Reservoir dam internal risk identification method and system
By combining muon detection technology with UAV lidar, the problem of complex iterative algorithms in the internal detection of reservoir dams has been solved, enabling efficient and accurate identification and targeted positioning of internal risks of dams, and providing clear engineering disposal targets.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-09
- Publication Date
- 2026-04-03
AI Technical Summary
Existing technologies for detecting the interior of reservoir dams suffer from problems such as complex iterative algorithms, large computational load, and high implementation difficulty. Furthermore, traditional detection methods cannot fully reflect the overall health status of the dam's internal structure and suffer from a disconnect between surface and internal information.
By employing muon detection technology combined with UAV lidar, the three-dimensional space of the dam is discretized into a regular voxel grid. Muon ray data is acquired using a muon detector. Combined with inversion algorithms and manual measurements, a three-dimensional density distribution map of the dam is drawn, and relatively low-density anomaly areas are identified.
It enables efficient and accurate identification of internal risks of the dam, shortens the detection cycle, lowers the technical threshold, provides clear target information, and provides a basis for subsequent engineering treatment.
Smart Images
Figure CN121786533A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to a method and system for identifying internal risks of a reservoir dam, belonging to the field of dam safety monitoring technology. Background Technology
[0002] As a crucial national water conservancy infrastructure, the structural safety of reservoir dams directly impacts the safety of life and property downstream and the stability of the regional economy and society. However, during long-term operation, dams are affected by various factors such as internal seepage, material aging, and geological changes, which can lead to hidden defects in their internal structure, such as cracks, voids, and loose areas. These internal risks are highly concealed and sudden, making them difficult to detect and warn of effectively using traditional manual inspections and surface monitoring methods. They represent a "blind spot" and a major source of risk in current dam safety management.
[0003] Currently, dam safety inspection technology still faces many severe challenges in practical applications. First, regarding internal detection capabilities, while traditional drilling methods can obtain some internal information, they are essentially point- or line-based detection methods with limited coverage, making it difficult to comprehensively reflect the overall health status of the dam's interior. Furthermore, this destructive testing method may damage the dam structure, and the high cost and long cycle of testing significantly reduce efficiency and economic viability.
[0004] Secondly, existing detection technologies often suffer from a disconnect between surface and internal information. For example, high-precision measurement technologies such as ground-based lidar and photogrammetry can accurately obtain information about the deformation of a dam's surface, but they cannot directly "see through" the damage to the internal structure. Consequently, it is difficult to establish a causal relationship between surface deformation and internal damage, leading to poor results in the location and assessment of potential hazards.
[0005] In addition, the terrain around reservoir dams is usually complex and diverse, with rugged terrain and steep slopes making it difficult to deploy traditional ground detection equipment and conduct comprehensive and efficient surveys, further increasing the difficulty and risks of detection.
[0006] Therefore, there is an urgent need to develop a new testing method that can achieve comprehensive, non-destructive testing of the internal structure of reservoir dams, while also being efficient and low-cost, in order to effectively compensate for the shortcomings of traditional technologies and improve the level of dam safety management and risk prevention capabilities.
[0007] Existing technologies have led to the development of schemes that use muon detection technology to identify and assess the safety risks of dams. Muon detection technology is a cutting-edge technology that uses muons in cosmic rays as natural probes to perform non-destructive imaging and monitoring of large, dense objects.
[0008] For dam risk identification, existing technologies involve iteratively calculating the density values of the muon attenuation flux using complex algorithms, repeatedly analyzing the values to discretize the dam into voxels. Then, a new 3D model of the dam's interior (containing density information at different locations) is generated by reassigning these values onto the dam's 3D model. While this iterative approach yields relatively accurate data, the complexity of the iterative algorithms and the sheer computational burden make implementation difficult. Summary of the Invention
[0009] The purpose of this invention is to provide a method and system for identifying risks inside reservoir dams, in order to solve the problem that the existing iterative algorithms are relatively complex and computationally intensive, resulting in high difficulty in final implementation.
[0010] To achieve the above objectives, the present invention includes: The present invention provides a method for identifying internal risks of a reservoir dam, comprising the following steps: 1) Obtain the three-dimensional outline of the dam and discretize its internal space into a regular voxel mesh; 2) Use a muon detector covering the entire dam to detect muon rays penetrating the dam and obtain muon observation data. The muon observation data includes the path length of each muon ray in each voxel and the corresponding muon flux attenuation for each muon ray. 3) Using voxel density as the unknown parameter and muon observation data as the target for inversion; treating the dam body as homogeneous, measuring the actual density of a region, and calculating the actual density of other regions based on the voxel density of that region and the voxel density of other regions. 4) Relatively low-density anomaly zones are considered as risk areas where the dam may pose a risk.
[0011] Furthermore, in step 1), a drone equipped with a lidar is used to perform a flight scan of the dam to obtain three-dimensional point cloud data of the dam; a three-dimensional surface mesh model of the dam is generated based on the three-dimensional point cloud data of the dam, thereby obtaining the three-dimensional outline of the dam.
[0012] Furthermore, in step 2), the muon detectors are deployed at the toe of the dam to detect the dam, the muon detectors are deployed more densely in areas suspected of being at risk, and the muon detectors are deployed evenly in other areas.
[0013] Furthermore, in step 4), a three-dimensional density distribution map of the dam is drawn based on the actual density of each area of the dam, and anomalies with relatively low density are found based on the three-dimensional density distribution map.
[0014] Furthermore, in step 3), the actual density of a region is obtained by manual measurement.
[0015] The present invention provides a risk identification system for the interior of a reservoir dam, comprising a data processing server, wherein the data processing server is used to execute instructions to achieve the following steps; 1) Obtain the three-dimensional outline of the dam and discretize its internal space into a regular voxel mesh; 2) Use a muon detector covering the entire dam to detect muon rays penetrating the dam and obtain muon observation data. The muon observation data includes the path length of each muon ray in each voxel and the corresponding muon flux attenuation for each muon ray. 3) Using voxel density as an unknown parameter and muon observation data as the target, inversion is performed; the dam body is considered homogeneous, the actual density of a region is obtained, and the actual density of other regions is calculated based on the voxel density of this region and the voxel density of other regions. 4) Relatively low-density anomaly zones are considered as risk areas where the dam may pose a risk.
[0016] Furthermore, it also includes a drone equipped with a lidar; in step 1), the drone performs a flight scan of the dam to obtain three-dimensional point cloud data of the dam; a three-dimensional surface mesh model of the dam is generated based on the three-dimensional point cloud data of the dam, thereby obtaining the three-dimensional outline of the dam.
[0017] Furthermore, the muon detectors are deployed at the toe of the dam to detect the dam, with the muon detectors being deployed more densely in areas suspected of being at risk, and the muon detectors being deployed evenly in other areas.
[0018] Furthermore, in step 4), a three-dimensional density distribution map of the dam is drawn based on the actual density of each area of the dam, and anomalies with relatively low density are found based on the three-dimensional density distribution map.
[0019] Furthermore, in step 3), the actual density of a region is obtained by manual measurement.
[0020] The beneficial effects of this invention are as follows: 1. Improved accuracy: The centimeter-level precision 3D model provided by the lidar fundamentally eliminates geometric model errors, making the calculation of muon path length extremely accurate and greatly improving the accuracy and efficiency of the calculation.
[0021] 2. By adopting a hierarchical intelligent detection strategy of "rapid general survey of the whole area + detailed diagnosis of key areas" (muon detectors are evenly distributed throughout the whole area and densely distributed in key areas), as well as the synchronous acquisition technology of detector array, the detection cycle that traditional methods require several months is shortened to several weeks or even several days, which greatly shortens the detection cycle.
[0022] 3. While meeting the requirements for risk identification, the combination of manual sampling and other methods simplifies the computational complexity and improves the overall computational efficiency.
[0023] 4. The final result is a three-dimensional "digital twin" that perfectly corresponds to the real dam, complete with internal density rendering, rather than an abstract numerical value or two-dimensional image. Engineering technicians can intuitively understand it without a professional geophysical background, greatly reducing the technical threshold and the difficulty of interpretation.
[0024] 5. Combined with manual detection, it can directly output the three-dimensional coordinates, approximate volume, and risk level of suspected abnormal areas, providing a clear target for subsequent engineering treatment. Attached Figure Description
[0025] Figure 1 This is a block diagram of the internal risk identification system for reservoir dams according to the present invention; Figure 2 This is a schematic diagram of a 3D point cloud model of a dam obtained by a drone's lidar scanning. Figure 3(a) is a schematic diagram of the layout of the Muon detector at the dam foot; Figure 3(b) is a schematic diagram of muon rays detected by the muon detector penetrating the dam. Figure 4 This is a data processing flowchart for determining whether a muon event is a valid muon event; Figure 5 This is a flowchart of the reservoir dam internal risk identification method of the present invention. Detailed Implementation
[0026] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be described in detail below with reference to the accompanying drawings and embodiments.
[0027] Method 1 for identifying internal risks of reservoir dams: The method for identifying internal risks of reservoir dams according to the present invention includes the following steps: 1. Data collection steps.
[0028] Using a lidar sensor mounted on a drone, a multi-view, high-overlap aerial scan of the dam body and surrounding slopes is performed to acquire high-precision 3D point cloud data of the dam. Alternatively, other methods can be used to collect 3D point cloud data or elevation data of the dam.
[0029] A muon detector array was deployed at the downstream toe of the dam to cover the dam for long-term continuous observation and to record muon flux data from different incident directions.
[0030] 2. Data processing and fusion modeling steps.
[0031] The lidar point cloud data is processed, including point cloud denoising, registration, filtering and classification, to generate a high-precision three-dimensional surface mesh model or a high-precision digital elevation model of the dam.
[0032] The 3D contour of the dam, obtained through a 3D surface mesh model or a digital elevation model, is used as a spatial constraint, and its internal space is discretized into a regular voxel (3D pixel) mesh.
[0033] Background subtraction and directional binning were performed on the muon flux data to obtain muon attenuation in different observation directions.
[0034] Since the signals recorded by the detector do not all originate from cosmic ray muons penetrating the dam, but also include some electrons, gamma rays, and environmental background noise, failing to subtract this background would severely underestimate the attenuation of muons (because the background signal artificially inflates the count value), leading to an underestimation of the density value and potentially masking the true low-density anomaly. Therefore, by triggering multiple scintillator units simultaneously (or in a predetermined order) within a pre-defined, extremely short time window, each triggered as a valid time candidate, and then performing a "geometric coincidence" judgment based on the muon incident direction and the dam's orientation, a muon event passing the "geometric judgment" is considered a valid muon event, thus completing the data background subtraction.
[0035] Orientation binning involves grouping all muons received by the detector according to their incident direction. Without orientation binning, mixing all fluxes together is equivalent to overlaying all projected images from a 3D scan into a blurry 2D image, losing most of the spatial information. Orientation binning divides the azimuth and zenith angles into several regions, assigning each detected muon event to its corresponding orientation bin based on its reconstructed zenith and azimuth angle values for 3D imaging.
[0036] Local subtraction and directional binning are mature existing technologies in the field of muon imaging, and will not be elaborated here.
[0037] 3. Dam Risk Detection Methods Using the geometric boundaries defined by the three-dimensional surface mesh model as fixed constraints, a forward projection matrix for muon tomography is constructed. Each element of this matrix represents the path length of a specific muon ray traveling through a voxel.
[0038] An inversion equation is established with voxel density as the unknown parameter and muon observation data as the target. Since muon attenuation is proportional to the average density along the observed path, assuming that the material of the entire dam body is homogeneous, we can calculate the density of other areas by manually measuring the density value of one area. This allows us to draw a three-dimensional density distribution map of the entire dam. We can infer the potential risk areas of the dam through relatively low-density anomaly areas. Combined with manual sampling and detection, we can locate and identify the risks inside the dam.
[0039] Method 2 for identifying internal risks of reservoir dams: Based on Implementation Method 1 of the Risk Identification Method for the Interior of Reservoir Dams, more specifically, the internal space of the three-dimensional contour of the dam is discretized into a voxel grid. For each muon observation direction, the path length of each muon ray passing through each voxel is calculated to form a projection matrix A. Each element of the projection matrix A represents the path length of a specific muon ray traveling through a certain voxel.
[0040] During measurements at the dam toe, the muon flux attenuation for each muon ray was calculated, μ. i If the muon flux decays along the i-th path, then μ i =-ln(N i / N0 i ), where N i This is a measured count, N0 i It is the expected background count without dam obstruction, μ i It is proportional to the average density of the material along the path.
[0041] The existing approach involves iteratively calculating the muon decay flux attenuation value using a complex algorithm, repeatedly iterating to determine the density value of the dam as discrete voxels. Then, the density information (containing density information at different locations) is reassigned onto the dam's 3D model. While this iterative method yields relatively accurate data, the complexity of the iterative algorithm and the large computational load make its final implementation quite challenging.
[0042] Since the flux decay of muons is proportional to the density along their path, assuming the dam being probed is homogeneous, an inversion equation is established using voxel density as the unknown parameter and muon observation data as the target. Further, by manually measuring the density value of one region, the density values of other regions can be inferred, allowing for the creation of a relative three-dimensional density distribution map of the entire dam. However, when calculating density, the material of the earth-rock dam needs to be limited to a reasonable range (e.g., 1.8 g / cm³ to 2.4 g / cm³) to ensure relatively accurate results. Subsequently, the relatively low-density anomalies in the dam's three-dimensional density distribution map can be used to infer potential risk areas within the dam. Combined with manual sampling and detection, this allows for the location and identification of risks within the dam. This provides clear targets for subsequent dam operation, maintenance, and engineering interventions.
[0043] Method 3 for identifying internal risks of reservoir dams: As a specific implementation method, the present invention provides a method for identifying internal risks of a reservoir dam, such as... Figure 5 As shown, it includes the following: The process involves using a drone to scan the dam with a lidar sensor to obtain 3D point cloud data; performing necessary processing on the point cloud data and using it to model the dam to obtain a 3D model; and then performing voxelization on the 3D model to obtain a voxelized 3D mesh.
[0044] Historical seepage data of the dam is analyzed to identify suspected hazardous areas. Muon detectors covering the entire dam are evenly deployed at the dam toe for comprehensive detection, while additional detectors are deployed more densely in suspected hazardous areas for focused monitoring. Data collected by the muon detectors is preprocessed and transmitted to the main control unit, which then performs data fitting.
[0045] Using the outline of the dam's 3D model as a constraint, calculate the path length of each muon ray within each voxel along the dam's 3D outline.
[0046] Calculate the directional flux decay of each muon ray, which is caused by muons passing through the dam.
[0047] Based on the reasonable density range set for the dam material as the density boundary condition, an inversion equation is established with voxel density as the unknown parameter and muon observation data as the target. The average density of a certain area is manually calculated, and the density of other areas is calculated based on the voxel density inverted from the voxels in that area.
[0048] A three-dimensional density distribution map of the dam's interior is fitted. Then, combined with manual detection, the relatively low-density areas in the three-dimensional density distribution map are verified to identify risk zones within the dam.
[0049] Implementation method of reservoir dam internal risk identification system: The reservoir dam internal risk identification system of this invention uses a dam model scanned by a drone equipped with a lidar as the data base, and employs muon non-destructive testing technology to invert the internal structure of the dam, thereby accurately identifying the risk areas inside the dam. The reservoir dam internal risk identification system includes a drone, lidar, and a front-end muon detector (including...). Figure 1 The front-end muon detector nodes 1 to n and the muon detector main control unit (i.e. Figure 1 The main control unit and data processing server in the system are structured as follows: Figure 1 As shown.
[0050] The drone is equipped with a lidar system and communicates with a data processing server to transmit lidar scan data. This allows for the rapid and efficient acquisition of high-precision 3D point cloud data of the dam and its surrounding terrain, such as... Figure 2 As shown, this provides baseline data for constructing the geometric profile of the dam.
[0051] Here, the front-end muon detectors and the muon detection main control unit are both categorized as the muon detection system. This system includes several front-end muon detector nodes evenly covering the dam being detected. The front-end muon detectors are responsible for the initial detection of muon events, generating digital signals, preprocessing data, and adding timestamps. The muon detection main control unit communicates with the data processing server and is responsible for collecting the data collected by the front-end muon detector nodes, determining whether a data event is valid based on predefined triggering logic, and packaging and uploading the data to the data processing server.
[0052] The data processing server is responsible for denoising, registering, and classifying the point cloud data from the lidar to generate a high-precision 3D mesh model or digital elevation model, thereby obtaining the 3D outline of the dam. At the same time, it stores and analyzes the data obtained from the main control unit of the muon detector, reconstructs the muon tracks, and then inverts the 3D density distribution inside the dam (3D outline).
[0053] The following section further explains how equipping drones with lidar improves the overall system accuracy.
[0054] Traditional methods often employ extremely simplified models, such as assuming the entire earth-rock dam is a regular cuboid or trapezoidal prism, which introduces significant errors. The role of a drone equipped with lidar is to directly and accurately measure the true path length of each muon within the dam body with centimeter-level precision. Especially for most earth-rock dams, the dam shape may be curved. The 3D model generated by the drone equipped with lidar fully captures all details of the dam body, including curvature, slope changes, walkways, and abutments. For the same muon trajectory, the system uses this model to perform ray projection calculations, precisely determining its entry and exit points on the dam body model.
[0055] Furthermore, errors in the position and orientation of the muon detector itself will directly lead to deviations in the recorded muon incident direction angle, resulting in errors in the calculation path. The UAV equipped with a lidar is itself a high-precision measurement tool. After generating the model, the optimal placement point for each detector can be directly planned on the model, providing the precise three-dimensional coordinates (X, Y, Z) of that point.
[0056] Furthermore, the inversion calculation of dam density requires dividing the interior of the dam body into a large number of voxels. If the voxel grid does not match the actual shape of the dam body, each voxel will contain both "dam body" and "air", resulting in excessive calculation errors and meaningless results. The 3D point cloud generated by the UAV equipped with LiDAR can discretize the dam into a large number of voxel grids, which strictly limits the calculation to the interior space of the dam body, avoids invalid calculations, and improves the efficiency and accuracy of the calculation.
[0057] This section provides further explanation of the muon detection system and the layout of the front-end muon detector nodes during detection.
[0058] In this embodiment, the front-end muon detector nodes use orthogonally arranged plastic scintillators as detection units. Historical risk data of the dam is collected before the entire detection process begins, and areas with suspected risks are targeted for focused detection. As shown in Figures 3(a) and 3(b), a detector array is set up at the dam toe. Detectors are placed at intervals of, for example, 30 meters for ordinary areas, and at intervals of, for example, 15 meters for areas with suspected risks. All front-end detectors are connected to the main control unit. Due to the long dam structure, optical fiber is used for transmission between the front-end muon detectors and the main control unit to improve transmission stability. Before starting detection of the entire dam at the dam toe, the detectors are placed horizontally in an unobstructed area to measure the muon distribution flux N0 and the background count of muons from each direction. This serves as the absolute benchmark for subsequent quantitative analysis.
[0059] Before reconstructing muon tracks, the data processing server needs to ensure that the received muon events are valid. The determination of a valid muon event needs to be completed at the front-end muon detector node and the main control unit. The determination criteria involve two levels, and the data processing flow is as follows: Figure 4 The steps shown are as follows: 1) All front-end muon detector nodes receive the raw muon signals; 2) First-level judgment: Threshold compliance judgment of the original muon signal is performed within the front-end muon detector node; If the condition is not met (compliance is negative), it is judged as dark noise or soft electrons, and the corresponding data is discarded; If the conditions are met (compliance is yes), a candidate event is generated and sent to the muon detection master unit. The candidate event includes a timestamp and the hit location on the front-end muon detector node. 4) Second-level judgment: The logic conformance judgment is triggered again within the main control unit of the muon detection; If it does not meet the threshold, it is judged as noise that happens to coincidentally meet the signal threshold, and the corresponding candidate case is discarded; If the event matches, it is confirmed as a valid muon event and sent to the data processing server for subsequent path reconstruction and muon imaging.
[0060] Other details regarding the determination of valid muon events are mature existing technologies and will not be elaborated upon in this implementation.
[0061] The method for analyzing internal risks of dams performed on a data processing server, which is also the method for identifying internal risks of reservoir dams in this invention: The 3D mesh model of the dam is discretized into a voxel (3D pixel) mesh. For each muon observation direction, the path length of each muon ray passing through each voxel is calculated to form a projection matrix A. Each element of the projection matrix A represents the path length of a specific muon ray in a certain voxel.
[0062] During measurements at the dam toe, the muon flux attenuation for each muon ray was calculated, μ. i If the muon flux decays along the i-th path, then μ i =-ln(N i / N0 i ), where N i This is a measured count, N0 i It is the expected background count without dam obstruction, μ i It is proportional to the average density of the material along the path.
[0063] The existing approach involves iteratively calculating the muon decay flux attenuation value using a complex algorithm, repeatedly iterating to determine the density value of the dam as discrete voxels. Then, the density information (containing density information at different locations) is reassigned onto the dam's 3D model. While this iterative method yields relatively accurate data, the complexity of the iterative algorithm and the large computational load make its final implementation quite challenging.
[0064] We know that the flux decay of muons is proportional to the density along their path. Assuming the dam we are investigating is a homogeneous earth-rock dam, we can establish an inversion equation with voxel density as the unknown parameter and muon observation data as the target. By manually measuring the density value of one area, we can infer the density values of other areas, and thus create a relative three-dimensional density distribution map of the entire dam. However, when calculating density, the material of the earth-rock dam needs to be limited to a reasonable range (e.g., 1.8 g / cm³ to 2.4 g / cm³) to ensure relatively accurate results. Then, we can infer potential risk areas of the dam from the relatively low-density anomalies in the dam's three-dimensional density distribution map. Combined with manual sampling and detection, this allows for the location and identification of risks within the dam. This provides clear targets for later dam operation, maintenance, and engineering intervention. For other types of dams, such as core-wall dams, we can perform manual measurements on the core wall and homogeneous soil areas separately to infer the density of other areas.
[0065] This invention provides an innovative technological system capable of achieving "transmissive, integrated, and highly efficient" dam detection. UAV lidar and cosmic ray muon detection technology offer new solutions for this purpose.
[0066] UAV lidar can quickly and accurately acquire high-resolution three-dimensional terrain and surface deformation data of the dam and its surrounding area, revealing the stability of the dam body from a macroscopic perspective.
[0067] Cosmic ray muon detection is an emerging non-destructive detection technology that uses the changes in the trajectory and flux of high-energy muons as they penetrate an object to invert the internal density distribution, much like performing a "CT scan" on a dam. It is extremely sensitive to low-density anomalies such as internal cavities and seepage zones.
[0068] However, each individual technology has its limitations: lidar can only observe the surface, while muon detection, although capable of penetrating the interior, has relatively low resolution and takes a long time to image large objects. Therefore, deeply integrating and complementing the information of these two technologies to form a collaborative detection capability that integrates "precise surface deformation monitoring" and "internal structural anomaly detection" is a cutting-edge development direction for identifying risks inside reservoir dams.
[0069] Therefore, the present invention provides a method and system for identifying internal risks of reservoir dams based on the fusion of UAV lidar and muon detection, which can efficiently, accurately and non-destructively detect the internal structure of reservoir dams, and provide clear targets for subsequent dam operation, maintenance and engineering treatment.
Claims
1. A method for identifying internal risks of a reservoir dam, characterized in that, Includes the following steps: 1) Obtain the three-dimensional outline of the dam and discretize its internal space into a regular voxel mesh; 2) Use a muon detector covering the entire dam to detect muon rays penetrating the dam and obtain muon observation data. The muon observation data includes the path length of each muon ray in each voxel and the corresponding muon flux attenuation for each muon ray. 3) Using voxel density as the unknown parameter and muon observation data as the target for inversion; treating the dam body as homogeneous, measuring the actual density of a region, and calculating the actual density of other regions based on the voxel density of that region and the voxel density of other regions. 4) Relatively low-density anomaly zones are considered as risk areas where the dam may pose a risk.
2. The method for identifying internal risks of a reservoir dam according to claim 1, characterized in that, In step 1), a drone equipped with a lidar is used to perform a flight scan of the dam to obtain three-dimensional point cloud data of the dam; a three-dimensional surface mesh model of the dam is generated based on the three-dimensional point cloud data of the dam, thereby obtaining the three-dimensional outline of the dam.
3. The method for identifying internal risks of a reservoir dam according to claim 1, characterized in that, In step 2), the muon detectors are deployed at the toe of the dam to detect the dam. The muon detectors are deployed more densely in areas of suspected risk to the dam, and the muon detectors are deployed evenly in other areas.
4. The method for identifying internal risks of a reservoir dam according to claim 1, characterized in that, In step 4), a three-dimensional density distribution map of the dam is drawn based on the actual density of each area of the dam, and anomalies with relatively low density are found based on the three-dimensional density distribution map.
5. The method for identifying internal risks of a reservoir dam according to claim 1, characterized in that, In step 3), the actual density of a region is obtained by manual measurement.
6. A risk identification system for the interior of a reservoir dam, characterized in that, Includes a data processing server, which is used to execute instructions to achieve the following steps; 1) Obtain the three-dimensional outline of the dam and discretize its internal space into a regular voxel mesh; 2) Use a muon detector covering the entire dam to detect muon rays penetrating the dam and obtain muon observation data. The muon observation data includes the path length of each muon ray in each voxel and the corresponding muon flux attenuation for each muon ray. 3) Using voxel density as an unknown parameter and muon observation data as the target, inversion is performed; the dam body is considered homogeneous, the actual density of a region is obtained, and the actual density of other regions is calculated based on the voxel density of this region and the voxel density of other regions. 4) Relatively low-density anomaly zones are considered as risk areas where the dam may pose a risk.
7. The reservoir dam internal risk identification system according to claim 6, characterized in that, It also includes drones equipped with lidar; in step 1), the drones perform aerial scanning of the dam to obtain three-dimensional point cloud data of the dam; a three-dimensional surface mesh model of the dam is generated based on the three-dimensional point cloud data of the dam, thereby obtaining the three-dimensional outline of the dam.
8. The reservoir dam internal risk identification system according to claim 6, characterized in that, The muon detectors are deployed at the toe of the dam to detect the dam. The muon detectors are deployed more densely in areas of suspected risk to the dam, and are deployed evenly in other areas.
9. The reservoir dam internal risk identification system according to claim 6, characterized in that, In step 4), a three-dimensional density distribution map of the dam is drawn based on the actual density of each area of the dam, and anomalies with relatively low density are found based on the three-dimensional density distribution map.
10. The reservoir dam internal risk identification system according to claim 6, characterized in that, In step 3), the actual density of a region is obtained by manual measurement.