Reservoir engineering safety management system and method

Abstract

The present application relates to the technical field of hydraulic engineering, and specifically relates to a reservoir engineering safety management system and method, comprising: obtaining a fissure tomographic image of a drainage gallery wall through a wall-penetrating ground penetrating radar, and synchronously collecting dynamic data of a phreatic line displacement of a distributed optical fiber sensor; performing space-time alignment on the two types of data, constructing a fissure space topology relationship and a seepage conduction path model, and generating a defect-seepage coupling atlas; using a coupling atlas analysis module and a space-time convolutional neural network module to identify the mutation trend of the phreatic line displacement rate, and triggering a gallery structure failure early warning signal; automatically generating a reinforcement work order after early warning, calling a UAV to carry out a high-definition image review task, and feeding back the review result to correct the seepage conduction path model, thereby realizing dynamic self-correction. The present application has the capabilities of precise modeling, intelligent early warning and real-time closed-loop regulation and control, and can significantly improve the structural safety management level of reservoir engineering.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of hydraulic engineering, and particularly relates to a reservoir engineering safety management system and method. BACKGROUND

[0002] As a major infrastructure related to national economy and people's livelihood, the structural safety of reservoir engineering directly affects flood control and storage, water resources regulation and safety and stability of downstream areas. Drainage gallery is a key structure in the gravity dam body or earth-rock dam body of the reservoir, which is an important channel for guiding dam seepage, relieving seepage pressure, preventing dam foundation piping and structural deterioration. During long-term operation, the gallery wall may produce microcracks due to factors such as concrete aging, penetration erosion and temperature stress, and form potential seepage channels under the action of high water seepage pressure, further inducing dam deformation, displacement anomaly and even overall instability.

[0003] The existing safety monitoring means of the reservoir gallery mainly rely on single-point infiltration line monitoring or discrete crack observation, which is difficult to realize the overall perception of the spatial distribution of wall cracks and seepage behavior. Although some methods introduce optical fiber sensors, they lack the ability to model the data structure and cannot reveal the dynamic coupling relationship between crack connectivity and seepage conduction path. In addition, most systems lack prediction mechanisms for sudden trends and real-time response capabilities, and rely on manual intervention for early warning, which has a lagging reaction and is difficult to dynamically correct the model, resulting in misjudgment, missed judgment and other problems, which limits its adaptability and practicality in high-risk hydraulic structure scenarios. SUMMARY

[0004] The present application provides a reservoir engineering safety management system and method, which realizes the collaborative monitoring, intelligent identification and early warning disposal of the structural hidden danger and seepage state of the drainage gallery.

[0005] The reservoir engineering safety management method comprises the following steps:

[0006] S1: obtaining a crack tomographic image inside the drainage gallery wall by a through-wall ground penetrating radar, and synchronously collecting dynamic data of the infiltration line displacement by a distributed optical fiber sensor arranged on the backwater surface of the retaining wall;

[0007] S2: spatiotemporally aligning the crack tomographic image and the dynamic data of the infiltration line displacement, constructing a seepage conduction path model based on the spatial topological relationship of the cracks, and outputting a defect-seepage coupling graph;

[0008] S3: inputting the defect-seepage coupling graph into an infiltration line mutation prediction model, and if the infiltration line mutation prediction model identifies that the infiltration line displacement rate exceeds the dynamic threshold in two consecutive monitoring periods and diffuses along the seepage conduction path, triggering a gallery structure failure early warning signal;

[0009] S4: Automatically generate a reinforcement work order including positioning coordinates according to the gallery structure failure early warning signal, simultaneously start a drone to review the early warning area with high-definition images, and feed back the review results to the seepage conduction path model for self-correction.

[0010] Optionally, the S1 comprises:

[0011] S11: Laying a scanning track grid of the wall-penetrating ground penetrating radar on the surface of the drainage gallery wall, controlling the wall-penetrating ground penetrating radar to move at a uniform speed of 0.2 m / s along the track grid to scan, and obtaining an initial crack tomography image inside the wall;

[0012] S12: According to the crack distribution density in the initial crack tomography image, determining the coordinates of the seepage line section that needs to be monitored on the backwater surface of the retaining wall, and activating the distributed optical fiber sensor corresponding to the section into a high-frequency sampling mode;

[0013] S13: Collecting seepage line displacement dynamic data of the key monitoring section through the activated distributed optical fiber sensor, and simultaneously controlling the wall-penetrating ground penetrating radar to perform secondary fine scanning on the high-density crack area to obtain a high-resolution crack tomography image;

[0014] S14: Fusing the high-frequency collected seepage line displacement dynamic data and the high-resolution crack tomography image to generate a synchronous monitoring data set with a time stamp.

[0015] Optionally, the S2 comprises:

[0016] S21: Performing a space-time alignment operation on the crack tomography image and the seepage line displacement dynamic data, unifying the spatial coordinates of the two types of data to the engineering absolute coordinate system through the Beidou space-time positioning chip, synchronizing the time stamp to the UTC standard time, and outputting the crack tomography image with unified coordinates and the seepage line displacement dynamic data with a synchronous time stamp;

[0017] S22: Extracting crack endpoint three-dimensional coordinates and opening parameters based on the crack tomography image with unified coordinates, and constructing a crack space topology relationship network graph;

[0018] S23: Mapping the seepage line displacement dynamic data with a synchronous time stamp to the crack space topology relationship network graph, calculating the seepage conduction probability of each connected edge under the action of the seepage line pressure gradient, and generating a seepage conduction path model with weight parameters;

[0019] S24: Based on the generated seepage conduction path model, outputting a defect-seepage coupling graph atlas labeled with conduction direction, risk weight, and corresponding seepage line displacement vector.

[0020] Optionally, the seepage line mutation prediction model comprises a coupling graph analysis module, a space-time convolutional neural network module, and a mutation probability calculation module.

[0021] Optionally, the S3 comprises:

[0022] S31: Extract the seepage conduction path weight, the infiltration line displacement vector and the conduction direction parameter in the defect-seepage coupling graph through the coupling graph analysis module, and generate structured prediction input data;

[0023] S32: Input the structured prediction input data into the spatio-temporal convolutional neural network module, fuse the historical working condition data to construct a three-dimensional feature tensor, and output a spatio-temporal variation feature matrix of the infiltration line displacement rate.

[0024] Optionally, the S3 further comprises:

[0025] S33: Perform residual analysis on the spatio-temporal variation feature matrix of the infiltration line displacement rate and the real-time collected infiltration line displacement dynamic data through the mutation probability calculation module, and generate the infiltration line displacement rate prediction value and the mutation probability index of each seepage conduction path;

[0026] S34: When the infiltration line displacement rate prediction value of the seepage conduction path exceeds the dynamic threshold in two consecutive monitoring periods, and the mutation probability index is greater than the preset diffusion threshold, trigger a gallery structure failure early warning signal.

[0027] Optionally, the S4 comprises:

[0028] S41: According to the gallery structure failure early warning signal, analyze the risk path positioning coordinates, call the historical reinforcement scheme matching the positioning coordinates from the emergency disposal preplan library, and generate a reinforcement work order including the positioning coordinates, the risk level and the recommended reinforcement measures;

[0029] S42: Send the positioning coordinates and the image acquisition instruction to the unmanned aerial vehicle control terminal, start the unmanned aerial vehicle to perform a high-definition image review task on the early warning area, and obtain millimeter-level precision review images including the crack width and the expansion direction.

[0030] Optionally, the S4 further comprises:

[0031] S43: Bind the high-definition image review result with the reinforcement work order to generate a comprehensive disposal report, push it to the mobile terminal of the responsible person and start a disposal countdown;

[0032] S44: Extract the crack expansion feature parameters in the high-definition image review result, feed them back to the seepage conduction path model to correct the weight coefficient of the fracture space topological relationship, and complete the model self-correction.

[0033] The reservoir engineering safety management system is used to realize the reservoir engineering safety management method, and comprises the following modules:

[0034] A crack image acquisition module is configured to control the through-wall ground penetrating radar to move along a scanning track grid, acquire a crack tomography image inside the wall of the drainage gallery, and synchronously collect spatial position and timestamp information during device operation;

[0035] A phreatic line displacement monitoring module is configured to activate the distributed optical fiber sensor arranged in the key monitoring section of the backwater surface of the retaining wall, and collect dynamic data of the phreatic line displacement;

[0036] A defect-seepage coupling atlas generation module is configured to perform space-time alignment on the crack tomography image and the dynamic data of the phreatic line displacement, construct a seepage conduction path model, and output a defect-seepage coupling atlas;

[0037] A phreatic line mutation prediction model module is configured to input the defect-seepage coupling atlas into a phreatic line mutation prediction model, fuse historical working condition data to generate a space-time variation feature matrix of the phreatic line displacement rate, and output a phreatic line displacement rate prediction value and a mutation probability index of each seepage conduction path, and trigger a gallery structure failure warning signal based on a dynamic threshold;

[0038] A reinforcement work order generation module is configured to analyze risk path positioning coordinates according to the gallery structure failure warning signal, call a historical reinforcement scheme matching the positioning coordinates from an emergency disposal plan library, and generate a reinforcement work order including positioning coordinates, risk level, and recommended reinforcement measures;

[0039] A UAV image review module is configured to control a UAV to perform a high-definition image review task according to the positioning coordinates in the reinforcement work order, collect millimeter-level review images including crack width and expansion direction, and extract crack expansion feature parameters;

[0040] A seepage conduction path model self-correction module is configured to feed back the crack expansion feature parameters to the seepage conduction path model, correct the weight coefficients in the crack space topological relationship, and realize model self-correction.

[0041] The beneficial effects of the present application are:

[0042] The present application cooperates through-wall ground penetrating radar and distributed optical fiber sensors to acquire crack tomography images and dynamic data of the phreatic line displacement inside the wall of the drainage gallery, and forms a timestamped synchronous monitoring data set by fusing high-resolution radar scanning and high-frequency optical fiber sampling, effectively improving the joint perception ability of micro defects and seepage state of the internal structure of the reservoir project, and realizing high-precision capture of potential instability hidden dangers.

[0043] The present application constructs a fissure space topological relation network graph based on space-time alignment, and superimposes the seepage conduction path model with weight parameters by adding the displacement data of the infiltration line, further outputs the defect-seepage coupling graph with the conduction direction, risk weight and displacement vector of the infiltration line, and the coupling graph analysis module and the space-time convolution neural network model can accurately predict the mutation trend and identify the structural failure risk in time, thereby effectively solving the problems of the ambiguous coupling relationship between the fissure and seepage and the early warning lag in the traditional monitoring method.

[0044] The present application automatically generates a reinforcement work order after mutation early warning, triggers the high-definition image review of the unmanned aerial vehicle on the early warning area, modifies the weight coefficient of the fissure space topological relation of the seepage conduction path model according to the crack expansion characteristic parameters, realizes the dynamic self-correction and feedback evolution of the model, constructs the closed-loop mechanism of structure identification, early warning judgment and model updating, and significantly enhances the intelligence, responsiveness and robustness of the safety management of the reservoir project. BRIEF DESCRIPTION OF DRAWINGS

[0045] In order to more clearly illustrate the technical solutions in the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description only illustrate the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0046] Figure 1 The method flowchart of the embodiment of the present application is shown in the figure.

[0047] Figure 2 The system flowchart of the embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0048] The present application will be described in detail below in combination with the drawings and specific embodiments. It should be noted that in order to make the embodiments more detailed, the following embodiments are the best, preferred embodiments, and other alternative ways can also be used by those skilled in the art to implement them; and the drawings are only used to more specifically describe the embodiments, and are not intended to specifically limit the present application.

[0049] It should be noted that in the specification, "one embodiment", "embodiment", "exemplary embodiment", "some embodiments" and the like indicate that the embodiments described can include specific features, structures or characteristics, but not necessarily every embodiment includes the specific features, structures or characteristics. In addition, when a specific feature, structure or characteristic is described in combination with an embodiment, it should be within the knowledge of those skilled in the art to realize this feature, structure or characteristic in combination with other embodiments (whether or not explicitly described).

[0050] In general, terms can be understood, at least partly, from usage in context. For example, the term "one or more," as used herein, can be used in the singular sense or in the plural sense depending, at least in part, on context. Additionally, the term "based on" can be understood as not necessarily requiring a set of exclusive factors, but, instead, can allow for existence of additional factors not expressly described, again depending, at least in part, on context.

[0051] As shown in Figure 1 The reservoir engineering safety management method comprises the following steps:

[0052] S1: Obtain the crack tomographic image inside the wall of the drainage gallery by the through-wall ground penetrating radar, and simultaneously collect the dynamic data of the seepage line position by the distributed optical fiber sensor arranged on the backwater surface of the retaining wall, specifically:

[0053] S11, through-wall ground penetrating radar scanning track grid arrangement and initial crack image acquisition: set a scanning track grid on the wall surface of the drainage gallery, the track grid is a rectangular grid structure, the horizontal and vertical spacings are both set to 0.3 meters, and the total length of the track is customized according to the size of the gallery wall, and a magnetic attraction type fixing structure is adopted to adapt to the concrete surface. The through-wall ground penetrating radar is installed on a track moving trolley, equipped with a high-frequency antenna array with a center frequency of 1.2 GHz, the antenna spacing is 5 cm, the linear polarization transmission mode is adopted, the penetration depth is ≥3 meters, and the crack identification resolution reaches 2 millimeters.

[0054] The ground penetrating radar performs full-coverage scanning along the track grid at a uniform speed of 0.2 m / s, uses an ultra-wideband pulse signal sending and echo receiving mechanism to image the area sensitive to the change of the dielectric constant inside the wall, and generates an initial crack tomographic image of the wall. The data acquisition process is real-time cached by an FPGA control module and transmitted to a field industrial computer for time domain inversion reconstruction, and the output is a gray-scale tomographic image grid matrix with a resolution of 256x256.

[0055] S12, key seepage section identification and sensor high-frequency activation: after obtaining the initial crack tomographic image, a density clustering algorithm is used to cluster and identify the high dielectric gradient area in the image, and the coordinates of the crack dense distribution section are extracted. The extracted coordinates are mapped to the retaining wall backwater surface area in the gallery structure diagram, and the key monitoring seepage line section is screened out. The corresponding section has a pre-buried distributed optical fiber sensor (using the BOTDA principle), and the system sends an activation instruction through a serial controller to make it enter a high-frequency sampling mode, and the sampling frequency is increased from 1 Hz in the basic mode to 10 Hz, improving the data timeliness and seepage change sensitivity.

[0056] S13, synchronization of high-frequency monitoring and secondary fine radar scanning: in the key monitoring section, real-time collection of the infiltration line displacement dynamic data fed back by the distributed optical fiber sensor, recording the Brillouin frequency shift change value of each sampling point of the sensing optical fiber, and converting it into the micro-strain data along the optical fiber axial direction. The system calculates the first derivative of the continuous sampling results as the displacement rate reference quantity.

[0057] At the same time, the control of the through-wall ground penetrating radar is repositioned to the area with dense crack distribution in the initial image for secondary scanning. The scanning in this round adopts a sampling strategy with lower speed (0.05 m / s) and higher pulse overlap rate (≥80%), further improving the signal-to-noise ratio and image resolution, and outputting a high-resolution crack tomography image with a resolution of 512x512.

[0058] S14, synchronization of monitoring data set fusion and time marking: the collected infiltration line displacement dynamic data and high-resolution crack tomography image are aligned according to the time stamp, and are organized using a unified data structure. Each set of data includes: time stamp (accurate to seconds), infiltration line displacement sensing point number and position coordinates, micro-strain value and displacement rate at the corresponding time point, crack image subgraph (sliced according to coordinate area) synchronously collected, maximum crack length in the crack image, density index, and other structural parameters.

[0059] Finally, a time-stamped synchronous monitoring data set is formed, which serves as the basic input data for subsequent seepage path modeling and failure warning.

[0060] S2: spatio-temporal alignment of crack tomography image and infiltration line displacement dynamic data, construction of seepage conduction path model based on crack spatial topological relationship, output of defect-seepage coupling atlas, specifically:

[0061] S21, spatio-temporal alignment operation: perform spatio-temporal alignment operation on the crack tomography image and the infiltration line displacement dynamic data output by S1. The system integrates a Beidou spatio-temporal positioning chip (model BD980) with centimeter-level three-dimensional spatial positioning accuracy and nanosecond-level time synchronization capability. The specific alignment process is as follows:

[0062] Spatial coordinate unification: map the two-dimensional pixel coordinates in the crack tomography image to the engineering absolute coordinate system (using E-N-U projection format) through the calibration conversion matrix, and also convert the positions of each sampling point collected in the distributed optical fiber sensor to the same coordinate system, ensuring one-to-one correspondence of the two types of data in the spatial dimension;

[0063] Time stamp synchronization: extract the original time stamps of the two types of data, and synchronize them to the UTC standard time output by the Beidou system to complete high-precision time alignment.

[0064] The output result is: crack tomography image with unified coordinates and infiltration line displacement dynamic data with synchronized time stamp.

[0065] S22, crack space topology relationship network graph construction: after completing the space-time alignment, the image analysis method based on structure edge extraction and topology skeleton extraction is used to analyze the crack tomography image in the unified coordinate:

[0066] The three-dimensional coordinates of the head and tail endpoints of each identifiable crack are extracted;

[0067] The opening parameter (unit: mm) of each crack is calculated by the image gray gradient inversion algorithm;

[0068] According to the spatial position proximity and geometric continuity, it is judged whether there is a connected relationship between adjacent endpoints.

[0069] Thus, a crack space topology relationship network graph is constructed, wherein:

[0070] The node represents the crack endpoint;

[0071] The edge represents the path between adjacent endpoints with potential seepage connectivity.

[0072] The network graph provides basic topological structure support for subsequent path modeling and risk propagation analysis.

[0073] S23, seepage conduction path model construction: seepage conduction path model generation

[0074] The synchronous time scale of the infiltration line displacement dynamic data is mapped into the above-mentioned crack space topology relationship network graph to construct the seepage conduction path model. The specific process is as follows:

[0075] The spatial position of each infiltration line sampling point is corresponded to the related crack node in the topology network graph;

[0076] The displacement difference and spatial distance between the sensing points are calculated to form an approximate infiltration line pressure gradient;

[0077] Combined with the crack opening and connection distance, a seepage conduction probability model of each connected edge under the action of the infiltration line pressure gradient is established.

[0078] The model is represented in the form of a weighted directed graph, defined as follows:

[0079] ;

[0080] Wherein, is a set of crack endpoints, is a set of connected edges between adjacent endpoints, is a set of edge weights, representing the seepage conduction probability, is a set of edge conduction directions, is a set of edge infiltration line displacement vectors. The calculation is:

[0081] ;

[0082] wherein, is an empirical adjustment coefficient, is the permeability coefficient of rock mass, is the average opening of fracture segment , is the spatial Euclidean distance between two endpoints, is the displacement difference between two sensing points of the saturation line, is the sampling time interval.

[0083] is defined as:

[0084] ;

[0085] wherein is the three-dimensional coordinate of the fracture endpoint, and sgn represents the direction.

[0086] is calculated by weighting the micro-strain direction and amplitude of the two connected sensing points, indicating the dominant deformation direction and size induced by seepage conduction on the current path.

[0087] The final seepage conduction path model is used for subsequent coupling graph construction and mutation identification.

[0088] S24: Defect-seepage coupling graph generation: according to the seepage conduction path model, output the final defect-seepage coupling graph. The graph includes the following elements:

[0089] Conduction direction: represented by in the seepage conduction path model;

[0090] Risk weight: calculated by comprehensive evaluation of edge weight and fracture opening;

[0091] Saturation line displacement vector: represented by in the seepage conduction path model, reflecting the local displacement trend;

[0092] The graph is organized and stored in the form of a graph database, with timestamp identification and spatial positioning capabilities, serving as the input basis for subsequent mutation identification in S3, while supporting self-correction operations according to review results in S4.

[0093] S3: Input the defect-seepage coupling graph into the saturation line mutation prediction model. If the saturation line mutation prediction model identifies that the saturation line displacement rate exceeds the dynamic threshold in two consecutive monitoring periods and spreads along the seepage conduction path, trigger the warning signal of gallery structure failure, specifically:

[0094] The infiltration line mutation prediction model comprises a coupling atlas analysis module, a space-time convolutional neural network module, and a mutation probability calculation module.

[0095] S31, structured prediction input data generation: the defect-seepage coupling atlas output by S2 is analyzed edge by edge by the coupling atlas analysis module, and key elements constituting the prediction model input are extracted, including:

[0096] seepage conduction path weight (edge weight in the seepage conduction path model ;

[0097] infiltration line displacement vector (representing the actual displacement direction and amplitude along the conduction path, labeled as );

[0098] conduction direction parameter (i.e., the directional edge direction vector defined in the atlas ).

[0099] For the convenience of neural network processing, the system encodes the above information into a structured data set, with the following format:

[0100] Each seepage conduction path as a sample unit

[0101] The feature fields include: , (split into 3-dimensional components), (split into 3-dimensional components), current timestamp, path start and end point coordinates.

[0102] The system updates this structured prediction input data for each monitoring period (such as 10-minute intervals) and provides it as an input sample sequence to the subsequent model.

[0103] S32, space-time feature tensor construction and feature extraction: the structured prediction input data generated in S31 is input into the space-time convolutional neural network module, which adopts a three-dimensional convolutional neural network architecture (3D-CNN) that integrates spatial structure features and time series dynamics. This network consists of the following three layers:

[0104] 1. Input layer: accepts an input tensor of the form , where is the number of seepage conduction path samples, is the channel dimension of each sample (including path weight, vector component, etc.), is the length of the continuous time window (such as 12 monitoring periods);

[0105] 2. Space-time convolutional layer: uses 3 The joint feature extraction is performed by the convolution kernel, and the joint pattern between the paths and the time series is extracted.

[0106] 3. Output layer: output a tensor form of the infiltration line displacement rate spatiotemporal variation feature matrix, with a dimension of , wherein each element represents the displacement rate feature value of the th flow conduction path at the th time step. In addition, the module fuses historical working condition data (such as synchronous water level, rainfall) into additional channels for convolution processing to improve the prediction accuracy.

[0107] S33, mutation probability index calculation and rate prediction value generation: through the mutation probability calculation module, the output infiltration line displacement rate spatiotemporal variation feature matrix is compared and analyzed with the real-time collected infiltration line displacement dynamic data path by path.

[0108] 1. Residual analysis: for each flow conduction path, in the latest two monitoring periods, the residual of the prediction value and the real-time observation value is calculated as follows:

[0109] ;

[0110] 2. Mutation probability index calculation: according to the residual value sequence and the historical fluctuation level, the exponential sliding statistical model (such as Z-score, normalized relative difference) is used to calculate the mutation probability index of the path, and the expression is as follows:

[0111] ;

[0112] wherein is the moving average of the residual sequence of the path , and is the moving standard deviation of the residual sequence of the path .

[0113] The final output result is: the infiltration line displacement rate prediction value of each flow conduction path and the corresponding mutation probability index .

[0114] S34, early warning judgment and signal triggering: when the following two conditions are met, the gallery structure failure early warning signal can be triggered:

[0115] In the last two monitoring periods, the infiltration line displacement rate prediction value of one or more flow conduction paths exceeds the dynamic threshold value (such as the historical 95% quantile) automatically updated by the system according to the history statistics;

[0116] The mutation probability index corresponding to the same path exceeds the system preset diffusion threshold (for example, the diffusion threshold is set to 2.5).

[0117] If the above conditions are met, the system generates a gallery structure failure early warning signal through the early warning interface and pushes it to the data fusion general control platform as the triggering basis of S4.

[0118] S4: Automatically generate a reinforcement work order including positioning coordinates according to the gallery structure failure early warning signal, simultaneously start a drone to review the high-definition image of the early warning area, and feed back the review result to the seepage conduction path model for self-correction, specifically:

[0119] S41, reinforcement work order generation: when the gallery structure failure early warning signal in step S3 is triggered, the system automatically analyzes the risk path positioning coordinates marked in the early warning signal and extracts its corresponding three-dimensional engineering absolute coordinates (E-N-U format). Then, the system starts the reinforcement response process, which includes the following operation steps:

[0120] Pre-plan matching: the system calls the preloaded emergency disposal pre-plan library, which stores historical reinforcement measures, construction records and response time information with spatial regions as indexes. Each pre-plan entry is bound to a center coordinate point, an applicable risk level interval, an applicable lithology environment label and a recommended disposal process;

[0121] Matching mechanism: a fast spatial matching algorithm based on KD-Tree is adopted to search for the historical pre-plan area with the closest spatial distance and the most matched risk level and lithology condition in the emergency disposal pre-plan library with the current risk path positioning coordinates as the query point. The matching process uses the following objective function:

[0122] ;

[0123] Wherein, represents the most matched historical pre-plan area, is the current risk path positioning coordinate, is the center coordinate of the th pre-plan in the emergency disposal pre-plan library, is the Euclidean distance, which represents the spatial proximity between the current risk path positioning coordinate and the historical pre-plan coordinate, represents the early warning level difference penalty term of the th historical area, if the historical early warning level of the area is lower than the current level, the value increases, represents the environmental factor difference degree of the th area, such as lithology and hydrological working conditions, which is usually obtained by a discrete scoring function (for example, completely matched is recorded as 0 and not matched is recorded as 1), and is the weighting coefficient, balancing the control space distance and the warning level / environment factor.

[0124] Reinforcement work order generation: output the reinforcement work order including the following fields:

[0125] Risk path positioning coordinates;

[0126] Risk level (such as Grade I / Grade II / Grade III, determined according to mutation probability index and path weight);

[0127] Recommended reinforcement measures (such as grouting sealing, carbon fiber reinforcement, double-layer composite lining);

[0128] Response time limit;

[0129] Responsible person assignment information.

[0130] S42, high-definition image review task execution: to verify the actual damage state of the warning area, the system sends the following review task parameters to the unmanned aerial vehicle control terminal:

[0131] Risk path positioning coordinates;

[0132] Execution instruction type: image collection;

[0133] Shooting requirements: millimeter-level resolution, with low-light condition adaptability;

[0134] Flight path planning mode: surround scanning + vertical fixed-point overhead shooting.

[0135] The system controls the unmanned aerial vehicle to carry high-resolution visible light + near-infrared composite cameras, and starts the high-definition image review task for the specified coordinate area. The collected images are transmitted to the on-site edge server in real time for processing. The processing process includes:

[0136] Crack width extraction: using multi-scale edge enhancement and sub-pixel fitting method, the resolution reaches 0.3mm;

[0137] Expansion direction calculation: based on the fitting of the main crack axis, output the direction vector and angle value;

[0138] Interference filtering: remove abnormal samples such as occlusion, reflection, and water marks.

[0139] Finally output the review image in image and structured data format for subsequent analysis and disposal.

[0140] S43, comprehensive disposal report generation and pushing: the system binds the reinforcement work order generated in S41 and the high-definition image review results obtained in S42, and synthesizes a structured comprehensive disposal report, which specifically includes:

[0141] Risk path positioning coordinates;

[0142] Risk level;

[0143] Crack width and propagation direction (with image captions);

[0144] Recommended reinforcement measures;

[0145] Response timeout countdown module (e.g., setting a 24-hour countdown);

[0146] Information on responsible entities and individuals;

[0147] Summary of historical records on the handling of similar risk areas.

[0148] The comprehensive handling report is pushed to the responsible person's mobile terminal, such as a mobile APP or a dedicated tablet client, through the DMS (Decision Management System) interface; at the same time, the system starts the corresponding handling countdown process, supporting supervision and scheduling, response tracking and automatic reminder mechanisms.

[0149] S44, Model Self-Correction: To improve the accuracy and dynamic adaptability of subsequent predictions, the system feeds back the crack propagation feature parameters from the high-resolution image verification results obtained in S42 to the previously constructed seepage conduction path model, performing a model self-correction operation, which specifically includes the following:

[0150] 1. Mapping and localization: Based on the coordinate information extracted from the crack image, the extended crack is mapped to the edge in the seepage conduction path model;

[0151] 2. Weighting coefficient correction: Weighting coefficients for the fracture spatial topology corresponding to this path. Dynamic corrections are performed using the following weighted update formula:

[0152] ;

[0153] in, This represents the weights in the original seepage conduction path model. Actual conduction intensity index derived from image crack width and propagation rate This indicates a modified fusion factor, controlling the weighting of new and old information.

[0154] 3. Synchronous update of direction vector: If the deviation between the extended direction of the image display and the transmission direction in the model exceeds a set threshold (e.g., 15°), the transmission direction parameter vector of that path will be updated synchronously.

[0155] 4. Topology fine-tuning: If new cracks or through-going cracks are found in the review images, the system will automatically add new nodes and new path edges to expand the original fissure space topology graph structure. The self-correction mechanism of this model provides more realistic and reliable data basis for the next round of infiltration line mutation prediction, realizing model closed-loop evolution driven by data.

[0156] As shown in Figure 2 The reservoir engineering safety management system is used to implement the reservoir engineering safety management method described above, and includes the following modules:

[0157] Fissure image acquisition module: used to control the through-wall ground penetrating radar to move along the scanning track grid, acquire the drainage gallery wall internal fissure tomography image, and synchronously collect the spatial position and timestamp information during equipment operation;

[0158] Infiltration line displacement monitoring module: used to activate the distributed optical fiber sensor arranged in the key monitoring section of the retaining wall backwater surface, and collect the infiltration line displacement dynamic data;

[0159] Defect-seepage coupling graph generation module: used to perform space-time alignment on the fissure tomography image and the infiltration line displacement dynamic data, construct a seepage conduction path model, and output a defect-seepage coupling graph;

[0160] Infiltration line mutation prediction model module: used to input the defect-seepage coupling graph into the infiltration line mutation prediction model, fuse historical working condition data to generate a space-time change feature matrix of the infiltration line displacement rate, and output the infiltration line displacement rate prediction value and mutation probability index of each seepage conduction path, and based on the dynamic threshold, trigger a gallery structure failure early warning signal;

[0161] Reinforcement work order generation module: used to analyze the risk path positioning coordinates according to the gallery structure failure early warning signal, call the emergency disposal preplan library to match the historical reinforcement scheme of the positioning coordinates, and generate a reinforcement work order including the positioning coordinates, risk level and recommended reinforcement measures;

[0162] Unmanned aerial vehicle image review module: used to control the unmanned aerial vehicle to perform high-definition image review tasks according to the positioning coordinates in the reinforcement work order, collect millimeter-level review images including crack width and expansion direction, and extract crack expansion feature parameters;

[0163] Seepage conduction path model self-correction module: used to feed back the crack expansion feature parameters to the seepage conduction path model, correct the weight coefficients in the fissure space topology relationship, and realize model self-correction.

[0164] The present application encompasses any alternatives, modifications, equivalent methods and solutions made to the essence and scope of the present application. In order to make the public have a thorough understanding of the present application, specific details are described in the following preferred embodiments of the present application, and the present application can also be fully understood without the description of these details to those skilled in the art. In addition, in order to avoid unnecessary confusion to the essence of the present application, well-known methods, processes, procedures, elements and circuits, etc. are not described in detail.

[0165] The above is only the preferred embodiment of the present application, and it should be pointed out that for ordinary skilled in the art, without departing from the principles of the present application, a number of improvements and refinements can also be made, which should be considered as the protection scope of the present application.

Claims

1. A method for safety management of a reservoir project, characterized by, The method comprises the following steps: S1: obtaining a crack tomography image inside a drainage gallery wall by a through-wall ground penetrating radar, and synchronously collecting dynamic data of a seepage line position by a distributed optical fiber sensor arranged on a backwater surface of a retaining wall; S2: performing space-time alignment on the crack tomography image and the dynamic data of the seepage line position, constructing a seepage conduction path model based on a crack spatial topological relationship, and outputting a defect-seepage coupling graph; specifically comprising: S21: performing space-time alignment on the crack tomography image and the dynamic data of the seepage line position, unifying the spatial coordinates of the two types of data to an engineering absolute coordinate system by a Beidou space-time positioning chip, synchronizing the time stamp to the UTC standard time, and outputting the crack tomography image with unified coordinates and the dynamic data of the seepage line position with a synchronous time stamp; S22: extracting crack endpoint three-dimensional coordinates and opening parameters based on the crack tomography image with unified coordinates, and constructing a crack spatial topological relationship network diagram; S23: mapping the dynamic data of the seepage line position with a synchronous time stamp to the crack spatial topological relationship network diagram, calculating seepage conduction probabilities of each connected edge under the action of a seepage line pressure gradient, and generating a seepage conduction path model with a weight parameter; S24: based on the generated seepage conduction path model, outputting a defect-seepage coupling graph with labeled conduction directions, risk weights and corresponding seepage line displacement vectors; S3: inputting the defect-seepage coupling graph into a seepage line mutation prediction model, and if the seepage line mutation prediction model identifies that the seepage line displacement rate exceeds a dynamic threshold in two consecutive monitoring periods and diffuses along the seepage conduction path, triggering a gallery structure failure warning signal; S4: automatically generating a reinforcement work order including positioning coordinates according to the gallery structure failure warning signal, synchronously starting a drone to review high-definition images of the warning area, and feeding back the review results to the seepage conduction path model for self-correction.

2. The method of claim 1, wherein The S1 comprises: S11: arranging a scanning track grid of the through-wall ground penetrating radar on the surface of the drainage gallery wall, controlling the through-wall ground penetrating radar to move at a uniform speed of 0.2 m / s along the track grid to scan, and obtaining an initial crack tomography image inside the wall; S12: determining the coordinates of the seepage line section that needs to be monitored on the backwater surface of the retaining wall according to the crack distribution density in the initial crack tomography image, and activating the distributed optical fiber sensor of the corresponding section to enter a high-frequency sampling mode; S13: collecting the dynamic data of the seepage line position of the key monitoring section by the activated distributed optical fiber sensor, and simultaneously controlling the through-wall ground penetrating radar to perform secondary fine scanning on the crack area to obtain a crack tomography image; S14: fusing the high-frequency collected dynamic data of the seepage line position and the crack tomography image to generate a time-stamped synchronous monitoring data set.

3. The method of claim 1, wherein The seepage line mutation prediction model comprises a coupling graph analysis module, a space-time convolutional neural network module and a mutation probability calculation module.

4. The method of claim 3, wherein The S3 comprises: S31: extracting the seepage conduction path weight, the seepage line displacement vector and the conduction direction parameter in the defect-seepage coupling graph by the coupling graph analysis module to generate structured prediction input data; S32: input the structured prediction input data into the spatio-temporal convolutional neural network module, fuse the historical working condition data to construct a three-dimensional feature tensor, and output a spatio-temporal variation feature matrix of the infiltration line displacement rate.

5. The method of claim 4, wherein The S3 further includes: S33: perform residual analysis on the spatio-temporal variation feature matrix of the infiltration line displacement rate and the real-time collected infiltration line displacement dynamic data by the mutation probability calculation module, to generate the infiltration line displacement rate prediction value and the mutation probability index of each seepage conduction path; S34: when the infiltration line displacement rate prediction value of the seepage conduction path exceeds the dynamic threshold in two consecutive monitoring periods, and the mutation probability index is greater than the preset diffusion threshold, trigger a gallery structure failure early warning signal.

6. The method of claim 5, wherein The S4 includes: S41: analyze the risk path positioning coordinates according to the gallery structure failure early warning signal, call the emergency disposal plan library to match the historical reinforcement scheme of the positioning coordinates, and generate a reinforcement work order including the positioning coordinates, risk level and recommended reinforcement measures; S42: send the positioning coordinates and image acquisition instructions to the unmanned aerial vehicle control terminal, start the unmanned aerial vehicle to perform a high-definition image review task on the early warning area, and obtain millimeter-level review images including crack width and expansion direction.

7. The method of claim 6, wherein The S4 further includes: S43: bind the high-definition image review result with the reinforcement work order to generate a comprehensive disposal report, push it to the mobile terminal of the responsible person and start a disposal countdown; S44: extract the crack expansion feature parameters in the high-definition image review result, feed back to the seepage conduction path model to correct the weight coefficient of the fracture space topological relationship, and complete the model self-correction.

8. A reservoir project safety management system for implementing the reservoir project safety management method according to any one of claims 1 to 7, characterized by, The system includes the following modules: A fracture image acquisition module is configured to control a through-wall ground penetrating radar to move along a scanning track grid, acquire a fracture tomography image inside a drainage gallery wall, and synchronously collect spatial position and time stamp information during equipment operation; An infiltration line displacement monitoring module is configured to activate a distributed optical fiber sensor arranged in a key monitoring section of the backwater surface of the retaining wall, and collect the infiltration line displacement dynamic data; A defect-seepage coupling graph generation module is configured to perform spatio-temporal alignment on the fracture tomography image and the infiltration line displacement dynamic data, construct a seepage conduction path model, and output a defect-seepage coupling graph; An infiltration line mutation prediction model module is configured to input the defect-seepage coupling graph into an infiltration line mutation prediction model, fuse historical working condition data to generate a spatio-temporal variation feature matrix of the infiltration line displacement rate, and output the infiltration line displacement rate prediction value and the mutation probability index of each seepage conduction path, and trigger a gallery structure failure early warning signal based on a dynamic threshold; A reinforcement work order generation module is configured to analyze risk path positioning coordinates according to the gallery structure failure early warning signal, call the emergency disposal plan library to match the historical reinforcement scheme of the positioning coordinates, and generate a reinforcement work order including the positioning coordinates, risk level and recommended reinforcement measures; An unmanned aerial vehicle image review module is configured to control the unmanned aerial vehicle to perform a high-definition image review task according to the positioning coordinates in the reinforcement work order, collect millimeter-level review images including crack width and expansion direction, and extract crack expansion feature parameters; The seepage conduction path model self-correction module is configured to feed back the fracture extension characteristic parameters to the seepage conduction path model, correct the weight coefficients in the fracture space topological relation, and realize model self-correction.

Images