A dam bank slope deformation trend early warning investigation method, system and early warning device

Abstract

The application discloses a dam bank slope deformation trend early warning investigation method, system and early warning device, and the method comprises the following steps: acquiring a plurality of SAR images of a target dam bank slope, constructing a three-dimensional deformation field, extracting a deformation gradient abnormal zone, identifying convex and concave deformation zones and forming a feature atlas; collecting spatiotemporal matching geological, hydrological and environmental multi-source signals, and constructing a deformation-signal correlation matrix through a dynamic time warping algorithm; respectively establishing mechanism early warning models for the deformation zones, outputting risk areas, and tracing hidden trouble causes based on the correlation matrix; the system comprises an image processing unit, a deformation analysis unit, a signal collection unit, a correlation construction unit, an early warning operation unit and a hidden trouble tracing unit, realizes full-process automatic linkage, and the early warning device comprises a collector, a processor, a memory and an early warning device, and guarantees data collection, operation storage and early warning output. The application realizes fine deformation identification, accurate risk early warning and hidden trouble tracing, and is suitable for long-term stability monitoring of various dam bank slopes.

Description

Technical Field

[0001] This invention relates to the field of dam safety monitoring technology, specifically to a method, system, and early warning device for early warning and investigation of dam bank slope deformation trends. Background Technology

[0002] As the core infrastructure of water conservancy projects, the stability of dam slopes directly determines the safety of the project and the flood control and water supply capacity of the surrounding area. With increasing service life, dam slopes are susceptible to risks such as slow creep, local instability, and even sudden landslides due to the combined effects of multiple factors, including geological changes (such as rock fissure expansion and fault activation), hydrological fluctuations (such as sudden rises and falls in reservoir water levels causing changes in the seepage field), and extreme weather disturbances (such as rainstorm infiltration leading to reduced soil shear strength). According to statistics from the Ministry of Water Resources, over 30% of dams in my country have varying degrees of slope deformation risks. Traditional monitoring methods, due to limited coverage and delayed early warning, are unable to identify hidden deformation risks in advance. Accidents in such cases can result in significant casualties and economic losses. Therefore, a technical solution capable of "large-scale monitoring, high-precision early warning, and accurate source tracing" of dam slope deformation is urgently needed to ensure the long-term safe operation of dams and prevent major water conservancy risks.

[0003] Traditional dam slope deformation monitoring technologies primarily rely on ground-based single-point monitoring, including manual inclinometers, GNSS single-point monitoring, and piezometer deployment. Manual inclinometers, which involve drilling and embedding inclinometer tubes, allow for periodic manual readings to obtain displacement data at different depths of the slope. Its advantages include high single-point measurement accuracy (down to millimeter level) and direct reflection of deep soil deformation characteristics, and it still finds application in localized monitoring of small dams. GNSS single-point monitoring, on the other hand, involves deploying GNSS receivers at key locations on the slope to acquire real-time changes in the three-dimensional coordinates of the points, providing all-weather monitoring capabilities and a data update frequency down to the minute level. However, these traditional technologies have significant limitations: First, they lack spatial coverage. Both manual inclinometers and GNSS are "point-based monitoring," acquiring data from only a limited number of points, failing to reflect the overall deformation field of the bank slope, and easily overlooking small-scale, hidden deformation areas (such as deformation caused by local cracks in the middle of the bank slope). Second, they have a strong early warning lag. Relying on manual data analysis, they struggle to capture deformation "acceleration inflection points" in real time, often only discovering risks when deformation has already reached a severe stage. Third, they suffer from a cost-efficiency imbalance. Manual inclinometers require a large amount of manpower for on-site operations and are greatly affected by weather and terrain, while GNSS equipment deployment is costly, and large-scale, dense deployment is not economically viable, making it difficult to meet the monitoring needs of the entire bank slope of large dams. Furthermore, traditional technologies mostly focus on "deformation data collection," lacking correlation analysis between deformation mechanisms and risk triggers, failing to provide accurate basis for hazard mitigation, and resulting in a high degree of blindness in the investigation work.

[0004] With the development of remote sensing technology and artificial intelligence, existing dam deformation monitoring technologies are gradually evolving towards "integrated air-space-ground" and "data-driven intelligence." On the one hand, InSAR (Synthetic Aperture Radar Interferometry) technology, with its advantages of large coverage and high spatiotemporal resolution, is widely used in dam deformation monitoring. By generating interferograms from multiple satellite SAR images, it can acquire deformation field data covering tens to hundreds of square kilometers with a spatial resolution of 10-30 meters, effectively compensating for the coverage deficiencies of traditional point-based monitoring. On the other hand, machine learning algorithms (such as LSTM and random forest) have been introduced into the field of deformation prediction, enabling advanced prediction of deformation trends by mining the correlation between deformation data and environmental and geological factors. However, existing technologies still have significant shortcomings: most InSAR applications only focus on "deformation field generation," failing to subdivide deformation characteristics based on the specific geological conditions of dam slopes (such as lithological differences and seepage effects), and thus unable to distinguish between different types of risks, such as "convex landslide risk deformation" and "concave settlement deformation." Furthermore, existing machine learning early warning models often use single deformation data as input, without integrating multi-source signals such as geological stress and groundwater seepage, leading to significant environmental interference affecting early warning accuracy (e.g., false alarms are prone to occur during periods of increased rainfall in the rainy season). In addition, existing technologies generally lack a "early warning-source tracing" closed loop, only able to identify risk areas but unable to accurately pinpoint the physical location and root cause of potential hazards, still requiring traditional methods such as large-scale drilling and manual inspections, which are inefficient and costly.

[0005] Chinese patent (CN114966685A) discloses a "method for monitoring and predicting dam deformation based on InSAR and deep learning". This technology obtains preliminary deformation information of the dam through InSAR, removes atmospheric delay phase by combining GNSS data, deploys corner reflectors in sparse areas of coherent points to improve monitoring accuracy, and finally uses LSTM network to predict future deformation. While this technology achieves integrated monitoring of InSAR and GNSS, improving the accuracy of deformation data to some extent, it suffers from several serious drawbacks: First, the monitoring focuses on the "overall deformation of the dam," without specifically designing for the critical vulnerable area of ​​the "bank slope." It fails to extract unique features of the bank slope, such as anomaly zones in deformation gradients and convex / concave deformation areas, making it difficult to meet the needs of refined risk identification. Second, the signal fusion dimension is singular, combining only GNSS data with environmental factors such as water level and temperature, without incorporating core signals directly reflecting bank slope stability, such as geological stress and strain, and groundwater seepage velocity. This results in an inability to establish a deep correlation between deformation and risk triggers, and weak anti-interference capabilities of the early warning model. Third, it lacks hazard tracing capabilities, only predicting future deformation and unable to trace back to the root causes of deformation (such as rock fissures and seepage channels) and their physical locations, thus failing to provide targeted support for subsequent treatment. Fourth, the technical process is complex and costly, requiring the additional deployment of corner reflectors and a GNSS monitoring network. This not only increases equipment and maintenance costs but is also limited by terrain (e.g., corner reflectors are difficult to deploy on steep bank slopes), hindering widespread application. Summary of the Invention

[0006] Based on the aforementioned technical problems, this application discloses a method, system, and early warning device for early warning and investigation of dam bank slope deformation trends. The method specifically includes:

[0007] Acquire multiple SAR images of the target dam bank slope;

[0008] A three-dimensional deformation field of the bank slope was constructed using dual-track InSAR technology, and deformation gradient anomaly zones were extracted by the deformation field gradient.

[0009] By identifying convex and concave deformation regions through the curvature of the deformation field, a deformation field feature map is formed;

[0010] Multi-source signal data that are spatiotemporally matched with the feature map of the deformation field are collected. The spatiotemporal similarity between the features of the deformation field and each source signal is calculated by the dynamic time warping algorithm, and the deformation-signal correlation matrix is ​​constructed.

[0011] Identify convex and concave deformation regions, and construct early warning models based on deformation mechanisms for each:

[0012] Real-time acquisition of SAR images and multi-source signal data of dam bank slopes; input of early warning models for corresponding deformation feature areas; output of risk warning areas.

[0013] By using a regional risk warning system and a constructed deformation-signal correlation matrix, the source of the associated signals can be traced back to identify the underlying causes of potential hazards.

[0014] Preferably, the construction of the three-dimensional deformation field of the bank slope specifically includes:

[0015] Multiple phases of ascending and descending orbit SAR images were acquired and preprocessed to obtain ascending and descending orbit SAR interferogram pairs for the same monitoring area. The line-of-sight phase difference of the ascending and descending orbit interferogram pairs was calculated, where the line-of-sight phase difference corresponding to the ascending orbit SAR image is denoted as . The line-of-sight phase difference corresponding to the de-orbiting SAR image is denoted as The formula for converting the line-of-sight phase difference into deformation in the corresponding direction is: , ,in This represents the deformation amount corresponding to the lifting direction. This represents the deformation amount corresponding to the direction of track descent. The wavelength of the Sentinel-1 SAR image. The radar incident angle of the up-orbit SAR image. The radar incident angle for the SAR image with reduced orbit;

[0016] Taking the dam axis direction as The horizontal direction of the axis and the vertical axis of the dam is The axis, the direction perpendicular to the horizontal plane is To calculate the deformation in each direction of the three-dimensional deformation field, we can simultaneously determine the following:

[0017]

[0018] in, for Axial deformation for Axial deformation for Axial deformation For the line-of-sight direction of the up-orbit SAR image and The included angle of the axis, For the line of sight of the SAR image with the orbit reduction The included angle of the axis, This represents the total deformation of the three-dimensional deformation field.

[0019] Preferably, the extraction of the deformation gradient anomaly band specifically involves:

[0020] Based on the constructed three-dimensional deformation field of the bank slope, the monitoring area is divided into grid cells, and the center coordinates of the grid cells are marked as... Obtain the three-dimensional deformation at the center of each mesh cell. , The gradient calculation method is used to calculate the gradient of each grid cell. axis, Deformation gradient in the axial direction , The formula is: , ;in, for Side length of axial mesh cells, for Side length of axial mesh cells, , Coordinates place Axial deformation , Coordinates place Axial deformation; calculate the overall deformation gradient for each mesh element. The formula is: The combined deformation gradient of each mesh element With average composite deformation gradient If a comparison is made, If a mesh cell forms a continuous region with at least a plurality of its adjacent mesh cells, then the current continuous region is determined to be a deformation gradient anomaly zone. This represents the gradient anomaly coefficient.

[0021] Preferably, the deformation field feature map includes the deformation rate, deformation direction, and overall deformation curvature for identifying convex and concave deformation regions;

[0022] The convex deformation region and the concave deformation region are determined by the combined deformation curvature. The division formula is: ,in , For grid cells exist - , - The curvature of the plane Then the region If the deformation in the direction of the curve is convex upwards, the current area is determined to be a convex deformation zone; otherwise, it is a concave deformation zone.

[0023] Preferably, the multi-source signal data includes geological signals, hydrological signals, and environmental signals; the geological signals are obtained by distributed optical fiber sensing to acquire stress and strain data at different depths of the bank slope; the hydrological signals are obtained by optical fiber piezometer to acquire groundwater seepage velocity and hydraulic head change data of the bank slope; and the environmental signals are obtained by meteorological station and soil moisture meter to acquire hourly rainfall and soil moisture content data.

[0024] Preferably, the construction of the deformation-signal correlation matrix is ​​specifically as follows:

[0025] The acquired multi-source signal data and deformation field feature data were preprocessed with spatiotemporal alignment to obtain time series with a length of [missing information]. Using a standardized dataset, the spatiotemporal similarity between deformed field features and source signals is calculated using the Dynamic Time Warping (DTW) algorithm. A DTW distance matrix is ​​constructed and converted into a similarity matrix for the time series of deformed field features. With arbitrary source signal time series The DTW distance formula is: ,in, These are the weight coefficients for the dynamic programming path. for The Middle The feature value at time 1, for The Middle The signal values ​​at each time point convert the DTW distance into spatiotemporal similarity. The formula is: Correlated signal pairs are selected, and a deformation-signal correlation matrix is ​​constructed with the deformation field feature type as the matrix row and the multi-source signal type as the matrix column.

[0026] Preferably, the early warning model based on deformation mechanism includes an early warning model for convex deformation zones and an early warning model for concave deformation zones;

[0027] The convex deformation zone early warning model uses stress-strain data and seepage velocity data from the associated signals as input layer variables, and applies the viscoelastic-plastic constitutive equation: Calculate the critical value of plastic deformation of rock mass, where For shear stress, Shear modulus For shear strain, The viscosity coefficient, For rock mass yield shear stress threshold, To determine the plastic hardening coefficient, a particle swarm optimization algorithm is used to optimize the model parameters, minimizing the mean square error between the predicted and actual deformation rates. The resulting value represents the deformation rate-risk correlation. As a condition for triggering an early warning, The convex warning coefficient;

[0028] The concave deformation zone early warning model uses soil moisture content and groundwater head data from the associated signals as input to calculate the pore water pressure dissipation rate, using the following formula: ,in Pore ​​water pressure, For time, The consolidation coefficient is . To achieve depth, the data, along with soil moisture content and groundwater head, are used as the input layer of a BP neural network, and the output layer is the settlement-risk index. Through training, the error between the model's predicted settlement and the measured settlement is brought within a preset threshold. The monthly increase threshold of settlement is set as the early warning trigger condition.

[0029] Preferably, the reverse tracing of the source of the associated signal to obtain the cause specifically involves:

[0030] The signal anomaly contribution is calculated by extracting the core associated signal type and similarity value corresponding to the maximum association similarity from the deformed-signal association matrix, using the following formula: ,in, The anomalous contribution of the core correlated signal, The maximum fluctuation value of the core correlation signal. For the first The correlation similarity of signals of the same class For the first The fluctuation value of the signal type For the total number of signal types, when At that time, it was determined to be the dominant cause of deformation. Deformation anomaly threshold; core associated signals include geological signals, hydrological signals and environmental signals, outputting the type of hazard inducing factors.

[0031] A dam bank slope deformation trend early warning and investigation system includes an image processing unit, a deformation analysis unit, a signal acquisition unit, a correlation construction unit, an early warning calculation unit, and a hidden danger tracing unit. Each unit is linked sequentially through a data interaction interface, specifically as follows:

[0032] The image processing unit preprocesses the acquired multi-period rise and fall SAR images of the target dam bank slope and outputs a standardized SAR image dataset, which is only transmitted to the deformation analysis unit.

[0033] The deformation analysis unit constructs a three-dimensional deformation field of the bank slope based on the standardized image output by the image processing unit and uses dual-track InSAR technology. It also extracts anomaly zones by deformation field gradient and identifies convex / concave deformation areas by deformation field curvature, forming a deformation field feature map, which is then transmitted to the signal acquisition unit and the association construction unit.

[0034] The signal acquisition unit acquires multi-source signals that are spatiotemporally matched with the feature map of the deformation field, and transmits the original signal data to the association construction unit.

[0035] The association construction unit uses a dynamic time warping algorithm to calculate the spatiotemporal similarity between the deformation field features and the multi-source signals, filters the associated signal pairs, constructs the deformation-signal association matrix, and transmits it to the early warning calculation unit and the hidden danger tracing unit.

[0036] The early warning calculation unit takes the convex / concave deformation areas identified by the deformation analysis unit, substitutes them into the early warning model based on the deformation mechanism, calculates the risk, outputs the risk, and transmits the risk early warning result to the hidden danger tracing unit.

[0037] The hidden danger tracing unit traces the source of core related signals and obtains the type of hidden danger cause by associating the association matrix of the association construction unit with the risk area of ​​the early warning calculation unit.

[0038] An early warning device includes a data collector, a processor, a memory, and an early warning unit, specifically:

[0039] The data acquisition device is used to acquire multi-phase SAR images and multi-source signal data of the target dam bank slope;

[0040] The memory is used to store SAR images and multi-source signal data acquired by the data acquisition device;

[0041] A processor is used to run executable instructions stored in memory;

[0042] The early warning device is used to receive risk warning results output by the processor and output risk and potential hazard information in the form of audible and visual alarms and a visual interface display.

[0043] Compared with the prior art, the technical solution of this application has the following technical effects:

[0044] This invention enables refined feature identification of dam slope deformation. By constructing a three-dimensional deformation field using dual-track InSAR technology and combining deformation field gradient and curvature analysis, it can accurately extract deformation gradient anomaly zones, clearly identify convex and concave deformation areas, and form a feature map containing deformation rate, direction, and comprehensive curvature. This breaks through the limitation of traditional monitoring that can only obtain single-point data, and fully presents the overall deformation pattern of the slope, providing accurate feature basis for subsequent risk assessment and effectively avoiding the omission of hidden deformation areas.

[0045] The deformation-signal correlation system constructed in this invention significantly improves the accuracy of risk correlation analysis. By using a dynamic time warping algorithm to calculate the spatiotemporal similarity between deformation field characteristics and multi-source signals from geology, hydrology, and environment, effective correlation signal pairs are selected to construct a correlation matrix, deeply binding deformation data with influencing factors. This solves the problem of traditional techniques lacking correlation between deformation mechanisms and causes, providing reliable correlation data support for subsequent early warning model training and hazard tracing, and making risk analysis more scientific.

[0046] The design of the zonal early warning model in this invention significantly improves the accuracy and targeting of early warnings. For different deformation mechanisms in convex and concave deformation zones, early warning models based on viscoelastic-plastic constitutive equations and pore water pressure dissipation formulas are constructed respectively. By combining particle swarm optimization algorithm and BP neural network to optimize model parameters, differentiated early warning trigger conditions are set to avoid the limitation of a single model being applicable to all deformation scenarios, effectively reducing the probability of false alarms and missed alarms, and providing accurate early warnings for different types of deformation risks.

[0047] This invention forms a "warning-source tracing" closed loop, significantly improving the efficiency of hazard handling. After outputting the risk warning area, it traces the source of the core related signals in reverse based on the correlation matrix, and accurately locates the type of hazard cause by combining the signal anomaly contribution formula. It does not require relying on large-scale blind investigation, providing a clear direction for subsequent handling, shortening the hazard investigation time, reducing the handling cost, and achieving efficient connection from risk identification to cause location, ensuring the timeliness and pertinence of dam bank safety handling.

[0048] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the preferred embodiments of this application are described in detail below with reference to the accompanying drawings.

[0049] The above and other objects, advantages and features of this application will become more apparent to those skilled in the art from the following detailed description of specific embodiments in conjunction with the accompanying drawings. Attached Figure Description

[0050] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In all drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts are not necessarily drawn to scale.

[0051] Based on the description of the figures and their corresponding technical content in the document, the titles of the figures are as follows:

[0052] Figure 1 A flowchart illustrating the method for early warning and investigation of dam bank slope deformation trends;

[0053] Figure 2 A schematic diagram of the architecture of the early warning model for concave deformation zones;

[0054] Figure 3 A schematic diagram of the architecture of the early warning model for convex deformation zones;

[0055] Figure 4 A unit architecture diagram of a dam bank slope deformation trend early warning and investigation system;

[0056] Figure 5 This is a schematic diagram of the components of a dam bank slope deformation trend early warning device. Detailed Implementation

[0057] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. In the following description, specific details such as specific configurations and components are provided merely to help fully understand the embodiments of this application. Therefore, those skilled in the art should understand that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of this application. In addition, for clarity and brevity, descriptions of known functions and structures are omitted in the embodiments.

[0058] It should be understood that the phrase "an embodiment" or "this embodiment" throughout the specification means that a specific feature, structure, or characteristic related to the embodiment is included in at least one embodiment of this application. Therefore, "an embodiment" or "this embodiment" appearing throughout the specification does not necessarily refer to the same embodiment. Furthermore, these specific features, structures, or characteristics can be combined in any suitable manner in one or more embodiments.

[0059] Furthermore, reference numerals and / or letters may be repeated in different examples within this application. Such repetition is for the purpose of simplification and clarity and does not in itself indicate a relationship between the various embodiments and / or settings discussed.

[0060] In this article, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can mean: A exists alone, B exists alone, and A and B exist simultaneously. The term " / and" in this article describes another type of relationship between related objects, indicating that two relationships can exist. For example, A / and B can mean: A exists alone, and A and B exist alone. In addition, the character " / " in this article generally indicates that the related objects before and after it are in an "or" relationship.

[0061] In this article, the term "at least one" is merely a description of the relationship between related objects, indicating that there can be three relationships. For example, "at least one of A and B" can mean: A exists alone, A and B exist simultaneously, or B exists alone.

[0062] It should also be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion.

[0063] Example 1

[0064] This embodiment mainly describes a method for early warning and investigation of dam bank slope deformation trends, such as... Figure 1 As shown, it includes:

[0065] Acquire multiple SAR images of the target dam bank slope;

[0066] A three-dimensional deformation field of the bank slope was constructed using dual-track InSAR technology, and deformation gradient anomaly zones were extracted by the deformation field gradient.

[0067] By identifying convex and concave deformation regions through the curvature of the deformation field, a deformation field feature map is formed;

[0068] Multi-source signal data that are spatiotemporally matched with the feature map of the deformation field are collected. The spatiotemporal similarity between the features of the deformation field and each source signal is calculated by the dynamic time warping algorithm, and the deformation-signal correlation matrix is ​​constructed.

[0069] Identify convex and concave deformation regions, and construct early warning models based on deformation mechanisms for each:

[0070] Real-time acquisition of SAR images and multi-source signal data of dam bank slopes; input of early warning models for corresponding deformation feature areas; output of risk warning areas.

[0071] By using a regional risk warning system and a constructed deformation-signal correlation matrix, the source of the associated signals can be traced back to identify the underlying causes of potential hazards.

[0072] Furthermore, the construction of the three-dimensional deformation field of the bank slope specifically includes: preprocessing the acquired Sentinel-1 ascending and descending orbit SAR images acquired over 12 consecutive months, including radiometric calibration, geometric correction and noise reduction, to obtain ascending and descending orbit SAR interferogram pairs for the same monitoring area;

[0073] Based on the phase unwrapping principle of dual-track InSAR technology, the line-of-sight (LOS) phase difference of the ascending and descending orbit interferogram pairs is calculated separately, where the line-of-sight phase difference corresponding to the ascending orbit SAR image is denoted as . The line-of-sight phase difference corresponding to the de-orbiting SAR image is denoted as The formula for converting the line-of-sight phase difference into deformation in the corresponding direction is:

[0074]

[0075]

[0076] In the formula, This represents the deformation amount (unit: mm) corresponding to the lifting direction. This represents the deformation amount (unit: mm) corresponding to the rail descent direction. The wavelength of the Sentinel-1 SAR image. The radar incident angle of the up-orbit SAR image. To determine the radar incident angle of the reduced-orbit SAR image, a three-dimensional coordinate system is established for the bank slope monitoring area. The X-axis is the direction along the dam axis, the Y-axis is the horizontal direction perpendicular to the dam axis, and the Z-axis is the direction perpendicular to the horizontal plane. The deformation quantities in each direction of the three-dimensional deformation field are obtained by simultaneously solving these systems.

[0077]

[0078] In the formula, for Axial deformation for Axial deformation for Axial deformation For the line-of-sight direction of the up-orbit SAR image and The included angle of the axis, For the line of sight of the SAR image with the orbit reduction The included angle of the axis, The total deformation of the three-dimensional deformation field is given by solving the system of equations to obtain the three-dimensional deformation field data covering the entire bank slope monitoring area.

[0079] Deformation gradient anomaly zones are extracted through deformation field gradient calculation. Specifically, based on the constructed three-dimensional deformation field of the bank slope, the monitoring area is uniformly divided into grid cells, with the center coordinates of each grid cell marked as follows: , , , They are respectively axis, The number of grid cells along the axis is determined, and the 3D deformation at the center of each grid cell is obtained. , ;

[0080] The gradient calculation method is used to calculate the gradient of each grid cell. axis, Deformation gradient in the axial direction, calculation Deformation gradient in the axial direction :

[0081]

[0082] calculate Deformation gradient in the axial direction :

[0083]

[0084] in, for Side length of axial mesh cells, for Side length of axial mesh cells, , Coordinates place Axial deformation , Coordinates place Axial deformation;

[0085] Calculate the combined deformation gradient for each mesh element. : ;

[0086] In the formula, For grid cells The overall deformation gradient at the location;

[0087] Calculate the average comprehensive deformation gradient of the entire monitoring area. The formula is: ;

[0088] The combined deformation gradient of each mesh element With average composite deformation gradient If a comparison is made, If a mesh cell forms a continuous region with at least a plurality of its adjacent mesh cells, then the current continuous region is determined to be a deformation gradient anomaly zone. This represents the gradient anomaly coefficient.

[0089] Furthermore, the deformation field feature map includes the deformation rate, deformation direction, and overall deformation curvature for identifying convex and concave deformation regions;

[0090] The convex deformation region and the concave deformation region are determined by the combined deformation curvature. The division formula is: ,in , For grid cells exist - , - The curvature of the plane Then the region If the deformation in the direction of the direction is convex upwards, the current area is determined to be a convex deformation area; otherwise, it is a concave deformation area.

[0091] Based on the three-dimensional deformation field of the bank slope, for each grid cell Calculate the curvature of the deformation field in the XZ and YZ planes respectively. The formula for the curvature of the deformation field in the XZ plane is:

[0092]

[0093] In the formula For grid cells Deformation curvature in the XZ plane, Let Z be the first derivative of the deformation in the Z direction with respect to the X direction. Let Z be the second derivative of the deformation in the Z direction with respect to the X direction.

[0094] The formula for the curvature of the deformation field of the YZ surface is:

[0095]

[0096] In the formula For grid cells Deformation curvature in the YZ plane, Let Z be the first derivative of the deformation in the Z direction with respect to the Y direction. Let be the second derivative of the deformation in the Z direction with respect to the Y direction;

[0097] Calculate the combined deformation curvature of the mesh elements The formula is: ,like If the value is greater than 0, and the Z-axis deformation in the corresponding region shows an upward convex distribution, then the region is determined to be a "convex deformation zone". If the value is less than 0 and the Z-direction deformation of the corresponding region is distributed in a downward concave manner, the region is determined to be a "concave deformation zone".

[0098] Calculate the deformation rate of each feature region The formula is: ;

[0099] In the formula , They are respectively , Time grid cell Z-axis deformation For time intervals;

[0100] Using the planar coordinates (X-axis, Y-axis) of the bank slope monitoring area as a basis, different colors are used to mark "deformation gradient anomaly zone" (red), "convex deformation zone" (orange), and "concave deformation zone" (blue), and the corresponding deformation rate is marked in each characteristic area. Deformation direction (indicated by arrows in the X / Y / Z direction of deformation synthesis) and overall deformation curvature This forms a visualized deformation field feature map.

[0101] Furthermore, the multi-source signal data includes geological signals, hydrological signals, and environmental signals; the geological signals acquire stress and strain data at different depths of the bank slope through distributed optical fiber sensing; the hydrological signals acquire groundwater seepage velocity and hydraulic head change data of the bank slope through optical fiber piezometers; and the environmental signals acquire hourly rainfall and soil moisture content data through meteorological stations and soil moisture meters.

[0102] Furthermore, the construction of the deformation-signal correlation matrix is ​​specifically as follows:

[0103] The acquired multi-source signal data and deformation field feature data were preprocessed with spatiotemporal alignment to obtain time series with a length of [missing information]. Using a standardized dataset, the spatiotemporal similarity between deformed field features and source signals is calculated using the Dynamic Time Warping (DTW) algorithm. A DTW distance matrix is ​​constructed and converted into a similarity matrix for the time series of deformed field features. With arbitrary source signal time series The DTW distance formula is: ,in, These are the weight coefficients for the dynamic programming path. for The Middle The feature value at time 1, for The Middle The signal values ​​at each time point convert the DTW distance into spatiotemporal similarity. The formula is: Correlated signal pairs are selected, and a deformation-signal correlation matrix is ​​constructed with the deformation field feature type as the matrix row and the multi-source signal type as the matrix column.

[0104] Furthermore, such as Figures 2-3 As shown, the early warning model based on deformation mechanism includes an early warning model for convex deformation zones and an early warning model for concave deformation zones;

[0105] The convex deformation zone early warning model uses stress-strain data and seepage velocity data from the associated signals as input layer variables, and applies the viscoelastic-plastic constitutive equation: Calculate the critical value of plastic deformation of rock mass, where For shear stress, Shear modulus For shear strain, The viscosity coefficient, For rock mass yield shear stress threshold, To determine the plastic hardening coefficient, a particle swarm optimization algorithm is used to optimize the model parameters, minimizing the mean square error between the predicted and actual deformation rates. The resulting value represents the deformation rate-risk correlation. As a condition for triggering an early warning, The convex warning coefficient;

[0106] The concave deformation zone early warning model uses soil moisture content and groundwater head data from the associated signals as input to calculate the pore water pressure dissipation rate, using the following formula: ,in Pore ​​water pressure, For time, The consolidation coefficient is . To achieve depth, the data, along with soil moisture content and groundwater head, are used as the input layer of a BP neural network, and the output layer is the settlement-risk index. Through training, the error between the model's predicted settlement and the measured settlement is brought within a preset threshold. The monthly increase threshold of settlement is set as the early warning trigger condition.

[0107] Furthermore, the reverse tracing of the source of the associated signal to obtain the trigger specifically involves:

[0108] The signal anomaly contribution is calculated by extracting the core associated signal type and similarity value corresponding to the maximum association similarity from the deformed-signal association matrix, using the following formula: ,in, The anomalous contribution of the core correlated signal, The maximum fluctuation value of the core correlation signal. For the first The correlation similarity of signals of the same class For the first The fluctuation value of the signal type For the total number of signal types, when At that time, it was determined to be the dominant cause of deformation. Deformation anomaly threshold; core associated signals include geological signals, hydrological signals and environmental signals, outputting the type of hidden danger cause;

[0109] The specific method for hazard location is as follows: If the core associated signal is a geological signal (stress-strain anomaly), based on the spatial coordinate mapping relationship of the distributed optical fiber sensor network, the three-dimensional coordinates of the sensor node corresponding to the abnormal stress-strain data are located, and the influence radius of the rock mass fracture is calculated using the following formula: In the formula, The radius of influence of the crack, The cumulative volume of shear strain (obtained by integrating stress-strain data). This represents the maximum strain value.

[0110] If the signal is a hydrological signal (abnormal seepage velocity), the location of the fiber optic piezometer should be considered in conjunction with the tracer transport time. The formula for calculating the flow velocity in the seepage channel is: In the formula, The actual flow velocity in the seepage channel is given by L, where L is the distance between the tracer injection point and the detection point. The difference in water head between two points is used to invert the spatial path of the channel;

[0111] If the signal is an environmental signal (soil moisture content anomaly), the diffusion coefficient of the moisture content anomaly is calculated based on the soil moisture meter coordinates, using the following formula: In the formula, The diffusion coefficient is... The change in moisture content is represented by A, where A is the area of ​​the abnormal region. For moisture content gradient, To determine the time span of abnormal water content diffusion, the boundary of surface hazards is determined by combining UAV thermal infrared imagery, and the cause type (geological / hydrological / environmental), 3D coordinates of the hazard, and parameters of its impact range are output.

[0112] This implementation details the four major technological innovations of this application: the construction of a three-dimensional deformation field through dual-track InSAR, the multi-source signal correlation matrix, the zoning mechanism early warning model, and the closed-loop hazard tracing. These innovations solve the core pain points of bank slope monitoring and realize a full-process upgrade from deformation monitoring to hazard prevention and control.

[0113] Example 2

[0114] This embodiment describes in detail a dam bank slope deformation trend early warning and investigation system, such as... Figure 4 As shown, it includes an image processing unit, a deformation analysis unit, a signal acquisition unit, a correlation construction unit, an early warning calculation unit, and a hidden danger tracing unit. Each unit forms a fully automated, mechanism-data-driven, and risk-cause closed-loop early warning and investigation chain, specifically:

[0115] The image processing unit preprocesses the acquired multi-period rise and fall SAR images of the target dam bank slope and outputs a standardized SAR image dataset, which is only transmitted to the deformation analysis unit.

[0116] The deformation analysis unit constructs a three-dimensional deformation field of the bank slope based on the standardized image output by the image processing unit and uses dual-track InSAR technology. It also extracts anomaly zones by deformation field gradient and identifies convex / concave deformation areas by deformation field curvature, forming a deformation field feature map, which is then transmitted to the signal acquisition unit and the association construction unit.

[0117] The signal acquisition unit acquires multi-source signals that are spatiotemporally matched with the feature map of the deformation field, and transmits the original signal data to the association construction unit.

[0118] The signal acquisition unit integrates core sensing modules by deploying a distributed multi-source heterogeneous sensor array, including distributed fiber optic sensors (deployed along different depths of the slope to synchronously collect stress and strain data), fiber optic piezometers (deployed at the toe of the slope and around the seepage barrier to monitor seepage velocity and water head changes in real time), and IoT soil moisture monitoring terminals (covering the slope surface in a grid pattern to collect soil moisture content and hourly rainfall). All sensors achieve spatiotemporal matching of data through timestamp synchronization technology to ensure that the collected multi-source signals and deformation field characteristics correspond accurately in the spatiotemporal dimension.

[0119] The association construction unit uses a dynamic time warping algorithm to calculate the spatiotemporal similarity between the deformation field features and the multi-source signals, filters the associated signal pairs, constructs the deformation-signal association matrix, and transmits it to the early warning calculation unit and the hidden danger tracing unit.

[0120] The early warning calculation unit takes the convex / concave deformation areas identified by the deformation analysis unit, substitutes them into the early warning model based on the deformation mechanism, calculates the risk, outputs the risk, and transmits the risk early warning result to the hidden danger tracing unit.

[0121] The hidden danger tracing unit traces the source of core related signals and obtains the type of hidden danger cause by associating the association matrix of the association construction unit with the risk area of ​​the early warning calculation unit.

[0122] This embodiment describes in detail the dam bank slope deformation trend early warning and investigation system. Through the core design of multi-unit collaboration, mechanism-driven, and closed-loop tracing, it achieves accurate identification of three-dimensional deformation fields and characteristic areas, breaks through the limitations of two-dimensional monitoring, adapts to different deformation mechanisms, accurately locates the causes, shortens the investigation time, and realizes an efficient closed loop from monitoring to treatment, significantly improving the safety control capabilities of dam bank slopes.

[0123] Example 3

[0124] This embodiment describes in detail an early warning device, such as... Figure 5As shown, the system includes a data collector, processor, memory, and early warning device, and is used for early warning and investigation of dam bank slope deformation trends. Specifically:

[0125] The data acquisition device is used to acquire multi-phase SAR images and multi-source signal data of the target dam bank slope;

[0126] The data acquisition system includes a SAR satellite data receiver, distributed fiber optic sensors, fiber optic piezometers, a meteorological monitor, and a soil moisture monitor. The SAR satellite data receiver receives multi-phase Sentinel-1 ascending and descending orbit SAR images. The distributed fiber optic sensors are deployed at different depths on the bank slope to collect stress and strain data. The fiber optic piezometers collect data on groundwater seepage velocity and hydraulic head changes. The meteorological monitor collects hourly rainfall data. The soil moisture monitor collects soil moisture content data. All collected data is transmitted to the processor.

[0127] A processor is used to execute executable instructions stored in memory.

[0128] The processor includes, but is not limited to, an industrial-grade central processing unit (CPU model: Intel Xeon Gold 6330) and a graphics processing unit (GPU model: NVIDIA A100). The central processing unit is used to perform SAR image preprocessing, dual-track InSAR three-dimensional deformation field calculation, dynamic time warping similarity analysis, and zonal early warning model parameter optimization (particle swarm optimization algorithm). The graphics processing unit is used to accelerate the visualization of the three-dimensional deformation field, early warning model calculation, and hazard location coordinate inversion. The processing results are transmitted to the data storage and early warning output device.

[0129] The memory is used to store SAR images and multi-source signal data collected by the collector, and to implement executable instructions, early warning model parameters and intermediate calculation data;

[0130] The data storage device uses a distributed solid-state drive array to store multi-period SAR image data, three-dimensional deformation field data, multi-source signal data, deformation-signal correlation matrix, early warning model parameters and calculation results. It supports data backup and historical query and can respond to read and write commands from the data processor.

[0131] The early warning device is used to receive the risk warning results output by the processor and output the risk and potential hazard information in the form of audible and visual alarms and visual interface display.

[0132] The early warning device includes a touch screen, an audible and visual alarm, and a wireless communication device. The touch screen is used to display the deformation field characteristic map, risk level, and location of potential hazards. The audible and visual alarm triggers the corresponding alarm signal according to the risk level. The wireless communication device is used to push early warning information and hazard data to the emergency command system, and can receive parameter adjustment instructions input by the user through the touch screen and feed them back to the data processor.

[0133] The above are merely preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. For those skilled in the art, the present invention can have various modifications and variations. Any changes, modifications, substitutions, integrations, and parameter changes made to these embodiments within the spirit and principles of the present invention, without departing from the principles and spirit of the present invention, through conventional substitutions or to achieve the same function, fall within the scope of protection of the present invention.

Claims

1. A dam bank slope deformation trend early warning and investigation method, characterized in that, The application relates to a dam slope risk early warning method based on SAR image and multi-source signal. The method comprises the following steps: SAR images of a target dam slope in multiple periods are acquired; a three-dimensional deformation field of the dam slope is constructed by adopting a double-track InSAR technology, and a deformation gradient anomaly zone is extracted by using the deformation field gradient; a convex deformation area and a concave deformation area are identified by using the deformation field curvature, and a deformation field feature map is formed; multi-source signal data which are matched with the deformation field feature map in time and space are collected, the time and space similarity between the deformation field features and the multi-source signals is calculated by using a dynamic time warping algorithm, and a deformation-signal correlation matrix is constructed; the convex deformation area and the concave deformation area are identified, and early warning models based on deformation mechanisms are respectively constructed; SAR images and multi-source signal data of the dam slope are acquired in real time, the early warning models of the corresponding deformation characteristic areas are input, and the areas of risk early warning are output; the source of the associated signal is traced back based on the constructed deformation-signal correlation matrix, and hidden trouble inducements are acquired; The multi-period ascending / descending track SAR images are preprocessed to obtain ascending / descending track SAR interferogram pairs of the same monitoring area, and the line-of-sight phase differences of the ascending / descending track interferogram pairs are calculated, wherein the line-of-sight phase difference corresponding to the ascending track SAR image is denoted as , and the line-of-sight phase difference corresponding to the descending track SAR image is denoted as The line-of-sight phase difference is converted into deformation in the corresponding direction, and the formula is as follows: , , wherein is the deformation in the ascending track direction, is the deformation in the descending track direction, is the wavelength of the Sentinel-1 SAR image, is the radar incidence angle of the ascending track SAR image, is the radar incidence angle of the descending track SAR image. The direction of the dam axis is taken as the x-axis The horizontal direction perpendicular to the dam axis is taken as the y-axis The direction perpendicular to the horizontal plane is taken as the z-axis The direction of the dam axis is taken as the x-axis, and the horizontal direction perpendicular to the dam axis is taken as the y-axis. wherein, is the axial deformation, is the axial deformation, is the axial deformation, is the angle between the line of sight of the ascending SAR image and the axis, is the angle between the line of sight of the descending SAR image and the axis, is the total deformation of the three-dimensional deformation field.

2. The dam bank slope deformation trend early warning and investigation method according to claim 1, characterized in that, the three-dimensional deformation field of the dam slope comprises the following steps: Based on the constructed three-dimensional deformation field of the bank slope, the monitoring area is divided into grid cells, and the center coordinates of the grid cells are marked as... Obtain the three-dimensional deformation at the center of each mesh cell. , The gradient calculation method is used to calculate the gradient of each grid cell. axis, Deformation gradient in the axial direction , The formula is: , ;in, for Side length of axial mesh cells, for Side length of axial mesh cells, , Coordinates place Axial deformation , Coordinates place Axial deformation; calculate the overall deformation gradient for each mesh element. The formula is: The combined deformation gradient of each mesh element With average composite deformation gradient If a comparison is made, If a mesh cell forms a continuous region with at least a plurality of its adjacent mesh cells, then the current continuous region is determined to be a deformation gradient anomaly zone. This represents the gradient anomaly coefficient.

3. The dam bank slope deformation trend early warning and investigation method according to claim 1, characterized in that, the deformation gradient anomaly zone is extracted, and the deformation gradient anomaly zone is extracted by adopting the following steps: The convex deformation region and the concave deformation region are combined by the deformation curvature The formula is: Wherein , The grid unit In - , - The change curvature of the plane, Then the region The deformation amount of the convex region is distributed upward, and the current region is determined as a convex deformation region, and vice versa.

4. The dam bank slope deformation trend early warning and investigation method according to claim 1, characterized in that, the deformation field feature map comprises deformation rates, deformation directions and comprehensive deformation curvatures of the convex deformation area and the concave deformation area which are identified; 5. The dam bank slope deformation trend early warning and investigation method according to claim 1 or 4, characterized in that, the multi-source signal data comprise geological signals, hydrological signals and environmental signals; the geological signals are stress and strain data of the dam slope in different depths which are acquired by using a distributed optical fiber sensor; the hydrological signals are underground water seepage velocity and water head change data of the dam slope which are acquired by using an optical fiber osmometer; and the environmental signals are hourly rainfall and soil moisture content data which are acquired by using a weather station and a soil moisture meter. The acquired multi-source signal data and deformation field feature data were preprocessed with spatiotemporal alignment to obtain time series with a length of [missing information]. Using a standardized dataset, the spatiotemporal similarity between deformed field features and source signals is calculated using the Dynamic Time Warping (DTW) algorithm. A DTW distance matrix is ​​constructed and converted into a similarity matrix for the time series of deformed field features. With arbitrary source signal time series The DTW distance formula is: ,in, These are the weight coefficients for the dynamic programming path. for The Middle The feature value at time 1, for The Middle The signal values ​​at each time point convert the DTW distance into spatiotemporal similarity. The formula is: Correlated signal pairs are selected, and a deformation-signal correlation matrix is ​​constructed with the deformation field feature type as the matrix row and the multi-source signal type as the matrix column.

6. The dam bank slope deformation trend early warning and investigation method according to claim 1, characterized in that, the deformation-signal correlation matrix is constructed by adopting the following steps: The convex deformation zone early warning model uses stress-strain data and seepage velocity data from the associated signals as input layer variables, and applies the viscoelastic-plastic constitutive equation: Calculate the critical value of plastic deformation of rock mass, where For shear stress, Shear modulus For shear strain, The viscosity coefficient, For rock mass yield stress threshold, To determine the plastic hardening coefficient, a particle swarm optimization algorithm is used to optimize the model parameters, minimizing the mean square error between the predicted and actual deformation rates. The resulting value represents the deformation rate-risk correlation. As a condition for triggering an early warning, The convex warning coefficient; The concave deformation area early warning model takes the soil moisture content and groundwater head data in the correlation signal as input, calculates the pore water pressure dissipation rate, and the formula is: wherein is the pore water pressure, is the time, is the consolidation coefficient, is the depth, and the soil moisture content and groundwater head data are taken as the input layer of the BP neural network, and the output layer is the settlement amount-risk index. Through training, the error between the iterative predicted settlement amount and the measured settlement amount reaches within the preset threshold, and the monthly increment threshold of the settlement amount is set as the early warning trigger condition.

7. The dam bank slope deformation trend early warning and investigation method according to claim 1, characterized in that, the early warning models based on the deformation mechanisms comprise a convex deformation area early warning model and a concave deformation area early warning model; The core associated signal type corresponding to the maximum correlation similarity and the similarity value are extracted from the deformation-signal correlation matrix to calculate the signal abnormal contribution degree, and the formula is: wherein, is the abnormal contribution degree of the core associated signal, is the maximum fluctuation value of the core associated signal, is the correlation similarity of the first signal, is the fluctuation value of the first signal, is the total number of signals, and when , it is determined that the deformation is the dominant cause, is the deformation anomaly threshold value; the core associated signal includes geological signals, hydrological signals and environmental signals, and the output hazard cause type.

8. A dam bank slope deformation trend early warning and investigation system suitable for the dam bank slope deformation trend early warning and investigation method of any one of claims 1-7, characterized in that, the source of the associated signal is traced back to acquire the inducements by adopting the following steps: The method comprises an image processing unit, a deformation analysis unit, a signal collection unit, a correlation construction unit, an early warning operation unit and a hidden trouble tracing unit, and the units are sequentially linked through a data interaction interface, and the units are specifically as follows: the image processing unit pre-processes the acquired multi-period ascending and descending track SAR images of the target dam slope, outputs a standardized SAR image data set, and only transmits the processed data to the deformation analysis unit; the deformation analysis unit constructs a three-dimensional deformation field of the dam slope based on the standardized image output by the image processing unit by adopting a double-track InSAR technology, extracts an anomaly zone by using the deformation field gradient, identifies a convex deformation area and a concave deformation area by using the deformation field curvature, forms a deformation field feature map, and transmits the deformation field feature map to the signal collection unit and the correlation construction unit; the signal collection unit collects multi-source signals which are matched with the deformation field feature map in time and space, and transmits original signal data to the correlation construction unit; the correlation construction unit calculates the time and space similarity between the deformation field features and the multi-source signals by using a dynamic time warping algorithm, screens associated signal pairs, constructs a deformation-signal correlation matrix, and transmits the deformation-signal correlation matrix to the early warning operation unit and the hidden trouble tracing unit; the early warning operation unit respectively inputs the convex deformation area and the concave deformation area which are identified by the deformation analysis unit into the early warning models based on the deformation mechanisms, calculates risks, and transmits the risk early warning results to the hidden trouble tracing unit. The hidden danger tracing unit traces the source of the core associated signal by associating the association matrix of the association construction unit with the risk area of the early warning operation unit, and obtains the hidden danger cause type.

9. A warning device, characterized in that The device comprises a collector, a processor, a memory and an early warning device, and is used to execute the dam bank slope deformation trend early warning and investigation method according to any one of claims 1-7, specifically comprising: The collector is used to collect multi-period SAR images and multi-source signal data of the target dam bank slope. The memory is used to store the SAR images and multi-source signal data collected by the collector, executable instructions for realizing the method, early warning model parameters and operation intermediate data. The processor is used to run the executable instructions stored in the memory, and realize the dam bank slope deformation trend early warning and investigation method according to any one of claims 1-7. The early warning device is used to receive the risk early warning result output by the processor, and output the risk and hidden danger cause information in the form of sound and light alarm and visual interface display.

Images