Dynamic fusion and intelligent early warning method and system for multi-source time sequence monitoring data of slope

Abstract

The application discloses a slope multi-source time sequence monitoring data dynamic fusion and intelligent early warning method and system, relates to the technical field of geological monitoring and early warning, and comprises the following steps: synchronously acquiring displacement vector data and stress tensor data of a region to be monitored, mapping the displacement vector data and the stress tensor data to a pre-constructed grid element based on a preset fusion weight; calculating a second-order strain gradient tensor field of the grid element, and extracting a rotation characteristic value representing a non-coordinated rotation degree; mapping and correlating the rotation characteristic value to the stress tensor data as a geometric constraint factor, calculating a real-time energy dissipation rate in the grid element, and identifying a strain localization strip by identifying the transfer characteristics of the real-time energy dissipation rate between adjacent grid elements; the beneficial effects are that the nonlinear mutation critical point of slope instability can be accurately identified, and the pain point of early warning lag in the early stage of disaster incubation in the prior art is effectively solved.

Description

Technical Field

[0001] This invention relates to the field of geological monitoring and early warning technology, and in particular to a method and system for dynamic fusion and intelligent early warning of multi-source time-series monitoring data of slopes. Background Technology

[0002] Slope instability, a serious engineering geological hazard, has an extremely complex evolution process, typically accompanied by the redistribution of stress fields within the soil and rock mass and the discontinuous evolution of macroscopic deformation fields. To effectively monitor slope conditions, existing technologies usually employ sensors such as displacement gauges, strain gauges, and pressure cells to acquire multi-source monitoring data, combined with numerical simulations or empirical models for stability evaluation. Real-time tracking of displacement vectors and stress states can provide disaster early warning support for engineering management departments, thereby reducing casualties and property losses.

[0003] However, existing monitoring and early warning systems rely heavily on empirical judgments of the absolute values ​​or rates of change of physical quantities. Their physical mechanisms are often based on the assumption of uniform deformation of continuous media. This model is difficult to accurately identify the nonlinear abrupt critical characteristics of the transformation of soil and rock masses from steady-state deformation to catastrophic instability. In particular, when facing the micromechanical behavior of the slip surface evolving from local initiation to global penetration, there is a lack of effective discrimination criteria and real-time feedback mechanisms. Because it is impossible to accurately define the instantaneous turning point of the change in the mechanical properties of the internal structure of the slope, existing technologies often have a lag in the timing of early warnings, making it difficult to provide reliable decision-making basis in the early stages of disaster incubation. Moreover, they are easily affected by environmental noise or non-physical fluctuations in the monitoring data, making it difficult to balance the robustness and timeliness of the early warning system. Summary of the Invention

[0004] In view of the above-mentioned state of the prior art, this application is hereby filed. Embodiments of this application provide a method and system for dynamic fusion and intelligent early warning of multi-source time-series monitoring data for slopes, which can accurately identify the nonlinear abrupt change critical point of slope instability, effectively solving the problem of delayed early warning in the early stages of disaster development in existing technologies.

[0005] According to one aspect of this application, a method for dynamic fusion and intelligent early warning of multi-source time-series monitoring data of slopes is provided, including:

[0006] Simultaneously acquire displacement vector data and stress tensor data of the area to be monitored, and map the displacement vector data and stress tensor data to pre-constructed mesh cells based on preset fusion weights;

[0007] Calculate the second-order strain gradient tensor field of the mesh element and extract the curl eigenvalues ​​characterizing the degree of non-coordinated rotation;

[0008] The curl eigenvalues ​​are used as geometric constraint factors to map and associate with the stress tensor data, the real-time energy dissipation rate within the grid cell is calculated, and strain localization stripes are identified by recognizing the transfer characteristics of the real-time energy dissipation rate between adjacent grid cells.

[0009] The stability of the strain localization strip is determined based on the acoustic tensor determinant. When the curl characteristic value exceeds the preset variation threshold and the calculated value of the acoustic tensor determinant decreases to the preset extreme value range, it is determined that the slope has a potential slip surface, and the bifurcation point of the potential slip surface initiation is captured.

[0010] The identification result is output based on the spatial connectivity trajectory of the bifurcation point;

[0011] The bifurcation points and their coordinates are fed back in real time to the mesh cell construction process. The topological connections of the mesh cells are dynamically reorganized using these coordinates, and the sampling frequency for acquiring the displacement vector data and stress tensor data in the next sampling cycle is adjusted synchronously.

[0012] Based on the identification results, the non-physical fluctuations of the stress tensor data in the current sampling period are corrected to obtain the mechanical compatibility deviation value. The fusion weight of the next sampling period is then adjusted based on the mechanical compatibility deviation value.

[0013] According to another aspect of this application, a dynamic fusion and intelligent early warning system for multi-source time-series monitoring data of slopes is provided, including:

[0014] Monitoring data mapping module: used to synchronously acquire displacement vector data and stress tensor data of the area to be monitored, and map the displacement vector data and stress tensor data to pre-constructed mesh cells based on preset fusion weights;

[0015] Feature extraction module: used to calculate the second-order strain gradient tensor field of the mesh element and extract curl feature values ​​that characterize the degree of non-coordinated rotation;

[0016] Localized stripe identification module: used to map and associate the curl feature value as a geometric constraint factor with the stress tensor data, calculate the real-time energy dissipation rate within the grid cell, and identify strain localized stripes by recognizing the transfer characteristics of the real-time energy dissipation rate between adjacent grid cells;

[0017] Bifurcation Capture Module: Used to determine the stability of the strain localization strip based on the acoustic tensor determinant. When the curl characteristic value exceeds the preset variation threshold and the calculated value of the acoustic tensor determinant decreases to the preset extreme value range, it is determined that the slope has a potential slip surface and the bifurcation point of the potential slip surface is captured.

[0018] Evolution output module: used to output recognition results based on the spatial connectivity trajectory of the bifurcation point;

[0019] Feedback adjustment module: used to feed back the bifurcation point and its position coordinates in real time to the mesh cell construction process, dynamically reorganize the topological connection relationship of the mesh cells using the position coordinates, and synchronously adjust the sampling frequency of the displacement vector data and stress tensor data to be acquired in the next sampling cycle; and

[0020] Based on the identification results, the non-physical fluctuations of the stress tensor data in the current sampling period are corrected to obtain the mechanical compatibility deviation value. The fusion weight of the next sampling period is then adjusted based on the mechanical compatibility deviation value.

[0021] According to another aspect of this application, an electronic device is provided, including a memory and a processor, the memory being used to store computer-executable instructions, and the processor being used to execute the computer-executable instructions, which, when executed by the processor, implement the steps of the method described above.

[0022] According to another aspect of this application, a computer storage medium is provided that stores computer-executable instructions thereon, which, when executed by a processor, implement the steps of the method described above.

[0023] Compared with existing technologies, the slope multi-source time-series monitoring data dynamic fusion and intelligent early warning method and system according to the embodiments of this application, by establishing a constitutive instability judgment criterion for the slip surface initiation stage, can sensitively capture the initiation signals of non-coordinated rotation and strain localization inside the rock and soil mass, thereby accurately identifying the nonlinear abrupt change critical point (bifurcation point) of slope instability, effectively solving the pain point of delayed early warning in the early stage of disaster in the existing technology; it can effectively eliminate false fluctuations caused by environmental noise and sensor anomalies, ensuring the energy conservation and logical consistency of the displacement field and stress field within the grid cell, and greatly improving the physical reliability of the multi-source monitoring data fusion;

[0024] By feeding back the location coordinates of the bifurcation point to the grid construction stage in real time, the topological connection relationship of the key stress area can be dynamically reorganized and the sampling frequency can be adjusted synchronously. This allows the system to maintain low energy consumption during the slope stabilization period and quickly lock key locations and improve data capture timeliness during the slip surface initiation period. While ensuring monitoring accuracy, this significantly improves the real-time response speed and computational efficiency of the early warning system. Attached Figure Description

[0025] The above and other objects, features, and advantages of this application will become more apparent from the more detailed description of the embodiments of this application in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of this application and form part of the specification. They are used together with the embodiments of this application to explain this application and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same components or steps.

[0026] Figure 1 This is a schematic diagram of the overall process of the dynamic fusion and intelligent early warning method for multi-source time-series monitoring data of slopes according to the present invention.

[0027] Figure 2 This is a schematic diagram illustrating the sampling frequency adjustment of the dynamic fusion and intelligent early warning method for multi-source time-series monitoring data of slopes according to the present invention.

[0028] Figure 3 This is a schematic diagram of the bifurcation point capture process of the dynamic fusion and intelligent early warning method for multi-source temporal monitoring data of slopes according to the present invention. Detailed Implementation

[0029] Hereinafter, exemplary embodiments according to this application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments of this application. It should be understood that this application is not limited to the exemplary embodiments described herein.

[0030] Example 1:

[0031] Reference Figures 1-3 As an embodiment of the present invention, a method for dynamic fusion and intelligent early warning of multi-source time-series monitoring data of slopes is provided:

[0032] Figure 1 The illustration shows a method for dynamic fusion and intelligent early warning of multi-source time-series monitoring data for slopes according to an embodiment of this application, including:

[0033] Simultaneously acquire displacement vector data and stress tensor data of the area to be monitored, and map the displacement vector data and stress tensor data to pre-constructed mesh cells based on preset fusion weights.

[0034] In this embodiment, the system first synchronously collects two types of time-series monitoring data using multiple sensors deployed in the slope monitoring area: one type is displacement vector data, typically acquired by GNSS receivers, total stations, or distributed fiber optic strain sensors, used to describe the three-dimensional displacement changes at various monitoring points on or inside the slope surface; the other type is stress tensor data, typically obtained by embedded earth pressure cells, stress gauges, or acoustic emission inversion methods, used to characterize the stress state and its spatiotemporal evolution within the soil and rock mass. To achieve spatiotemporal alignment and unified representation of these two types of heterogeneous data, the system pre-constructs a two-dimensional or three-dimensional grid cell system covering the entire monitoring area. The system can use regular structured grids (such as rectangular or hexahedral units) or adaptive unstructured grids based on slope topography and geological structure characteristics. Each grid unit has a unique spatial identifier and topological connectivity. During the mapping of monitoring data to grid units, the system introduces a fusion weighting mechanism to coordinate the spatial inconsistencies between displacement and stress fields. Specifically, for any grid unit e, assuming it contains m displacement monitoring points and n stress monitoring points, the system assigns weighting coefficients to displacement vector data and stress tensor data based on factors such as the distance between the monitoring points and the unit centroid, sensor accuracy confidence level, and historical data stability. and ,satisfy ;

[0035] Specifically, the data mapping process is performed as follows:

[0036] Displacement vector data mapping: Mapping the displacement vectors of all displacement monitoring points within grid cell e. By performing Kriging interpolation or inverse distance weighted averaging based on its spatial location, the representative displacement vector of the grid cell can be obtained. Then multiply by the displacement fusion weight The weighted displacement contribution value is obtained using the following formula:

[0037] ;

[0038] Stress tensor data mapping: Stress tensors measured at all stress monitoring points within mesh element e. By performing volume averaging or stress tensor interpolation, the average stress tensor of the mesh element can be obtained. Then multiply by the stress fusion weight The weighted stress contribution value is obtained using the following formula:

[0039] ;

[0040] Finally, the displacement-stress state within mesh element e after fusion is expressed by the following coupling form:

[0041] ;

[0042] in, ⊕ represents the displacement-stress state after fusion, and ⊕ represents the tensor superposition operation of the two types of data at the element level. Specifically, it means that the displacement vector and stress tensor are stored together in the attribute set of the mesh element according to the preset data structure.

[0043] The second-order strain gradient tensor field of the mesh element is calculated, and the curl eigenvalues ​​characterizing the degree of non-coordinated rotation are extracted.

[0044] In this embodiment, after mapping displacement vector data and stress tensor data to mesh elements, the system calculates the second-order strain gradient tensor field of each mesh element e based on the fused displacement vector data within that mesh element e, and extracts the curl characteristic value representing the degree of non-cooperative rotation of deformation in that region. Specifically, for any mesh element e, the first-order strain tensor of the region represented by that mesh element is first calculated using finite element shape function interpolation or finite difference methods based on its nodal displacements or internal displacement field distribution. This process belongs to the conventional strain analysis technique in continuum mechanics, and its specific formulas will not be elaborated here; the second-order strain gradient tensor field is derived from the above-mentioned first-order strain tensor. The spatial gradient is obtained, and the specific formula is as follows:

[0045] ;

[0046] in, The second-order strain gradient tensor field is a third-order tensor, whose components are denoted as . , The spatial gradient operator represents the partial derivatives of each component of the first-order strain tensor along the spatial coordinate directions. For tensor product operations;

[0047] Second-order strain gradient tensor field The component form of the expression is:

[0048] ;

[0049] in, For the first-order strain tensor Specific components, Indicates the direction of the k-th spatial coordinate. This indicates the specific component. Along the space Rate of change of direction;

[0050] Construct a second-order tensor Its weight Defined by the following formula:

[0051] ;

[0052] in, All are three-dimensional Levi-Civita symbols (also known as permutation tensors). Represents spatial coordinates Find the partial derivatives;

[0053] For second-order tensors Perform eigenvalue decomposition:

[0054] ; ;

[0055] in, For second-order tensors The p-th eigenvector represents a specific spatial direction. It corresponds to the eigenvector. eigenvalues;

[0056] Finally, curl eigenvalues The value chosen is the one with the largest absolute value among the three eigenvalues ​​mentioned above.

[0057] ;

[0058] This curl characteristic value It is a non-negative scalar whose magnitude directly and quantitatively characterizes the degree of uncoordinated rotation of the soil and rock mass at the location of the grid cell, as revealed by the strain gradient field.

[0059] By mapping and associating curl eigenvalues ​​with stress tensor data as geometric constraint factors, the real-time energy dissipation rate within the grid cell is calculated.

[0060] This association is achieved through a scalar geometric constraint mapping function, which uses curl eigenvalues. As input, output a scalar coefficient for correcting stress, and the components of the geometrically corrected stress tensor. It is given by the following formula:

[0061] ;

[0062] in, Original stress tensor data at mesh element e The amount, For geometric constraint mapping functions, , For reference curl eigenvalues, The value of can be determined based on historical data analysis or typical constitutive relations of the monitored slope soil and rock materials. It represents a critical level of curl characteristic that can cause a significant change in the mechanical response of the material. This indicates the relative strength of the current degree of uncoordinated rotation relative to the reference critical level;

[0063] Components of the geometrically corrected stress tensor and the strain rate tensor of the mesh element e The real-time energy dissipation rate of this grid cell is calculated using the following formula:

[0064] ;

[0065] in, For real-time energy dissipation rate, The rate of change of the strain component with time, " indicates the double dot product operation of a tensor;

[0066] Furthermore, strain localization stripes are identified by recognizing the transfer characteristics of real-time energy dissipation rates between adjacent grid cells. Specifically:

[0067] First, calculate the energy transfer intensity from grid cell e to any of its adjacent grid cells n, using the following formula:

[0068] ;

[0069] in, This represents the rate of change of energy dissipation from element e to element n in the spatial direction, i.e., the energy transfer intensity. and These are the real-time energy dissipation rates of grid cell e and its adjacent grid cell n, respectively. This refers to the spatial distance between the centroids of two grid cells.

[0070] Traversing all mesh elements, the following criteria are used to identify the nascent regions of strain localization stripes: If a mesh element e satisfies its own real-time energy dissipation rate... The energy transfer intensity is significantly higher than the average value of the entire monitoring area (e.g., more than 1.5 times the average value), and there are at least two adjacent grid cells (n1, n2) in different directions, such that the energy transfer intensity pointing to the grid cell e is significantly higher than the average value of the entire monitoring area (e.g., more than 1.5 times the average value). and If the value is significantly positive (indicating that energy is converging towards this unit from multiple directions), then the grid unit e is marked as a candidate unit for strain localization stripe. The system then clusters and connects all spatially continuous or approximately linearly arranged candidate units for strain localization stripe to form the identified strain localization stripe. This strain localization stripe physically characterizes the spatial trajectory of severe energy dissipation and continuous energy flow within the slope, and is a clear indicator of potential slip surface incubation.

[0071] Stability is determined for strain localization strips based on acoustic tensor determinant. When the curl eigenvalue exceeds the preset variation threshold and the calculated value of acoustic tensor determinant decreases to the preset extreme value range, it is determined that a potential slip surface has been generated on the slope, and the bifurcation point of the potential slip surface initiation is captured.

[0072] In this embodiment, the preset variation threshold is dynamically determined based on the statistical distribution of curl characteristic values ​​during the stable operation phase of the slope. Specifically, the system continuously acquires historical monitoring sequences of the slope in a quasi-static equilibrium state, extracts the curl characteristic values ​​of each grid unit within N consecutive sampling periods, and calculates their statistical mean and standard deviation. The preset variation threshold is set as the sum of the statistical mean and k times the standard deviation, where the coefficient k is selected between 2 and 3 according to the safety level of the slope. The physical meaning of this threshold is to define the normal disturbance boundary of the non-coordinated rotation magnitude. When the real-time curl characteristic value exceeds this boundary, it is determined that a local rotation variation exceeding the statistical noise range has occurred inside the material.

[0073] The preset extreme value range is determined based on the critical criterion of material constitutive stability. First, the system obtains the initial value of the acoustic tensor determinant of the mesh element in its initial undisturbed state, as a benchmark for material integrity. According to the Hill stability criterion in continuum mechanics, when the acoustic tensor determinant decreases from a positive value to zero, it means that the wave propagation speed disappears and the material becomes unstable and bifurcates. Therefore, the lower limit of the preset extreme value range is set to 0, and the upper limit is set to the initial benchmark value. times ( (This is a scaling factor, with a value between 0.01 and 0.05). The physical meaning of this range is to lock in the critical state where the material's resistance stiffness is almost lost. When the value of the acoustic tensor determinant falls into this range, it is determined that a potential slip surface with qualitative changes in mechanical properties has formed inside the slope.

[0074] Figure 3 This is a schematic diagram of the bifurcation point capture process in the dynamic fusion and intelligent early warning method for multi-source time-series monitoring data of slopes according to the present invention; it should be noted that the capture of bifurcation points includes:

[0075] The first-order strain tensor is obtained by performing spatial partial derivative calculations on the displacement vector data within the mesh element. Based on stress tensor data and first-order strain tensor, the second-order work increment within the mesh element is calculated in real time. The specific formula is as follows:

[0076] ;

[0077] in, The second-order work increment within the grid cell. The stress tensor data of mesh element e within the current sampling period. The increment relative to the previous sampling period, The first-order strain tensor of the mesh element e within the same sampling period The increment;

[0078] Based on the numerical evolution of the acoustic tensor determinant, and combined with the instantaneous extreme point where the second-order work increment changes from positive to negative, the constitutive relation of the grid cell is determined to be unstable and bifurcated.

[0079] It should be further explained that the determination of unstable forking includes:

[0080] Based on stress tensor data and the first-order strain tensor, the tangent stiffness matrix of the mesh element is reconstructed in real time. Specifically, to determine constitutive instability, the system reconstructs the tangent stiffness matrix of mesh element e in real time. This tangent stiffness matrix describes the material stiffness characteristics under the current stress-strain state, and its components are derived from the stress tensor data. and first-order strain tensor Determining the tangent modulus through the constitutive model (such as an elastoplastic model) is a routine calculation for those skilled in the art and will not be elaborated here.

[0081] Eigenvalue decomposition is performed on the tangent stiffness matrix to obtain the minimum eigenvalue that characterizes the material's resistance attenuation state. Specifically, eigenvalue decomposition is performed on the tangent stiffness matrix to obtain all its eigenvalues, and the minimum eigenvalue is found. This minimum eigenvalue characterizes the material's resistance attenuation state in the current loading direction; its value is close to zero or negative, which means that the material loses stiffness in that direction, and is a mathematical representation of instability.

[0082] The evolution trajectory of the minimum eigenvalue is monitored in real time. When the value of the minimum eigenvalue decreases from a positive value to a zero or negative value, and the derivative of the second-order work increment with respect to time changes from a negative value to a zero value, the constitutive relation of the grid cell is determined to be unstable and bifurcerated.

[0083] Specifically, the evolution of two key indicators is monitored simultaneously: the acoustic tensor is a matrix constructed based on the tangent stiffness matrix and the normal vector of the current strain localization strip, and the value of its determinant decreasing to zero or negative is a classic criterion for the possible occurrence of strain localization; the change of the second-order work increment from positive to negative means that the material begins to absorb rather than dissipate energy, which is a sign of dynamic instability.

[0084] The criteria for determining instability and bifurcation are as follows: at grid cell e, the following three conditions must be met simultaneously: ① the calculated value of the acoustic tensor determinant decreases to a preset extreme value range (e.g., close to zero or less than zero); ② the derivative of the second-order work increment with respect to time changes from negative to zero (i.e., the turning point where the energy absorption rate changes from increasing to stable); ③ the value of the minimum eigenvalue of the tangent stiffness matrix decreases from positive to zero or negative.

[0085] When a grid cell is detected to simultaneously meet the above three conditions for the first time in real time, it is determined that the constitutive relation of the grid cell has become unstable and bifurcated, and this moment is recorded as the initial moment of the unstable bifurcation.

[0086] The initial moment of the instability bifurcation is determined as the bifurcation point initiation moment, and the spatial location where the curl feature value corresponding to the initial moment exceeds the preset variation threshold is extracted to capture the bifurcation point of potential slip surface initiation.

[0087] Specifically, the system extracts the curl eigenvalue at the initial moment. The spatial coordinates of all grid cells exceeding the preset variation threshold are the bifurcation points where potential slip surfaces initiate.

[0088] The recognition result is output based on the spatial connectivity trajectory of the bifurcation point, and the specific implementation is as follows:

[0089] First, record all captured bifurcation points according to the centroid coordinates of their corresponding grid cells, forming a set of bifurcation point spatial coordinates. Where M is the total number of branching points. Let be the three-dimensional coordinates of the k-th bifurcation point, derived from the centroid of the grid cell where the bifurcation occurred. Since bifurcation points originate from discrete grid cells, the system determines the spatial connectivity between bifurcation points based on the topological connections of the grid cells pre-constructed in the monitoring area. Specifically, if the grid cells corresponding to two bifurcation points are adjacent in the grid topology (i.e., sharing a face, edge, or corner), then these two bifurcation points are considered directly connected in space. Then, using the bifurcation points as nodes and the adjacency relationships as edges, an undirected graph G=(V,E) is constructed, where the node set V corresponds to the set of bifurcation points. The edge set E contains all directly connected pairs of branching vertices. Then, a depth-first search or breadth-first search algorithm is used to traverse the graph and identify all connected components. Each connected component constitutes an initial connected path for a branching vertex, denoted as . , where L is the number of branching points in the path;

[0090] For each initial connected path The system sorts the points in the path according to the temporal order of the bifurcation points (i.e., the initial moment of the unstable bifurcation) or spatial adjacency to ensure the continuity of the trajectory in time and space. To further obtain a smooth potential slip surface trajectory, the system uses cubic B-spline curves to perform spatial interpolation and smoothing on the sorted discrete bifurcation points, generating a continuous spatial curve. Based on the fitted spatial curve or discrete path point sequence, the system calculates the geometric characteristic parameters of each potential slip surface trajectory, including: trajectory length: obtained by accumulating the Euclidean distance between adjacent discrete points or by numerically integrating the spatial curve; trajectory orientation: the spatial curve is projected onto the horizontal plane, and its principal axis direction is calculated (which can be obtained through principal component analysis or fitted straight lines); trajectory curve inclination angle: the spatial curve is projected onto the vertical profile, and its average inclination angle is calculated.

[0091] The above analysis results are integrated into a structured identification output, which includes: the number of identified potential slip surfaces; the set of bifurcation points and their spatial coordinates for each potential slip surface; a description of the connectivity trajectory of each potential slip surface, including the discrete point sequence and the parameters of the fitted B-spline curve; and the key geometric parameters of each potential slip surface, including the length, orientation, and inclination of the spatial curve.

[0092] The bifurcation point and its location coordinates are fed back to the grid cell construction process in real time, and the topological connection relationship of the grid cells is dynamically reorganized using the location coordinates;

[0093] In this embodiment, the system obtains the set of position coordinates of all bifurcation points within the current period. Based on these coordinates, the system identifies the regions where bifurcation points are spatially clustered, i.e., the key regions for the initiation and evolution of potential slip surfaces. Subsequently, the system initiates dynamic reorganization of the mesh topology. The goal of dynamic reorganization is to: in regions with dense bifurcation points, refine the mesh cell division to improve the analytical capability for subtle changes in strain and stress fields in these regions; simultaneously, in stable regions far from bifurcation points, maintain or appropriately sparse the mesh to optimize overall computational efficiency. In specific implementation, based on a preset densification criterion (e.g., the number of bifurcation points in a neighborhood with radius R centered on any mesh cell exceeds a threshold N), the system performs subdivision operations on mesh cells that meet the conditions. The subdivision algorithm can decompose existing mesh cells into smaller sub-cells by inserting new nodes, and update the entire region's mesh cell set and its topological connection relationships accordingly (i.e., record the connection information between each cell and its adjacent cells). This process utilizes mature mesh adaptive technology to ensure the correct inheritance and mapping of physical field (such as displacement and stress) information after mesh re-division.

[0094] Simultaneously adjust the sampling frequency for acquiring displacement vector data and stress tensor data in the next sampling cycle;

[0095] While completing the dynamic reorganization of the mesh topology, the system differentiates the sampling frequency of displacement vector data and stress tensor data acquisition for each region based on the spatial distribution characteristics of the bifurcation points; the adjustment rules are as follows:

[0096] For areas containing bifurcation points or within dynamically reorganized and densified meshes, the system marks them as critical monitoring areas. For displacement and stress sensors deployed in these areas, the sampling frequency of their next sampling cycle is increased to a preset high-frequency mode (e.g., from the usual 1 time / minute to 10 times / minute) to achieve faster timing of instability processes. For areas without bifurcation points and without densified meshes, the system marks them as general monitoring areas. The sampling frequency of sensors in these areas can remain unchanged or be reduced to a preset low-frequency mode (e.g., 1 time / 10 minutes) to reduce overall system energy consumption and data redundancy.

[0097] The specific adjustment instructions for the sampling frequency are generated by the system's central controller based on the latest grid cell topology and sensor position mapping, and then sent to the corresponding data acquisition terminals for execution. Through the aforementioned feedback adjustment mechanism, the system achieves a closed loop of "monitoring-identification-feedback-optimization." Its technical advantages are: enabling monitoring and computing resources to dynamically focus on critical areas where slope instability is occurring; maintaining low-power operation during the slope stabilization period; and rapidly improving the spatiotemporal resolution of critical areas during the slip surface initiation and propagation phases. This significantly improves the timeliness and operational efficiency of the entire monitoring system while ensuring early warning accuracy.

[0098] Based on the identification results, the non-physical fluctuations of the stress tensor data within the current sampling period are corrected to obtain the mechanical compatibility deviation value. Specifically, the system uses the actual mechanical state of the slope revealed by the identified strain localization bands and bifurcation points to perform physical rationality checks and corrections on the original stress tensor data acquired within the current sampling period. This is to eliminate non-physical fluctuations introduced by sensor noise, transmission interference, or environmental fluctuations, and quantify the degree of correction to obtain the mechanical compatibility deviation value. The specific implementation is as follows:

[0099] The set of grid cells occupied by the strain localization strips and their bifurcation points is defined as the real subdomain of the mechanical state at the current moment. The real subdomain of this mechanical state This reflects the region within the slope where significant mechanical responses (high energy dissipation, instability bifurcation) actually occur. The system assumes that this is the true subdomain of the mechanical state. Within this framework, the stress state retrieved from monitoring data and the mechanical state inferred from the identification results should satisfy the basic principles of energy conservation and force balance. Based on this, the core of the correction process is to utilize the true subdomain of the mechanical state. The overall mechanical response (e.g., the average stress or energy dissipation within the domain estimated based on the identification results) serves as a benchmark to constrain and correct the original stress tensor data of each mesh element within the subdomain.

[0100] Computational mechanical state real subdomain Volume average reference stress tensor Its weight By using the raw stress tensor data of all mesh elements e within this subdomain The volume-weighted average is obtained using the following formula:

[0101] ;

[0102] in, Let e ​​be the volume of the grid cell. For the real subdomain of mechanical state The total volume;

[0103] For the real subdomain of mechanical state For each mesh cell e within the system, the system takes its original stress tensor data. To the volume-averaged reference stress tensor After performing a contraction correction, the corrected stress tensor is obtained. The specific formula is as follows:

[0104] ;

[0105] in, It is a correction weighting factor associated with grid cell e, and its value is related to the real-time energy dissipation rate of that grid cell. or curl characteristic value Correlation occurs when these indicators of intense local activity are higher. The smaller the value (i.e., the more trust is placed in the original data of that grid cell); conversely, when these metrics are very low, The larger the value; ,in, To adjust the parameters, Ωtr represents the average energy dissipation rate of the real subdomain of the mechanical state;

[0106] The formula for calculating the mechanical compatibility deviation value is as follows:

[0107] ;

[0108] in, This is the mechanical compatibility deviation value (i.e., the relative error of the Frobenius norm of the stress tensor before and after correction). The Frobenius norm of the tensor, representing the mechanical compatibility deviation value. The size directly quantifies the original stress data of the mesh element to satisfy the true subdomain of mechanical state. The degree of correction required for overall harmony. The larger the value, the more significant the non-physical fluctuations contained in the original monitoring data of that point;

[0109] The fusion weights for the next sampling period are adjusted based on the mechanical compatibility deviation value.

[0110] Specifically, adjusting the fusion weights for the next sampling period includes:

[0111] Based on the topological connectivity of the mesh cells, the spatial variational rate for calculating the mechanical compatibility deviation between adjacent mesh cells is given by the following formula:

[0112] ;

[0113] in, The mechanical compatibility deviation value is the rate of change in space along the direction from grid cell e to grid cell n. and These are the mechanical compatibility deviation values ​​between mesh element e and its adjacent mesh element n, respectively.

[0114] By weighting the spatial variation rate with the feature length of the grid cell, a coordination bias field characterizing the degree of spatial inconsistency in the monitoring data is generated. The specific formula is as follows:

[0115] ;

[0116] in, To coordinate the deviation field, It is the characteristic length of the grid cell e. The set of all adjacent cells of grid cell e. It is the number of adjacent grid cells;

[0117] It should be noted that the construction of the physical credibility index includes:

[0118] The increment of external force work in the mesh element within the current sampling period is calculated based on displacement vector data, and the increment of internal effect variable energy within the mesh element is calculated based on stress tensor data and first-order strain tensor. The difference between the increment of external force work and the increment of internal effect variable energy is calculated to obtain the energy conservation residual. The specific implementation is as follows:

[0119] First, calculate the energy conservation residual of grid cell e within the current sampling period. :

[0120] Based on displacement vector data, the work done by external forces (such as gravity and boundary forces) acting on the element on the current displacement increment is calculated, i.e., the increment of external force work. Based on the stress tensor data of this unit with the first-order strain tensor increment Calculate the change in its internal energy (strain energy), that is, the increase in internal effect energy. ;

[0121] The formula for calculating the residual due to energy conservation is as follows:

[0122] ;

[0123] An energy normalization coefficient is constructed based on the ratio of the energy conservation residual to the increase in external work. The specific formula is as follows:

[0124] ;

[0125] in, It is a very small positive number, used to prevent division by zero errors;

[0126] The physical reliability index is determined based on the distribution of the energy normalization coefficient. The physical reliability index is negatively correlated with the absolute value of the energy normalization coefficient, as shown in the following formula:

[0127] ;

[0128] in, The physical reliability index of grid cell e. and To adjust the parameters;

[0129] Based on the distribution gradient of the physical reliability index within the grid cells, the gain adjustment amount of the fusion weights for the next sampling period is determined, and the specific implementation is as follows:

[0130] First, calculate the spatial gradient magnitude of the physical credibility index. The specific formula is as follows:

[0131] ;

[0132] in, This is the physical reliability index for grid cell n;

[0133] The gain adjustment of the fusion weights is based on the spatial gradient magnitude. and current credibility Specifically, regarding physical credibility indicators Its inherently low (poor data reliability) and its spatial gradient magnitude For larger regions (where credibility changes drastically), the fusion weights need to be adjusted more aggressively to suppress the influence of untrusted data and adapt to its spatial variations. The specific formula is as follows:

[0134] ;

[0135] in, As the global learning rate, the fusion weights of the grid cell e in the next sampling period will eventually be updated based on the current weights and this adjustment; for example, for displacement fusion weights... The update can be represented as And then normalization was performed to ensure Through this closed-loop feedback mechanism, the fusion weights are adaptively optimized according to the physical reliability status of the monitoring data.

[0136] The initial design goal of this scheme is to resolve the fundamental contradiction between "spatial and temporal non-uniformity" and "fixed configuration of monitoring resources" in the evolution of slope disasters. The concept is to break away from the rigid mode of uniform frequency and precision across the entire area in traditional monitoring, and instead establish a dynamic response system with physical induction factors as the core. By using the identified bifurcation points and strain localization strips as spatial guidance signals, the topological connection relationship of the computational grid is dynamically reorganized in reverse, and the asynchronous adjustment of the sensor sampling frequency is triggered simultaneously. The establishment of this logic enables the system to automatically focus the spatial resolution of displacement and stress on the potential instability area from the underlying algorithm level. At the same time, the principle of "mechanical compatibility" is introduced to cross-validate multi-source data, thereby constructing a monitoring closed loop that can self-evolve as the mechanical state of the slope evolves.

[0137] Through a feedback strategy that focuses on demand, the system maintains low-frequency acquisition and sparse grid during the slope stabilization period, while rapidly improving the spatiotemporal resolution of key areas during the slip surface initiation period, achieving ultimate optimization of monitoring energy efficiency. Secondly, this scheme innovatively uses the volume-averaged reference stress of the real subdomain of mechanical state as a physical anchor point, realizing self-healing correction of non-physical fluctuations in the original stress data, which greatly improves the robustness of the early warning system to sensor noise and environmental fluctuation interference. Most importantly, by calculating the physical reliability index based on energy conservation residuals and adjusting the fusion weight accordingly, it ensures that the deep fusion results of multi-source data in the next sampling period can restore the real mechanical response inside the slope to the greatest extent, thus providing a higher order of early warning accuracy and reliability than single data discrimination.

[0138] Traditional equal-weighted fusion methods or fixed-frequency sampling methods lack verification of the physical logic of the data, making them prone to false alarms or missed alarms when sensors are damaged or the environment changes abruptly. In contrast, this solution quantifies the degree of data distortion by using mechanical compatibility deviation values, which has clearer mechanical boundary constraints compared to purely mathematical filtering methods such as Kalman filtering. Furthermore, compared to conventional encryption strategies based on macroscopic displacement threshold triggering, this solution uses the bifurcation point, an early microscopic physical signal, as the trigger source, successfully advancing the early warning time from the middle and late stages of a disaster to its initial stage, fundamentally solving the industry problem of delayed disaster signal identification under complex geological conditions.

[0139] Figure 2 This is a schematic diagram illustrating the sampling frequency adjustment of the slope multi-source time-series monitoring data dynamic fusion and intelligent early warning method of the present invention;

[0140] This application further proposes a method that includes:

[0141] Based on the connectivity trajectory of the bifurcation points, the local curvature sequence and curvature change rate sequence along the potential slip surface direction between adjacent bifurcation points are calculated. Specifically, the connectivity trajectory of the bifurcation points is represented as a parametric curve in the spatial coordinate system. Where s is the arc length of the trajectory, the specific formula is as follows:

[0142] ;

[0143] ;

[0144] in, It is a local curvature sequence. Let be the first derivative of the connected locus, and let represent the tangent direction of the connected locus. Let be the second derivative of the connected trajectory, representing the rate of change of direction of the connected trajectory during propagation. For the curvature change rate sequence, For the differential increment of local curvature, The differential increment of the arc length of the connected trajectory;

[0145] By using the local curvature sequence and the curvature change rate sequence as geometric morphological feature parameters, and co-mapping them with the real-time energy dissipation rate, a curvature-dissipation coupling intensity distribution field along the propagation direction of the potential slip surface is constructed.

[0146] It should be noted that the construction of the curvature-dissipation coupling intensity distribution field includes:

[0147] The local curvature sequence and the curvature change rate sequence are mapped to a curvature tensor, as shown in the following formula:

[0148] ;

[0149] in, Let be the curvature tensor, which characterizes the intensity of curvature evolution of the slip surface in space. and These are the unit normal vector and unit tangent vector of the connected trajectory of the bifurcation point at the current point, respectively;

[0150] The spatial second-order gradient of the real-time energy dissipation rate is calculated, and the centripetal gradient vector characterizing the strain energy dissipation accumulation state is determined based on the distribution density of the spatial second-order gradient. The specific formula is as follows:

[0151] ;

[0152] in, The centripetal gradient vector characterizes the state of strain energy dissipation and accumulation. Spatial Laplace operation representing real-time energy dissipation rate;

[0153] The tensor inner product shrinking algorithm is invoked to couple the curvature tensor with the centripetal gradient vector, generating a curvature-dissipation coupling intensity distribution field. Let the intensity value at any point in the curvature-dissipation coupling intensity distribution field be... The specific formula is as follows:

[0154] ;

[0155] in, For the tensor and vector contraction operation, it represents the modulation effect of geometric curvature on energy accumulation. The magnitude of the curvature-dissipation coupling strength distribution field is positively correlated with the absolute value of the curvature change rate sequence and the magnitude of the centripetal gradient vector.

[0156] Based on the curvature-dissipation coupling strength distribution field, the system identifies energy focusing transition zones where the local gradient of the coupling strength exceeds a preset variation threshold. Specifically, the system performs gradient calculations on each sampling point in the curvature-dissipation coupling strength distribution field, and when its magnitude... The following equation is used to determine the energy focusing transition zone:

[0157] Preset mutation threshold;

[0158] in, The spatial evolution rate vector of the curvature-dissipation coupling strength distribution field. Let be the magnitude of the spatial evolution rate vector. , and These represent the intensity values ​​at any point in the curvature-dissipation coupling intensity distribution field. Partial derivatives along the three orthogonal axes;

[0159] When the coupling strength gradient of the energy focusing transition zone exceeds a preset critical gradient threshold, the system performs the following actions:

[0160] The energy focusing transition zone was identified as the priority monitoring anchor region, and the attenuation direction of the coupling strength gradient (i.e., the spatial evolution rate vector) was used as the basis for this monitoring. The grid cells are non-uniformly subdivided in the direction of the gradient (intensity value). Increase node density at the point of fastest change;

[0161] Simultaneously increase the sampling frequency of displacement vector data and stress tensor data acquired in the priority monitoring anchor area, and superimpose the geometric position and intensity value of the energy focusing transition zone into the data structure of the connected trajectory of the bifurcation point in real time;

[0162] In the next sampling period, when determining the evolution of potential slip surfaces, the system not only needs to find geometrically connected paths, but also needs to calculate the line integrals on each path. Prioritize searching for the path that maximizes the integral value. This physical constraint forces the slip surface identification results to shift towards the region where energy is focused and curvature changes abruptly, thereby achieving real-time additional constraints on morphological evolution.

[0163] In the context of complex mechanics, the instability and connection of slip surfaces are not random, but tend to develop along the direction of most severe energy dissipation and most significant geometric abrupt change. In order to capture this pattern, this embodiment no longer starts from a single displacement trajectory, but proposes a prediction operator guided by both geometry and physics. By mapping the local curvature sequence into a tensor and performing a contraction operation with the centripetal gradient vector that characterizes the energy collapse trend, a curvature-dissipation coupling intensity distribution field is constructed. The spatial directionality of the field gradient is used to predict the most likely expansion path of the slip surface, thereby changing the monitoring perspective from a lagging description of the existing trajectory to an advanced prediction of the evolution trend.

[0164] By constructing a coupling intensity distribution field and identifying the energy focusing transition zone, the danger zone where strain energy is extremely concentrated and the geometry is about to undergo drastic changes can be locked in advance, thereby achieving accurate prediction and adaptive mesh densification of the priority monitoring anchor area. Secondly, by using the maximum value path of the line integral as an additional constraint condition, the path deviation that may be caused by noise interference in the traditional identification algorithm is forcibly corrected, ensuring that the slip surface trajectory identification results are close to physical reality (energy focusing and curvature abrupt change). This constraint mechanism based on the superposition of physical fields enables the system to maintain extremely high trajectory capture stability when facing complex and ever-changing internal slope structures, and the dynamic frequency adjustment mechanism ensures high-fidelity recording of catastrophic transient signals, significantly improving the real-time evolution analysis capability of the entire monitoring system.

[0165] Traditional trajectory extrapolation methods rely primarily on mathematical spline interpolation for geometric prediction, completely ignoring the underlying dynamic driving forces. This makes them prone to prediction failure when the slip surface turns or bifurcates. While conventional pure energy criterion methods can identify high-energy regions, they lack geometric constraints such as local curvature and cannot provide specific evolution vector directions. This scheme couples the two through a tensor shrinking algorithm, which not only fills the logical gaps of a single criterion but also, compared to neural network predictions based on black-box algorithms, has completely transparent computational logic and dimensional consistency. Its results have clear physical credibility and exhibit stronger robustness in dealing with discontinuous deformation and abrupt changes in soil and rock masses.

[0166] Example 2:

[0167] This is one embodiment of the present invention, which differs from the previous embodiment in that:

[0168] A dynamic fusion and intelligent early warning system for multi-source time-series monitoring data of slopes, including:

[0169] Monitoring data mapping module: used to synchronously acquire displacement vector data and stress tensor data of the area to be monitored, and map the displacement vector data and stress tensor data to pre-constructed mesh cells based on preset fusion weights;

[0170] Feature extraction module: used to calculate the second-order strain gradient tensor field of the mesh element and extract curl feature values ​​that characterize the degree of non-coordinated rotation;

[0171] Localization stripe identification module: It is used to map and associate curl feature values ​​as geometric constraint factors with stress tensor data, calculate the real-time energy dissipation rate within the grid cell, and identify strain localization stripes by identifying the transfer characteristics of the real-time energy dissipation rate between adjacent grid cells.

[0172] Bifurcation Capture Module: Used to determine the stability of strain localization strips based on acoustic tensor determinant. When the curl eigenvalue exceeds the preset variation threshold and the calculated value of the acoustic tensor determinant decreases to the preset extreme value range, it determines that the slope has a potential slip surface and captures the bifurcation point where the potential slip surface begins.

[0173] Evolution output module: Used to output recognition results based on the spatial connectivity trajectory of the bifurcation points;

[0174] Feedback adjustment module: Used to feed back the bifurcation points and their coordinates in real time during the mesh cell construction process. It dynamically reorganizes the topological connections of the mesh cells using these coordinates and simultaneously adjusts the sampling frequency for acquiring displacement vector and stress tensor data in the next sampling cycle; and

[0175] Based on the identification results, the non-physical fluctuations of the stress tensor data in the current sampling period are corrected to obtain the mechanical compatibility deviation value. The fusion weight of the next sampling period is then adjusted based on the mechanical compatibility deviation value.

[0176] Example 3:

[0177] In one embodiment of the present invention, which differs from the previous embodiment, the electronic device includes one or more processors and a memory.

[0178] A processor can be a central processing unit (CPU) or other form of processing unit with data processing and / or instruction execution capabilities, and can control other components in an electronic device to perform desired functions.

[0179] The memory may include one or more computer program products, which may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. Volatile memory may include, for example, random access memory (RAM) and / or cache memory. Non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc.

[0180] In one example, the electronic device may also include input devices and output devices, which are interconnected via a bus system and / or other forms of connection mechanisms (not shown). In addition, depending on the specific application, the electronic device may include any other suitable components.

[0181] Example 4:

[0182] Embodiments of this application may also be computer-readable storage media storing computer program instructions thereon, which, when executed by a processor, cause the processor to perform the steps described in the "Exemplary Methods" section above according to the various embodiments of this application.

[0183] Computer-readable storage media may take the form of any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may, for example, include, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatuses, or devices, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: electrical connections having one or more wires, portable disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0184] The basic principles of this application have been described above with reference to specific embodiments. However, it should be noted that the advantages, benefits, and effects mentioned in this application are merely examples and not limitations, and should not be considered as essential features of each embodiment of this application. Furthermore, the specific details disclosed above are for illustrative and facilitative purposes only, and are not limitations. These details do not restrict the application from being implemented using the specific details described above.

[0185] The block diagrams of devices, apparatuses, devices, and systems involved in this application are merely illustrative examples and are not intended to require or imply that they must be connected, arranged, or configured in the manner shown in the block diagrams. As those skilled in the art will recognize, these devices, apparatuses, devices, and systems can be connected, arranged, and configured in any manner. Words such as “comprising,” “including,” “having,” etc., are open-ended terms meaning “including but not limited to,” and are used interchangeably with them. The terms “or” and “and” as used herein refer to the terms “and / or,” and are used interchangeably with them unless the context clearly indicates otherwise. The term “such as” as used herein refers to the phrase “such as but not limited to,” and is used interchangeably with it.

[0186] It should also be noted that in the apparatus, equipment, and methods of this application, the components or steps can be disassembled and / or recombined. These disassemblies and / or recombinations should be considered as equivalent solutions of this application.

[0187] The above description of the disclosed aspects is provided to enable any person skilled in the art to make or use this application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the general principles defined herein can be applied to other aspects without departing from the scope of this application. Therefore, this application is not intended to be limited to the aspects shown herein, but rather to be accorded the widest scope consistent with the principles and novel features disclosed herein.

[0188] The above description has been given for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of this application to the forms disclosed herein. Although numerous exemplary aspects and embodiments have been discussed above, those skilled in the art will recognize certain variations, modifications, alterations, additions, and sub-combinations thereof.

Claims

1. A method for dynamic fusion and intelligent early warning of multi-source time-series monitoring data for slopes, characterized in that, include: Simultaneously acquire displacement vector data and stress tensor data of the area to be monitored, and map the displacement vector data and stress tensor data to pre-constructed mesh cells based on preset fusion weights; Calculate the second-order strain gradient tensor field of the mesh element and extract the curl eigenvalues ​​characterizing the degree of non-coordinated rotation; The curl eigenvalues ​​are used as geometric constraint factors to map and associate with the stress tensor data, the real-time energy dissipation rate within the grid cell is calculated, and strain localization stripes are identified by recognizing the transfer characteristics of the real-time energy dissipation rate between adjacent grid cells. The stability of the strain localization strip is determined based on the acoustic tensor determinant. When the curl characteristic value exceeds the preset variation threshold and the calculated value of the acoustic tensor determinant decreases to the preset extreme value range, it is determined that the slope has a potential slip surface, and the bifurcation point of the potential slip surface initiation is captured. The identification result is output based on the spatial connectivity trajectory of the bifurcation point; The bifurcation point and its position coordinates are fed back to the construction process of the mesh cell in real time. The topological connection relationship of the mesh cell is dynamically reorganized using the position coordinates, and the sampling frequency of the displacement vector data and stress tensor data to be obtained in the next sampling period is adjusted synchronously. as well as Based on the identification results, the non-physical fluctuations of the stress tensor data in the current sampling period are corrected to obtain the mechanical compatibility deviation value. The fusion weight of the next sampling period is then adjusted based on the mechanical compatibility deviation value.

2. The method for dynamic fusion and intelligent early warning of multi-source time-series monitoring data for slopes according to claim 1, characterized in that: The method further includes: Based on the connectivity trajectory of the bifurcation points, calculate the local curvature sequence and curvature change rate sequence between adjacent bifurcation points along the potential slip surface direction; The local curvature sequence and the curvature change rate sequence are used as geometric morphological feature parameters and are co-mapped with the real-time energy dissipation rate to construct the curvature-dissipation coupling intensity distribution field of the potential slip surface along the propagation direction. Based on the curvature-dissipation coupling strength distribution field, identify the energy focusing transition zone where the local gradient of coupling strength exceeds a preset variation threshold; When the coupling strength gradient of the energy focusing transition zone exceeds the preset critical gradient threshold, the energy focusing transition zone is determined as the priority monitoring anchor area, and the grid cells are non-uniformly refined and subdivided based on the attenuation direction of the coupling strength gradient. Simultaneously increase the sampling frequency of displacement vector data and stress tensor data acquired within the priority monitoring anchor zone; and The geometric position and intensity value of the energy focusing transition zone are superimposed in real time onto the connection trajectory of the bifurcation point as an additional constraint condition for determining the morphological evolution of the potential slip surface in the next sampling period.

3. The method for dynamic fusion and intelligent early warning of multi-source time-series monitoring data for slopes according to claim 2, characterized in that: The construction of the curvature-dissipation coupling intensity distribution field includes: Map the local curvature sequence and the curvature change rate sequence to a curvature tensor; Calculate the spatial second-order gradient of the real-time energy dissipation rate, and determine the centripetal gradient vector characterizing the strain energy dissipation accumulation state based on the distribution density of the spatial second-order gradient. The curvature tensor and the centripetal gradient vector are coupled using the tensor inner product shrinking algorithm to generate a curvature-dissipation coupling intensity distribution field; wherein the magnitude of the curvature-dissipation coupling intensity distribution field is positively correlated with the absolute value of the curvature change rate sequence and the magnitude of the centripetal gradient vector.

4. The method for dynamic fusion and intelligent early warning of multi-source time-series monitoring data for slopes according to claim 1, characterized in that: The capture of the bifurcation point includes: The spatial partial derivatives of the displacement vector data within the grid cell are calculated to obtain the first-order strain tensor. Based on the stress tensor data and the first-order strain tensor, the second-order work increment within the mesh cell is calculated in real time. Based on the numerical evolution of the acoustic tensor determinant, and combined with the instantaneous extreme point where the second-order work increment changes from positive to negative, it is determined that the constitutive relation of the grid cell has become unstable and bifurcated. The initial moment of the instability and bifurcation is determined as the bifurcation point initiation moment, and the spatial position where the curl feature value corresponding to the initial moment exceeds the preset variation threshold is extracted to capture the bifurcation point of the potential slip surface initiation.

5. The method for dynamic fusion and intelligent early warning of multi-source time-series monitoring data for slopes according to claim 4, characterized in that: The determination of the unstable bifurcation includes: Based on the stress tensor data and the first-order strain tensor, the tangent stiffness matrix of the mesh element is reconstructed in real time. The tangent stiffness matrix is ​​decomposed into eigenvalues ​​to obtain the minimum eigenvalues ​​that characterize the material's resistance attenuation state. The evolution trajectory of the minimum eigenvalue is monitored in real time. When the value of the minimum eigenvalue decreases from a positive value to a zero or negative value, and the derivative of the second-order work increment with respect to time changes from a negative value to a zero value, it is determined that the constitutive relation of the grid cell has become unstable and bifurcated.

6. The method for dynamic fusion and intelligent early warning of multi-source time-series monitoring data for slopes according to claim 4, characterized in that: The adjustment of the fusion weights for the next sampling period includes: Based on the topological connection relationship of the grid cells, the spatial variation rate of the mechanical compatibility deviation value between adjacent grid cells is calculated; The spatial variation rate is weighted and mapped to the feature length of the grid cell to generate a coordination deviation field that characterizes the degree of spatial inconsistency in the monitoring data. A physical reliability index is constructed based on the energy conservation residual of each grid cell in the coordinated deviation field. Based on the distribution gradient of the physical reliability index within the grid cell, the gain adjustment amount of the fusion weight for the next sampling period is determined.

7. The method for dynamic fusion and intelligent early warning of multi-source time-series monitoring data for slopes according to claim 6, characterized in that: The construction of the physical credibility index includes: The increment of external force work in the mesh element during the current sampling period is calculated based on the displacement vector data, and the increment of internal effect variable energy in the mesh element is calculated based on the stress tensor data and the first-order strain tensor. Calculate the difference between the increment of external force work and the increment of internal effect energy change to obtain the energy conservation residual; An energy normalization coefficient is constructed based on the ratio of the energy conservation residual to the increase in external force work. The physical reliability index is determined based on the distribution of the energy normalization coefficient, wherein the physical reliability index is negatively correlated with the absolute value of the energy normalization coefficient.

8. A dynamic fusion and intelligent early warning system for multi-source time-series monitoring data of slopes, characterized in that, include: Monitoring data mapping module: used to synchronously acquire displacement vector data and stress tensor data of the area to be monitored, and map the displacement vector data and stress tensor data to pre-constructed mesh cells based on preset fusion weights; Feature extraction module: used to calculate the second-order strain gradient tensor field of the mesh element and extract curl feature values ​​that characterize the degree of non-coordinated rotation; Localized stripe identification module: used to map and associate the curl feature value as a geometric constraint factor with the stress tensor data, calculate the real-time energy dissipation rate within the grid cell, and identify strain localized stripes by recognizing the transfer characteristics of the real-time energy dissipation rate between adjacent grid cells; Bifurcation Capture Module: Used to determine the stability of the strain localization strip based on the acoustic tensor determinant. When the curl characteristic value exceeds the preset variation threshold and the calculated value of the acoustic tensor determinant decreases to the preset extreme value range, it is determined that the slope has a potential slip surface and the bifurcation point of the potential slip surface is captured. Evolution output module: used to output recognition results based on the spatial connectivity trajectory of the bifurcation point; Feedback adjustment module: used to feed back the bifurcation point and its position coordinates to the construction process of the mesh cell in real time, dynamically reorganize the topological connection relationship of the mesh cell using the position coordinates, and synchronously adjust the sampling frequency of the displacement vector data and stress tensor data to be acquired in the next sampling period; as well as Based on the identification results, the non-physical fluctuations of the stress tensor data in the current sampling period are corrected to obtain the mechanical compatibility deviation value. The fusion weight of the next sampling period is then adjusted based on the mechanical compatibility deviation value.

9. An electronic device comprising a memory and a processor, characterized in that: The memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions, which, when executed by the processor, implement the steps of the method as described in any one of claims 1 to 7.

10. A computer storage medium storing computer-executable instructions thereon, characterized in that: When the computer-executable instructions are executed by a processor, they implement the steps of the method as described in any one of claims 1 to 7.

Images