A water conservancy facility safety monitoring method and device based on multi-modal data and a medium

Abstract

The embodiment of the specification discloses a water conservancy facility safety monitoring method and device based on multi-modal data, equipment and medium, relates to the field of engineering detection technology for solving the monitoring lag problem, and the method comprises the following steps: obtaining configuration description data corresponding to the current water conservancy facility through multi-dimensional engineering data of the current water conservancy facility; based on the multi-modal data obtained by the current sensor and the configuration description data, multi-physical field coupling calculation is carried out to obtain a dynamic physical field; wherein the dynamic physical field is time series data for indicating real-time physical quantities and their change rates of each spatial position of the current water conservancy facility; the abnormal water conservancy facility area is determined by identifying the abnormal position of the physical quantity in the dynamic physical field; the configuration description data corresponding to the abnormal water conservancy facility area is updated by using the monitoring data of the obtained abnormal water conservancy facility area, and the updated configuration description data is obtained; and the current water conservancy facility is adjusted based on the updated configuration description data.

Description

Technical Field

[0001] This specification relates to the field of engineering testing technology, and in particular to a method, equipment and medium for safety monitoring of water conservancy facilities based on multimodal data. Background Technology

[0002] Water conservancy facilities are a general term for various engineering facilities built to control, regulate, and utilize surface water and groundwater in nature to achieve the purposes of harm control and benefit generation. With the rapid development of water conservancy infrastructure construction, the safe operation of water conservancy facilities such as dams, dikes, and sluices is of vital importance for flood control and disaster reduction, water resource allocation, and ecological environmental protection. Therefore, safety monitoring of water conservancy facilities is necessary to ensure safety.

[0003] Existing technologies, such as purely data-driven methods based on statistical learning or general large models, while adept at uncovering data patterns, lack physical interpretability and struggle to reveal the complex mechanical response mechanisms of hydraulic facilities. Purely mechanism-driven methods, such as traditional finite element analysis, while based on physical equations, suffer from significant deviations between simulation results and actual engineering conditions due to simplified boundary conditions, fixed material parameters, and the inability to assimilate on-site monitoring data in real time. Furthermore, current multimodal data from various sources and with varying levels of precision are largely isolated, making it difficult to provide effective support for current safety assessments.

[0004] In terms of safety monitoring, traditional methods mainly rely on fixed sensors deployed at key sections of water conservancy facilities to collect monitoring data. However, this method has a weak correlation between monitoring data and physical mechanisms, making it difficult to obtain the true stress state of water conservancy facilities. Furthermore, point-based monitoring data cannot form a continuous physical field, resulting in insufficient global situational awareness. Accurate analysis is difficult after local anomalies occur, and the evolution path of the anomaly cannot be predicted. Response methods relying on human experience are lagging and cannot provide timely guidance for the maintenance of water conservancy facilities. Summary of the Invention

[0005] To address the aforementioned technical problems, this specification provides one or more embodiments of a method, equipment, and medium for monitoring the safety of water conservancy facilities based on multimodal data.

[0006] One or more embodiments of this specification employ the following technical solutions: This specification provides one or more embodiments of a method for safety monitoring of water conservancy facilities based on multimodal data, the method comprising: By using multi-dimensional engineering data of the current water conservancy facilities, configuration description data corresponding to the current water conservancy facilities is obtained; wherein, the configuration description data is a set of data used to describe the geometric structure, physical attributes and boundary conditions of the current water conservancy facilities; Based on the multimodal data obtained from the current sensors and the configuration description data, multiphysics coupling calculations are performed to obtain a dynamic physical field; wherein, the dynamic physical field is time series data of real-time physical quantities and their rate of change used to indicate the spatial location of the current water conservancy facility; By identifying the abnormal locations of physical quantities in the dynamic physical field, the area of ​​abnormal water conservancy facilities can be determined. By acquiring the monitoring data of the abnormal water conservancy facility area, the configuration description data corresponding to the abnormal water conservancy facility area is updated to obtain the updated configuration description data; Based on the updated configuration description data, the current water conservancy facilities are adjusted accordingly.

[0007] Optionally, in one or more embodiments of this specification, configuration description data corresponding to the current water conservancy facility is obtained through multi-dimensional engineering data of the current water conservancy facility, specifically including: Obtain multi-dimensional engineering data of current water conservancy facilities; wherein, the multi-dimensional engineering data includes at least: spatial geometric data, construction parameter data, and historical operation and maintenance data; Based on the spatial geometric data, a digital twin model corresponding to the current water conservancy facility is constructed; Based on the correspondence between the construction parameter data, the historical operation and maintenance data, and the spatial location data, the construction parameter data and the historical operation and maintenance data are mapped to the digital twin model to obtain the spatial correlation between the spatial location data, the construction parameter data, and the historical operation and maintenance data. Based on the spatial relationship, the configuration description data corresponding to the current water conservancy facility is obtained.

[0008] Optionally, in one or more embodiments of this specification, multi-physics coupling calculations are performed based on the multimodal data obtained by the current sensor and the configuration description data to obtain a dynamic physical field, specifically including: The multimodal data is acquired by a first sensor deployed at the key section of the current water conservancy facility; wherein, the multimodal data acquired by the first sensor deployed at the key section of the current water conservancy facility includes at least real-time physical quantities at each spatial location; The multimodal data is converted into boundary conditions and load excitations of the configuration description data based on the transformation function of the pre-stored mapping table. The boundary conditions and load excitations are compared with the pre-stored characteristic values ​​of multiple typical working conditions to obtain the physical field coupling attributes corresponding to the current water conservancy facility based on the comparison similarity; wherein, the physical field coupling attributes include: displacement field type, stress field type and seepage field type; The corresponding reference physical field is matched according to the physical field coupling properties, and the dynamic physical field is obtained by calculating the deviation between the multimodal data and the reference physical field.

[0009] Optionally, in one or more embodiments of this specification, a corresponding reference physical field is matched according to the physical field coupling attribute to obtain a dynamic physical field based on the deviation between the multimodal data and the reference physical field, specifically including: According to the physical field coupling attribute, a reference physical field that matches the physical field coupling attribute of the current water conservancy facility is retrieved from the pre-stored physical field library; wherein, the reference physical field is a standard physical quantity distribution field corresponding to different typical working conditions obtained based on historical measured data; The difference between the current sensing data and the historical reference data corresponding to the reference physical field is calculated to obtain a real-time deviation vector. Based on the real-time deviation vector, the situation correction coefficient at the current moment is calculated using a linear interpolation algorithm. The baseline physical field is corrected according to the situation correction coefficient to obtain the corrected physical field, and the corrected physical fields of each time step are arranged in chronological order to obtain the dynamic physical field of the time series.

[0010] Optionally, in one or more embodiments of this specification, before performing multi-physics coupling calculations based on the multimodal data obtained from the current sensor and the configuration description data to obtain the dynamic physical field, the method further includes: Extract the engineering structure data from the multi-dimensional engineering data; wherein, the engineering structure data includes: geometric feature data and geological condition data; Based on the geometric feature data, abrupt change regions of the current water conservancy facility are identified; wherein, the abrupt change region is the region where the geometric shape of the current project undergoes abrupt change; Based on the geological condition data, the geologically weak zones of the current water conservancy facilities are identified; wherein, the geologically weak zones include lithological contact surfaces, fault fracture zones, and areas where weak interlayers are distributed; Based on the spatial intersection of the abrupt change region and the geologically weak zone, the risk area of ​​the current water conservancy facility is determined; Based on the hydraulic monitoring data corresponding to the risk area, the key sections of the current water conservancy facilities are determined.

[0011] Optionally, in one or more embodiments of this specification, identifying abnormal hydraulic facility areas by recognizing anomalous locations in the dynamic physical field specifically includes: Extract the physical quantity response values ​​of each spatial location in the dynamic physical field at the current moment, as well as the time derivatives corresponding to the physical quantity response values; the physical quantity response values ​​include: displacement values, stress values, and seepage pressure values ​​at each location of the current hydraulic facility; The physical quantity response value is compared with a preset static safety threshold, and the time derivative is compared with a preset dynamic rate of change threshold to identify abnormal locations in the dynamic physical field. The topological connection relationship of the current water conservancy facilities is determined based on the configuration description data, and the spatial connectivity of the abnormal location is determined based on the topological connection relationship. The abnormal locations are initially clustered based on the spatial connectivity, and then secondary clustering is performed based on physical properties to obtain multiple connected abnormal clusters. Calculate the multiphysics coupling correlation degree of each of the anomalous clusters to determine whether the anomalous clusters have multiphysics coupling failure characteristics; Anomalies exhibiting the characteristics of multi-field coupling failure, or anomalies with a single physical quantity exceeding a preset static safety threshold, are identified as the abnormal hydraulic facility area.

[0012] Optionally, in one or more embodiments of this specification, the configuration description data corresponding to the abnormal water conservancy facility area is updated using the monitoring data of the abnormal water conservancy facility area to obtain updated configuration description data, specifically including: A second sensor is located in the area of ​​the abnormal water conservancy facility to acquire monitoring data of the area based on the second sensor; wherein the sampling density of the second sensor is higher than that of the first sensor, and the second sensor performs mobile inspections in the area of ​​the abnormal water conservancy facility; Based on the spatial boundary of the abnormal water conservancy facility area, obtain the configuration description data to be updated, and map the monitoring data to the spatial location corresponding to each water conservancy equipment component in the configuration description data to be updated based on spatial coordinates, so as to obtain the update input data of the configuration description data to be updated; Using the updated input data as the target, the material constitutive parameters of the water conservancy facilities in the configuration description data to be updated are corrected through iterative inversion until the residual between the simulation result and the updated input data is less than a preset residual threshold, and the corrected local constitutive parameters are obtained. The local constitutive parameters are updated to the configuration description data to be updated, and the updated configuration description data is obtained.

[0013] Optionally, in one or more embodiments of this specification, the current water conservancy facilities are adjusted based on the updated configuration description data, specifically including: The inference engine pre-installed in the configuration description data is invoked to automatically generate multiple progressive load and working condition combinations based on the current physical field state of the abnormal water conservancy facility area; Based on the updated configuration description data, a multiphysics coupling forward deduction is performed on the load and working condition combination to obtain the evolution trend of each physical field; wherein, the evolution trend of the physical field is obtained based on time step iteration; Traverse the simulation data corresponding to each time step in the simulation results to obtain the spatial locations of key water conservancy facilities whose physical quantity response values ​​are greater than the preset static safety threshold. The spatial locations of key water conservancy facilities in adjacent time steps are connected according to the time sequence to obtain the development path; the risk level of the abnormal water conservancy facility area is determined according to the length of the development path and the difference between the physical quantity response value and the preset threshold. Based on the risk level, the system calls upon the matching early warning and control information from the pre-set control strategy library to feed back the early warning and control information to the current water conservancy facilities, thereby achieving closed-loop safety management and control.

[0014] This specification provides one or more embodiments of a water conservancy facility safety monitoring device based on multimodal data. The device includes: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform any of the methods described above.

[0015] This specification provides one or more embodiments of a non-volatile storage medium storing computer-executable instructions capable of performing any of the methods described above.

[0016] The above-described at least one technical solution adopted in the embodiments of this specification can achieve the following beneficial effects: By constructing a digital twin model of water conservancy facilities using configuration description data and combining it with multiphysics coupling calculations, discrete sensor data is expanded into a continuous dynamic physical field covering all spatial locations of the water conservancy facilities. This helps improve monitoring coverage and reduce blind spots. By identifying the abnormal locations of physical quantities in the dynamic physical field, abnormal water conservancy facility areas can be determined. This can be done based on continuous physical field data, freeing anomaly identification from the limitations of fixed-point sensor placement and solving the problem that traditional point-based monitoring methods cannot detect anomalies at non-measuring point locations. Monitoring data updates the local configuration description data, ensuring that the configuration description data remains synchronized with the actual state of the project. This avoids calculation deviations caused by outdated model parameters and achieves dynamic self-evolution of the model. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments or prior art of this specification, the drawings used in the description of the embodiments or prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this specification. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort. In the drawings: Figure 1 A schematic diagram of a method for monitoring the safety of water conservancy facilities based on multimodal data, provided in the embodiments of this specification; Figure 2 This is a structural schematic diagram of a water conservancy facility safety monitoring device based on multimodal data, provided as an embodiment of this specification. Detailed Implementation

[0018] This specification provides an embodiment of a method, equipment, and medium for safety monitoring of water conservancy facilities based on multimodal data.

[0019] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this specification, and not all embodiments. Based on the embodiments of this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this specification.

[0020] like Figure 1 As shown in the figure, this specification provides a flowchart illustrating a method for monitoring the safety of water conservancy facilities based on multimodal data. Based on... Figure 1 As can be seen, in one or more embodiments of this specification, a method for safety monitoring of water conservancy facilities based on multimodal data specifically includes the following steps: S101: Obtain the configuration description data corresponding to the current water conservancy facility through multi-dimensional engineering data of the current water conservancy facility; wherein, the configuration description data is a data set used to describe the geometric structure, physical attributes and boundary conditions of the current water conservancy facility.

[0021] The current multi-dimensional engineering data of water conservancy facilities includes, but is not limited to, geometric structural data, physical attribute data, and various loads and constraints borne by water conservancy facilities during operation, such as normal water storage level, design flood level, and check flood level, which are derived from design drawings, survey reports, and construction records.

[0022] As can be seen from the above, multi-dimensional engineering data comes from diverse sources and may also have different formats. Therefore, in order to convert these multi-dimensional engineering data from heterogeneous data into unified structured data that can be recognized by computers, thereby solving the data silo problem, the embodiments of this specification perform structuring processing on the multi-dimensional engineering data of current water conservancy facilities to generate unified configuration description data.

[0023] Specifically, in one feasible embodiment of this specification, configuration description data corresponding to the current water conservancy facility is obtained through multi-dimensional engineering data of the current water conservancy facility, specifically including: Acquire multi-dimensional engineering data for current water conservancy facilities. This multi-dimensional engineering data includes at least: spatial geometric data, construction parameter data, and historical operation and maintenance data. Spatial geometric data originates from design drawings, BIM models, or 3D laser scanning data of the water conservancy facilities. Examples include the geometric information of the current water conservancy facilities, such as dam outline dimensions, dam height, dam crest width, upstream and downstream dam slopes, gallery location and dimensions, and dam foundation excavation outline. Construction parameter data comes from construction records and other data, including parameters such as concrete grade, elastic modulus, pouring temperature, and compaction degree for each material zone. Historical operation and maintenance data is contained in monitoring records or inspection logs during the operation of the water conservancy facilities, such as historical water level change records, historical deformation monitoring data, historical seepage observation data, and records of abnormal events and their handling measures.

[0024] After obtaining multi-dimensional engineering data, a digital twin model corresponding to the current water conservancy facility is constructed based on the spatial geometric data. Specifically, in one feasible approach, the acquired spatial geometric data can be preprocessed, and then the processed spatial geometric data can be transformed into the same coordinate system. Based on the spatial geometric data in the unified coordinate system, existing methods such as parametric modeling can be used to construct the architectural model of the water conservancy facility, and the model can be discretized into a grid to generate a digital twin model. At this time, the digital twin model only contains the geometric shape and grid topology information of the water conservancy facility, without associated material properties and historical operation and maintenance information.

[0025] Therefore, based on the correspondence between the construction parameter data and the historical operation and maintenance data, and the spatial location data in the digital twin model, the construction parameter data and historical operation and maintenance data are mapped to the corresponding locations in the digital twin model. That is, different material zones are divided in the digital twin model, such as the dam body concrete zone, the dam foundation rock zone, and the anti-seepage wall zone. The material attribute parameters in the construction parameter data are associated with the corresponding zones, and the locations of abnormal events and areas with large historical deformations in the historical operation and maintenance data are associated with the corresponding locations in the model using spatial coordinates. Through the above mapping, the spatial correlation between spatial geometric data, construction parameter data, and historical operation and maintenance data is obtained. In other words, each spatial location in the digital twin model is associated with its geometric shape, material physical properties, historical risks, and boundary conditions. Therefore, by structurally encapsulating the digital twin model with established spatial correlations, material properties, historical risks, and boundary conditions used to describe the current geometric structure, physical properties, and boundary conditions of the water conservancy facilities, a unified configuration description data can be formed.

[0026] S102: Based on the multimodal data obtained by the current sensors and the configuration description data, perform multi-physics coupling calculations to obtain a dynamic physical field; wherein, the dynamic physical field is time-series data of real-time physical quantities and their rate of change used to indicate the spatial location of the current water conservancy facility.

[0027] Among them, dynamic physical field refers to time series data that indicates the real-time physical quantities and their rate of change at various spatial locations of the current water conservancy facilities. In other words, it covers various spatial locations of the current water conservancy facilities and includes physical quantities such as displacement, stress, and seepage pressure. The time series data, arranged based on time steps, can reflect the dynamic evolution of the physical quantities.

[0028] Furthermore, in one or more embodiments of this specification, before performing multi-physics coupling calculations based on the multimodal data and configuration description data obtained from the current sensor to obtain the dynamic physical field, the method further includes: The engineering structural data is extracted from multi-dimensional engineering data. This structural data includes geometric feature data and geological condition data. Geometric feature data originates from the design drawings or BIM models of the hydraulic facilities within the multi-dimensional engineering data, and at least includes geometric information such as dam outline dimensions, dam height, dam crest width, upstream and downstream dam slopes, gallery location and dimensions, and dam thickness variation locations. Geological condition data can be derived from geological survey reports, and at least includes geological information such as dam foundation lithology distribution, fault location and attitude, dense joint and fracture zones, and distribution of weak interlayers.

[0029] Geometric mutation regions refer to the locations where the geometric shape of a water conservancy facility undergoes abrupt changes. Therefore, by traversing the geometric feature data of each adjacent location, regions where the difference in vector geometric feature data exceeds a preset value can be identified as mutation regions of the current water conservancy facility, such as regions where the thickness of the dam body suddenly changes.

[0030] Geologically weak zones refer to areas in the rock mass of a foundation or dam abutment that have poor mechanical properties. Therefore, in order to scientifically determine the placement of sensors and ensure the representativeness and validity of the monitoring data, the geologically weak zones of the current water conservancy facilities were identified based on geological condition data. Specifically, geologically weak zones include, but are not limited to, lithological contact surfaces, fault fracture zones, and areas of weak interlayers. In one feasible approach, the aforementioned areas marked in the geological survey report can be mapped to a geometric model using spatial coordinates to form a spatial distribution set of geologically weak zones as the geologically weak zones of the current water conservancy facilities.

[0031] The geometrically abrupt change regions and geologically weak zones identified above are spatially superimposed and analyzed in a unified spatial coordinate system. The intersection of the point set of the geometrically abrupt change region and the point set of the geologically weak zone is the region that has both geometrical abrupt change and geologically weak zone. At this time, the risk level of the region is higher. Therefore, the spatial intersection region of the two is regarded as the risk area of ​​the current water conservancy facilities.

[0032] Hydraulic monitoring data includes information on head distribution, seepage pressure, and hydraulic gradient in the area under operating conditions such as design water level, normal storage water level, and check flood level. Within the identified risk areas, locations with large hydraulic gradients or high seepage pressures are further screened and designated as critical sections. Identifying the sections with the most unfavorable hydraulic conditions within the risk area helps ensure that sensors can effectively capture key characteristics of the structural response, avoiding arbitrary and blind sensor deployment and providing a reliable data foundation for subsequent multiphysics coupling calculations.

[0033] Specifically, in one or more embodiments of this specification, multi-physics coupling calculations are performed based on the multimodal data obtained from the current sensor and the configuration description data to obtain the dynamic physical field, specifically including the following implementation process: Multimodal data is acquired using first sensors deployed at key sections of the current water conservancy facilities. This multimodal data, acquired through the first sensors deployed at key sections of the current water conservancy facilities, includes at least real-time physical quantities at various spatial locations.

[0034] The multimodal data is transformed into boundary conditions and load excitations for configuration description data based on a transformation function using a pre-stored mapping table. Here, load excitation refers to the set of data on external actions or driving forces acting on the hydraulic facility structure, causing it to respond.

[0035] The pre-stored mapping table records the spatial coordinate correspondence between each sensor and its corresponding node or unit in the configuration description data, as well as the transformation function that converts physical quantity values ​​into boundary conditions. For example, the displacement value measured by the displacement gauge is converted into the displacement constraint boundary of the corresponding node, the pressure value measured by the piezometer is converted into the head boundary of the corresponding node, and the water level value measured by the level gauge is converted into the distributed water pressure load on the upstream surface.

[0036] The transformed boundary conditions and load excitations are compared with the pre-stored characteristic values ​​of multiple typical operating conditions. Based on the similarity of the comparison, the type of operating condition of the current water conservancy facility is determined, and then the corresponding physical field coupling properties are obtained. The characteristic values ​​of typical operating conditions include at least the characteristic parameters such as water level, water pressure distribution, and temperature field distribution under the normal water level condition, the design flood level condition, and the check flood level condition.

[0037] Specifically, to quantify the similarity between the current state and each typical operating condition, in the embodiments of this specification, for the j-th typical operating condition, its deviation from the typical operating condition is:

[0038] in, , as well as These represent the current water level, representative temperature, and representative seepage pressure, respectively. , as well as These are the water level, temperature, and seepage pressure values ​​corresponding to the j-th typical working condition, respectively. , as well as These are the weighting coefficients for each physical quantity. The specific values ​​of each weighting coefficient can be preset according to the type and sensitivity of the water conservancy facility. After obtaining the deviation value, the similarity between the current state and each typical working condition is obtained based on the reciprocal of the deviation value. Thus, the typical working condition with the highest similarity is selected as the matching working condition for the current water conservancy facility. A deviation value of 0 indicates that the current state is consistent with each typical working condition, with the highest similarity.

[0039] For example: the current upstream water level of a certain water conservancy facility =87.5m, representative value of dam body temperature =28℃. Pre-stored typical operating conditions include normal water level operating conditions ( =85.0m, =25℃) and design flood level conditions ( =92.5m, =30℃). Preset weighting coefficients. =0.7, =0.3, =0.

[0040] At this point, calculate the deviation between the current state and each typical operating condition, that is, the deviation from the normal water level operating condition: At this point, the similarity to the normal water level condition is approximately 0.377.

[0041] And the deviation from the design flood level conditions: The similarity to the design flood level condition at this point is approximately 0.244.

[0042] Because 0.377 The value is 0.244, so the current state is more similar to the normal water level condition. Therefore, the normal water level condition is selected as the matching condition for the current water conservancy facility, and the physical field coupling attributes and reference physical field corresponding to this condition are obtained for subsequent deviation calculation and dynamic physical field generation.

[0043] If the combined deviation of the current feature value from all typical working conditions is greater than the preset similarity threshold, it indicates that the previous working condition is an atypical working condition, and the corresponding reference physical field cannot be directly matched. In this case, one feasible approach is to select the two typical working conditions with the highest similarity, and perform weighted interpolation on their corresponding reference physical fields using the reciprocal of the deviation value to generate a temporary reference physical field suitable for the current working condition. For example, in a certain scenario, assume: Let the maximum and second-highest similarity be the temporary reference physics fields, respectively:

[0044] in, and These are the reference physical fields for the corresponding working conditions.

[0045] Based on the matched working condition type, the corresponding physical field coupling attributes are obtained. These physical field coupling attributes describe the physical field types that require special attention under that working condition, including at least displacement field type, stress field type, and seepage field type. For example, under high water level conditions, seepage pressure increases and effective stress decreases, requiring a focus on the coupling effect between the seepage field and displacement field; under sudden temperature drop conditions, the dam surface contraction is constrained by internal forces, generating stress, requiring a focus on the temperature stress field; under seismic conditions, the full coupling effect of displacement field, stress field, and seepage field must be considered simultaneously.

[0046] Based on the obtained physical field coupling properties, a matching reference physical field is retrieved from the pre-stored physical field library. The reference physical field refers to the standard physical quantity distribution field corresponding to a specific typical working condition obtained based on historical measured data. It records the standard displacement value, stress value and seepage pressure value of each spatial location in the entire area of ​​the water conservancy facility under the working condition.

[0047] The deviation between the current multimodal data acquired by the sensors and the historical reference data at the corresponding locations in the reference physical field is calculated. Specifically, for each sensor's spatial location, the difference between the current measured value and the historical reference value is calculated to form a real-time deviation vector. Based on this deviation vector, the reference physical field is corrected, adjusting the physical quantity values ​​of the corresponding region in the reference physical field to the measured values ​​or interpolated results of the measured values. The correction results at different times are arranged in chronological order to obtain the dynamic physical field.

[0048] This process maps sensor data from a limited number of points to a configuration description data space, and combines multi-physics coupling calculations to expand the sensor data into a continuous physical field distribution covering the entire area of ​​the water conservancy facility. This effectively solves the problem of monitoring blind spots and helps to comprehensively grasp the overall safety status of the project. By identifying the current operating condition, matching the pre-stored benchmark physical field, and correcting based on measured deviations, the dynamic physical field at the current moment can be quickly obtained. Compared with complete finite element reanalysis, this reduces the computational load. For atypical operating conditions, temporary benchmark situation fields can also be generated through interpolation, ensuring that a usable dynamic physical field can be obtained under any operating condition. In this process, the dynamic physical field is calculated and arranged step-by-step based on real-time sensor data. The dynamic physical field at each moment reflects the true state of the water conservancy facility at that moment, and it can be continuously updated as new data arrives. This allows the system to promptly detect abnormal trends in physical quantities, providing more sufficient evidence for early warning and avoiding the problem of missing the best early warning opportunity due to data lag.

[0049] Specifically, in one or more embodiments of this specification, a corresponding reference physical field is matched according to the physical field coupling properties, and a dynamic physical field is obtained by calculating the deviation between the multimodal data and the reference physical field. The specific process includes the following: Based on the physical field coupling properties obtained in the preceding steps, a reference physical field matching the physical field coupling properties of the current hydraulic facility is retrieved from a pre-stored physical field database. The reference physical field is a standard physical quantity distribution field corresponding to different typical operating conditions, obtained based on historical measured data.

[0050] By calculating the difference between the current sensor data and the historical reference data corresponding to the reference physical field, the real-time deviation vector of the current sensor data relative to the historical reference data is obtained. Specifically, if the measured physical quantity at a given location is read from the current sensor data (e.g., a displacement gauge reading of 12.5 mm), and the historical reference value for the same spatial coordinate location is read from the reference physical field (e.g., a historical reference displacement value of 10.0 mm), the difference between the two is calculated, yielding a real-time deviation value of 12.5 mm - 10.0 mm = 2.5 mm. The real-time deviation values ​​of all sensor positions are then organized according to the sensor's spatial coordinates to form a vector consisting of several discrete point deviation values, which serves as the real-time deviation vector.

[0051] Due to the limited number of sensors, the real-time deviation vector only exists in discrete spatial locations where sensors are deployed. For spatial locations without sensors, the deviation value cannot be directly obtained. Therefore, in this embodiment, based on the real-time deviation vector, a linear interpolation algorithm is used to calculate the situation correction coefficient at the current moment, thereby realizing the calculation of the deviation value for each spatial location in the current water conservancy facility. That is, using the sensor location as a known point and the real-time deviation value of each known point as a known quantity, a spatial interpolation function is constructed. For each spatial location in the current water conservancy facility, based on its spatial distance from each known point, the deviation estimate of that point is calculated using inverse distance weighting or linear interpolation methods, thereby obtaining the corrected physical field by correcting the reference physical field according to the situation correction coefficient. The corrected physical field is then arranged in chronological order for each time step to obtain the dynamic physical field of the time series.

[0052] The situation correction coefficient is a quantitative indicator used to describe the degree of deviation between the current measured physical quantity value and the historical benchmark value at each spatial location within the current water conservancy facility. It can be understood that the situation correction coefficient distribution field is a continuous distribution function of deviation estimates covering all spatial locations within the current water conservancy facility, with its domain being all spatial points of the water conservancy facility and its range being the deviation estimates at each point.

[0053] Specifically, the process of obtaining the situation correction coefficient is as follows: The spatial location of each sensor is taken as a known control point. For each known control point, its deviation value is known, calculated as the real-time deviation value obtained by the difference between the current measured value and the historical reference value. For example, sensor A is located at spatial location (x1, y1, z1), and its deviation value is ΔV1; sensor B is located at spatial location (x2, y2, z2), and its deviation value is ΔV2. This continues, with N sensors corresponding to N known control points and their deviation values. Using the spatial location and deviation value of the known control points as input, a spatial interpolation function is constructed. This interpolation function describes the relationship between the deviation value of any point in space and the deviation values ​​of its neighboring known control points. Specifically, for any point Q(x, y, z) to be estimated in the entire domain, its deviation estimate is obtained:

[0054] in, Let Q be the estimated deviation value of the point to be estimated. Let be the deviation value of the i-th known control point. Let f be the spatial distance from point Q to the i-th known control point, and let f be the power exponent, preferably p=2. Calculate using the Euclidean distance formula .

[0055] By repeating the above calculations for all spatial locations within the entire region, the deviation estimates for each point can be obtained, thus forming a continuous situation correction coefficient distribution field covering the entire area of ​​abnormal hydraulic facilities. After obtaining the situation correction coefficient distribution field, it is applied to the historical baseline situation field, and each location in the historical baseline situation field is corrected point by point. Specifically, for any point P(x, y, z) in space, the historical baseline physical quantity value of that point is read from the historical baseline situation field. The deviation estimate for that point is read from the situation correction coefficient distribution field. The historical baseline physical quantity value at that point is calculated together with the deviation estimate to obtain the estimated physical quantity value at the current moment. For example, an additive correction can be used to obtain the estimated physical quantity value.

[0056] Alternatively, multiplication correction can be used to obtain the estimated value of the physical quantity:

[0057] After performing the above correction operations on all points within the entire region, a safety situation field that matches the current measured data and reflects the true state of the water conservancy facilities at the current moment is obtained, which is called the corrected safety situation field.

[0058] Then, the above process is repeated step by step. At each time step, the current sensor data for that time step is acquired, the real-time deviation vector for that time step is calculated, the situation correction coefficient distribution field for that time step is calculated through spatial interpolation, and the historical baseline situation field is corrected to obtain the corrected security situation field for that time step. The corrected security situation fields obtained from different time steps are arranged sequentially in chronological order to form a time series dataset containing data from multiple time steps. This dataset is called the dynamic physical field of the time series.

[0059] This process, using limited sensor data as input, employs a spatial interpolation algorithm to convert discrete point position deviations into continuous deviations across the entire spatial domain, resolving the issue of missing position deviation values ​​for sensorless locations. Furthermore, this process eliminates the need for a complete finite element reanalysis of sensorless locations; interpolation alone allows for rapid estimation of the global deviation distribution, significantly reducing computational overhead and meeting the response speed requirements of real-time monitoring. Simultaneously, by repeatedly executing time steps and arranging the results, a dynamic physical field reflecting the evolution of physical quantities over time is obtained, providing a complete data foundation for subsequent analysis of anomalous evolution trends.

[0060] S103: By identifying the abnormal locations of physical quantities in the dynamic physical field, the area of ​​abnormal water conservancy facilities is determined.

[0061] By utilizing the dynamic physical field generated in step S102, and analyzing the physical quantity values ​​and their variation characteristics at each spatial location, areas that may pose safety risks are automatically identified and designated as abnormal water conservancy facility areas.

[0062] The dynamic physical field includes the displacement, stress, and seepage pressure values ​​of all spatial points within the entire hydraulic facility at the current moment, as well as the rate of change of each physical quantity. These physical quantity data are the basis for determining whether the engineering structure is in a normal state. Therefore, by identifying the abnormal locations of physical quantities in the dynamic physical field, clustering can be used to determine that the area where the abnormal location is located is the abnormal hydraulic facility area.

[0063] Specifically, in one or more embodiments of this specification, identifying abnormal hydraulic facility areas by recognizing abnormal locations in a dynamic physical field includes: The dynamic physical field is a data sequence arranged by time steps, recording the physical quantity data of each spatial point in the entire hydraulic facility at each time step. In the embodiments of this specification, the physical quantity response values ​​of each spatial location in the dynamic physical field at the current moment, and the corresponding time derivatives of the physical quantity response values, are extracted. The physical quantity response values ​​are the specific values ​​of each physical quantity, including: displacement values, stress values, and seepage pressure values ​​at each location of the hydraulic facility. The time derivative corresponding to the physical quantity response value refers to the rate of change of the physical quantity response value with time, reflecting whether the physical quantity at that location is tending towards stability or is changing at an accelerating rate. The time derivative is:

[0064] V(t) is the physical quantity response value at the current position, V(t-Δt) is the physical quantity response value at the previous time, and Δt is the time interval between the current time and the previous time.

[0065] For example, in a specific instance, if the current displacement of a certain position is 12.5 mm and the displacement at the previous moment was 12.0 mm, and the time interval is 1 hour, then the time derivative of the displacement at that position is 0.5 mm / hour, indicating that the displacement is increasing at a rate of 0.5 mm per hour.

[0066] After extracting the physical quantity response values ​​and time derivatives at each spatial location, they are compared with preset thresholds. Specifically, the physical quantity response values ​​are compared with preset static safety thresholds, and the time derivatives are compared with preset dynamic rate of change thresholds. The static safety threshold, used to determine whether the current state exceeds safety limits, is a fixed value preset based on engineering experience such as hydraulic facility design specifications and material strength limits. For example, for displacement values, a static safety threshold of 15mm can be set, indicating that a displacement exceeding 15mm is considered abnormal. For stress values, the static safety threshold is set to the design value of concrete tensile strength, such as 2.0MPa. For seepage pressure values, the static safety threshold is set to the maximum allowable seepage pressure at that location, such as 0.3MPa. The dynamic rate of change threshold is used to determine whether the physical quantity changes too rapidly; it is a preset rate of change limit based on engineering experience and statistical laws. For example, setting the displacement change rate threshold to 0.3 mm / hour means that a displacement growth rate exceeding 0.3 mm / hour is considered an abnormal trend, and setting the seepage pressure change rate threshold to 0.05 MPa / hour means that a rapid increase in seepage pressure is considered to indicate a risk of seepage failure.

[0067] By comparing the physical quantity response value with a preset static safety threshold and the time derivative with a preset dynamic rate of change threshold, if at least one physical quantity exceeds the static threshold or the rate of change of one physical quantity exceeds the dynamic threshold, then the location is identified as an abnormal location in the dynamic physical field.

[0068] After identifying anomalous locations, it is necessary to determine the spatial connectivity between these locations in order to aggregate scattered anomalous points into connected anomalous regions. Since the configuration description data records the current three-dimensional spatial coordinate system of the water conservancy facilities and the geometric adjacency relationships between each spatial location, the Euclidean distance in three-dimensional space is calculated for any two identified anomalous spatial locations. If this distance is less than a preset connectivity threshold, the two anomalous locations are considered directly connected. If a series of anomalous locations exist where the distance between any two adjacent anomalous locations is less than the connectivity threshold, these anomalous locations constitute a connected path. For any two anomalous locations, if there is at least one connected path consisting of consecutive adjacent anomalous locations, the two anomalous locations are considered spatially connected. In this way, each anomalous location is divided into several connected groups, and each connected group corresponds to a spatially continuous anomalous region.

[0069] Anomaly locations are initially clustered based on the obtained spatial connectivity to group all spatially connected anomaly locations into the same initial anomaly cluster. After initial clustering, several initial anomaly clusters are obtained, each corresponding to a spatially contiguous region. Then, secondary clustering is performed based on physical attributes to further analyze the physical attribute composition of the anomaly locations within each cluster. If two initial anomaly clusters are not directly connected in space, but their spatial distance is less than a preset merging threshold (e.g., less than 5 meters), and the main anomaly physical quantities within the two clusters are of the same type, then these two initial anomaly clusters are merged into one anomaly cluster, resulting in multiple connected anomaly clusters, each corresponding to a candidate anomaly water conservancy facility area.

[0070] The multiphysics coupling correlation degree of each anomaly cluster is calculated to determine whether the anomaly cluster exhibits multiphysics coupling failure characteristics. Multiphysics coupling failure characteristics refer to the simultaneous occurrence of two or more physical field anomalies within an anomaly cluster, with a physical causal relationship between these anomalies. For example, seepage pressure anomalies lead to a decrease in effective stress, which in turn causes displacement anomalies; or displacement anomalies lead to structural cracking, which in turn causes seepage anomalies. Such multiphysics linkage anomalies have a higher risk level than single physical field anomalies.

[0071] Specifically, the formula for calculating the multiphysics coupling correlation degree of each anomaly cluster is as follows:

[0072] in, This represents the number of locations with abnormal displacement within the cluster. The number of stress anomaly locations within the cluster. The number of locations with abnormal seepage within the cluster. This refers to the number of locations within a cluster that simultaneously possess two or more anomaly types.

[0073] For each anomalous cluster, the number of displacement anomalies, stress anomalies, and seepage anomalies within the cluster, as well as the number of locations exhibiting two or more anomaly types simultaneously, are counted. This count is then calculated as the ratio of the number of locations exhibiting two or more anomaly types simultaneously to the sum of the number of locations exhibiting each anomaly type. A larger value indicates a more significant multi-field coupling characteristic within the anomalous cluster. Based on this analysis, anomalous clusters exhibiting multi-field coupling failure characteristics, or anomalous clusters where a single physical quantity exceeds a preset static safety threshold, are identified as anomalous hydraulic facility areas.

[0074] This comprehensive judgment process effectively reduces false alarms caused by occasional exceedances of thresholds in a single physical quantity, while avoiding missed alarms due to multiple interconnected anomalies even when a single physical quantity does not exceed the threshold. Furthermore, it compares not only the current physical quantity response value with the static safety threshold but also the physical quantity's time derivative with the dynamic rate of change threshold. Even if the current physical quantity has not exceeded the safety threshold, if its rate of change is too rapid, it can still be identified as an anomaly, helping to detect potential risks early and gain warning time. After identifying anomaly locations, spatial connectivity is used for clustering, grouping spatially connected anomaly locations into anomaly clusters. Only connected anomaly areas forming a certain spatial scale are ultimately identified as anomaly water conservancy facility areas, effectively avoiding false alarms caused by noise interference. The initial clustering based on spatial connectivity, followed by secondary clustering based on physical attributes, also avoids the incorrect splitting of areas that should belong to the same anomaly area due to small gaps, making the division of anomaly areas more consistent with engineering practice.

[0075] S104: By acquiring the monitoring data of the abnormal water conservancy facility area, update the configuration description data corresponding to the abnormal water conservancy facility area to obtain the updated configuration description data.

[0076] The configuration description data is pre-constructed based on the design drawings, construction parameters, and historical operation and maintenance data of the water conservancy facilities, reflecting the geometric structure, physical properties, and boundary conditions of the facilities under normal conditions. However, when anomalies occur during the operation of the water conservancy facilities, the material properties or geometry of the area may have changed. For example, concrete aging leads to a decrease in the elastic modulus, crack formation leads to an increase in the permeability coefficient, or long-term load action leads to the accumulation of local deformation. At this time, the original configuration description data is no longer accurate. Therefore, if the configuration description data is used for subsequent analysis, it will cause the model calculation results to deviate from the actual engineering state. Therefore, in the embodiments of this specification, the configuration description data corresponding to the abnormal water conservancy facility area is updated by obtaining the monitoring data of the abnormal water conservancy facility area to obtain the updated configuration description data.

[0077] Specifically, in one or more embodiments of this specification, the configuration description data corresponding to the abnormal water conservancy facility area is updated using monitoring data of the abnormal water conservancy facility area to obtain updated configuration description data, specifically including: Once the area of ​​abnormal hydraulic facilities is identified, a second sensor located within that area is selected and deployed to perform intensive detection of the area, acquiring monitoring data. This second sensor has a higher sampling density than the first sensor and performs mobile inspections within the abnormal area. This includes, but is not limited to, drone-mounted visual monitoring units, robot-mounted ground-penetrating radar units, and adjustable fiber optic sensing units. Unlike the fixed first sensor, the second sensor can perform mobile inspections within the area based on its location and extent, collecting high-precision monitoring data from different angles and positions.

[0078] Based on the spatial boundary of the abnormal hydraulic facility area, i.e., the minimum bounding box range of the area in three-dimensional space, configuration description data corresponding to the area to be updated is extracted from the configuration description data. This configuration description data includes the current geometric mesh information, material property parameters, and boundary conditions at each spatial location within the abnormal area. Monitoring data collected by the second sensor is mapped to the corresponding spatial location in the configuration description data to be updated, according to the spatial coordinate correspondence, to obtain the update input data for the configuration description data to be updated. Specifically, in one feasible approach, for each monitoring data point, based on its spatial coordinates, a spatial data point with the same coordinates or the closest distance is found in the configuration description data to be updated, and the physical quantity value of the monitoring data is assigned to the corresponding spatial data point to form the update input data. This update input data includes measured physical quantity values ​​at multiple spatial locations within the abnormal area, serving as a target reference for subsequent parameter correction.

[0079] With the goal of updating the input data, the material constitutive parameters of the hydraulic facilities in the configuration description data to be updated are corrected through iterative inversion until the residual between the simulation results and the updated input data is less than a preset residual threshold, thus obtaining the corrected local constitutive parameters. In other words, the current initial constitutive parameters in the configuration description data to be updated are used as initial values. Based on these constitutive parameters, forward calculations are performed to simulate the physical quantity responses, such as displacement, stress, and seepage pressure, at various spatial locations in the anomalous region under the same boundary conditions and load excitation, obtaining simulation results. The simulation results are compared with the target values ​​at the corresponding locations in the updated input data, and the residuals, i.e., the difference values, are calculated. If the residual is greater than the preset residual threshold, the constitutive parameters are adjusted using optimization methods such as gradient descent, genetic algorithms, or Bayesian inversion, based on the direction and magnitude of the residual, generating new parameter combinations. No specific limitations are imposed on the optimization methods here. Based on the updated constitutive parameters, forward calculations are performed again, and the residuals are recalculated. This iterative process is repeated until the residual between the simulation results and the target values ​​is less than the preset residual threshold. The constitutive parameters obtained at this point are the optimal parameters that enable the model calculation results to closely match the measured data. These optimal parameters are then used as the corrected local constitutive parameters. The obtained local constitutive parameters are then used to update the corresponding positions in the configuration description data to be updated, resulting in the updated configuration description data.

[0080] By updating the local configuration description data corresponding to the abnormal hydraulic facilities area, rather than rebuilding or reanalyzing the global configuration description data, the computational load is significantly reduced and the update efficiency is improved. High-precision measured data collected by a second sensor is used to correct the material constitutive parameters of the abnormal area through an iterative inversion algorithm, ensuring that the model remains synchronized with the actual engineering state and avoiding calculation deviations caused by outdated model parameters.

[0081] S105: Based on the updated configuration description data, make feedback adjustments to the current water conservancy facilities.

[0082] The updated configuration description data reflects the current true material properties and geometric state of the abnormal water conservancy facility area, and can be used to automatically execute or assist manual decision-making based on the water conservancy project control system to make feedback adjustments to the current water conservancy facilities.

[0083] Specifically, in one or more embodiments of this specification, the current water conservancy facilities are adjusted based on the updated configuration description data, including: The system invokes a pre-configured simulation engine to automatically generate multiple progressive load and load case combinations based on the current physical field state of the anomalous hydraulic facility area. The current physical field state refers to the complete physical quantity distribution and dynamic changes of the anomalous hydraulic facility area at the current moment. Progressive loads refer to loads with gradually increasing intensity, such as gradually rising from the normal water level to the design flood level and then to the check flood level. Load case combinations refer to the superposition of different load types, such as the simultaneous action of high water level and low temperature, or the simultaneous action of high water level and earthquake. The purpose of generating multiple combinations is to simulate the development of each anomaly under different conditions to assess the adverse consequences or subsequent development of the anomaly.

[0084] Specifically, in one feasible approach, the simulation engine is a pre-installed software module within the configuration description data, providing external calling services via an application programming interface (API). The configuration description data includes the executable code path of the simulation engine, the interface function names, and the format definitions of the required input parameters. When the system executes this step, it starts the simulation engine by calling the predefined interface function and passes the current configuration description data of the hydraulic facility as the basic input parameter to the simulation engine. The process by which the simulation engine automatically generates multiple progressive load and working condition combinations based on the current physical field state of the abnormal hydraulic facility area is as follows: The simulation engine first reads the current physical field state of the abnormal hydraulic facility area from the updated configuration description data, including the displacement values, stress values, seepage pressure values, the rate of change of each physical quantity, and the current load conditions at each spatial location in the area. Then, the simulation engine identifies the currently dominant load type. If the anomaly is mainly manifested as seepage pressure anomaly, then water level load is used as the progressive load. If the anomaly is mainly manifested as temperature stress, then temperature load is used as the progressive load. If the anomaly is mainly manifested as excessive deformation, then the combined progressive load of water level and temperature is considered. Taking water level load as an example, the simulation engine reads the current upstream water level as the starting value for the progressive load, determines the upper limit of the progressive load based on the design parameters of the hydraulic facilities, and automatically generates a progressive water level sequence according to a preset step size, such as gradually increasing the water level from the current level to the check flood level. Simultaneously, the simulation engine generates various load case combinations, such as the combination of progressive water level with current temperature, the combination of progressive water level with extreme temperature, the combination of progressive water level with seismic load, and the combination of progressive current water level and temperature. The generated progressive load and load case combinations are organized into a simulation task queue. Each task includes the load type and intensity sequence, load case combination method, initial state data, simulation duration, and time step, ready to execute multiphysics coupled forward simulation sequentially or in parallel.

[0085] Based on the updated configuration description data, a multiphysics coupled forward inference is performed on each of the generated load and operating condition combinations. Forward inference refers to using the current state as the initial condition and calculating the physical field response of the hydraulic facility step-by-step according to a set load change sequence. The output results for each time step include the displacement, stress, and seepage pressure values ​​at each spatial location of the hydraulic facility at that time step. These results are arranged in chronological order to form the evolution trend of each physical field.

[0086] This involves reading the load values ​​and types corresponding to the current time step from the load change sequence. Load types include water level loads, temperature loads, and seismic loads. Depending on the load type, a corresponding processing method is applied. For water level loads, a water pressure distribution that increases linearly with water depth is generated on the upstream surface of the model based on the water level value. For temperature loads, temperature boundary conditions are generated on the model surface or internal nodes based on the temperature value. Regardless of the load type, it is converted into load conditions on the model boundaries. The generated water pressure distribution, temperature boundary conditions, or seismic inertial forces are then assigned to the corresponding locations in the digital twin model. The load assignment method depends on the load type: distributed loads are applied to the model surface, volume loads are applied to all elements, and temperature boundaries are applied to specified nodes. Each loading location obtains its corresponding load value based on its spatial coordinates. The preset displacement constraints in the digital twin model remain unchanged in this step. The load conditions of the current time step, the fixed boundary constraints, and the calculation results of the previous time step are submitted to the solver. At this point, the solver, based on the updated material constitutive parameters, calculates the displacement, stress, and seepage pressure values ​​at various spatial locations within the digital twin model under a given load. The calculation process simultaneously considers the interactions between the displacement field, stress field, and seepage field. After the solution is completed, the displacement, stress, and seepage pressure values ​​at each spatial location are extracted from the calculation results, organized according to spatial coordinates to form the displacement field, stress field, and seepage field for the current time step, and these data are saved as the output for the current time step. The displacement field, stress field, and seepage field calculated in the current time step are used as the initial state for the next time step, and the time pointer is advanced to the next time step. This process is repeated for each load value in the load change sequence until the last load value is processed or the preset total simulation time is reached. The output results of all time steps are arranged in chronological order to form the simulation results for each physical field.

[0087] Based on the simulation results, the spatial locations and development paths of key water conservancy facilities with abnormal development are identified within the abnormal water conservancy facility area. Based on these locations and development paths, the risk level is determined, and corresponding early warning and control information is generated. Specifically: The simulation data corresponding to each time step in the simulation results is traversed to identify spatial locations where the physical quantity response value is greater than a preset static safety threshold, which are then used as the spatial locations of critical water conservancy facilities. Next, the spatial locations of critical water conservancy facilities in adjacent time steps are connected according to chronological order to obtain a development path. Based on the length of the development path and the difference between the physical quantity response value and the preset threshold, the risk level of the abnormal water conservancy facility area is determined; generally, the longer the path and the larger the difference, the higher the risk level.

[0088] In one approach, the risk level of an abnormally risky water conservancy area is determined based on the following scoring formula:

[0089] Where L is the development path length, representing the total spatial path length traversed by the anomaly from its starting position to its current extent, in meters; To preset the maximum impact distance, it is pre-set based on the scale of the water conservancy facility and terrain conditions. For example, for dams with a height exceeding 100m, It can be preset to 200m. D is the maximum difference between the physical quantity response value and the preset static safety threshold. The maximum permissible over-limit value can be preset and set according to a certain proportion of the static safety threshold. as well as For the weighting coefficients, satisfying ,in α The path length weight is used for weighting. β Weighting for the degree of exceeding the limit. β Usually, a larger value is taken (e.g.) α =0.4, β =0.6) to highlight the severity of the exceedance.

[0090] In another approach, the risk level of an abnormally risky water conservancy area can be obtained by superimposing the historical risk levels corresponding to the length and the historical risk levels corresponding to the difference. Based on the risk level, matching early warning and control information from a pre-set control strategy library is invoked, and this information is fed back to the current water conservancy facilities to achieve closed-loop safety management.

[0091] By employing multiphysics coupling forward extrapolation, using the current physical field state as initial conditions, and progressively calculating the distribution of displacement, stress, and seepage fields at future time points according to a progressive load sequence, the evolution trend of anomalies can be predicted in advance over the next few hours to days. This addresses the problem of traditional monitoring methods only issuing alarms after an anomaly occurs and failing to predict its future development, thus allowing time for emergency response. Spatial locations where the physical quantity response values ​​at each time step exceed the static safety threshold are extracted from the extrapolation results as critical locations. Connecting these critical locations in chronological order to form development paths helps clarify key information for safety management. Automatic matching of corresponding early warning and control information based on the determined risk level enables rapid generation of response suggestions when an anomaly occurs, effectively shortening the time from anomaly detection to decision-making and response.

[0092] like Figure 2 As shown in the diagram, this specification provides a structural schematic diagram of a water conservancy facility safety monitoring device based on multimodal data. Figure 2 As can be seen from one or more embodiments of this specification, a water conservancy facility safety monitoring device based on multimodal data includes: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform any of the methods described above.

[0093] In one or more embodiments of this specification, a non-volatile storage medium storing computer-executable instructions is also provided, characterized in that the computer-executable instructions are capable of executing any of the methods described above.

[0094] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments of apparatus, devices, and non-volatile computer storage media are basically similar to the method embodiments, so the descriptions are relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0095] The foregoing has described specific embodiments of this specification. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are possible or may be advantageous.

[0096] The above description is merely one or more embodiments of this specification and is not intended to limit this specification. Various modifications and variations can be made to the one or more embodiments of this specification by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of one or more embodiments of this specification should be included within the scope of the claims of this specification.

Claims

1. A method for safety monitoring of water conservancy facilities based on multimodal data, characterized in that, The method includes: By using multi-dimensional engineering data of the current water conservancy facilities, configuration description data corresponding to the current water conservancy facilities is obtained; wherein, the configuration description data is a set of data used to describe the geometric structure, physical attributes and boundary conditions of the current water conservancy facilities; Based on the multimodal data obtained from the current sensors and the configuration description data, multiphysics coupling calculations are performed to obtain a dynamic physical field; wherein, the dynamic physical field is time series data of real-time physical quantities and their rate of change used to indicate the spatial location of the current water conservancy facility; By identifying the abnormal locations of physical quantities in the dynamic physical field, the area of ​​abnormal water conservancy facilities can be determined. By acquiring the monitoring data of the abnormal water conservancy facility area, the configuration description data corresponding to the abnormal water conservancy facility area is updated to obtain the updated configuration description data; Based on the updated configuration description data, the current water conservancy facilities are adjusted accordingly. Based on the multimodal data obtained from the current sensors and the configuration description data, multiphysics coupling calculations are performed to obtain the dynamic physical field, specifically including: The multimodal data is acquired by a first sensor deployed at the key section of the current water conservancy facility; wherein, the multimodal data acquired by the first sensor deployed at the key section of the current water conservancy facility includes at least real-time physical quantities at each spatial location; The multimodal data is converted into boundary conditions and load excitations of the configuration description data based on the transformation function of the pre-stored mapping table. The boundary conditions and load excitations are compared with the pre-stored characteristic values ​​of multiple typical working conditions to obtain the physical field coupling attributes corresponding to the current water conservancy facility based on the comparison similarity; wherein, the physical field coupling attributes include: displacement field type, stress field type and seepage field type; Based on the physical field coupling properties, a corresponding reference physical field is matched, and the dynamic physical field is obtained by calculating the deviation between the multimodal data and the reference physical field. By using the monitoring data of the abnormal water conservancy facility area, the configuration description data corresponding to the abnormal water conservancy facility area is updated to obtain the updated configuration description data, specifically including: A second sensor is located in the area of ​​the abnormal water conservancy facility to acquire monitoring data of the area based on the second sensor; wherein the sampling density of the second sensor is higher than that of the first sensor, and the second sensor performs mobile inspections in the area of ​​the abnormal water conservancy facility; Based on the spatial boundary of the abnormal water conservancy facility area, obtain the configuration description data to be updated, and map the monitoring data to the spatial location corresponding to each water conservancy equipment component in the configuration description data to be updated based on spatial coordinates, so as to obtain the update input data of the configuration description data to be updated; Using the updated input data as the target, the material constitutive parameters of the water conservancy facilities in the configuration description data to be updated are corrected through iterative inversion until the residual between the simulation result and the updated input data is less than a preset residual threshold, and the corrected local constitutive parameters are obtained. The local constitutive parameters are updated to the configuration description data to be updated, and the updated configuration description data is obtained.

2. The method for safety monitoring of water conservancy facilities based on multimodal data according to claim 1, characterized in that, By using multi-dimensional engineering data of the current water conservancy facilities, the configuration description data corresponding to the current water conservancy facilities is obtained, specifically including: Obtain multi-dimensional engineering data of current water conservancy facilities; wherein, the multi-dimensional engineering data includes at least: spatial geometric data, construction parameter data, and historical operation and maintenance data; Based on the spatial geometric data, a digital twin model corresponding to the current water conservancy facility is constructed; Based on the correspondence between the construction parameter data, the historical operation and maintenance data, and the spatial location data, the construction parameter data and the historical operation and maintenance data are mapped to the digital twin model to obtain the spatial correlation between the spatial location data, the construction parameter data, and the historical operation and maintenance data. Based on the spatial relationship, the configuration description data corresponding to the current water conservancy facility is obtained.

3. The method for safety monitoring of water conservancy facilities based on multimodal data according to claim 1, characterized in that, Based on the physical field coupling properties, a corresponding reference physical field is matched, and a dynamic physical field is obtained by calculating the deviation between the multimodal data and the reference physical field. Specifically, this includes: According to the physical field coupling attribute, a reference physical field that matches the physical field coupling attribute of the current water conservancy facility is retrieved from the pre-stored physical field library; wherein, the reference physical field is a standard physical quantity distribution field corresponding to different typical working conditions obtained based on historical measured data; The difference between the current sensing data and the historical reference data corresponding to the reference physical field is calculated to obtain a real-time deviation vector. Based on the real-time deviation vector, the situation correction coefficient at the current moment is calculated using a linear interpolation algorithm. The baseline physical field is corrected according to the situation correction coefficient to obtain the corrected physical field, and the corrected physical fields of each time step are arranged in chronological order to obtain the dynamic physical field of the time series.

4. The method for safety monitoring of water conservancy facilities based on multimodal data according to claim 1, characterized in that, Before performing multi-physics coupling calculations to obtain the dynamic physical field based on the multimodal data obtained from the current sensors and the configuration description data, the method further includes: Extract the engineering structure data from the multi-dimensional engineering data; wherein, the engineering structure data includes: geometric feature data and geological condition data; Based on the geometric feature data, abrupt change regions of the current water conservancy facility are identified; wherein, the abrupt change region is the region where the geometric shape of the current water conservancy facility undergoes abrupt change; Based on the geological condition data, the geologically weak zones of the current water conservancy facilities are identified; wherein, the geologically weak zones include lithological contact surfaces, fault fracture zones, and areas where weak interlayers are distributed; Based on the spatial intersection of the abrupt change region and the geologically weak zone, the risk area of ​​the current water conservancy facility is determined; Based on the hydraulic monitoring data corresponding to the risk area, the key sections of the current water conservancy facilities are determined.

5. The method for safety monitoring of water conservancy facilities based on multimodal data according to claim 1, characterized in that, By identifying anomalous locations in the dynamic physical field, areas of abnormal hydraulic facilities are determined, specifically including: Extract the physical quantity response values ​​of each spatial location in the dynamic physical field at the current moment, as well as the time derivatives corresponding to the physical quantity response values; the physical quantity response values ​​include: displacement values, stress values, and seepage pressure values ​​at each location of the current hydraulic facility; The physical quantity response value is compared with a preset static safety threshold, and the time derivative is compared with a preset dynamic rate of change threshold to identify abnormal locations in the dynamic physical field. Based on the configuration description data, the spatial connectivity of the abnormal location is determined; The abnormal locations are initially clustered based on the spatial connectivity, and then secondary clustering is performed based on physical properties to obtain multiple connected abnormal clusters. Calculate the multiphysics coupling correlation degree of each of the anomalous clusters to determine whether the anomalous clusters have multiphysics coupling failure characteristics; Anomalies exhibiting the characteristics of multi-field coupling failure, or anomalies with a single physical quantity exceeding a preset static safety threshold, are identified as the abnormal hydraulic facility area.

6. The method for safety monitoring of water conservancy facilities based on multimodal data according to claim 1, characterized in that, Based on the updated configuration description data, feedback adjustments are made to the current water conservancy facilities, specifically including: The inference engine pre-installed in the configuration description data is invoked to automatically generate multiple progressive load and working condition combinations based on the current physical field state of the abnormal water conservancy facility area; Based on the updated configuration description data, a multiphysics coupling forward deduction is performed on the load and working condition combination to obtain the evolution trend of each physical field; wherein, the evolution trend of the physical field is obtained based on time step iteration; Traverse the extrapolation data corresponding to each time step in the evolution trend to obtain the spatial locations where the physical quantity response value is greater than the preset static safety threshold as the spatial locations of key water conservancy facilities; The development path is obtained by connecting the spatial locations of key water conservancy facilities in adjacent time steps based on the time sequence; the risk level of the abnormal water conservancy facility area is determined according to the length of the development path and the difference between the physical quantity response value and the preset static safety threshold. Based on the risk level, the system calls upon the matching early warning and control information from the pre-set control strategy library to feed back the early warning and control information to the current water conservancy facilities, thereby achieving closed-loop safety management and control.

7. A water conservancy facility safety monitoring device based on multimodal data, characterized in that, The device includes: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method described in any one of claims 1-6.

8. A non-volatile storage medium storing computer-executable instructions, characterized in that, The computer-executable instructions are capable of performing the method described in any one of claims 1-6.

Images