Building spoil slope material component rapid investigation and deformation damage prediction method

Abstract

The invention relates to a building spoil slope material component rapid investigation and deformation damage prediction method comprising the following steps: obtaining a target multi-view image through aerial photography of an unmanned aerial vehicle, preprocessing the target multi-view image, and extracting material component characteristic information comprising material component proportion information, actual size distribution, directionality indexes, edge complexity and heterogeneous degree; a discrete element heterogeneous slope model reflecting actual material information is constructed in combination with three-dimensional point cloud data, and deformation damage prediction results such as a slip plane development track, particle displacement field distribution and settlement deformation are obtained through gravity load simulation. Safe, efficient and accurate material component analysis is achieved through the unmanned aerial vehicle aerial photography and image recognition technology, the heterogeneous model constructed based on actual material component information can accurately simulate the slope deformation and damage process, the personnel exposure risk and investigation cost in a high-risk area are remarkably reduced, the spatial and temporal distribution characteristics of the material components are dynamically adapted, and the material component analysis accuracy is improved. And the global dynamic analysis requirement is met.

Description

Technical Field

[0001] This invention relates to the field of construction waste soil slope technology, specifically to a method for rapid investigation of material composition and prediction of deformation and failure of construction waste soil slopes. Background Technology

[0002] Construction waste slopes in mountainous areas are typically composed of construction debris, engineering waste, and other waste materials. These materials are characterized by their wide range of sources, diverse compositions, and disordered packing, resulting in dynamic updates of material composition, loose structure, and significant differences in physical and mechanical properties. Rapid and accurate analysis of the material composition of construction waste slopes and prediction of slope deformation and failure characteristics can provide basic data for operation and maintenance managers and researchers, and is also an important guarantee for the full-cycle safety management of such sites.

[0003] Currently, the determination of material composition on construction waste slopes mainly relies on on-site observation or indoor geotechnical tests through on-site sampling at limited points. This method is inefficient, costly, and highly susceptible to factors such as topography, and it has problems such as difficulty in characterizing the material conditions of the entire slope area. Furthermore, the loose and easily collapsible nature of construction waste slopes poses a high operational risk to researchers during on-site investigations and sampling. In addition, existing studies are mostly based on homogeneous models to analyze or predict the deformation and failure characteristics of construction waste slopes, and the results obtained have certain deviations from the actual situation. Summary of the Invention

[0004] In view of the above problems, the present invention provides a method for rapid investigation of material composition and deformation and failure prediction of construction waste slopes, which solves the problems of low efficiency, high cost, and great influence of topographic and other factors in the investigation of material composition of traditional construction waste slopes, making it difficult to characterize the material situation of the entire slope area, as well as the large deviation between the existing deformation and failure prediction of construction waste slopes and the actual situation.

[0005] To achieve the above objectives, this application provides a method for rapid investigation of material composition and prediction of deformation and failure of construction waste slopes, including:

[0006] Collect multi-view images of the target slope according to preset aerial photography parameters, including flight altitude, heading overlap rate, lateral overlap rate, and camera tilt angle;

[0007] The multi-view images are filtered and preprocessed to obtain standard slope images. The preprocessing includes brightness adjustment, contrast adjustment and image partitioning and cropping. The standard slope images contain global information of the slope area to be analyzed and clearly identifiable material boundaries.

[0008] RGB threshold segmentation and binarization were performed on standard slope images to extract material composition feature parameters, including the proportion of each material component, actual size distribution information, orientation index information, edge complexity and degree of heterogeneity.

[0009] Based on the material composition characteristic parameters and combined with the 3D point cloud data obtained by reconstructing multi-view images, a discrete element heterogeneous slope model is constructed. The discrete element heterogeneous slope model should reflect the basic characteristic parameters of the actual material information and the actual slope geometric morphology characteristic parameters. The basic characteristic parameters include the material content ratio, particle morphology characteristics and structural distribution characteristics, while the actual slope geometric morphology characteristic parameters include slope height and slope angle parameters.

[0010] Gravity loads are applied to a discrete element heterogeneous slope model to simulate and obtain slope deformation and failure prediction results. The slope deformation and failure prediction results include the slip surface development trajectory, particle displacement field distribution, and settlement deformation.

[0011] In some embodiments, preset aerial photography parameters are obtained through the following steps:

[0012] Obtain the original geometric features of the target slope, including slope height, slope length, and slope angle;

[0013] Determine the initial flight altitude range based on the original geometric morphological characteristics;

[0014] Based on image resolution parameters, an adjusted flight altitude is generated within the initial flight altitude range;

[0015] Obtain the slope topography complexity of the target slope, and set the parameters of the forward overlap rate and the lateral overlap rate accordingly. The shooting paths under the forward overlap rate parameter and the shooting paths under the lateral overlap rate parameter are orthogonal to each other.

[0016] Calculate the camera tilt angle to ensure that the matching error between the camera tilt angle and the slope normal direction meets the preset error range;

[0017] Based on the on-site environmental information, the parameters of flight altitude, forward overlap rate, lateral overlap rate, and camera tilt angle are jointly optimized to generate the optimized flight altitude, forward overlap rate, lateral overlap rate, and camera tilt angle, which are the preset aerial photography parameters.

[0018] In some embodiments, acquiring multi-view images of the target slope according to preset aerial photography parameters includes:

[0019] By using a drone equipped with a high-resolution camera, aerial surveying tasks are carried out according to the flight altitude, and flight control commands corresponding to the final flight altitude are established simultaneously during the aerial surveying tasks.

[0020] Based on the heading overlap rate and lateral overlap rate parameters, generate UAV three-dimensional spatial flight path distribution information that meets the preset overlap rate requirements;

[0021] Adjust the gimbal of the high-resolution camera according to the camera tilt angle so that the optical axis of the high-resolution camera matches the slope normal of the target slope.

[0022] Under the preset aerial survey parameters, confirm whether the flight safety of the UAV meets the preset safety threshold under real-time environmental information. If not, suspend the aerial survey mission and adjust the flight altitude.

[0023] If so, the high-resolution camera will be triggered to capture images according to the preset acquisition frequency to obtain multi-view images that meet the requirements of spatiotemporal synchronization.

[0024] In some embodiments, the multi-view images are filtered and preprocessed to obtain standard slope images, including:

[0025] Based on the original RGB channel values ​​of the multi-view image, an adaptive histogram equalization algorithm is used to adjust the brightness in the HSV color space to obtain a brightness-adjusted image.

[0026] The brightness-adjusted image is subjected to contrast enhancement processing using an illumination compensation algorithm to obtain a brightness-enhanced image, including:

[0027] Based on the three-dimensional point cloud computing slope normal field corresponding to the multi-view image, establish an incident light angle compensation model.

[0028] A brightness-enhanced image is obtained by non-linearly stretching the brightness component of the brightness-adjusted image in the Lab color space.

[0029] The slope boundary is identified based on the characteristics of the top and bottom lines of the target slope, and the corrected image is obtained by performing perspective distortion correction and geometric deformation compensation on the brightness-enhanced image in combination with UAV positioning data.

[0030] The corrected image is processed using an edge sharpening algorithm to generate a standard slope image that meets the requirements for material composition identification.

[0031] In some embodiments, RGB thresholding and binarization are performed on standard slope images to extract material component feature parameters, including:

[0032] Based on the RGB channel values ​​of standard slope images, an adaptive threshold segmentation algorithm is used to segment materials such as tile fragments, steel bars, and wood separately to obtain binarized images of each material.

[0033] Feature parameters are extracted from the binarized images of each material, including:

[0034] Calculate the content ratio of each material component based on the pixel area ratio of the material;

[0035] The dimensional characteristics of each material component are statistically analyzed based on the equivalent diameter algorithm, and the actual size distribution information is calculated.

[0036] By combining the orientation rose diagram with the anisotropy index to quantify the aspect ratio and principal direction angle of material particles, orientation index information is obtained.

[0037] The fractal dimension is calculated using the box counting method to quantify the edge complexity of material particles;

[0038] The spatial distribution characteristics of materials are analyzed based on the semi-variogram function to obtain the degree of heterogeneity.

[0039] Based on the material composition content ratio information, actual size distribution information, orientation index information, edge complexity, and degree of heterogeneity, material composition characteristic parameters are generated, and a set of material composition characteristic parameters is established.

[0040] In some embodiments, the 3D point cloud data obtained by reconstructing multi-view images includes:

[0041] An initial 3D point cloud is generated based on multi-view imagery using a multi-view stereo matching algorithm.

[0042] The initial 3D point cloud is sequentially processed by statistical outlier filtering, radius filtering, and feature-preserving filtering based on slope threshold to obtain a filtered 3D point cloud.

[0043] Based on the camera tilt angle and flight altitude, a projection geometry correction model is established to perform lens distortion correction, projection deformation compensation, and illumination non-uniformity correction on the filtered 3D point cloud to obtain the corrected 3D point cloud.

[0044] In some embodiments, a discrete element heterogeneous slope model is constructed based on material composition characteristic parameters and combined with three-dimensional point cloud data reconstructed from multi-view images, including:

[0045] A numerical model of the slope is constructed using point cloud registration and surface reconstruction algorithms based on 3D point cloud data;

[0046] Based on the material composition characteristic parameters, the proportion of each material component in the slope model is determined and denoted as the material content proportion.

[0047] Based on the actual size distribution and orientation index in the material composition characteristic parameters, the particle morphology characteristics are obtained by using a multi-sphere aggregation algorithm or a block feature algorithm.

[0048] Based on the marginal complexity and heterogeneity of the material composition characteristic parameters, the structural distribution characteristics are obtained by box counting.

[0049] By coupling the material content ratio, particle morphology characteristics, and structural distribution characteristics into the slope numerical model, and by assigning discrete element contact parameters, a discrete element heterogeneous slope model with real material content ratio and distribution characteristics is established.

[0050] In some embodiments, a gravity load is applied to a discrete element heterogeneous slope model to simulate and obtain slope deformation and failure prediction results. These prediction results include the slip surface development trajectory, particle displacement field distribution, and settlement deformation, including:

[0051] Initial stress equilibrium calculations were performed on the discrete element heterogeneous slope model, and the servo control method was used to bring the discrete element heterogeneous slope model to the initial static equilibrium state.

[0052] Gravity loads are applied gradually, and dynamic calculations are performed using the explicit time integration method. The formation and development trajectory of the slip surface are tracked in real time through contact force chain evolution analysis to obtain the development trajectory of the slip surface.

[0053] Simultaneously, the position change data of each particle unit during the calculation process are recorded, and the particle displacement field distribution is generated through vector field visualization technology, and the statistical characteristics of the displacement vector in each region are calculated.

[0054] Based on the vertical displacement data of surface monitoring points of the discrete element heterogeneous slope model, the settlement deformation surface is constructed using the Kriging interpolation method, and the settlement deformation amount and its spatial distribution characteristics in different regions are calculated.

[0055] By coupling analysis of the slip surface development trajectory, particle displacement field distribution, and settlement deformation, the prediction results of slope deformation and failure are obtained.

[0056] Unlike existing technologies, the above technical solution acquires multi-view images of the target slope through UAV aerial photography and preprocesses them. Image recognition technology is then used to extract material composition feature parameters, including material composition percentage, actual size distribution, orientation index, edge complexity, and degree of heterogeneity. Combined with 3D point cloud data, a discrete element heterogeneous slope model reflecting the actual material information and distribution is constructed. Gravity load simulation yields characteristic parameters such as the slip surface development trajectory, particle displacement field distribution, and settlement deformation, ultimately generating slope deformation and failure prediction results. This technical solution achieves safe, efficient, and accurate material composition analysis through UAV aerial photography and image recognition technology, quantifying and analyzing comprehensive information such as material content, geometric dimensions, and structural characteristics. The heterogeneous model constructed based on actual material composition accurately simulates the slope deformation and failure process, overcoming the significant bias of traditional homogeneous models and providing a reliable basis for engineering design. It significantly reduces the risk of personnel exposure and investigation costs in high-risk areas, dynamically adapts to the spatiotemporal distribution characteristics of material composition, and meets the needs of dynamic analysis across the entire region. It has wide applicability and can be extended to various loose deposits.

[0057] The above description of the invention is merely an overview of the technical solution of this application. In order to enable those skilled in the art to better understand the technical solution of this application and to implement it based on the description and drawings, and to make the above-mentioned objectives and other objectives, features and advantages of this application easier to understand, the following description is provided in conjunction with the specific embodiments and drawings of this application. Attached Figure Description

[0058] The accompanying drawings are only used to illustrate the principles, implementation methods, applications, features, and effects of specific embodiments of the present invention and other related contents, and should not be considered as limitations on this application.

[0059] In the accompanying drawings of the instruction manual:

[0060] Figure 1 This is a flowchart illustrating the method for material composition identification and deformation / damage prediction in a specific implementation.

[0061] Figure 2 The route planning diagram described in the specific implementation method;

[0062] Figure 3 A typical aerial photograph of the implementation site described in the specific implementation method and its preprocessed result;

[0063] Figure 4 This is a diagram illustrating the image recognition process and results for area A2 of the implementation site as described in the specific implementation method.

[0064] Figure 5 This is a diagram showing the material composition identification results at the implementation site as described in the specific implementation method.

[0065] Figure 6 This is a real-world model of the implementation site described in the specific implementation method;

[0066] Figure 7 A schematic diagram of the discrete element model of the implementation site described in the specific implementation method;

[0067] Figure 8 This is a diagram showing the development trend of the potential sliding surface of the slope at the implementation site as described in the specific implementation method.

[0068] Figure 9 This is a comparison diagram of the simulated slope slip surface of the implementation site and its actual shape, as described in the specific implementation method. Detailed Implementation

[0069] To illustrate the possible application scenarios, technical principles, implementable specific solutions, and achievable objectives and effects of this application in detail, the following description, in conjunction with the listed specific embodiments and accompanying drawings, provides a detailed explanation. The embodiments described herein are merely illustrative of the technical solutions of this application and are therefore intended to limit the scope of protection of this application.

[0070] In this document, the term "embodiment" means that a specific feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The term "embodiment" appearing in various places throughout the specification does not necessarily refer to the same embodiment, nor does it specifically limit its independence or connection with other embodiments. In principle, in this application, as long as there are no technical contradictions or conflicts, the technical features mentioned in each embodiment can be combined in any way to form corresponding implementable technical solutions.

[0071] Unless otherwise defined, the technical terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the use of related terms herein is merely for the purpose of describing particular embodiments and is not intended to limit this application.

[0072] In the description of this application, the term "and / or" is used to describe the logical relationship between objects, indicating that three relationships can exist. For example, A and / or B means: A exists, B exists, and A and B exist simultaneously. Additionally, the character " / " in this document generally indicates that the preceding and following objects have an "or" logical relationship.

[0073] The processor described in the embodiments of this application can be implemented by hardware, firmware, software, or a combination thereof. It can be a circuit, one or more of an application-specific integrated circuit (ASIC), a digital signal processor (DSP), a digital signal processing device (DSPD), a programmable logic device (PLD), a field-programmable gate array (FPGA), a central processing unit (CPU), a controller, a microcontroller, or a microprocessor. It also includes other physical, biological, or chemical structures that can implement the same or equivalent functions as the processors listed above, such as biological neurons, quantum computing units, DNA computing units, etc., so that the processor can execute some or all of the steps in the computer program or method involved in the various embodiments of this application, or any combination of the steps mentioned therein.

[0074] The computer program involved in the embodiments can be stored in a computer device readable storage medium, which includes, but is not limited to, disks, magnetic tapes, magnetic cards, floppy disks, flash memory, optical disks, optical cards, read-only memory (ROM), random access memory (RAM), erasable programmable ROM (EPROM), and electrically erasable programmable ROM (EEPROM), etc., and also includes other biological, physical, or chemical structures that can achieve the same or equivalent functions as the storage media listed above, such as DNA, RNA, proteins, and other units with information storage capabilities. In specific embodiments, the storage medium involved can be one of the above-mentioned media types, or a combination of the above-mentioned media types. In different embodiments, the computer program involved in the embodiments can be centrally stored in a single medium, or distributed and stored in multiple media. The memory containing the computer device readable storage medium can be non-volatile memory or random access memory. These computer device readable storage media can be built into the device, or can be connected to the device involved in the embodiments as an external device or part of an external device. In some embodiments, the memory having a computer device readable storage medium is deployed locally; in other embodiments, the memory may be deployed remotely from the processor, for example, as a network-attached memory accessed via RF circuitry or an external port and a communication network, wherein the communication network may be the Internet, one or more intranets, a local area network (LAN), a wide area network (WLAN), a storage area network (SAN), or a suitable combination thereof, as long as computer device access to the memory is enabled. Furthermore, the computer program involved in the embodiments may be stored in plaintext / ciphertext form, or it may be designed as training data, integrated and recombined through model training and implicitly stored in the parameter states of a deep neural network or other machine learning model.

[0075] Construction waste slopes in mountainous areas are typically composed of construction debris, engineering waste, and other waste materials. These materials are characterized by their wide range of sources, diverse compositions, and disordered packing, resulting in dynamic updates of material composition, loose structure, and significant differences in physical and mechanical properties. Rapid and accurate analysis of the material composition of construction waste slopes and prediction of slope deformation and failure characteristics can provide basic data for operation and maintenance managers and researchers, and is also an important guarantee for the full-cycle safety management of such sites.

[0076] Currently, the determination of material composition on construction waste slopes mainly relies on on-site observation or indoor geotechnical tests through limited on-site sampling. This method is inefficient, costly, and highly susceptible to factors such as topography, making it difficult to characterize the material conditions across the entire slope area. Furthermore, the loose and easily collapsible nature of construction waste slopes poses a high operational risk to researchers during on-site investigations and sampling. In addition, existing studies are mostly based on homogeneous models to analyze or predict the deformation and failure characteristics of construction waste slopes, resulting in some discrepancies between the obtained results and the actual situation. Moreover, studies considering the impact of material composition and its heterogeneity on the deformation and failure of construction waste slopes in mountainous areas are also rare.

[0077] The traditional application of material composition investigation and stability analysis for construction waste slopes has the following main drawbacks:

[0078] (1) The field investigation of construction waste slopes is inefficient, costly, and greatly affected by topographical factors. In addition, construction waste slopes are loose and prone to collapse and sliding, which poses a high safety risk to researchers when they work on site.

[0079] (2) Sampling analysis of construction waste slope mainly involves indoor geotechnical tests through on-site sampling at limited locations to determine the composition and proportion of each material on the construction waste slope. However, there are problems such as difficulty in characterizing the material situation of the entire slope area.

[0080] (3) Existing studies are mostly based on homogeneous models to analyze or predict the deformation and failure characteristics of construction waste slopes. The results obtained are somewhat different from the actual situation. Furthermore, there are few studies that consider the influence of material composition and its heterogeneity on the deformation and failure of construction waste slopes in mountainous areas.

[0081] In view of the shortcomings of existing technologies, the purpose of this invention is to provide a method for rapid investigation of the material composition and prediction of deformation and failure of construction waste slopes.

[0082] The main purpose of this invention is to address the problems of low efficiency, high cost, and significant influence from topographical factors in traditional construction waste slope material composition surveys, which make it difficult to characterize the material conditions of the entire slope area. Furthermore, it aims to solve the problem of high operational risks for researchers during on-site investigations and sampling when construction waste slopes are characterized by looseness and susceptibility to landslides. This invention introduces UAV aerial photography and image recognition technology into the investigation of construction waste slope material composition, enabling safe, rapid, high-precision, and accurate quantitative analysis of material composition information across the entire research area, avoiding manual operations in high-risk areas.

[0083] The secondary objective of this invention is to address the problem that traditional research on the deformation and failure characteristics of construction waste slopes generally neglects the influence of slope material composition and heterogeneity. By making full use of UAV aerial photography and image recognition data, a discrete element heterogeneous model that conforms to the actual material composition and slope geometry is constructed to realize the simulation analysis and prediction of the deformation and failure characteristics of construction waste slopes, and to provide an intuitive basis for the design and construction of such slopes.

[0084] For details, please refer to Figures 1 to 9 This embodiment provides a method for rapid investigation of material composition and prediction of deformation and failure of construction waste slopes, including:

[0085] Collect multi-view images of the target slope according to preset aerial photography parameters, including flight altitude, heading overlap rate, lateral overlap rate, and camera tilt angle;

[0086] The multi-view images are filtered and preprocessed to obtain standard slope images. The preprocessing includes brightness adjustment, contrast adjustment and image partitioning and cropping. The standard slope images contain global information of the slope area to be analyzed and clearly identifiable material boundaries.

[0087] RGB threshold segmentation and binarization were performed on standard slope images to extract material composition feature parameters, including the proportion of each material component, actual size distribution information, orientation index information, edge complexity and degree of heterogeneity.

[0088] Based on the material composition characteristic parameters and combined with the 3D point cloud data obtained by reconstructing multi-view images, a discrete element heterogeneous slope model is constructed. The discrete element heterogeneous slope model should reflect the basic characteristic parameters of the actual material information and the actual slope geometric morphology characteristic parameters. The basic characteristic parameters include the material content ratio, particle morphology characteristics and structural distribution characteristics, while the actual slope geometric morphology characteristic parameters include slope height and slope angle parameters.

[0089] Gravity loads are applied to a discrete element heterogeneous slope model to simulate and obtain slope deformation and failure prediction results. The slope deformation and failure prediction results include the slip surface development trajectory, particle displacement field distribution, and settlement deformation.

[0090] In this embodiment, multi-view images of the target slope are acquired according to preset aerial photography parameters, including flight altitude, forward overlap rate, lateral overlap rate, and camera tilt angle, i.e., the acquisition angle matching the slope normal direction. Optionally, the flight altitude H is determined by the formula H=f·δ / a, where f is the camera focal length, δ is the ground resolution, and a is the camera sensor pixel size, ensuring that the image resolution meets the requirements of subsequent analysis; the forward and lateral overlap rates are both set to be no less than 70% to improve image matching accuracy; the absolute value of the camera tilt angle is complementary to the slope angle, so that the camera optical axis is as perpendicular to the slope as possible, reducing perspective distortion and image occlusion problems. Optionally, in complex terrain conditions, a grid-like or five-directional aerial photography path can be adopted, and the aerial photography path and flight altitude can be dynamically adjusted during flight to cope with sudden situations, such as sudden changes in wind speed or electromagnetic interference. For example, the UAV used should be equipped with an RTK / PPK positioning system (positioning accuracy ±1cm+1ppm), the camera sensor size should be no less than 1 inch and the single pixel size should be ≤3μm; the flight stability should meet the hovering accuracy of ±0.1m and the wind resistance level ≥7 (corresponding to wind speed of 13.9-17.1m / s). In actual operation, a safety margin can be set according to the formula H=1.2×(f·δ / a) to prevent the collision risk caused by terrain undulations.

[0091] In this embodiment, the multi-view image is preprocessed to obtain a standard slope image. Preprocessing includes brightness adjustment, contrast adjustment, and image cropping to ensure that the standard slope image contains comprehensive information about the slope area to be analyzed and clearly identifiable material boundaries. Brightness and contrast adjustments aim to enhance the RGB differences between different materials, for example, by optimizing image quality using Photoshop's Levels tool. Image cropping divides the large-scale image into several sub-regions (such as A1-A6) according to the actual needs of the target area, facilitating subsequent refined analysis.

[0092] Preferably, for localized shadows or low-contrast areas, manual coloring or Adaptive Histogram Equalization (CLAHE) technology can be used to further improve the discernibility of material boundaries. Specifically, when using Adaptive Histogram Equalization (CLAHE) technology, the grid size is set to 8×8, and the contrast is limited to 2.0; each clipping partition retains a 5% overlap band (e.g., regions A1 and A2 overlap by 30 pixels) to ensure seamless connection of analysis results between adjacent regions.

[0093] In this embodiment, RGB threshold segmentation and binarization are performed on the standard slope image to extract material component characteristic parameters, including material component content ratio information, actual size distribution information, orientation index information, edge complexity, and heterogeneity. RGB threshold segmentation is implemented based on the PCAS system. Specifically, Gaussian filtering is performed before binarization to eliminate noise interference. The pore throat closure radius and minimum recognition area should be analyzed and set according to the specific situation of the material components in the image. When the pixel area of ​​the material component to be identified differs greatly from the matrix, the minimum recognition area can be set to 10 to 25 times the pore throat closure radius (e.g., when the minimum recognition area is 50 pixels, the pore throat closure radius is set to 2 pixels). When the pixel area of ​​the material component to be identified differs little from the matrix, the minimum recognition area can be set to 1 to 0.5 times the pixel diameter corresponding to the pore throat closure radius (e.g., when the minimum recognition area is 1 pixel, the pore throat closure radius is set to 2 pixels). The fractal dimension is calculated using box counting. By setting specific RGB ranges for materials (e.g., ceramic tiles R: 190-220, G: 200-230, B: 210-240) and global grayscale thresholds (e.g., 172 ± tolerance 25), images are binarized to distinguish different material components. Actual size distribution information is obtained through the conversion relationship between pixels and actual size (Ct=c·R, St=s·R²), where Ct and St are the actual perimeter and area, c and s are the pixel perimeter and area, and R is the number of pixels in the image, ensuring the accuracy of geometric parameters. Orientation indices (probability entropy 0.975), edge complexity (shape factor 0.648), and heterogeneity (fractal dimension 1.296) are calculated using the statistical analysis function of the PCAS system. Optionally, for metal materials such as steel bars, a stricter tolerance range (10-30) can be set to improve recognition accuracy.

[0094] In this embodiment, a discrete element method (DEM) heterogeneous slope model is constructed based on material composition characteristic parameters and 3D point cloud data reconstructed from multi-view images. This model possesses fundamental characteristic parameters reflecting the actual material distribution, including material content percentage, particle morphology characteristics, and structural distribution characteristics. The 3D point cloud data is generated using specialized software (such as DJITerra or Context Capture) and further converted into a DEM model suitable for numerical simulation. Material content characteristics are directly derived from image recognition results (e.g., ceramic tile percentage 11.21%), particle morphology characteristics (e.g., shape factor 0.648) guide particle generation rules in the DEM model, and structural distribution characteristics (e.g., fractal dimension 1.296) define the heterogeneous distribution of the model. Preferably, a multi-particle-size mixed placement algorithm can be used to simulate the loose structural characteristics of miscellaneous fill soil (88%).

[0095] In this embodiment, a gravity load is applied to a discrete element heterogeneous slope model to simulate and obtain the slope deformation and failure prediction results, including the slip surface development trajectory, particle displacement field distribution, and settlement deformation. Gravity loading is implemented using particle discrete element software (such as PFC or MatDEM), and the model material parameters (such as density, normal stiffness, and friction coefficient) are set based on laboratory tests and existing research. The slip surface development trajectory is determined by analyzing the particle displacement field (such as...). Figure 8 The particle displacement field distribution (obtained from 500,000 time steps of evolution) is used to assess the local stability of the slope, while the settlement deformation (e.g., 0.2 m) reflects the overall compressibility of the slope. Optionally, GPU-accelerated computation (e.g., MatDEM) can significantly improve the simulation efficiency of large-scale particle systems.

[0096] In this embodiment, the deformation and failure prediction results of the construction waste slope are determined based on the simulation results of the discrete element heterogeneous slope model. The prediction results include the potential slip range, the maximum slip distance (e.g., 3.118m), and the slip surface morphology (e.g., ...). Figure 9 (Comparison with actual morphology). Optionally, the accuracy of the prediction results can be ensured by comparing the degree of fit between the simulated slip surface and the actual slip surface. Furthermore, this embodiment can be extended to the stability assessment of other loose deposits (such as mine spoil heaps or landfills), requiring only adjustment of the material RGB threshold library to adapt to different application scenarios. Specifically, the slip surface error can be further quantified using Hausdorff distance (required ≤0.15H, where H is the slope height); the settlement verification needs to be compared with InSAR monitoring data (error ≤5%), and when extended to mine spoil heaps, near-infrared band (700-1000nm) needs to be collected to distinguish the degree of oxidation of coal gangue.

[0097] In some preferred embodiments, aerial photography (using a DJI Phantom 4 RTK) and image recognition (PCAS system) were employed. By setting aerial photography parameters of 70m altitude, 80% overlap, and -60° tilt angle, 238 multi-view images of construction waste soil were acquired. After preprocessing (brightness and contrast adjustment, and cropping to A1-A6 regions), RGB threshold segmentation (RGB for ceramic tiles: 190-220 / 200-230 / 210-240) and binarization (threshold 172, tolerance 25) were used to extract material composition characteristic parameters (such as ceramic tile percentage 11.21%, probability entropy 0.975, shape factor 0.648, and fractal dimension 1.296). Combined with 3D point cloud data (generated via DJI Terra), a discrete element heterogeneous model was constructed to simulate the slip surface development trajectory (maximum slip 3.118m) and settlement (0.2m) after gravity loading. The final prediction results showed a significant reduction in error compared to the actual slip morphology.

[0098] This embodiment combines UAV aerial photography with image recognition to achieve rapid (completed within 60 minutes of a single aerial survey), safe (avoiding manual entry into high-risk areas), and high-precision (recognition error ≤10%) investigation of the material composition of construction spoil slopes. It is particularly effective in quantifying and analyzing the geometric distribution (length / width / perimeter) and structural features (fractal dimension, shape factor, probability entropy) of heterogeneous materials such as tile fragments. The constructed discrete element heterogeneous model accurately predicts slope deformation and failure characteristics, reducing the prediction deviation of slip distance compared to traditional homogeneous models. Optionally, this method can be extended to material composition identification and deformation and failure prediction in mine spoil heaps, landfills, etc., adapting to different scenarios by adjusting the RGB threshold library.

[0099] In some embodiments, preset aerial photography parameters are obtained through the following steps:

[0100] Obtain the original geometric features of the target slope, including slope height, slope length, and slope angle;

[0101] Determine the initial flight altitude range based on the original geometric morphological characteristics;

[0102] Based on image resolution parameters, an adjusted flight altitude is generated within the initial flight altitude range;

[0103] Obtain the slope topography complexity of the target slope, and set the parameters of the forward overlap rate and the lateral overlap rate accordingly. The shooting paths under the forward overlap rate parameter and the shooting paths under the lateral overlap rate parameter are orthogonal to each other.

[0104] Calculate the camera tilt angle to ensure that the matching error between the camera tilt angle and the slope normal direction meets the preset error range;

[0105] Based on the on-site environmental information, the parameters of flight altitude, forward overlap rate, lateral overlap rate, and camera tilt angle are jointly optimized to generate the optimized flight altitude, forward overlap rate, lateral overlap rate, and camera tilt angle, which are the preset aerial photography parameters.

[0106] In this embodiment, the preliminary acquisition of the geometric features of the target slope includes slope height, slope length, and slope angle. These parameters are the basic data for aerial survey planning. Slope height and slope length determine the range of the aerial survey area, while the slope angle directly affects the calculation of the camera tilt angle. For example, the slope height of the target slope is set to approximately 25m, and the slope angle is between 36.6° and 40.2°. These measured data provide an accurate basis for the calculation of aerial survey parameters. Optionally, for irregular slopes, the slope angle values ​​of different sections can be obtained by segmented measurement.

[0107] In this embodiment, when determining the initial flight altitude range based on the slope's geometric characteristics, both image resolution and slope safety distance requirements must be considered. Specifically, the flight altitude calculation formula is H = f·δ / a, where f is the camera focal length, δ is the ground resolution, and a is the camera sensor pixel size. This embodiment organically combines equipment parameters with terrain features to ensure image acquisition quality. Preferably, for slopes with a height of 25m, a safety buffer height can be set to avoid collisions between the UAV and the slope.

[0108] In this embodiment, when adjusting the flight altitude based on image resolution parameters, a balance must be struck between resolution requirements and flight efficiency. For example, the image sensor has at least 20 million effective pixels, which corresponds to specific ground resolution requirements. This embodiment establishes a dynamic adjustment mechanism for flight altitude and image quality, allowing for flexible setting of the flight altitude according to different survey accuracy requirements. Optionally, a lower flight altitude can be used in key survey areas to obtain higher resolution images.

[0109] In this embodiment, when setting the forward overlap rate and lateral overlap rate according to the complexity of the slope terrain, preferably, both the forward overlap rate and the lateral overlap rate are at least 70%. For areas with complex terrain, the overlap rate can be appropriately increased to 80% to ensure image matching accuracy. This embodiment establishes a correlation rule between the overlap rate and terrain features, and the design that the shooting paths for the forward overlap rate and the lateral overlap rate are orthogonal ensures the comprehensiveness of image acquisition.

[0110] In this embodiment, when calculating the camera tilt angle, the absolute value of the camera tilt angle should be complementary to the average slope angle of the construction waste slope. For a slope angle of 40.2°, the camera tilt angle should be set to -49.8° (considering the negative sign indicates the shooting direction) to minimize the angle between the camera optical axis and the slope normal direction. The preset error range is usually controlled within ±2° to ensure image quality.

[0111] In this embodiment, when jointly optimizing parameters based on the on-site environment, practical factors such as wind direction and electromagnetic interference conditions need to be comprehensively considered. Specifically, emergency pause conditions include: RTK lockout > 3s or battery voltage < 3.6V / cell; dynamic adjustment rules: when wind speed > 15m / s, flight altitude is increased by 20%; when illumination < 10000 lux, supplementary lighting mode is activated. Aerial photography operations require a clear understanding of the flight environment, including altitude and terrain conditions. This embodiment establishes a multi-parameter collaborative optimization mechanism to dynamically adjust the aerial photography plan based on on-site environmental conditions.

[0112] This embodiment fully presents the process of determining aerial photography parameters: the initial flight altitude is calculated based on the slope geometry (slope height 25m, slope angle 36.6°-40.2°) (using H=f·δ / a), the heading and lateral overlap rates are set (at least 70%), the camera tilt angle is determined (complementary to the slope angle), and finally the final aerial photography parameters are generated by optimizing according to the site environment (such as wind speed and electromagnetic conditions).

[0113] This embodiment ensures high-quality slope images are obtained through a systematic parameter calculation process; camera tilt angle settings based on slope angle matching effectively reduce image distortion; a dynamic optimization mechanism adapts to complex site environments; and an overlap rate of at least 70% guarantees image matching accuracy. This significantly improves the efficiency and accuracy of investigating the composition of construction waste slope materials, while reducing the risks associated with manual on-site operations.

[0114] In some embodiments, acquiring multi-view images of the target slope according to preset aerial photography parameters includes:

[0115] By using a drone equipped with a high-resolution camera, aerial surveying tasks are carried out according to the flight altitude, and flight control commands corresponding to the final flight altitude are established simultaneously during the aerial surveying tasks.

[0116] Based on the heading overlap rate and lateral overlap rate parameters, generate UAV three-dimensional spatial flight path distribution information that meets the preset overlap rate requirements;

[0117] Adjust the gimbal of the high-resolution camera according to the camera tilt angle so that the optical axis of the high-resolution camera matches the slope normal of the target slope.

[0118] Under the preset aerial survey parameters, confirm whether the flight safety of the UAV meets the preset safety threshold under real-time environmental information. If not, suspend the aerial survey mission and adjust the flight altitude.

[0119] If so, the high-resolution camera will be triggered to capture images according to the preset acquisition frequency to obtain multi-view images that meet the requirements of spatiotemporal synchronization.

[0120] In this embodiment, when performing aerial surveying tasks using a UAV equipped with a high-resolution camera, the final flight altitude is determined considering both slope geometry and image resolution requirements. The UAV generates control commands corresponding to its altitude in real time during flight, ensuring the accuracy of the flight trajectory. This embodiment utilizes a dynamic correlation mechanism between flight altitude and control commands, enabling the UAV to adapt to flight stability requirements under different altitude conditions. Preferably, the flight control system can combine RTK / PPK positioning to achieve centimeter-level positioning accuracy.

[0121] In this embodiment, the shooting paths with forward overlap and the shooting paths with lateral overlap together constitute the three-dimensional spatial flight path distribution information. By setting a preset overlap requirement (at least 70%), it is ensured that adjacent images have a sufficient number of matching points of the same name. This embodiment extends two-dimensional flight path planning to three-dimensional space, fully considering the three-dimensional characteristics of slope terrain. Optionally, for steep slope areas, the overlap rate can be appropriately increased to 80% to improve the image matching success rate under complex terrain.

[0122] In this embodiment, adjusting the gimbal angle of the high-resolution camera to match the optical axis direction with the slope normal direction is a crucial step. The camera tilt angle is set according to the principle of being complementary to the slope angle, effectively reducing image perspective distortion and obstructed areas. This embodiment achieves this through a real-time dynamic adjustment mechanism of the gimbal, ensuring optimal shooting angles are maintained at different waypoints. Preferably, the gimbal control system can combine inertial measurement unit (IMU) data to achieve sub-degree-level angle control accuracy.

[0123] In this embodiment, the determination of actual environmental conditions includes an overall assessment of lighting and wind speed, providing assurance for flight safety. Lighting conditions can be used to adjust camera shooting modes and exposure parameters to ensure consistent image quality; wind speed conditions are used to assess flight stability risks.

[0124] In this embodiment, the assessment of UAV flight safety takes into account both UAV performance and image quality requirements. If UAV flight safety cannot be guaranteed under real-time environmental conditions, the aerial photography mission can be paused and the flight altitude adjusted to ensure equipment and data security.

[0125] In this embodiment, the preset acquisition frequency balances image coverage and data redundancy. Precise time synchronization control ensures that the multi-view images meet spatiotemporal consistency requirements. This embodiment strictly binds the triggering of shooting to flight position information, ensuring that the image of each waypoint location carries precise spatial coordinate information. Preferably, the acquisition frequency can be automatically adjusted according to flight speed to guarantee that the preset overlap rate requirement is met.

[0126] In some specific application embodiments, drones equipped with high-resolution cameras (effective pixels not less than 20 million) and RTK / PPK positioning systems are selected to perform aerial surveying tasks according to pre-calculated flight altitude, forward overlap rate, and lateral overlap rate (all not less than 70%). During flight, the gimbal is precisely controlled to maintain a complementary relationship between the camera tilt angle and the slope angle (e.g., setting the tilt angle to -60° when the slope angle is 30°), ensuring that the angle between the camera optical axis and the slope normal is minimized. Simultaneously, environmental conditions such as wind speed and lighting conditions are pre-determined. If the drone's flight safety cannot be guaranteed under real-time environmental conditions, the aerial surveying task is paused and the flight altitude is adjusted. The final acquired multi-view images meet the spatiotemporal synchronization requirements, providing a data foundation for subsequent image recognition and modeling. The entire aerial surveying process strictly follows the preset acquisition frequency and flight path planning (such as a "tic-tac-toe" or "five-way" pattern) to ensure complete image coverage and compliance with quality requirements.

[0127] This embodiment expands the data dimensions of traditional field surveys and sampling by using high-resolution cameras, enabling the acquisition of richer material characteristic information; predictive on-site environmental conditions significantly improves the safety and reliability of field operations; an automatic adjustment mechanism ensures that image data meeting requirements can be obtained under various environmental conditions; and precise spatiotemporal synchronization control provides a high-quality data foundation for subsequent 3D modeling. This comprehensively improves the efficiency and accuracy of construction waste slope surveys while reducing operational risks.

[0128] In some embodiments, the multi-view images are filtered and preprocessed to obtain standard slope images, including:

[0129] Based on the original RGB channel values ​​of the multi-view image, an adaptive histogram equalization algorithm is used to adjust the brightness in the HSV color space to obtain a brightness-adjusted image.

[0130] The brightness-adjusted image is subjected to contrast enhancement processing using an illumination compensation algorithm to obtain a brightness-enhanced image, including:

[0131] Based on the three-dimensional point cloud computing slope normal field corresponding to the multi-view image, establish an incident light angle compensation model.

[0132] A brightness-enhanced image is obtained by non-linearly stretching the brightness component of the brightness-adjusted image in the Lab color space.

[0133] The slope boundary is identified based on the characteristics of the top and bottom lines of the target slope, and the corrected image is obtained by performing perspective distortion correction and geometric deformation compensation on the brightness-enhanced image in combination with UAV positioning data.

[0134] The corrected image is processed using an edge sharpening algorithm to generate a standard slope image that meets the requirements for material composition identification.

[0135] In this embodiment, an adaptive histogram equalization algorithm is used to adjust the brightness in the HSV color space based on the original RGB channel values ​​of the multi-view image. After converting the RGB image to the HSV space, histogram equalization is performed only on the luminance component (V channel) based on local regions. Optionally, an adaptive window size can be set to balance detail enhancement and noise suppression, ultimately resulting in a brightness-adjusted image with a more uniform brightness distribution.

[0136] In this embodiment, an incident light angle compensation model is established based on the slope normal vector field obtained from three-dimensional point cloud computing to eliminate illumination differences across different slope directions. Simultaneously, the luminance component is nonlinearly stretched in the Lab color space; optionally, an S-curve is used to adjust the dynamic range of luminance, achieving contrast optimization without altering chromaticity information and generating a luminance-enhanced image.

[0137] In this embodiment, a geometric correction model is constructed based on the identified features of the slope crest and toe lines, combined with UAV positioning data. The geometric correction model simultaneously compensates for perspective distortion and geometric deformation caused by terrain undulations. Optionally, polynomial transformation is used to achieve precise geometric correction of the image, resulting in a spatially accurate corrected image.

[0138] In this embodiment, the feature region image is processed using an edge sharpening algorithm to improve the clarity of material boundaries by enhancing high-frequency components. Optionally, the sharpening intensity is adaptively adjusted based on material texture features to generate a standard slope image that meets the requirements for fine identification of material components.

[0139] This embodiment achieves the transformation of multi-view images into standard analysis data through a complete image preprocessing workflow. Color space conversion and adaptive enhancement improve the visual effect of the images, geometric correction based on 3D information ensures spatial accuracy, intelligent segmentation and feature extraction focus on key analysis areas, and sharpening processing enhances detail recognition.

[0140] In some embodiments, RGB thresholding and binarization are performed on standard slope images to extract material component feature parameters, including:

[0141] Based on the RGB channel values ​​of standard slope images, an adaptive threshold segmentation algorithm is used to segment materials such as tile fragments, steel bars, and wood separately to obtain binarized images of each material.

[0142] Feature parameters are extracted from the binarized images of each material, including:

[0143] Calculate the content ratio of each material component based on the pixel area ratio of the material;

[0144] The dimensional characteristics of each material component are statistically analyzed based on the equivalent diameter algorithm, and the actual size distribution information is calculated.

[0145] By combining the orientation rose diagram with the anisotropy index to quantify the aspect ratio and principal direction angle of material particles, orientation index information is obtained.

[0146] The fractal dimension is calculated using the box counting method to quantify the edge complexity of material particles;

[0147] The spatial distribution characteristics of materials are analyzed based on the semi-variogram function to obtain the degree of heterogeneity.

[0148] Based on the material composition content ratio information, actual size distribution information, orientation index information, edge complexity and heterogeneity degree, material composition characteristic parameters are generated, and a set of material composition characteristic parameters is established.

[0149] In this embodiment, an adaptive threshold segmentation algorithm is used to segment materials such as ceramic tile fragments, reinforcing steel bars, and wood based on the RGB channel values ​​of standard slope images. The segmentation threshold range is dynamically adjusted according to the color characteristics of different materials. Specifically, a wider tolerance range is used for ceramic tile fragments to accommodate differences in surface glaze color, a narrower threshold range is used for reinforcing steel bars to suppress oxidation interference, and wood is segmented using hue components. Morphological processing is used to eliminate fine noise, achieving effective separation of multiple materials against complex backgrounds.

[0150] In this embodiment, the content ratio of each material component is calculated based on the pixel area ratio of the material. The size characteristics of each material component are statistically analyzed using an equivalent diameter algorithm to convert the actual size distribution information. The binarized regions of each material component are extracted sequentially through global threshold segmentation, and the total pixel area of ​​each material component is calculated to obtain the content ratio. At the same time, the connected pore regions are marked, and the pixel size is converted into a physical scale by calibrating the image resolution to generate a histogram of material particle size distribution. The equivalent diameter distribution of material particles is statistically analyzed, and the actual size distribution information of material particles can be calculated and statistically obtained.

[0151] In this embodiment, the aspect ratio and principal orientation angle of material particles are quantified by combining the directional rose diagram and the anisotropy index to obtain orientation index information. The dominant orientation of material particles is detected through parameter space voting to generate a directional rose diagram. The material particles are then skeletonized, and principal component analysis is performed on the skeleton pixel coordinates to calculate the major axis direction and anisotropy index. For linear materials such as reinforcing bars, an orientation consistency index can also be calculated to reflect the compaction or accumulation characteristics during construction.

[0152] In this embodiment, the box counting method is used to calculate the fractal dimension, which quantifies the edge complexity of material particles. A logN(ε)-log(1 / ε) double logarithmic coordinate system is established within the scale range of 4-256 pixels. The fractal dimension value is obtained by fitting the slope using the least squares method, thereby representing the edge complexity of material particles.

[0153] In this embodiment, the spatial distribution characteristics of materials are analyzed based on semi-variogram analysis to obtain the degree of heterogeneity. The variogram value γ(h) under different hysteresis distances (h) is calculated, and the range, sill value, and nugget effect are obtained by fitting a spherical model. The range reflects the autocorrelation range of the pore space, the sill value characterizes the degree of heterogeneity, and the nugget effect indicates microscopic randomness.

[0154] This embodiment achieves accurate extraction of various types of construction waste materials through adaptive threshold segmentation. Combined with multi-dimensional feature parameter calculation methods, it comprehensively characterizes the content ratio, geometric morphology, and structural properties of the materials. The content ratio information of the material components characterizes the composition information of the material, the size distribution and orientation index reflect the macroscopic distribution law of the material, and the edge complexity and heterogeneity capture the differences in microstructure.

[0155] In some embodiments, the 3D point cloud data obtained by reconstructing multi-view images includes:

[0156] An initial 3D point cloud is generated based on multi-view imagery using a multi-view stereo matching algorithm.

[0157] The initial 3D point cloud is sequentially processed by statistical outlier filtering, radius filtering, and feature-preserving filtering based on slope threshold to obtain a filtered 3D point cloud.

[0158] Based on the camera tilt angle and flight altitude, a projection geometry correction model is established to perform lens distortion correction, projection deformation compensation, and illumination non-uniformity correction on the filtered 3D point cloud to obtain the corrected 3D point cloud.

[0159] In this embodiment, an initial 3D point cloud is generated based on multi-view imagery using a multi-view stereo matching algorithm. By establishing dense correspondences between images, pixel-level disparity is calculated using polar geometric constraints and photometric consistency criteria. For the texture feature differences in the slope scene, the algorithm adaptively optimizes the matching strategy: in areas with weak texture, spatial continuity is enhanced through probability diffusion; in high-frequency feature areas, gradient constraints are combined to preserve geometric details. Optionally, a semi-global matching algorithm can be used to optimize the disparity calculation process, achieving preliminary reconstruction from 2D images to 3D point clouds.

[0160] In this embodiment, the initial 3D point cloud is sequentially processed by statistical outlier filtering, radius filtering, and feature-preserving filtering based on a slope threshold. Statistical outlier filtering removes isolated noise points, radius filtering eliminates local density anomalies, and slope threshold filtering preserves slope topographic features while removing interference points such as vegetation. Optionally, color information can be combined to assist classification filtering, resulting in a cleaner filtered 3D point cloud.

[0161] In this embodiment, a projection geometry correction model is established based on the camera tilt angle and flight altitude to perform lens distortion correction, projection deformation compensation, and illumination non-uniformity correction on the filtered 3D point cloud. This embodiment considers perspective distortion of aerial images and scale changes caused by terrain undulations. Optionally, a polynomial fitting method is used to optimize the correction parameters to obtain a corrected 3D point cloud with higher geometric accuracy.

[0162] This embodiment achieves complete reconstruction from multi-view images to high-quality 3D point cloud data through a systematic point cloud processing workflow. Adaptive stereo matching ensures the integrity of the initial reconstruction, multi-level filtering improves data quality, and geometric correction compensates for the inherent errors of the acquisition system. This embodiment is particularly suitable for 3D modeling of complex terrains such as construction spoil slopes, effectively handling the modeling difficulties of heterogeneous material distribution, and balancing the needs of automated processing and feature preservation.

[0163] In some embodiments, a discrete element heterogeneous slope model is constructed based on material composition characteristic parameters and combined with three-dimensional point cloud data reconstructed from multi-view images, including:

[0164] A numerical model of the slope is constructed using point cloud registration and surface reconstruction algorithms based on 3D point cloud data;

[0165] Based on the material composition characteristic parameters, the proportion of each material component in the slope model is determined and denoted as the material content proportion.

[0166] Based on the actual size distribution and orientation index in the material composition characteristic parameters, the particle morphology characteristics are obtained by using a multi-sphere aggregation algorithm or a block feature algorithm.

[0167] Based on the marginal complexity and heterogeneity of the material composition characteristic parameters, the structural distribution characteristics are obtained by box counting.

[0168] By coupling the material content ratio, particle morphology characteristics, and structural distribution characteristics into the slope numerical model, and by assigning discrete element contact parameters, a discrete element heterogeneous slope model with real material content ratio and distribution characteristics is established.

[0169] In this embodiment, a slope numerical model is constructed using point cloud registration and surface reconstruction algorithms on the 3D point cloud data obtained through multi-view image reconstruction. Optionally, radial basis functions can be used to interpolate and complete missing areas of the point cloud to ensure the integrity of the geometric framework.

[0170] In this embodiment, based on the actual size distribution and orientation index in the material composition characteristic parameters, a multi-sphere aggregation algorithm or a block feature algorithm is used to obtain the particle morphology characteristics. The particle gradation range is determined according to the size distribution parameters, and the orientation index is used to constrain the distribution of particles along their long axis. Optionally, a chain-like sphere combination method is used for linear materials such as reinforcing bars to more realistically reflect their geometric characteristics.

[0171] In this embodiment, the structural distribution characteristics are obtained through box counting based on the edge complexity and heterogeneity of the material composition characteristic parameters. The box counting method is used to calculate the fractal properties of the material edges, determine the internal structural complexity, and reflect the spatial variation law of the material's microstructure.

[0172] In this embodiment, the material content ratio, particle morphology characteristics, and structural distribution characteristics are coupled to the slope numerical model. A discrete element heterogeneous slope model with realistic material distribution characteristics is generated by assigning discrete element contact parameters. Appropriate friction coefficients and adhesion parameters are set according to the differences in material type. Optionally, gradual parameter settings are used for the interface transition region to simulate actual contact characteristics.

[0173] This embodiment achieves multi-scale modeling of the heterogeneous characteristics of slopes through multi-source data fusion. The geometric numerical model accurately recreates the macroscopic morphology of the slope, the material content ratio reflects the proportion of constituent materials, and the particle morphology and structural distribution characteristics characterize the microscopic properties of the materials at different scales. The discrete element model can more realistically simulate the mechanical behavior of mixed materials in construction waste slopes, especially showing better characterization of deformation and failure processes at heterogeneous interfaces. The model parameters are derived from actual detection data, avoiding the large errors caused by traditional homogenization assumptions, and providing a more reliable technical method for predicting slope deformation and failure.

[0174] In some embodiments, a gravity load is applied to a discrete element heterogeneous slope model to simulate and obtain slope deformation and failure prediction results. These prediction results include the slip surface development trajectory, particle displacement field distribution, and settlement deformation, including:

[0175] Initial stress equilibrium calculations were performed on the discrete element heterogeneous slope model, and the servo control method was used to bring the discrete element heterogeneous slope model to the initial static equilibrium state.

[0176] Gravity loads are applied gradually, and dynamic calculations are performed using the explicit time integration method. The formation and development trajectory of the slip surface are tracked in real time through contact force chain evolution analysis to obtain the development trajectory of the slip surface.

[0177] Simultaneously, the position change data of each particle unit during the calculation process are recorded, and the particle displacement field distribution is generated through vector field visualization technology, and the statistical characteristics of the displacement vector in each region are calculated.

[0178] Based on the vertical displacement data of surface monitoring points of the discrete element heterogeneous slope model, the settlement deformation surface is constructed using the Kriging interpolation method, and the settlement deformation amount and its spatial distribution characteristics in different regions are calculated.

[0179] By coupling analysis of the slip surface development trajectory, particle displacement field distribution, and settlement deformation, the prediction results of slope deformation and failure are obtained.

[0180] In this embodiment, initial stress equilibrium calculations are performed on the discrete element heterogeneous slope model, and a servo control method is used to bring the model to an initial static equilibrium state. By dynamically adjusting boundary constraints or particle displacement rates, the internal stress of the system gradually converges to the target stress field distribution. Optionally, a damping coefficient can be set to control the convergence speed, ensuring that an equilibrium state is achieved within a reasonable calculation time, thus eliminating potential stress anomalies during model initialization.

[0181] In this embodiment, an explicit time integration method is used for dynamic calculations during the gradual application of gravity load. By monitoring the evolution of the contact force chain in real time, key regions of force chain fracture and reorganization are identified, thereby tracing the formation and development trajectory of the slip surface. Optionally, combining particle rotation analysis can more accurately determine the shear deformation region.

[0182] In this embodiment, the positional changes of each particle unit during the calculation process are recorded, and the particle displacement field distribution is generated using vector field visualization technology. The statistical characteristics of the displacement vectors in each region are calculated, including parameters such as average displacement and displacement direction consistency. Optionally, a streamline visualization method can be used to more intuitively display the particle motion trend.

[0183] In this embodiment, a settlement deformation surface is constructed using Kriging interpolation based on the vertical displacement data of surface monitoring points. This embodiment considers spatial correlation and can reconstruct a continuous settlement distribution field from discrete monitoring point data. Optionally, the interpolation parameters can be optimized by combining variogram analysis to obtain the settlement deformation and its spatial distribution characteristics in different regions.

[0184] In this embodiment, a complete prediction result for slope deformation and failure is obtained by comprehensively analyzing data from three aspects: the development trajectory of the slip surface, the distribution of particle displacement field, and the amount of settlement deformation. The correspondence between the slip surface location and the abrupt change region of the displacement field is analyzed, and the correlation between settlement distribution and internal deformation mode is evaluated. Optionally, spatiotemporal evolution curves of deformation parameters can be established to describe the full-process characteristics of slope deformation.

[0185] This embodiment achieves a multi-dimensional characterization of slope deformation and failure prediction through systematic numerical simulation and feature extraction methods. Initial stress balance calculation ensures the rationality of the simulation, and the dynamic load application process realistically reflects the slope response under gravity. Slip surface tracking reveals the potential failure mechanism, particle displacement field analysis reveals internal deformation characteristics, and settlement deformation calculation quantifies the degree of surface deformation. The coupled analysis of these three methods provides a comprehensive basis for slope failure prediction.

[0186] The above embodiments provide a method for rapid investigation of material composition and deformation / failure prediction of construction waste slopes, combining UAV aerial photography, image recognition, and heterogeneous discrete element simulation. This method has significant application and technical support value in engineering practice. Compared to existing technologies, its advantages are:

[0187] (1) In terms of the investigation of material composition of construction waste slopes, this technical solution organically combines UAV aerial photography and image recognition technology to achieve safe, efficient and accurate analysis of slope material composition. Compared with traditional manual on-site investigation and limited sampling methods, this new method is safe, intelligent, low-cost, high-precision and accurate. It can achieve comprehensive quantitative analysis of material information, including material content ratio, geometric size distribution (length, width, perimeter, area, etc.), and structural characteristic parameters such as probability entropy, shape factor, and fractal dimension. The identification error is controlled within 10%, which can provide rich data for the prediction of deformation and failure of construction waste slopes.

[0188] (2) Regarding the prediction of deformation and failure of construction waste slopes, this technical solution combines UAV aerial surveying and image recognition results to construct a discrete element heterogeneous model of the slope that conforms to the actual material composition and slope geometry, which can accurately simulate and predict the deformation and failure process of construction waste slopes. Compared with existing studies that mostly analyze or predict slope deformation and failure characteristics based on homogeneous models, the research results obtained will have certain deviations from the actual situation. This technical solution comprehensively considers the influence of material composition and its heterogeneity on the deformation and failure of construction waste slopes, providing a more reliable basis for the design and construction of slope engineering, and obtaining the extremely irregular deformation and failure characteristics of slopes containing ceramic tile construction waste. In addition, the simulation and comparison analysis of the actual morphology of the slope slip surface in the embodiment verifies the accuracy and reliability of this technical solution.

[0189] (3) In terms of engineering safety and economic efficiency, this technical solution makes full use of UAV aerial surveying technology, which significantly reduces the risk of personnel exposure and the cost of investigation and sampling in high-risk areas. Construction spoil slopes are loose and prone to collapse and landslides. The traditional determination of the material composition of construction spoil sites is mainly based on on-site investigation and indoor geotechnical tests by sampling at limited points. This method requires a large number of personnel, has high time and economic costs, is greatly affected by factors such as terrain, and poses a high safety risk to workers. This technical solution significantly improves the efficiency of material composition identification, is more suitable for the characteristics of the spatiotemporal distribution of material composition and the dynamic updating of terrain conditions in construction spoil sites, and can better meet the needs of dynamic analysis of material composition in the entire spoil site. Moreover, the number of UAV aerial survey flights and routes can be dynamically adjusted according to the planning and construction of the construction spoil site to obtain rich spatiotemporal four-dimensional information on the material stockpiling of the spoil site, providing solid basic data for operation and maintenance management personnel and researchers.

[0190] (4) The technical advantages and beneficial effects of this technical solution are also reflected in its wide applicability. This technical solution is not only applicable to construction spoil slopes, but can also be extended to material composition investigation and deformation and failure prediction of various multi-source heterogeneous loose accumulations such as mine spoil heaps and landfills. From parameter optimization in the design stage to dynamic monitoring in the filling construction stage, and early warning management in the operation and maintenance period, this technical solution can provide comprehensive technical support, realize accurate control of the entire life cycle of artificial fillings such as construction spoil slopes, and provide application technical support for digital operation and maintenance management and disaster prevention and mitigation of complex sites such as construction spoil heaps.

[0191] The aforementioned technical solution acquires multi-view images of the target slope through UAV aerial photography and preprocesses them. It then extracts material composition characteristic parameters, including material component content ratios, actual size distribution, orientation indices, edge complexity, and heterogeneity. These parameters are combined with 3D point cloud data to construct a discrete element heterogeneous slope model reflecting the actual material distribution. Gravity load simulation yields slope deformation and failure prediction results, such as the slip surface development trajectory, particle displacement field distribution, and settlement deformation. This technical solution achieves safe, efficient, and accurate material composition analysis through UAV aerial photography and image recognition technology. It can quantify and analyze comprehensive information such as material content, geometric dimensions, and structural characteristics. The heterogeneous model constructed based on actual material composition information can accurately simulate the slope deformation and failure process, overcoming the bias problems of traditional homogeneous models and providing a reliable basis for engineering design. It significantly reduces the risk of personnel exposure and investigation costs in high-risk areas, dynamically adapts to the spatiotemporal distribution characteristics of material composition, and meets the needs of dynamic analysis across the entire region. It has wide applicability and can be extended to various loose deposits.

[0192] Finally, it should be noted that although the above embodiments have been described in the text and drawings of this application, this should not limit the scope of patent protection of this application. Any technical solutions that are based on the essential concept of this application and utilize the content described in the text and drawings of this application, resulting in equivalent structural or procedural substitutions or modifications, as well as the direct or indirect application of the technical solutions of the above embodiments to other related technical fields, are all included within the scope of patent protection of this application.

Claims

1. A method for rapid investigation of material composition and prediction of deformation and failure of construction waste slopes, characterized in that, include: Multi-view images of the target slope are acquired according to preset aerial photography parameters, including flight altitude, heading overlap rate, lateral overlap rate, and camera tilt angle. The multi-view images are filtered and preprocessed to obtain standard slope images. The preprocessing includes brightness adjustment, contrast adjustment and image partitioning and cropping. The standard slope images contain global information of the slope area to be analyzed and clearly identifiable material boundaries. The standard slope image is subjected to RGB threshold segmentation and binarization to extract material component feature parameters, which include the proportion of each material component, actual size distribution information, orientation index information, edge complexity and degree of heterogeneity. Based on the material composition characteristic parameters, and combined with the three-dimensional point cloud data reconstructed from the multi-view images, a discrete element heterogeneous slope model is constructed. The discrete element heterogeneous slope model should reflect the basic characteristic parameters of the actual material information and the actual slope geometric morphology characteristic parameters. The basic characteristic parameters include the material content ratio, particle morphology characteristics, and structural distribution characteristics. The actual slope geometric morphology characteristic parameters include slope height and slope angle parameters. Gravity load is applied to the discrete element heterogeneous slope model to simulate and obtain slope deformation and failure prediction results. The slope deformation and failure prediction results include the slip surface development trajectory, particle displacement field distribution and settlement deformation.

2. The method for rapid investigation of material composition and prediction of deformation and failure of construction waste slopes according to claim 1, characterized in that, The preset aerial photography parameters are obtained through the following steps: Obtain the original geometric features of the target slope, including slope height, slope length, and slope angle; The initial flight altitude range is determined based on the original geometric features. Based on image resolution parameters, an adjusted flight altitude is generated within the initial flight altitude range; The slope topography complexity of the target slope is obtained, and parameters for the forward overlap rate and the lateral overlap rate are set accordingly. The shooting paths under the forward overlap rate parameter and the shooting paths under the lateral overlap rate parameter are orthogonal to each other. Calculate the camera tilt angle so that the matching error between the camera tilt angle and the slope normal direction meets the preset error range; Based on the on-site environmental information, the parameters for adjusting flight altitude, forward overlap rate, lateral overlap rate, and camera tilt angle are jointly optimized to generate the optimized flight altitude, forward overlap rate, lateral overlap rate, and camera tilt angle, which are the preset aerial photography parameters.

3. The method for rapid investigation of material composition and prediction of deformation and failure of construction waste slopes according to claim 2, characterized in that, Collect multi-view images of the target slope according to preset aerial photography parameters, including: The drone is equipped with a high-resolution camera to perform aerial surveying tasks at the stated flight altitude, and flight control commands corresponding to the final flight altitude are established simultaneously during the execution of the aerial surveying tasks. Based on the heading overlap rate and lateral overlap rate parameters, generate UAV three-dimensional spatial flight path distribution information that meets the preset overlap rate requirements; Based on the camera tilt angle, adjust the gimbal of the high-resolution camera so that the optical axis direction of the high-resolution camera matches the slope normal direction of the target slope. Under the preset aerial survey parameters, confirm whether the flight safety of the UAV meets the preset safety threshold under real-time environmental information. If not, suspend the aerial survey mission and adjust the flight altitude. If so, the high-resolution camera will be triggered to capture images according to the preset acquisition frequency to obtain multi-view images that meet the requirements of spatiotemporal synchronization.

4. The method for rapid investigation of material composition and prediction of deformation and failure of construction waste slopes according to claim 1, characterized in that, The multi-view images are filtered and preprocessed to obtain standard slope images, including: Based on the original RGB channel values ​​of the multi-view image, an adaptive histogram equalization algorithm is used to adjust the brightness in the HSV color space to obtain a brightness-adjusted image. The brightness-adjusted image is subjected to contrast enhancement processing using an illumination compensation algorithm to obtain a brightness-enhanced image, including: Based on the three-dimensional point cloud computing slope normal vector field corresponding to the multi-view image, an incident light angle compensation model is established. A brightness-enhanced image is obtained by non-linearly stretching the brightness component of the brightness-adjusted image in the Lab color space. The slope boundary is identified based on the characteristics of the top and bottom lines of the target slope, and the corrected image is obtained by performing perspective distortion correction and geometric deformation compensation on the brightness-enhanced image in combination with UAV positioning data. The corrected image is processed using an edge sharpening algorithm to generate the standard slope image that meets the requirements for material composition identification.

5. The method for rapid investigation of material composition and prediction of deformation and failure of construction waste slopes according to claim 1, characterized in that, The standard slope image is subjected to RGB threshold segmentation and binarization to extract material component feature parameters, including: Based on the RGB channel values ​​of the standard slope image, an adaptive threshold segmentation algorithm is used to segment materials such as tile fragments, steel bars, and wood to obtain binarized images of each material. Feature parameters are extracted from the binarized images of each material, including: Calculate the content ratio of each material component based on the pixel area ratio of the material; The dimensional characteristics of each material component are statistically analyzed based on the equivalent diameter algorithm, and the actual size distribution information is calculated. By combining the orientation rose diagram with the anisotropy index to quantify the aspect ratio and principal direction angle of material particles, orientation index information is obtained. The fractal dimension is calculated using the box counting method to quantify the edge complexity of material particles; The spatial distribution characteristics of materials are analyzed based on the semi-variogram function to obtain the degree of heterogeneity. The material composition feature parameters are generated based on the material composition content ratio information, actual size distribution information, orientation index information, edge complexity, and degree of heterogeneity, and a material composition feature parameter set is established.

6. The method for rapid investigation of material composition and prediction of deformation and failure of construction waste slopes according to claim 1, characterized in that, The 3D point cloud data obtained by reconstructing the multi-view images includes: An initial 3D point cloud is generated based on the multi-view imagery using a multi-view stereo matching algorithm. The initial 3D point cloud is sequentially subjected to statistical outlier filtering, radius filtering, and feature-preserving filtering based on slope threshold to obtain a filtered 3D point cloud. Based on the camera tilt angle and flight altitude, a projection geometry correction model is established to perform lens distortion correction, projection deformation compensation, and illumination non-uniformity correction on the filtered 3D point cloud to obtain the corrected 3D point cloud.

7. The method for rapid investigation of material composition and prediction of deformation and failure of construction waste slopes according to claim 1, characterized in that, Based on the material composition characteristic parameters and combined with the three-dimensional point cloud data reconstructed from the multi-view images, a discrete element heterogeneous slope model is constructed, including: A numerical model of the slope is constructed using point cloud registration and surface reconstruction algorithms on the aforementioned three-dimensional point cloud data; Based on the material composition characteristic parameters, the proportion of each material component in the slope model is determined and denoted as the material content proportion. Based on the actual size distribution and orientation index in the material composition characteristic parameters, the particle morphology characteristics are obtained using a multi-sphere aggregation algorithm or a block feature algorithm. Based on the edge complexity and heterogeneity of the material composition characteristic parameters, the structural distribution characteristics are obtained by box counting. The material content ratio, particle morphology characteristics, and structural distribution characteristics are coupled to the slope numerical model. By assigning discrete element contact parameters, a discrete element heterogeneous slope model with real material content ratio and distribution characteristics is established.

8. The method for rapid investigation of material composition and prediction of deformation and failure of construction waste slopes according to claim 1, characterized in that, Gravity load is applied to the discrete element heterogeneous slope model to simulate and obtain slope deformation and failure prediction results. These prediction results include the slip surface development trajectory, particle displacement field distribution, and settlement deformation, including: The discrete element heterogeneous slope model is subjected to initial stress equilibrium calculation, and the discrete element heterogeneous slope model is brought to the initial static equilibrium state by servo control method. Gravity loads are applied gradually, and dynamic calculations are performed using the explicit time integration method. The formation and development trajectory of the slip surface are tracked in real time through contact force chain evolution analysis to obtain the development trajectory of the slip surface. Simultaneously, the position change data of each particle unit during the calculation process are recorded, and the particle displacement field distribution is generated through vector field visualization technology, and the statistical characteristics of the displacement vector in each region are calculated. Based on the vertical displacement data of surface monitoring points of the discrete element heterogeneous slope model, the settlement deformation surface is constructed using the Kriging interpolation method, and the settlement deformation amount and its spatial distribution characteristics in different regions are calculated. By coupling analysis of the slip surface development trajectory, particle displacement field distribution, and settlement deformation, the slope deformation and failure prediction results are obtained.

Images