modeling and restoration method for underground powerhouses of hydropower stations of the present invention.

[0031] Compared with existing technologies, the beneficial effects of the above-described method of the present invention are as follows:

[0032] This invention divides the point cloud region into polar coordinate grids and calculates feature values ​​that accurately reflect the intensity of local surface anisotropy by combining the relative curvature difference and angular position weight of each data point. This method can effectively quantify the complex gradual transition process of the spiral shell from a circular to an elliptical cross-section, thereby accurately identifying key fine structural regions such as the spiral shell before filtering. This provides a reliable basis for subsequent adaptive filtering and ensures that the geometric details of these regions are preserved to the greatest extent during the modeling process. Furthermore, this invention constructs comprehensive feature values ​​by fusing structural feature values ​​that characterize the spatial gradual transition law and frequency domain feature values ​​that characterize the spiral periodicity. This comprehensive feature value can accurately distinguish complex fine structures similar to the spiral shell in hydropower stations from other simple structures, providing a decisive quantitative basis for subsequent adaptive and differentiated point cloud filtering strategies. Finally, this invention achieves adaptive filtering of hydropower station point clouds by dynamically combining comprehensive feature values ​​with voxel filtering algorithms. That is, fine grids are used to preserve details for key complex structures such as the spiral shell, while coarse grids are used to efficiently simplify simple regions, improving the accuracy and efficiency of hydropower station underground powerhouse modeling. Attached Figure Description

[0033] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings of the embodiments will be briefly described below. Obviously, the drawings described below only relate to some embodiments of the present invention and are not intended to limit the present invention.

[0034] Figure 1 A flowchart illustrating a preferred embodiment of the invention;

[0035] Figure 2 A schematic diagram of the comprehensive feature value extraction process for a preferred embodiment of the invention. Detailed Implementation

[0036] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of the BIM and VR-based visualization modeling and reconstruction method for underground powerhouses of hydropower stations proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0037] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0038] Example 1: As Figure 1-2 As shown in this embodiment, the method for visual modeling and reconstruction of underground powerhouses in hydropower stations based on BIM and VR includes the following steps:

[0039] Step S1: Obtain the 3D point cloud of the underground powerhouse of the hydropower station.

[0040] To achieve high-fidelity reconstruction of the underground powerhouse of a hydropower station in a VR scene, the foundation lies in acquiring complete and accurate 3D point clouds. Therefore, point cloud acquisition is a crucial prerequisite for the entire workflow. This embodiment utilizes a drone equipped with a LiDAR system and employs oblique flight technology to completely cover the underground powerhouse area from above, acquiring 3D point clouds of the underground powerhouse. This ensures that the acquired point clouds fully represent the underground powerhouse, preventing incomplete VR scene generation due to missing point clouds.

[0041] Step S2: By quantifying the curvature anisotropic distribution pattern of each cluster in space and its spiral periodicity characteristics in the frequency domain, the comprehensive characteristic value of each cluster is calculated.

[0042] Hydroelectric generators, turbines, and pipelines are the main structures in the underground powerhouse of a hydroelectric power station. Their shape resembles a snail's shell, hence the term "volute." The cross-section of the volute is mostly circular, with an elliptical cross-section at the tail. There is a volute wrap angle, typically 345° or 350°, between the inlet and nose sections. This complex geometry gives its point cloud data significant features that can be recognized by algorithms. Specifically:

[0043] At the level of curvature and normal vector, the curvature of the volute exhibits a clear regularity. In the circular cross-sectional area with a large radius at the inlet, the curvature shows the same direction; while as the spiral path extends towards the elliptical cross-section with a small radius at the nose, the curvature shows anisotropy, that is, the difference in the degree of bending in different directions gradually increases. At the same time, the normal vector of the point cloud also shows a regular deflection along the spiral path.

[0044] At the overall structural and frequency domain levels, the spiral path of the volute is essentially a quasi-periodic structure. This structure causes the spatial distribution of its three-dimensional point cloud to exhibit a spiral-shaped periodic change, and the local point density also fluctuates stably. When the curvature distribution in its circumferential direction is regarded as a signal, the signal will exhibit strong harmonic characteristics in the frequency domain, and the energy is mainly concentrated near a specific frequency determined by the volute wrap angle (such as 345°).

[0045] In summary, the global curvature anisotropy of the point cloud in the volute region is not randomly distributed, but rather exhibits a pattern that gradually transitions from isotropy to anisotropy, demonstrating a clear spatial regularity. Therefore, this embodiment quantifies the spatial curvature anisotropy distribution pattern of each cluster and its spiral periodic characteristics in the frequency domain, comprehensively calculating the comprehensive feature value of each cluster. This feature value is then used to identify and quantify the volute structure. A schematic diagram of the comprehensive feature value extraction process provided in this embodiment is shown below. Figure 2 As shown, the specific construction process of the comprehensive feature value is as follows:

[0046] S2.1: Cluster all data points in the 3D point cloud, and perform surface fitting on each data point and all data points in its neighborhood within each cluster. Based on the difference between the two principal curvatures of each data point on the fitted surface, determine the relative curvature difference value of each data point. Divide the region where each cluster is located into polar radius-polar angle grids. Based on the difference between the polar angle of each data point in each grid cell and the median of the polar angle interval corresponding to the grid cell, determine the angle weight of each data point in each grid cell. Combined with the relative curvature difference, determine the curvature anisotropic feature value of each grid cell.

[0047] Since the most important structure in the underground powerhouse of a hydropower station is the turbine generator, all data points in the three-dimensional point cloud of the underground powerhouse are first clustered. In this embodiment, the DBSCAN density clustering algorithm is used to cluster the data points. The distance metric is the Euclidean distance between the three-dimensional coordinates of the data points, and the number of clusters is determined by the elbow rule. All clusters are output.

[0048] The DBSCAN clustering algorithm and the elbow rule are both well-known techniques. The specific process of using the DBSCAN clustering algorithm to cluster data points and the specific process of using the elbow rule to determine the number of clusters will not be described in detail.

[0049] Furthermore, in this embodiment, surface fitting is performed on each data point and all data points in its neighborhood within each cluster. Based on the difference between the two principal curvatures of each data point on the fitted surface, the relative curvature difference value of each data point is determined. Specifically:

[0050] In each cluster, each data point and all data points in its neighborhood are used as a surface fitting algorithm to fit a surface to obtain the fitted surface.

[0051] It should be noted that there are many commonly used surface fitting algorithms. In this embodiment, a surface fitting algorithm based on the least squares method is used to fit the surface of each data point and its neighborhood. In practical applications, as other implementation methods, implementers may also use other surface fitting methods according to specific circumstances. This embodiment does not impose any special restrictions.

[0052] Among them, the surface fitting algorithm based on the least squares method is a well-known technique, and the specific process of using it to fit the surface of data points and their neighborhood will not be described in detail.

[0053] It should be noted that the process of dividing the neighborhood range of each data point in this embodiment is as follows: a circular area with radius R as the center of each data point is used as the neighborhood of each data point. In this embodiment, the radius R is 1, and the unit of radius R is consistent with the point cloud unit, usually m. In actual application, as other implementation methods, implementers can also set the radius size or define the neighborhood division process according to the specific situation. This embodiment does not impose any special restrictions.

[0054] Furthermore, in this embodiment, the absolute difference between the two principal curvatures on the fitted surface for each data point is divided by the sum of the absolute values ​​of the two principal curvatures, and this is taken as the relative curvature difference value for each data point.

[0055] The method for calculating the principal curvature is a well-known technique and will not be elaborated further.

[0056] Based on the relative curvature difference value of each data point, it can be understood that the relative curvature difference value is used to characterize the geometric complexity and non-uniformity of the surface of each data point and its neighborhood. It reflects the difference in the degree of curvature of the surface in the two principal directions near each data point. If the relative curvature difference value of the current data point is larger, it indicates that the curvature anisotropy of the current data point's neighborhood is stronger. For example, if a data point is located on the surface of a volute that transitions from a circle to an ellipse, or at the edge of a pipe or equipment, its two principal curvatures are one large and one small, resulting in a significant difference. This indicates the drastic change in the geometric shape and structural complexity of the current data point and its neighborhood, which is crucial for the subsequent identification of fine structures such as volutes.

[0057] Conversely, the smaller the relative curvature difference of the current data point, the weaker the curvature anisotropy of the current data point's neighborhood. For example, if a data point is located in a large-radius circular area of ​​the volute or on a flat ground or wall, and its two principal curvatures are similar in magnitude with a difference close to zero, it indicates that the geometry of the current data point and its neighborhood is simpler and smoother, and its geometric structure is more spherical or planar, thus contributing less to the subsequent identification of complex and fine structures.

[0058] Furthermore, in this embodiment, the region where each cluster is located is divided into polar radius-polar angle grids. Based on the difference between the polar angle of each data point in each grid cell and the median of the corresponding polar angle interval of the grid cell, the angle weight of each data point in each grid cell is determined. Specifically:

[0059] In this embodiment, the 3D point cloud corresponding to the region where each cluster is located is transformed into a polar coordinate system. The coordinate axes containing the polar radius and polar angle are equally divided to obtain various polar radius intervals and polar angle intervals. Since the total range of the polar angle is 0°~360°, this embodiment equally divides the total range of the polar angle into 12 intervals, each with a length of 30°. The total range of the polar radius is equally divided into 8 intervals, and the polar angle intervals and polar radius intervals are sorted and numbered in ascending order to obtain... Grid size.

[0060] The process of converting rectangular coordinates to polar coordinates is a well-known technique and will not be elaborated further.

[0061] Furthermore, in this embodiment, the angle weight of each data point in each grid cell is determined based on the difference between the polar angle of each data point in each grid cell and the median of the corresponding polar angle interval of the grid cell. Specifically:

[0062] As one implementation method, in this embodiment, the angle weight of data point j within grid cell i is... The expression is: In the formula, Indicates the polar angle of data point j within grid cell i; This represents the median of the polar angle interval corresponding to grid cell i; The value represents the length of the preset polar angle interval; exp() represents an exponential function with the natural constant as the base.

[0063] It should be noted that, based on the above division of polar angle intervals, the length of the preset polar angle interval can be determined. It is 30°.

[0064] Based on the angle weight of each data point within each grid cell, it can be understood that the angle weight is used to assign different importance to the curvature characteristics of a data point in the polar coordinate grid according to its position, reflecting the effectiveness of the data point's contribution to the overall characteristics of its grid cell. If the difference between the polar angle of data point j in grid cell i and the median of the corresponding polar angle interval of grid cell i is larger, it indicates that the current data point is located in the edge region of the grid. At this time, the angle weight will decrease sharply, indicating that the curvature characteristics of data point i are less representative of the grid center region, and its contribution is suppressed.

[0065] Conversely, if the difference between the polar angle of data point j in grid cell i and the median of the corresponding polar angle interval of grid cell i is smaller, it means that the current data point is closer to the center region of the grid. At this time, the angle weight will approach the maximum value. This indicates that the curvature characteristics of data point j are highly representative of the center region of the grid cell, thus ensuring that the calculated grid cell feature values ​​can more accurately reflect the geometric characteristics of the grid center rather than the edge.

[0066] Furthermore, this embodiment determines the curvature anisotropy characteristic value of each grid cell based on the angle weight of each data point within each grid cell and in conjunction with the relative curvature difference. Specifically:

[0067] In this embodiment, the product of the normalized relative curvature difference value and the normalized angle weight value of each data point in each grid cell is calculated, and the mean of the product of all data points in each grid cell is used as the curvature anisotropic feature value of each grid cell.

[0068] Based on the curvature anisotropy eigenvalues ​​of each grid cell, it can be understood that the curvature anisotropy eigenvalues ​​are used to comprehensively characterize the weighted average geometric complexity of all point clouds within the polar coordinate grid cell, reflecting the strength of surface anisotropy in the spatial region corresponding to the grid cell. If the relative curvature difference of the data points within the current grid cell is larger and the angle weight is larger, the corresponding curvature anisotropy eigenvalue is larger, indicating that the region represented by the current grid cell as a whole has strong curvature anisotropy, suggesting that the current grid cell may correspond to a complex structural region, such as the elliptical cross-section of a volute.

[0069] Conversely, if the relative curvature difference values ​​of the data points within the current grid cell are generally small, or even if there are a few data points with large relative curvature differences but small angle weights, the corresponding curvature anisotropy characteristic value will be smaller. This indicates that the curvature anisotropy of the region represented by the current grid cell is very weak overall, suggesting that the geometry of the region is simpler and more uniform, possibly corresponding to simple structures such as flat ground, walls, or large-radius circular areas of a volute.

[0070] Thus, this embodiment divides the point cloud region into a polar coordinate grid and calculates feature values ​​that accurately reflect the intensity of local surface anisotropy by combining the relative curvature difference and angular position weight of each data point. This method can effectively quantify the complex gradual change process of the volute from a circular to an elliptical cross-section, thereby accurately identifying key fine structural regions such as the volute before filtering, providing a reliable basis for subsequent adaptive filtering, and ensuring that the geometric details of these regions are preserved to the greatest extent during the modeling process.

[0071] S2.2: Evaluate the spatial distribution regularity of the curvature anisotropic eigenvalues ​​of all grid cells within the region of each cluster to determine the structural eigenvalues ​​of each cluster; based on the energy distribution of the curvature anisotropic eigenvalues ​​of all grid cells within the region of each cluster in the frequency domain, determine the frequency domain eigenvalues ​​of each cluster, and combine them with the structural eigenvalues ​​to determine the comprehensive eigenvalues ​​of each cluster.

[0072] As a continuous helical structure with a fixed wrap angle (e.g., 345°), the cross-sectional geometry of the volute exhibits a regular evolution along the helical path. If its circumferential curvature distribution is abstracted as a periodic signal, this signal will display significant harmonic characteristics in the frequency domain. Specifically, its spectral energy will be highly concentrated on specific harmonic orders determined by the volute wrap angle (345°), thus forming clear and identifiable frequency domain characteristics.

[0073] Therefore, based on the above analysis, this embodiment determines the structural characteristic values ​​of each cluster by evaluating the spatial distribution regularity of the anisotropic curvature eigenvalues ​​of all grid cells within the region of each cluster; it also determines the frequency domain characteristic values ​​of each cluster by analyzing the energy distribution of the anisotropic curvature eigenvalues ​​of all grid cells within the region of each cluster in the frequency domain, and combines these with the structural characteristic values ​​to determine the comprehensive characteristic values ​​of each cluster. The specific process is as follows:

[0074] First, this embodiment determines the structural characteristic values ​​of each cluster by evaluating the spatial distribution regularity of the anisotropic curvature eigenvalues ​​of all grid cells within the region where each cluster is located. Specifically:

[0075] Within each cluster region, a curvature anisotropy matrix is ​​constructed based on the curvature anisotropy eigenvalues ​​of all grid cells. The rows of the curvature anisotropy matrix represent the indices of the polar angle intervals corresponding to the grid cells, and the columns represent the indices of the polar radius intervals. The elements of the curvature anisotropy matrix are the curvature anisotropy eigenvalues, thus yielding a... The matrix.

[0076] Furthermore, the curvature anisotropic matrix is ​​used as the input to the singular value decomposition algorithm, and all singular values ​​are output. The sum of the first preset number of singular values ​​in the descending order of all singular values ​​is used as the structural feature value of each cluster.

[0077] It should be noted that the preset quantity is set manually. In this embodiment, the preset quantity is 3. In actual application, as other implementation methods, implementers can also set it according to specific circumstances. This embodiment does not impose any special restrictions.

[0078] Singular value decomposition (SVD) is a well-known technique, and the specific process of using it to calculate the singular values ​​of a matrix will not be elaborated here.

[0079] Based on the structural eigenvalues ​​of each cluster, it can be understood that the structural eigenvalues ​​quantify the degree of structuring and complexity of curvature anisotropy within the cluster. If the structural eigenvalue of the current cluster is larger, it indicates that the energy contained in the first few main modes in the curvature anisotropy matrix is ​​more concentrated, indicating that the curvature anisotropy distribution of the current cluster has strong structure and regularity. For example, a complete volute has a very clear curvature anisotropy gradient pattern from the inlet to the nose, which is more likely to be the region where complex and fine structures such as pipes and turbines with elliptical cross-section features of the volute are located, requiring fine filtering.

[0080] Conversely, if the structural eigenvalue of the current cluster is smaller, it indicates that the energy in the curvature anisotropy matrix is ​​dispersed in many tiny patterns, suggesting that the curvature anisotropy distribution of the current cluster is chaotic and lacks macroscopic regularity. For example, a mixed region containing noise and multiple unrelated components or a simple wall surface does not have an obvious global pattern in its curvature change, so it is more likely to be a simple structure such as the ground or wall, and a stronger filter can be used to simplify the data.

[0081] Furthermore, this embodiment determines the frequency domain characteristic values ​​of each cluster based on the energy distribution of the anisotropic curvature eigenvalues ​​of all grid cells within the region of each cluster in the frequency domain. Specifically:

[0082] In this embodiment, all elements of each column in the curvature anisotropic matrix of the region where each cluster is located are used as input to the fast Fourier transform algorithm, and the spectrum of each column of elements in the frequency domain is output. The information entropy of all amplitudes of all elements in the frequency domain of the curvature anisotropic matrix of the region where each cluster is located is calculated and denoted as the first feature value of each column.

[0083] Furthermore, based on the preset volute envelope angle, the harmonic order in the frequency domain signal corresponding to each column element of the curvature anisotropic matrix is ​​calculated. The absolute value of the amplitude at the harmonic order corresponding to the preset volute envelope angle is calculated and denoted as the second eigenvalue of each column. The specific calculation expression for the harmonic order is as follows: In the formula, The preset volute wrap angle is used to calculate the harmonic order of the frequency domain signal corresponding to the element in the a-th column of the curvature anisotropy matrix; A represents the preset volute wrap angle, the value of which in this embodiment is... M represents the number of rows in the matrix; in this embodiment, M=12. This represents the floor function.

[0084] Furthermore, in this embodiment, the third characteristic value of each column is determined based on the difference between the second characteristic value of each column and its adjacent previous column. Specifically, in this embodiment, the absolute difference between the second characteristic value of each column and its adjacent previous column is calculated, and the negative of the absolute difference is used as the independent variable of an exponential function with the natural constant as the base. The resulting exponential function value is used as the third characteristic value of each column.

[0085] Based on the first, second, and third eigenvalues, we can understand that the first eigenvalue characterizes the concentration of energy distribution of curvature anisotropic eigenvalues ​​in the frequency domain. The smaller the first eigenvalue of the current column in the curvature anisotropic matrix, the more concentrated the spectral energy is in a few frequency components, the stronger the signal regularity, indicating that the curvature anisotropic distribution of the current column has strong periodicity, which is a typical volute spiral structure. The second eigenvalue is used to directly quantify the harmonic intensity in the curvature anisotropic matrix that matches the volute wrap angle. The larger the second eigenvalue of the current column, the higher the energy of the signal corresponding to the element in the current column at the characteristic frequency corresponding to 345°. This strongly proves that the curvature distribution in this region has a periodicity that is completely consistent with the volute wrap angle, and the greater the possibility that this region is a volute structure. The third eigenvalue is used to measure the continuity of characteristic frequency amplitude between adjacent radial rings. The closer the third eigenvalue of the current column is to 1, the closer the characteristic amplitudes of adjacent column elements are, the smoother the change, and the more it corresponds to the characteristic of the volute structure being continuous and smooth in the radial direction.

[0086] Furthermore, in this embodiment, the first, second, and third eigenvalues ​​of each column are used to form an eigenvector. All eigenvectors of the curvature anisotropic matrix are used as input to the Bayesian algorithm. A feature weight is assigned to the first eigenvalue, which is the proportion of the first eigenvalue to the sum of the first, second, and third eigenvalues. A spatial Gaussian kernel is also set. M and N represent the number of rows and columns of the curvature anisotropic matrix, respectively. max() represents the maximum value function. The gain coefficient is set to 0.5. Finally, the output probability value is used as the frequency domain feature value of each cluster.

[0087] Among them, the Bayesian algorithm is a well-known technique, and the specific process of using it to perform probability evaluation on the eigenvectors of the curvature anisotropic matrix will not be elaborated here.

[0088] Based on the frequency domain feature values ​​of each cluster, it can be understood that the frequency domain feature values ​​are used to determine the probability that a cluster conforms to the spiral periodicity of a volute. The frequency domain feature values ​​integrate the regularity of the spectrum, the intensity of the characteristic frequency, and the continuity of the structure. If the frequency domain feature value of the current cluster is larger, it indicates that the current cluster exhibits strong regularity, high characteristic amplitude, and high phase continuity in the frequency domain. This indicates that the region corresponding to the current cluster matches the characteristics of the volute in the frequency domain, and that the region corresponding to the current cluster is the location of the complex and fine structure of the elliptical cross section of the volute, such as pipes, turbines, generators, etc. When using the voxel filtering algorithm to denoise this region, a finer voxel grid size is used to preserve the original geometric details of the point cloud to the greatest extent.

[0089] Conversely, if the frequency domain eigenvalue of the current cluster is smaller, it indicates that the current cluster fails to simultaneously satisfy the characteristics of strong regularity, high characteristic amplitude, and high phase continuity in the frequency domain. This suggests that the frequency domain characteristics of the region corresponding to the current cluster differ significantly from the ideal model of the volute. For example, its spectral energy is dispersed, it lacks strong harmonic components that match the volute's wrap angle, or there are discontinuous abrupt changes in the structure in the radial direction. Therefore, it can be determined as a non-critical region with a simple structure, such as a flat ground, a regular cylinder, or random noise. When using the voxel filtering algorithm to denoise this region, a larger voxel grid size can be used to efficiently simplify the data and remove redundancy.

[0090] Furthermore, this embodiment determines the comprehensive feature value of each cluster based on the structural feature value and frequency domain feature value of each cluster, specifically:

[0091] In this embodiment, the result of forward fusion of the structural feature values ​​and frequency domain feature values ​​of each cluster is used as the comprehensive feature value of each cluster.

[0092] It should be understood that positive fusion refers to combining two or more indicators through addition or multiplication to obtain a comprehensive indicator, thereby more comprehensively and accurately assessing a phenomenon or problem. This fusion method is not limited to simple arithmetic operations, but can also include more complex statistical models and analytical methods. Implementers can choose according to specific circumstances, and this embodiment does not impose any special restrictions.

[0093] Preferably, in this embodiment, the mean of the forward fusion of the time-domain feature values ​​and frequency-domain feature values ​​of each cluster is used as the comprehensive feature value of each cluster. In practical applications, implementers may also adopt forward fusion methods such as sum, product or other weighting methods according to specific circumstances. This embodiment does not impose any special restrictions.

[0094] Based on the comprehensive feature values ​​of each cluster, it can be understood that the comprehensive feature values ​​are used to integrate global information in the time domain and local periodic information in the frequency domain to perform a final and comprehensive geometric complexity assessment of each cluster, characterizing the comprehensive possibility that the cluster is a complex structure like a volute. If the structural feature value of the current cluster is larger and the frequency domain feature value is larger, it indicates that the region where the current cluster is located has a regular gradual change from the same direction to the opposite direction similar to that of a volute, indicating that the region where the current cluster is located is a complex structure with the elliptical cross-sectional characteristics of a volute.

[0095] Conversely, if the structural and frequency domain characteristic values ​​of the current cluster are both small, or at least one is significantly low, it indicates that the region where the current cluster is located either lacks a macroscopic, regular gradual change from the same direction to the opposite direction, or does not possess a microscopic, spiral periodicity that matches the volute's wrap angle, or neither. This comprehensively reflects that the region's geometry is simple, disordered, or irregular, such as a flat ground, a regular cylindrical structure, or messy noise points. Therefore, it can be determined as a simple region that is not a volute.

[0096] Thus, this embodiment constructs a comprehensive feature value by integrating structural feature values ​​that characterize the gradual change of space with frequency domain feature values ​​that characterize the spiral periodicity. This comprehensive feature value can accurately distinguish between complex and fine structures similar to volutes in hydropower stations and other simple structures, providing a decisive quantitative basis for the subsequent implementation of adaptive and differentiated point cloud filtering strategies.

[0097] Step S3: Based on the comprehensive feature values, optimize the voxel filtering algorithm to filter the 3D point cloud of the underground powerhouse of the hydropower station, and perform visualization modeling of the underground powerhouse of the hydropower station based on the filtered 3D point cloud.

[0098] Based on the comprehensive eigenvalues ​​obtained in step S2, these eigenvalues ​​are used to adjust the voxel grid size in the voxel filtering algorithm. Specifically:

[0099] As one implementation method, in this embodiment, the voxel grid size when filtering cluster m in the voxel filtering algorithm is... The expression is: ; This represents the normalized value of the comprehensive eigenvalues ​​of cluster m; , These represent the upper and lower limits of the preset voxel mesh size, respectively.

[0100] It should be noted that the upper and lower limits of the preset voxel mesh size are 20 and 5 respectively in this embodiment. In actual applications, as other implementation methods, implementers can also set them according to specific circumstances. This embodiment does not impose any special restrictions.

[0101] Based on the voxel grid size, it can be understood that if the comprehensive feature value of the current cluster is larger, it means that the current cluster is more likely to correspond to a complex structure similar to a spiral shell, such as a hydroelectric power station, and details need to be preserved. Therefore, the voxel grid size at the calculation site is smaller. This means that when performing voxel filtering, a very fine grid will be used to preserve the original geometric details of the point cloud to the greatest extent.

[0102] Conversely, the smaller the comprehensive feature value of the current cluster, the more likely the current cluster corresponds to a simple, non-critical area, such as a flat ground or a regular wall, without needing to retain too much detail. Therefore, the calculated voxel grid size will be larger, which means that a coarser grid will be used when performing voxel filtering to efficiently remove redundant point clouds and reduce the amount of data.

[0103] Furthermore, the three-dimensional point cloud of the underground powerhouse of the hydropower station is filtered based on the optimized voxel filtering algorithm to obtain the filtered three-dimensional point cloud.

[0104] The process of using voxel filtering algorithm to filter 3D point clouds is a well-known technique and will not be described in detail here.

[0105] In OurBIM software, the filtered 3D point cloud is used to create a parametric BIM model based on its outline and spatial position. Necessary lightweight processing is then applied to the BIM model, such as merging small components and optimizing texture maps. For structures in the BIM model, such as hydroelectric generators and pipelines, IFC standard attribute information (e.g., component name, model, material) is added. Furthermore, a VR scene is set up (e.g., defining scene brightness, setting the initial viewing angle, and planning the user's roaming path in VR). The processed BIM model and VR scene are then imported into the VR environment via the OurVR module. Basic interactive functions (e.g., roaming settings, information query) are configured in the VR engine to generate a VR scene of the hydroelectric power station's underground powerhouse.

[0106] The principle of using OurBIM and OurVR software to visualize the underground powerhouse of a hydropower station is a well-known technology and will not be elaborated further.

[0107] Thus, this embodiment achieves adaptive filtering of hydropower station cloud by dynamically combining comprehensive feature values ​​with voxel filtering algorithm. Specifically, it uses fine mesh to preserve details for key complex structures such as spiral shells, and uses coarse mesh to efficiently simplify simple areas, thereby improving the accuracy and efficiency of hydropower station underground powerhouse modeling.

[0108] Example 2:

[0109] The computer-readable storage medium of this embodiment stores a computer program that, when executed by a processor, implements the steps in the BIM and VR-based visualization modeling and restoration method for underground powerhouses of hydropower stations in Embodiment 1.

[0110] The computer-readable storage medium in this embodiment can be an internal storage unit of the terminal, such as the terminal's hard disk or memory; the computer-readable storage medium in this embodiment can also be an external storage device of the terminal, such as a plug-in hard disk, smart memory card, secure digital card, flash memory card, etc. equipped on the terminal; furthermore, the computer-readable storage medium can include both the terminal's internal storage unit and external storage devices.

[0111] The computer-readable storage medium of this embodiment is used to store computer programs and other programs and data required by the terminal. The computer-readable storage medium can also be used to temporarily store data that has been output or will be output.

[0112] Example 3:

[0113] The computer device of this embodiment includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, it implements the steps in the BIM and VR-based visualization modeling and restoration method for underground powerhouses of hydropower stations in Embodiment 1.

[0114] In this embodiment, the processor can be a central processing unit, or other general-purpose processors, digital signal processors, application-specific integrated circuits, off-the-shelf programmable gate arrays or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc. The memory can include read-only memory and random access memory, and provides instructions and data to the processor. A portion of the memory can also include non-volatile random access memory. For example, the memory can also store device type information.

[0115] Those skilled in the art will understand that the content disclosed in the embodiments can be provided as a method, system, or computer program product. Therefore, this solution can take the form of a hardware embodiment, a software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this solution can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage and optical storage) containing computer-usable program code.

[0116] This solution is described with reference to flowchart illustrations and / or block diagrams of methods and computer program products according to embodiments of this solution. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing device, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0117] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0118] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0119] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0120] The examples described herein are merely preferred embodiments of the invention and are not intended to limit the concept and scope of the invention. Any modifications and improvements made by those skilled in the art to the technical solutions of the invention without departing from the design concept of the invention should fall within the protection scope of the invention.

[0121] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the specific embodiments described above. The specific embodiments and descriptions in the specification are merely for further illustrating the principles of the invention. Various changes and modifications can be made to the present invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the claims and their equivalents.

Claims

1. A method for visual modeling and reconstruction of underground powerhouses in hydropower stations based on BIM and VR, characterized in that, Includes the following steps: Obtain the 3D point cloud of the underground powerhouse of the hydropower station; Cluster all data points in the 3D point cloud; perform surface fitting on each data point and all data points in its neighborhood within each cluster; determine the relative curvature difference value of each data point based on the difference between the two principal curvatures on the fitted surface; divide the region where each cluster is located into polar radius-polar angle grids; determine the angle weight of each data point in each grid cell based on the difference between the polar angle of each data point in each grid cell and the median of the polar angle interval corresponding to the grid cell; and determine the curvature anisotropic feature value of each grid cell in combination with the relative curvature difference. The spatial distribution regularity of the curvature anisotropic eigenvalues ​​of all grid cells within the region of each cluster is evaluated to determine the structural eigenvalues ​​of each cluster. Based on the energy distribution of the anisotropic curvature eigenvalues ​​of all grid cells within the region of each cluster in the frequency domain, the frequency domain eigenvalues ​​of each cluster are determined, and combined with the structural eigenvalues, the comprehensive eigenvalues ​​of each cluster are determined. Based on the comprehensive feature values, the voxel filtering algorithm is optimized to filter the 3D point cloud of the underground powerhouse of the hydropower station, and the underground powerhouse of the hydropower station is visualized and modeled based on the filtered 3D point cloud.

2. The method as described in claim 1, characterized in that, The distance metric used in the clustering process is the Euclidean distance between the three-dimensional coordinates of the data points.

3. The method as described in claim 1, characterized in that, The relative curvature difference value for each data point is the result of the absolute difference between the two principal curvatures on the fitted surface and the sum of the absolute values ​​of the two principal curvatures.

4. The method as described in claim 1, characterized in that, The expression for the angle weight of each data point within each grid cell is: ； In the formula, This represents the angle weight of data point j within grid cell i; Represents the polar angle of data point j within grid cell i; This represents the median of the polar angle interval corresponding to grid cell i; The value represents the length of the preset polar angle interval; exp() represents an exponential function with the natural constant as the base.

5. The method as described in claim 4, characterized in that, The method for determining the curvature anisotropic eigenvalues ​​of each grid element is as follows: Calculate the product of the normalized relative curvature difference value and the normalized angle weight value for each data point in each grid cell. Take the mean of the product of all data points in each grid cell as the curvature anisotropic characteristic value of each grid cell.

6. The method as described in claim 1, characterized in that, The method for determining the structural feature values ​​of each cluster is as follows: Within each cluster region, a curvature anisotropy matrix is ​​constructed based on the curvature anisotropy eigenvalues ​​of all grid cells, where the elements in the curvature anisotropy matrix are curvature anisotropy eigenvalues. The curvature anisotropic matrix is ​​used as the input to the singular value decomposition algorithm. All singular values ​​are output, and the sum of the first preset number of singular values ​​in the descending order of all singular values ​​is used as the structural feature value of each cluster.

7. The method as described in claim 6, characterized in that, The method for determining the frequency domain eigenvalues ​​of each cluster is as follows: Calculate the information entropy of all amplitudes of all elements in the frequency domain of the curvature anisotropic matrix of each column in the region where each cluster is located, and denote it as the first eigenvalue of each column. Based on the preset volute encirclement angle, calculate the harmonic order in the frequency domain signal corresponding to each column element of the curvature anisotropic matrix, calculate the absolute value of the amplitude at the harmonic order corresponding to the preset volute encirclement angle, and record it as the second characteristic value of each column. The third characteristic value of each column is determined based on the difference between the second characteristic value of each column and its adjacent preceding column; The first, second, and third eigenvalues ​​of each column are used to form an eigenvector. All eigenvectors of the curvature anisotropic matrix are used as inputs to the Bayesian algorithm, and the output probability values ​​are used as the frequency domain eigenvalues ​​of each cluster.

8. The method as described in claim 7, characterized in that, The comprehensive characteristic value of each cluster is the result of the positive fusion of the structural characteristic value and the frequency domain characteristic value of each cluster.

9. The method as described in claim 1, characterized in that, The optimized voxel filtering algorithm specifically involves adjusting the voxel grid size when filtering cluster m. The expression is: ； This represents the normalized value of the comprehensive eigenvalues ​​of cluster m; , These represent the upper and lower limits of the preset voxel mesh size, respectively.

10. A visualization modeling and reconstruction device for underground powerhouses of hydropower stations based on BIM and VR, characterized in that: The device is used to perform the steps in the BIM and VR-based visualization modeling and restoration method for underground powerhouses of hydropower stations as described in any one of claims 1 to 9.

Images