A dam seepage state visualization inversion method, device and medium
By constructing a damped incremental response model and ridge regression algorithm, combined with an exponential time decay operator, a permeability cloud map of the hydraulic response performance index (HREI) is generated, which solves the problems of rainfall noise and reservoir scheduling interference in dam seepage monitoring, and realizes accurate positioning and visualization of seepage risk.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- ANHUI & HUAI RIVER WATER RESOURCES RES INST
- Filing Date
- 2026-02-13
- Publication Date
- 2026-06-05
Smart Images
Figure CN122153835A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of water conservancy engineering data processing technology, specifically to a method, equipment, and storage medium for visual inversion of dam seepage behavior based on multi-source response decoupling and physical constraints. Background Technology
[0002] As a key hub for flood control, irrigation, and power generation in a river basin, the structural safety of reservoir dams is directly related to public safety. Among the many dam failure modes, seepage failure (such as piping, soil erosion, and contact scouring) accounts for a very high proportion. Therefore, real-time monitoring of the seepage field by piezometers buried inside the dam body and using the monitoring data to infer the permeability of the dam's interior are core means to ensure the safe operation of dams.
[0003] Existing methods for monitoring and analyzing dam seepage mainly rely on statistical models (such as the HST model) or simple cross-correlation analysis based on the Pearson correlation coefficient. However, these traditional methods have the following significant technical limitations in practical engineering applications: (1) Weak anti-interference ability: The piezometers on the surface or in the shallow part of the dam are easily affected by heavy rainfall. Rainfall infiltration will cause the piezometer reading to rise temporarily. At this time, traditional methods often cannot effectively decouple the "false high pressure caused by rainfall" from the "real high pressure caused by reservoir water infiltration", which is prone to false alarms. (2) Difficulty in identifying physical lag: Affected by reservoir scheduling or tides, reservoir water level often fluctuates periodically. At this time, pure data-driven correlation analysis is prone to falling into the "multiple solutions trap", that is, it is impossible to distinguish whether the current pressure fluctuation is caused by the current reservoir water level or by the reservoir water level tens of hours ago (lag ambiguity), which makes it impossible to accurately calculate the true rate of pressure transmission.
[0004] In addition, existing technologies mostly focus on the prediction of single-point values and lack a comprehensive evaluation index that can integrate the "driving factors", "transmission rate" and "model confidence", making it difficult to intuitively and visually present whether there are concentrated seepage channels or seepage hazards inside the dam.
[0005] To address this issue, this application proposes a method for visualizing and inverting the seepage behavior of dams, which can effectively eliminate rainfall noise and periodic water level interference, accurately locate deep seepage hazards, and achieve visual early warning, thereby solving the aforementioned technical problems. Summary of the Invention
[0006] The main objective of this invention is to provide a method for visualizing and inverting the seepage behavior of dams, so as to solve the technical problems mentioned in the background art.
[0007] The present invention solves the above-mentioned technical problems by adopting the following technical solutions: A method for visualizing and inverting the seepage behavior of a dam involves performing the following steps using computer equipment: Step S1. Construct a damping incremental response model: Obtain historical monitoring time series data of the dam, simulate the damping effect of the medium through sliding window averaging, calculate the first-order difference sequence, and establish an incremental model with water level change rate and rainfall rate as inputs and seepage pressure change rate as output; Step S2. Construct the spatiotemporal lag feature matrix: Based on the preset maximum lag time window, construct a joint matrix containing reservoir water level lag features and rainfall lag features with multiple time steps; Step S3. Apply physical distance decay constraint: Define an exponential time decay operator to weight and scale the reservoir water level lag features in the joint matrix, thereby introducing Darcy's law time priority constraint into the feature space; Step S4. Perform multi-source decoupling inversion calculation: Construct a ridge regression model, use L2 regularization term combined with a set of physical constraints to train the incremental model, and perform decoupling calculation to obtain the reservoir water level sensitivity coefficient vector and rainfall sensitivity coefficient vector; Step S5. Index Calculation and Behavior Cloud Map Generation: Based on two sets of sensitivity coefficient vectors, the primary response peak is identified to determine the physical lag time, and the hydraulic response performance index (HREI) is constructed by combining the signal-to-noise ratio. Finally, the index is mapped to the dam profile using spatial interpolation technology to generate a permeability behavior cloud map that reflects the distribution of regional seepage risk.
[0008] Preferably, the incremental model construction process in step S1 specifically includes: Set the length of the sliding window Regarding time series Aligned original reservoir water level sequence Rainfall sequence and osmotic pressure sequence A moving average method is used to simulate the damping effect of the dam medium, eliminating high-frequency measurement noise, and a damping incremental response model is constructed. The moving window averaging formula is as follows:
[0009]
[0010] in, The sliding window length is used to simulate the hysteresis smoothing characteristics of the medium. For smoothed timing The osmotic pressure value below, For smoothed timing The water level below, The timing for the sliding window is The original reservoir seepage pressure value below, The timing for the sliding window is The original reservoir water level value; The first-order difference of the smoothed sequence is calculated using the following formula:
[0011]
[0012] in, This is the seepage pressure value calculated using the first-order difference method. This is the water level value calculated using the first-order difference method. Finally, invalid data in the difference sequence, such as NaN values and infinite values generated during the difference calculation, are removed to construct a time series sample set containing only the effective rate of change as an incremental model. Preferably, the construction process of the joint matrix in step S2 includes: Based on the physical dimensions and medium characteristics of the dam, the maximum physical lag search window for the reservoir water level is set. and the maximum impact window of rainfall ; For the current moment Construct each containing the past The eigenvector of the rate of change of water level at each time point is: and including the past The feature vector of the rate of change of rainfall at each time step is: ; The two sets of feature vectors are concatenated horizontally, and all time samples are stacked vertically to form a joint matrix with high-dimensional features. .
[0013] Preferably, the specific operation process for weighted scaling of the reservoir water level lag features in the joint matrix in step S3 includes: Construct a set of exponential time decay operators for the joint matrix. The columns that lag behind reservoir water levels are weighted and then weighted to achieve weighted scaling. The weighted scaling formula is as follows:
[0014] in, For time sequence Lower hysteresis The original values of the reservoir water level characteristics at each time step For time sequence The eigenvalues after weighted scaling The preset physical attenuation coefficient; Here, an exponential time decay operator is constructed. Used to construct physical distance decay constraints on the joint matrix The columns that lag behind the reservoir water level are weighted for the joint matrix. Medium lag Hourly water level characteristic column, multiply all its elements by Ultimately, this forces the regression model to converge to the short-lag path solution in the solution space where multicollinearity exists, thus eliminating long-lag misjudgments caused by periodic water level fluctuations. At this point, by artificially reducing the numerical amplitude of the distant lag feature, when the subsequent regression model attempts to utilize the distant lag feature (such as... When fitting the current seepage pressure, a very large regression coefficient must be assigned to offset the effect of the attenuation operator. This will lead to a significant increase in the regularization penalty term of the model. Therefore, the optimization algorithm will tend to prioritize the "proximal hysteresis feature" that has not been significantly attenuated, thereby forcing the inversion result to converge to the physical shortest conduction path.
[0015] Preferably, the specific calculation process for obtaining the reservoir water level sensitivity coefficient vector and the rainfall sensitivity coefficient vector through decoupling calculation in step S4 includes: Before model training, for each measurement point, the non-empty effective time periods in its seepage pressure change rate sequence are identified, and a binary mask matrix is generated. ; Based on the weighted feature matrix and target vector The objective function for ridge regression is constructed as follows:
[0016] in, The regularization strength; Only the joint matrix corresponding to the valid time steps in the mask matrix is selected. The row components are used in ridge regression training to adapt to situations where the sensor experiences intermittent failures or data loss. The regression coefficient vector is solved using the least squares method during training. ; The ridge regression coefficients obtained through training are used to construct the ridge regression coefficient vector. Decomposed into a subset of coefficients corresponding to the reservoir water level and the corresponding subset of rainfall coefficients This allows us to separate the independent contribution rates of the reservoir water level component to the seepage pressure change and the independent contribution rate of the rainfall component to the seepage pressure change.
[0017] Preferably, in step S5, a primary peak locking strategy is adopted in the process of determining the physical lag time. The primary peak locking strategy specifically includes: Calculate the absolute value sequence of the reservoir water level sensitivity coefficient vector. ; Identify the set of local peaks in the sequence whose normalized energy is greater than a preset threshold; The peak with the smallest time step in the set is selected as the primary peak, and the energy centroid in the neighborhood of the primary peak is calculated. The time corresponding to the centroid is defined as the physical true lag time of the measuring point. .
[0018] Preferably, the Hydraulic Response Effectiveness Index (HREI) in step S5 is a comprehensive indicator used to quantitatively evaluate the connectivity and purity of local seepage channels in a dam, and its specific calculation formula is as follows:
[0019]
[0020]
[0021]
[0022] in, The hydraulic response performance index. The multi-source decoupling ratio is determined by the total sensitivity of the reservoir water level. Overall sensitivity to rainfall The ratio is determined to characterize the purity of reservoir water components in the seepage driving force and to suppress spurious signals caused by rainfall interference. This is the denominator correction coefficient for the multi-source decoupling ratio. The time efficiency factor is the physical real time lag. They are inversely proportional and are used to characterize the agility of pressure transmission. The constant coefficient of the time term, This is the correction factor for the denominator of the time term. For model confidence factors, For model fit, This is the correction factor for the confidence term. The goodness of fit of the ridge regression model on incremental data is determined to characterize the statistical reliability of the inversion results.
[0023] Preferably, the specific operation process for generating the permeability cloud map using spatial interpolation technology in step S5 includes: Obtain the spatial coordinates of all measuring points on the dam body profile. and its corresponding HREI value; Using Kriging interpolation or inverse distance weighted interpolation algorithms, a continuous HREI scalar field is constructed, and the HREI index of each measuring point is mapped to the two-dimensional profile coordinate system of the dam to generate a thermal distribution cloud map, in which areas with high HREI values are identified as dominant leakage channels or high-risk areas. Plot the HREI index distribution curves along the seepage path of the dam at each measuring point; If the curve shows a monotonically decreasing trend, it is judged as normal Darcy flow; If the curve shows a numerical rebound in the downstream area, it is determined that there is a risk of seepage around the source or concentrated leakage in that area.
[0024] In another aspect, the present invention also discloses a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps of the method described above.
[0025] In another aspect, the present invention also discloses a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the method described above.
[0026] As can be seen from the above technical solution, the present invention provides a method for visualizing and inverting the seepage behavior of dams. Compared with the prior art, the present invention has the following advantages: 1. This invention, by constructing an incremental model and combining it with the ridge regression algorithm, can mathematically separate the contribution rates of rainfall components and reservoir water level components in seepage pressure changes, thereby effectively solving the false alarm problem caused by "spurious correlation" at shallow measuring points during heavy rainfall.
[0027] 2. By introducing an exponential time decay operator in feature engineering, this invention can force the model to follow the time priority principle of Darcy's law, thereby effectively eliminating long-lag misjudgments caused by periodic fluctuations in reservoir water level.
[0028] 3. This invention constructs a hydraulic response performance index (HREI) by integrating signal purity, transmission rate, and model confidence. This can not only visually display the dominant seepage channels (high HREI zone) and seepage barriers (low HREI zone) inside the dam through the final generated permeability cloud map, but also achieve accurate quantification of dam seepage risk and rapid location of dominant seepage channels and potential hazard areas based on the hydraulic response performance index (HREI).
[0029] It should be understood that the descriptions in this section are not intended to identify key or essential features of embodiments of the invention, nor are they intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Of course, implementing any product of the invention does not necessarily require achieving all of the advantages described above simultaneously. Attached Figure Description
[0030] The accompanying drawings, which form part of this application, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an undue limitation of the invention. In the drawings: Figure 1 This is a schematic diagram of the overall process of the present invention; Figure 2 This is a comparison curve of HREI index decay along the path generated in an embodiment of the present invention; Figure 3 This is a comparative cloud map (HREI thermal distribution map) of the dam profile of Reservoir A in this embodiment of the invention, where the red area is used to represent areas with high seepage risk; Figure 4 This is a comparative cloud map of the permeability of the dam profile of Reservoir B in this embodiment of the invention (HREI thermal distribution map), where the red area is used to represent the area with high seepage risk. Figure 5 This is a curve showing the decrease in HREI index along the flow path for reservoir A in this embodiment of the invention. Figure 6 This is a curve showing the decrease in HREI index along the flow path for reservoir B in this embodiment of the invention. Detailed Implementation
[0031] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Unless otherwise specified, the embodiments and features in the embodiments of this application can be combined with each other. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0032] For details in the embodiments, please refer to Figures 1 to 6 .
[0033] The dam seepage behavior visualization inversion method proposed in this invention first constructs a damped incremental response model and performs physical damping smoothing and first-order difference processing on the monitoring data. Then, it constructs a spatiotemporal lag feature matrix, builds a joint input containing multi-step long lag features, and introduces an exponential time decay operator to apply physical distance decay constraints and weighted penalties for long lag features. On this basis, it performs multi-source decoupled inversion calculation and uses a ridge regression model to separate the independent contribution coefficients of reservoir water level and rainfall. Finally, it identifies the primary response peak based on the sensitivity coefficient to lock the physical lag time and constructs an original Hydro-Response Efficiency Index (HREI) to generate a seepage behavior cloud map.
[0034] In a specific embodiment, such as Figure 1 As shown, the procedure includes the following steps: L1. Constructing an incremental damping response model: Obtain historical monitoring time series data of the dam, simulate the damping effect of the medium through sliding window averaging, calculate the first-order difference sequence, and establish an incremental model with water level change rate and rainfall rate as inputs and seepage pressure change rate as output.
[0035] This step specifically includes: L11. Multi-source data acquisition and alignment: Collect historical monitoring data of the dam, including reservoir water level time series. Rainfall time series And multiple piezometer observation sequences distributed at different locations on the dam body. All data is resampled and aligned according to a uniform time granularity.
[0036] L12. Physical Damping Simulation (Pre-smoothing): Considering the damping and filtering effects of the earth-rock dam medium on water pressure transmission, and that the original measurements contain high-frequency measurement noise, direct difference would amplify the noise. Therefore, a sliding window averaging process is first applied to the sequence:
[0037]
[0038] in The length of the sliding window (taken in the specific implementation of this embodiment) (hours), simulating the hysteresis smoothing characteristics of the medium.
[0039] L13. Incremental sequence calculation: Calculate the first difference of the smoothed data to construct an incremental model that reflects the "rate of change":
[0040]
[0041] Regarding rainfall Since it is itself in incremental form, it is directly used as input after smoothing. At this point, the model's objective changes from "predicting absolute seepage pressure" to "predicting the instantaneous change in seepage pressure".
[0042] L14. Effective Cleaning: By removing NaN values and infinite values generated during the difference calculation, a sample set containing only valid incremental data is obtained.
[0043] At this point, by constructing a damped incremental response model, physical smoothing and differential processing are performed on the non-stationary monitoring sequence, effectively extracting transient change characteristics.
[0044] L2. Constructing the spatiotemporal lag feature matrix: Based on the preset maximum lag time window, construct a joint matrix containing reservoir water level lag features and rainfall lag features with multiple time steps.
[0045] This step aims to construct a high-dimensional feature space capable of capturing the seepage hysteresis effect of dams, specifically including: L21. Lag window definition: Based on the physical dimensions and medium characteristics of the dam, the maximum physical lag search window for the reservoir water level is set. (48 hours is taken in the specific implementation of this embodiment) and the maximum impact window of rainfall. (24 hours are taken during the specific implementation of this embodiment).
[0046] L22. Reservoir water level lag characteristic structure: For the current moment , constructing a system containing the past The feature vector of the rate of change of water level at each time point:
[0047] L23. Characteristics of rainfall lag: Similarly, construct the rainfall feature vector:
[0048] L24. Joint matrix assembly: The aforementioned feature vectors are concatenated horizontally, and all time samples are stacked vertically to form a high-dimensional feature matrix. .
[0049] L3. Apply physical distance decay constraint: Define an exponential time decay operator to weight and scale the reservoir water level lag feature in the joint matrix, thereby introducing the time priority constraint of Darcy's law into the feature space.
[0050] This step aims to introduce prior knowledge of Darcy's law regarding time priority to solve the problem of misjudgment due to long lags, specifically including: L31. Define the exponential time decay operator: Construct a time step that varies with the time delay The weight vector increases and decreases exponentially. Its element calculation formula is:
[0051] in The attenuation coefficient is 0.05 in the specific implementation of this embodiment. Lag hours ( ).
[0052] L32. Feature-weighted scaling: For the characteristic matrix The columns with lagging reservoir water levels are weighted. Hourly water level characteristic column, multiply all its elements by :
[0053] At this point, the constraint mechanism needs to be explained: By artificially reducing the numerical amplitude of the far-lagging feature, when the subsequent regression model attempts to utilize the far-lagging feature (such as... When fitting the current seepage pressure, a very large regression coefficient must be assigned to offset the effect of the attenuation operator. This will lead to a significant increase in the model's regularization penalty term. Therefore, the optimization algorithm here will tend to prioritize the "proximal hysteresis feature" that has not been significantly attenuated, thereby forcing the inversion results to converge to the physically shortest conduction path.
[0054] At this point, in the process of constructing the high-dimensional spatiotemporal lag feature matrix, the introduction of an exponential time decay operator in feature engineering can force the model to follow the time priority principle of Darcy's law and impose physical distance constraints that conform to Darcy's law, thereby effectively eliminating long-lag misjudgments caused by the periodic fluctuation of reservoir water level and fundamentally eliminating the pseudo-correlation interference of long-lag features.
[0055] L4. Perform multi-source decoupling inversion calculation: Construct a ridge regression model, use L2 regularization term combined with physical constraints to train the incremental model, and perform decoupling calculation to obtain the reservoir water level sensitivity coefficient vector and rainfall sensitivity coefficient vector; This step utilizes statistical learning methods to separate the contributions of different driving factors, specifically including: L41. Dynamic Mask Generation: For each piezometer measuring point Check its target vector The validity of the result. Generate a binary mask. Only periods with complete data are retained for training to accommodate intermittent sensor failures.
[0056] L42. Ridge Regression Model Training: Based on the weighted feature matrix and target vector Construct a ridge regression algorithm to optimize the objective function:
[0057] in The regularization strength is given. The regression coefficient vector is solved using the least squares method. .
[0058] L43. Sensitivity Decoupling: The obtained coefficient vector Decomposed into a subset of coefficients corresponding to the reservoir water level and the corresponding subset of rainfall coefficients This achieves decoupling of multi-source responses.
[0059] By constructing an incremental model and combining it with the ridge regression algorithm, the contribution rates of rainfall and reservoir water level components to seepage pressure changes can be mathematically separated, thereby effectively solving the false alarm problem caused by "spurious correlation" at shallow measuring points during heavy rainfall.
[0060] L5. Index Calculation and Behavior Cloud Map Generation: Based on the sensitivity coefficient vector, the primary response peak is identified to determine the physical lag time, and the hydraulic response performance index (HREI) is constructed by combining the signal-to-noise ratio. The index is then mapped to the dam profile using spatial interpolation technology to generate a permeability behavior cloud map that reflects the distribution of regional seepage risk.
[0061] This step transforms mathematical parameters into engineering-readable physical metrics and visualizations, specifically including: L51. Primary peak lock-in (physical lag time determination): Calculate the absolute value sequence of reservoir water level coefficients Instead of using the global centroid method, the search algorithm identifies the first significant peak position in the sequence with a normalized energy greater than 0.6. Take the neighborhood of this peak (e.g., The centroid of the coefficient within (hours) is taken as the physical true lag time of the measuring point. .
[0062] L52. Construction of the Hydraulic Response Performance Index (HREI): The hydraulic response performance index (HREI) is used as a comprehensive evaluation index, and the formula is as follows:
[0063] In practice: Decoupling ratio: The larger the value, the more dominant the reservoir water level is, and the less it is affected by rainfall.
[0064] Time item: The larger the value, the faster the transmission and the stronger the potential penetration.
[0065] Confidence terms: To assess the model's goodness of fit, ensure that the results are statistically significant.
[0066] L53. Spatial Interpolation and Contour Mapping: Obtain the spatial coordinates of all measuring points on the dam body profile. And its corresponding HREI value. A continuous HREI scalar field is constructed using cubic spline interpolation or kriging interpolation algorithms.
[0067] L54. Visualization Output: Generate permeability cloud map: Use red and blue thermal color scales to render the HREI field. Red areas represent high seepage risk (high response, short hysteresis), and blue areas represent safety (low response or long hysteresis).
[0068] Generate friction loss curve: Extract HREI values from measuring points along the seepage path and plot the curve. If the curve shows monotonous decay, it is considered normal; if the curve rebounds (tails up) downstream, the system will automatically issue a "circumferential seepage / concentrated leakage" warning.
[0069] At this point, based on the exponential time decay operator, the independent contribution rates of reservoir water level and rainfall to seepage pressure change are mathematically separated using the multi-source decoupling inversion algorithm. Based on the primary peak locking strategy and the hydraulic response efficiency index (HREI), a permeability cloud map that intuitively reflects the distribution of hidden dangers in the deep part of the dam can be generated.
[0070] Therefore, this step constructs the Hydraulic Response Effectiveness Index (HREI) by integrating signal purity, transmission rate, and model confidence. This not only allows for the intuitive display of dominant seepage channels (high HREI zone) and seepage barriers (low HREI zone) inside the dam through the final generated permeability cloud map, but also enables the accurate quantification of dam seepage risk and the rapid location of dominant seepage channels and potential hazard areas based on the Hydraulic Response Effectiveness Index (HREI).
[0071] In summary, this method integrates prior physical knowledge with statistical learning algorithms to achieve "denoising, decoupling, localization, and visualization" of dam seepage behavior, significantly improving the accuracy and anti-interference capability of dam safety monitoring in complex environments.
[0072] In one embodiment, taking reservoir A as an example, combined with Figure 3 and Figure 5 As can be seen, this invention can clearly distinguish the seepage health status of different cross sections: exist Figure 3 In the middle section, section 1 (columns 1-1 to 1-4 on the left) shows a transition from dark blue to light blue, and the HREI value is generally low.
[0073] correspond Figure 5 The blue curve in the figure shows that the HREI index of section 1 exhibits a significant monotonic exponential decay characteristic from upstream (20m) to downstream (80m, HREI≈0).
[0074] This conforms to the laminar flow law of porous media described by Darcy's law, indicating that the seepage prevention curtain in this area is working normally, the pressure head is effectively dissipated in the dam body, and there is no risk of leakage.
[0075] Abnormal cross-sectional characteristics (cross-section 2): In Figure 3 In the middle section, the bottom of section 2 (right side 2-1 to 2-4) shows a bright dark red color (HREI>110), indicating that there is extremely strong reservoir water level connectivity at this location.
[0076] More importantly, observation Figure 5 The orange curve in the figure shows that this section not only has an extremely high response value at the upstream end (HREI≈115), but also does not decay to 0 at the downstream end (60m-80m interval) like section 1. Instead, it exhibits a "tail lift" phenomenon (rebounding from 0 to around 13). This "tail lift" phenomenon cannot be captured by traditional correlation coefficient analysis.
[0077] In summary, this invention eliminated rainfall interference through multi-source decoupling, confirming that the rebound was not caused by surface rainfall, but rather by physical signals indicating the existence of deep seepage channels or concentrated seepage zones.
[0078] In one embodiment, taking reservoir B as an example, combined with Figure 4 and Figure 6 As shown, this embodiment further demonstrates the diagnostic capabilities of the present invention in complex flow fields: like Figure 6 As shown by the pink curve, the HREI index at section 4 shows a standard decreasing trend with increasing distance (from 2.6 to 0), indicating that the area is in a normal seepage prevention state.
[0079] Inverse gradient anomaly pattern (sections 2 and 3): combined Figure 4 thermal distribution and Figure 6 The curve trend reveals that section 2 (orange curve) and section 3 (green curve) exhibit an "inverse gradient" phenomenon. That is, the HREI index increases with distance from the upstream.
[0080] This inverse gradient phenomenon, where the downstream response is stronger than the upstream response, usually indicates the presence of reverse water pressure support or lateral seepage around the dam downstream, directly driven by the reservoir water level. If only the original water level data is considered, the high downstream water level can easily mask the seepage characteristics.
[0081] In summary, this invention captures this anomalous physical field distribution through the quantification of the HREI index, thereby providing clear spatial targeting information for risk mitigation and reinforcement.
[0082] Now, combining the above examples, draw as follows: Figure 2The HREI index decay curve shown is used to illustrate and output the difference between normal sections (monotonically decaying) and abnormal sections (downstream rebound).
[0083] In another aspect, the present invention also discloses a computer-readable storage medium storing a computer program, which, when executed by a processor, causes the processor to perform the steps of the method described above.
[0084] In another aspect, the present invention also discloses a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the steps of the method described above.
[0085] In another embodiment provided in this application, a computer program product containing instructions is also provided, which, when run on a computer, causes the computer to execute any of the dam seepage behavior visualization inversion methods described in the above embodiments.
[0086] It is understood that the system provided in the embodiments of the present invention corresponds to the method provided in the embodiments of the present invention, and the explanation, examples and beneficial effects of the relevant content can be referred to the corresponding parts of the above method.
[0087] This application also provides an electronic device, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, communication interface, and memory communicate with each other via the communication bus. Memory, used to store computer programs; The processor, when executing the program stored in memory, implements the above-mentioned method for visualizing and inverting the seepage behavior of dams.
[0088] The communication bus mentioned in the above-mentioned electronic devices can be a standard bus for interconnecting peripheral components or an extended industrial standard structure bus, etc. This communication bus can be divided into address bus, data bus, control bus, etc.
[0089] The communication interface is used for communication between the aforementioned electronic devices and other devices.
[0090] The memory may include random access memory or non-volatile memory, such as at least one disk storage device. Optionally, the memory may also be at least one storage device located remotely from the aforementioned processor.
[0091] The processors mentioned above can be general-purpose processors, including central processing units, network processors, etc.; they can also be digital signal processors, application-specific integrated circuits, field-programmable gate arrays or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.
[0092] In the above embodiments, implementation can be achieved entirely or partially through software, hardware, firmware, or any combination thereof. When implemented using software, it can be implemented entirely or partially in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired or wireless means. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device such as a server or data center that integrates one or more available media. The available medium can be a magnetic medium, an optical medium, or a semiconductor medium, etc.
[0093] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
[0094] Furthermore, it should be noted that if any directional indication (such as up, down, left, right, front, back, etc.) is involved in the embodiments of the present invention, the directional indication is only used to explain the relative positional relationship and movement of each component in a specific posture. If the specific posture changes, the directional indication will also change accordingly.
[0095] Furthermore, if the embodiments of this invention involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the meaning of "and / or" throughout the text includes three parallel solutions; for example, "A and / or B" includes solution A, solution B, or a solution where both A and B are satisfied simultaneously. Furthermore, in the embodiments of this invention, "multiple" refers to two or more. Moreover, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed by this invention.
Claims
1. A method for visualizing and inverting the seepage behavior of a dam, characterized in that, include: Step S1. Obtain historical monitoring time series data of the dam, simulate the medium damping effect by means of sliding window averaging, calculate the first-order difference sequence, and establish an incremental model with water level change rate and rainfall rate as input and seepage pressure change rate as output; Step S2. Preset the maximum lag time window and construct a joint matrix containing reservoir water level lag characteristics and rainfall lag characteristics with multiple time steps; Step S3. Define an exponential time decay operator to perform weighted scaling on the reservoir water level lag characteristics in the joint matrix; Step S4. Construct a ridge regression model, use L2 regularization term combined with a set of physical constraints to train the incremental model, and decouple the calculation to obtain the reservoir water level sensitivity coefficient vector and the rainfall sensitivity coefficient vector. Step S5. Based on two sets of sensitivity coefficient vectors, identify the primary response peak to determine the physical lag time, and construct a hydraulic response performance index by combining the signal-to-noise ratio. Finally, use spatial interpolation technology to map the index to the dam profile to generate a permeability cloud map and a friction loss curve.
2. The dam seepage behavior visualization inversion method as described in claim 1, characterized in that, The incremental model construction process in step S1 specifically includes: Set the length of the sliding window Regarding time series Aligned original reservoir water level sequence Rainfall sequence and osmotic pressure sequence A moving average is used to simulate the damping effect of the dam medium and eliminate high-frequency measurement noise. The formula for this moving window averaging is as follows: in, The length of the sliding window. For smoothed timing The osmotic pressure value below, For smoothed timing The water level below, The timing for the sliding window is The original reservoir seepage pressure value below, The timing for the sliding window is The original reservoir water level value; The first-order difference of the smoothed sequence is calculated using the following formula: in, This is the seepage pressure value calculated using the first-order difference method. This is the water level value calculated using the first-order difference method. Finally, invalid data in the difference series are removed to construct a time series sample set containing only the effective rate of change as an incremental model.
3. The method for visualizing and inverting the seepage behavior of dams as described in claim 2, characterized in that, The process of constructing the joint matrix in step S2 includes: Based on the physical dimensions and medium characteristics of the dam, the maximum physical lag search window for the reservoir water level is set. and the maximum impact window of rainfall ; For the current moment Construct each containing the past The eigenvector of the rate of change of water level at each time point is: and including the past The feature vector of the rate of change of rainfall at each time step is: ; The two sets of feature vectors are concatenated horizontally, and all time samples are stacked vertically to form a joint matrix with high-dimensional features. .
4. The dam seepage behavior visualization inversion method as described in claim 3, characterized in that, The specific operation process for weighted scaling of the reservoir water level lag features in the joint matrix in step S3 includes: Construct a set of exponential time decay operators for the joint matrix. Weighted scaling exists, and the weighted scaling formula is as follows: in, For time sequence Lower hysteresis The original values of the reservoir water level characteristics at each time step For time sequence The eigenvalues after weighted scaling This is the preset physical attenuation coefficient.
5. The method for visualizing and inverting the seepage behavior of a dam as described in claim 2, characterized in that, The specific calculation process for obtaining the reservoir water level sensitivity coefficient vector and the rainfall sensitivity coefficient vector through decoupling calculation in step S4 includes: Before model training, for each measurement point, the non-empty effective time periods in its seepage pressure change rate sequence are identified, and a binary mask matrix is generated. ; Based on the weighted feature matrix and target vector The ridge regression objective function is constructed as follows: in, The regularization strength; Only the joint matrix corresponding to the valid time steps in the mask matrix is selected. The row components are used in ridge regression training to adapt to situations where the sensor experiences intermittent failures or data loss. The regression coefficient vector is solved using the least squares method during training. ; The ridge regression coefficients obtained through training are used to construct the ridge regression coefficient vector. Decomposed into a subset of coefficients corresponding to the reservoir water level and the corresponding subset of rainfall coefficients This allows us to separate the independent contribution rates of the reservoir water level component to the seepage pressure change and the independent contribution rate of the rainfall component to the seepage pressure change.
6. The method for visualizing and inverting the seepage behavior of a dam as described in claim 5, characterized in that, In step S5, a primary peak locking strategy is adopted during the determination of the physical lag time. The primary peak locking strategy specifically includes: Calculate the absolute value sequence of the reservoir water level sensitivity coefficient vector. ; Identify the set of local peaks in the sequence whose normalized energy is greater than a preset threshold; The peak with the smallest time step in the set is selected as the primary peak, and the energy centroid in the neighborhood of the primary peak is calculated. The time corresponding to the centroid is defined as the physical true lag time of the measuring point. .
7. The method for visualizing and inverting the seepage behavior of a dam as described in claim 5, characterized in that, The specific calculation formula for constructing the hydraulic response performance index in step S5 is as follows: in, The hydraulic response performance index. The multi-source decoupling ratio is determined by the total sensitivity of the reservoir water level. Overall sensitivity to rainfall The ratio is determined. This is the denominator correction coefficient for the multi-source decoupling ratio. The time efficiency factor is the physical real time lag. Inversely proportional The constant coefficient of the time term, This is the correction factor for the denominator of the time term. For model confidence factors, For model fit, This is the correction coefficient for the confidence term.
8. The method for visualizing and inverting the seepage behavior of a dam as described in claim 5, characterized in that, The specific operation process for generating the permeability behavior cloud map using spatial interpolation technology in step S5 includes: Obtain the spatial coordinates of all measuring points on the dam body profile. and its corresponding HREI value; Using Kriging interpolation or inverse distance weighted interpolation algorithms, a continuous HREI scalar field is constructed, and the HREI index of each measuring point is mapped to the two-dimensional profile coordinate system of the dam to generate a thermal distribution cloud map, in which areas with high HREI values are identified as dominant leakage channels or high-risk areas. Plot the HREI index distribution curves along the seepage path of the dam at each measuring point; If the curve shows a monotonically decreasing trend, it is judged as normal Darcy flow; If the curve shows a numerical rebound in the downstream area, it is determined that there is a risk of seepage around the source or concentrated leakage in that area.
9. A computer-readable storage medium, characterized in that, The device stores a computer program that, when executed by a processor, causes the processor to perform the steps of the method as described in any one of claims 1 to 8.
10. A computer device, characterized in that, It includes a memory and a processor, the memory storing a computer program that, when executed by the processor, causes the processor to perform the steps of the method as described in any one of claims 1 to 7.