Water conservancy project operation and maintenance management system based on digital twinning

Abstract

The application relates to the technical field of structure perception, and particularly relates to a water conservancy project operation and maintenance management system based on digital twinning, which comprises the following steps: S1: acquiring structure monitoring data and environmental disturbance data, performing time rhythm analysis, identifying a non-synchronous interval, generating a time-space synchronous driving sequence based on an interpolation prediction and time window alignment method, and inputting the driving sequence into a digital twinning system to generate an initial engineering response situation; S2: detecting a local response cavity in the initial engineering response situation, complementing response data of the cavity area through a residual reasoning and neighborhood response tensor propagation algorithm, and outputting a reconstructed structure response situation; S3: extracting a plurality of potential risk transmission paths, calculating dominant causality and transmission potential values of the paths, and screening out key dominant risk paths; and generating specific operation and maintenance control strategies for control key points. The application realizes the transformation from passive monitoring to active control, and has high engineering practicability and deployment value.

Description

Technical Field

[0001] This invention relates to the field of structural sensing technology, and in particular to a water conservancy project operation and maintenance management system based on digital twins. Background Technology

[0002] Current safety monitoring systems for hydraulic engineering structures primarily rely on a combination of statically deployed sensor networks and regular manual inspections to obtain structural status data. Although concepts such as multi-source sensor deployment and digital twins have been gradually applied to risk management of large hydraulic structures in recent years, significant differences in the data frequency, physical quantity types, and spatial distribution of different sensors often lead to problems such as asynchronous time steps, missing response values, or incomparable physical quantities during the fusion modeling process. This severely restricts the dynamic reconstruction of structural response status and the accurate identification of risk trends. Furthermore, in the face of sensor blind spots or abnormal response zones caused by sudden environmental disturbances (such as floods, vibrations, or equipment failures), existing technologies lack effective data completion mechanisms and often rely on experience-based judgment or offline retrospective analysis, which is insufficient to meet the real-time and continuous requirements of hydraulic structure operation status.

[0003] On the other hand, existing risk identification methods based on rule thresholds or expert knowledge mostly focus on the anomaly monitoring of local structural points, while lacking systematic modeling of the overall risk transmission mechanism of the structure. Especially under complex structural topologies, the path of risk diffusion from source nodes to key parts lacks calculable support. In addition, the formulation of control strategies often deviates from causal reasoning of risk paths and identification of intermediate nodes, resulting in a lack of precision and pertinence in operation and maintenance measures. In summary, there is an urgent need for an intelligent perception and control method for hydraulic structures that integrates multi-source data-driven approaches, dynamic response modeling, topological residual reasoning, and risk path identification. Summary of the Invention

[0004] This invention provides a water conservancy project operation and maintenance management system based on digital twins, which improves the monitoring integrity, the interpretability of the transmission mechanism, and the accuracy of the response strategy adaptation capability of large-scale water conservancy projects under complex disturbance scenarios.

[0005] A water conservancy project operation and maintenance management system based on digital twins includes the following functions: S1: Acquire structural monitoring data and environmental disturbance data of the water conservancy project; perform time rhythm analysis on the structural monitoring data and environmental disturbance data to identify asynchronous intervals where their update cycles do not overlap; based on interpolation prediction and time window alignment methods, fill and align the data in the asynchronous intervals to generate a spatiotemporally synchronized driving sequence, and input the driving sequence into the digital twin system. The digital twin system simulates and generates the initial engineering response state based on the spatiotemporally synchronized driving sequence. S2: Receive the initial engineering response status, detect local response gaps in the initial engineering response status caused by missing or abnormal data; map the local response gaps to the structural topology network and historical monitoring trajectory of the water conservancy project, and complete the response data of the gap area through residual inference and neighborhood response tensor propagation algorithm, and output the reconstructed structural response status. S3: Based on the response status of the reconstructed structure, extract multiple potential risk transmission paths; calculate the dominant causality and transmission potential of each path, and select the key dominant risk paths accordingly; generate specific operation and maintenance control strategies for the control key points on the dominant risk paths, and output them to the operation and maintenance execution module to control the water conservancy facilities.

[0006] Optionally, S1 includes: S11: Obtain structural monitoring data through a sensor network deployed on the hydraulic engineering structure, the structural monitoring data including at least displacement, stress, seepage pressure and vibration data; at the same time, obtain environmental disturbance data through meteorological and hydrological monitoring stations and remote data interfaces, the environmental disturbance data including at least rainfall, water level, flow velocity and wind speed data; S12: Perform spectral analysis or periodic verification on the structural monitoring data and environmental disturbance data respectively to identify the first dominant update cycle of the structural monitoring data and the second dominant update cycle of the environmental disturbance data; compare the first dominant update cycle and the second dominant update cycle, and mark the time period in which the two cycles are out of phase and the data points are missing as the asynchronous interval. S13: Within the asynchronous interval, interpolation and alignment processing strategies are adopted for data sequences with different update frequencies to generate a spatiotemporally synchronized driving sequence. S14: Input the spatiotemporally synchronized driving sequence into the digital twin system, which has a built-in finite element analysis model of the water conservancy project, and simulate and generate the initial engineering response situation by running the finite element analysis model.

[0007] Optionally, S13 includes: S131: For data sequences with low update frequency, a data completion process based on interpolation prediction is used. By constructing a smooth change curve between adjacent known data points, missing data points are filled in, and the temporal continuity of data points is restored. S132: For data sequences with high update frequency, a time window-based moving average method is used for data alignment within the asynchronous interval. By weighting and averaging multiple historical data points within a fixed time window, short-term fluctuations are suppressed, and the stability and rhythm consistency of the data are improved.

[0008] Optionally, S14 includes: S141: The spatiotemporal synchronization drive sequence after interpolation and time alignment processing is input into the digital twin system. The digital twin system has an embedded structural response modeling module for water conservancy projects. The modeling module calls a preset finite element analysis model or physical mechanism model based on the topological characteristics and stress boundary conditions of the engineering structure to simulate the dynamic response process of the water conservancy project structure under the influence of external environmental disturbances. S142: Based on the multidimensional monitoring data in the spatiotemporal synchronous driving sequence, perform time-series deduction of structural response evolution and output initial response status information, which includes structural displacement distribution, stress state, vibration characteristics and seepage pressure response.

[0009] Optionally, S2 includes: S21: Receive the initial engineering response status and mark the continuous regions in the status data where the response value exceeds the preset response threshold range or the gradient change changes abruptly as potential local response holes. S22: Map the local response void to the structural topology network of the hydraulic engineering project, which is constructed by the geometric connection relationship and mechanical transmission path of the engineering structure; at the same time, retrieve the normal operation data pattern of the void area under historical and similar environmental disturbances in the historical monitoring trajectory database. S23: Calculate the residual between the observed value of the current void region and the expected value obtained by inference based on the historical data pattern using the residual inference algorithm, and distribute it to the adjacent network nodes; S24: Based on the neighborhood response tensor propagation algorithm, according to the constitutive relation and boundary conditions of the structural material, complete the response data of the void region and output the reconstructed structural response state.

[0010] Optionally, the residual inference algorithm constructs a reference response pattern similar to the current environmental disturbance conditions based on historical monitoring data, compares the reference response pattern with the actual observed response values ​​in the neighborhood of the void region to obtain the residual, and distributes the residual to the adjacent nodes of the void region according to the structural topology network.

[0011] Optionally, the neighborhood response tensor propagation algorithm uses local residuals as source terms to construct tensor propagation paths in the structural topology network. Based on the constitutive relation and boundary conditions of the structural material, the response quantities between adjacent nodes are weighted and diffused. Through multiple iterations until the tensor field converges, a continuous and balanced response tensor distribution is formed.

[0012] Optionally, S3 includes: S31: Based on the reconstructed structural response status, identify multiple risk source nodes whose response values ​​exceed the warning threshold; based on the structural mechanics model and hydraulic transmission model of the water conservancy project, construct a risk transmission directed graph with the risk source nodes as the starting point and the key engineering facilities as the ending point; S32: In the directed graph of risk transmission, a path search algorithm is used to extract multiple potential risk transmission paths from each risk source node to the key engineering facilities; S33: Calculate the dominant causality and transmission potential of each path; S34: Based on the preset screening strategy, select at least one path with the highest comprehensive score of dominant causality and transmission potential as the key dominant risk path. S35: Identify nodes or connecting edges with high intermediation centrality on the dominant risk path and determine them as key control points; based on the simulation results of applying control measures to the key control points using the digital twin system, generate specific operation and maintenance control strategies and output them to the operation and maintenance execution module.

[0013] Optionally, S33 includes: S331: The dominant causality is quantified by calculating the transfer entropy or Granger causality strength of the response variables between adjacent nodes on the path, and multiplying them together on the path; S332: The transmission potential value is obtained by calculating the weighted product of the risk levels of each node on the path, wherein the node risk level is determined by the magnitude of the node's response value exceeding the threshold and its importance in the structure.

[0014] Optionally, the transfer entropy is used to measure the direction and intensity of information flow in the response sequence between nodes, and the Granger causality strength is obtained by performing causality tests on the historical response sequences of adjacent nodes in the path and extracting significance statistics as causal strength.

[0015] The beneficial effects of this invention are: This invention introduces a rhythmic analysis mechanism for structural monitoring data and environmental disturbance data to automatically identify asynchronous intervals with inconsistent update cycles. It also employs a combination of interpolation prediction and moving average to process low-frequency and high-frequency data separately, ensuring the consistency and physical comparability of multi-source data across time scales. Unlike traditional methods that simply align sampling points, this invention generates a spatiotemporally synchronized driving sequence while preserving the intrinsic characteristics of the signal. When input into the structural digital twin model, it can obtain an initial response state with higher fidelity, improving the model's dynamic perception and prediction capabilities. It is particularly suitable for response simulation under sudden strong disturbances.

[0016] This invention constructs a mechanical topology network for hydraulic engineering structures, combines historical monitoring trajectories, and utilizes residual inference algorithms and neighborhood response tensor propagation mechanisms to dynamically fill in void areas even when response data is missing or abruptly changes, maintaining the continuity of the structural response field. Furthermore, by calculating multiple indices of causality and transmission potential, it identifies the dominant causal relationships and key intermediate control points along the risk transmission path. Combined with digital twin simulation, it generates precise control strategies, achieving a shift from passive monitoring to active control. This invention possesses high engineering practicality and deployment value. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only for this invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart of a method according to an embodiment of the present invention; Figure 2 This is a logical framework diagram of an embodiment of the present invention. Detailed Implementation

[0019] The present invention will now be described in detail with reference to the accompanying drawings and specific embodiments. Those skilled in the art may employ other alternative methods to implement some well-known technologies; moreover, the accompanying drawings are only for more specific description of the embodiments and are not intended to specifically limit the present invention.

[0020] like Figures 1-2 As shown, a water conservancy project operation and maintenance management system based on digital twins includes the following functions: S1: Acquire structural monitoring data and environmental disturbance data of the water conservancy project; perform time rhythm analysis on the structural monitoring data and environmental disturbance data to identify asynchronous intervals where their update cycles do not overlap; based on interpolation prediction and time window alignment methods, fill and align the data in the asynchronous intervals to generate a spatiotemporally synchronized driving sequence, and input the driving sequence into the digital twin system. The digital twin system simulates and generates the initial engineering response state based on the spatiotemporally synchronized driving sequence. S1 specifically includes: S11: Acquire structural monitoring data, including displacement data, through a sensor network deployed at key parts of the hydraulic engineering structure. Stress data osmotic pressure data and vibration data Simultaneously, environmental disturbance data is acquired through meteorological and hydrological monitoring stations and remote data interfaces. This environmental disturbance data includes rainfall. Water level Flow rate and wind speed This results in two types of time-series signal sequences, specifically including: Structural monitoring data sequence set : ; Environmental disturbance data sequence set : ; S12: Set of structural monitoring data sequences Spectral analysis (Fast Fourier Transform, FFT) was performed to obtain the first dominant update period of the structural monitoring data. For the set of environmental disturbance data sequences Periodicity testing (Lomb-Scargle periodogram) was performed to obtain the second dominant update period of the environmental disturbance data. A criterion for determining asynchronous intervals is introduced, and the asynchronous interval is determined as follows: If within the time interval Above: If so, it is determined to be an asynchronous interval. ; in, For the instantaneous periodic phase of the structured data, For the instantaneous periodic phase of environmental data, The maximum phase deviation tolerance is set, with a value ranging from 30° to 45°. A phase difference within 30° is generally considered to indicate that the rhythms are basically consistent, while a difference exceeding 45° will seriously affect the physical consistency of timing alignment. This is the threshold for the proportion of missing data. If the missing data ratio exceeds 0.1, it may affect the fitting quality; exceeding 0.3 often leads to unstable interpolation. The range of values ​​is... Balancing robustness and tolerance This is an asynchronous interval; In the operation and maintenance management of water conservancy projects, it is often necessary to process data streams from different sources simultaneously, such as structural monitoring data and environmental disturbance data. However, due to differences in acquisition methods, transmission paths, and recording mechanisms, these data typically suffer from inconsistent sampling frequencies, data upload delays, or acquisition intervals, resulting in them being asynchronous on the timeline and forming "asynchronous intervals." Without unified identification and alignment, this will severely impact the modeling accuracy and reliability of response judgments in the digital twin system.

[0021] To this end, this step designs a method based on update rhythm analysis, which aims to identify the dominant update cycles of structural data and environmental data as a whole, and to determine the problem intervals with significant temporal inconsistencies between the data based on cycle phase and missing data.

[0022] Specifically, firstly, frequency domain analysis or periodicity checks are performed on structural monitoring data and environmental disturbance data respectively to extract the most representative dominant update cycles. These dominant cycles reflect the temporal rhythm characteristics of the data. Structural data typically changes slowly and has long cycles, while environmental data is affected by weather, traffic flow, etc., and has a higher update frequency. After identifying the dominant cycles, the instantaneous update phase of the two data sequences is monitored in real time within a sliding time window. If the update rhythms of the two types of data differ significantly within a certain time period (i.e., the phase shift exceeds a preset threshold), or if there are a large number of missing data points within that time period (e.g., the proportion of missing sampling points is too high), then this time period is considered to be an asynchronous interval. This interval will be used for interpolation repair and alignment processing in subsequent steps, taking into account both rhythm compatibility and data integrity, to ensure that the generated driving data can truly reflect the dynamic coupling relationship between the structure and the environment, thereby providing a stable and reliable input foundation for the digital twin system.

[0023] S13: After identifying the asynchronous intervals between structural monitoring data and environmental disturbance data, it is necessary to uniformly fill in and align missing or time-misaligned data to avoid inconsistencies in time-driven data causing errors in subsequent model inferences. Considering the update frequency and characteristic differences of various monitoring data, interpolation prediction and moving average methods are used to preprocess data of different frequencies to form the final spatiotemporal synchronization driving sequence. The preprocessing specifically includes two types of cases: (1) For structural or environmental data sequences with low update frequency (such as daily or hourly sampling): the time interval between data points is long. If there are missing values ​​in the asynchronous interval, it will directly affect the continuity of the dynamic process. In order to reconstruct its time distribution curve, the cubic spline interpolation algorithm is used to fit the function between adjacent data points and fill in the missing values, as shown in: ; in, This represents the time point value to be interpolated. Using timestamps for adjacent known data points can preserve the changing trend of the original data and provide a smooth, continuous transition effect, making it suitable for describing the evolutionary behavior of slowly changing data such as structural responses or water levels. This represents the cubic spline interpolation algorithm used to fill in missing points; (2) For data sequences with high update frequency (such as sampling per minute or more densely): there may be slight time shifts in the data within asynchronous intervals, and direct interpolation may introduce errors. Therefore, a moving average method based on a time window is used for data alignment. By backtracking multiple data points within a certain window width from the current time point, the mean is calculated to suppress local fluctuations and make it consistent with the rhythm of low-frequency data, as expressed as: ; in, This represents the sliding window size, with a value ranging from 5 to 15. The sampling time interval, It is the smoothed value after alignment; By employing the two processing methods described above, data completion and rhythm alignment are achieved for all asynchronous intervals, forming a combined sequence of structural and environmental data on a unified time axis, represented as follows: ; in, This indicates that the spatiotemporal synchronous driving sequence constructed according to a unified time step serves as the direct input for subsequent digital twin system simulation, effectively improving the simulation accuracy and risk prediction capability of the dynamic evolution process of water conservancy projects. S14: Will The data is input into a multiphysics coupled model of a hydraulic engineering project built on a digital twin platform to virtually simulate the structural response. The model's calculation outputs constitute a set of time-related parameters. The set of engineering response variables constitutes the initial engineering response state of a water conservancy project, expressed as: ; ; in, Represents the structural response displacement vector. For structural velocity, For acceleration response, For the quality matrix, Here is the damping matrix. Here is the stiffness matrix. The external environment activation function originates from the synchronization drive sequence. It represents the dynamic response distribution of a water conservancy project under current structural and environmental conditions, and is used for subsequent risk diagnosis and operation and maintenance strategy formulation.

[0024] In the construction of a digital twin system for water conservancy projects, to accurately reflect the dynamic response state of the engineering entity under actual operating conditions, it is necessary to use physical modeling methods to transform multi-source spatiotemporal driving data into predicted results of structural behavior evolution. This step proposes a response simulation method based on a structural dynamics model to generate an initial engineering response state with physical constraints. The spatiotemporal synchronous driving sequence constructed in the previous step is used as the system input variable and introduced into the dynamic solution model within the digital twin system. This model typically selects a suitable physical modeling framework based on the characteristics of the simulated water conservancy project (such as dams, gates, pumping stations, slopes, etc.), including a finite element analysis model, which can simulate the stress-strain response, seepage diffusion behavior, and vibration characteristics of the structure.

[0025] By continuously solving the external environment excitation function, the digital twin system can output the initial engineering response status covering spatial distribution and temporal evolution, including the displacement response field for identifying the overall deformation trend of the structure, the stress-strain tensor distribution for assessing the stress on key components, the vibration characteristic parameters for detecting local abnormal fluctuations, and the seepage pressure and pore pressure changes for judging the accumulation state of hidden dangers in slopes and embankments. It not only combines the dual mechanism of structural mechanics constraints and environmental data-driven approaches, but also makes the simulation results not only have real physical meaning, but also reflect the impact of external disturbance inputs on the engineering structure in real time. This provides a highly reliable input foundation for subsequent response gap filling, risk path deduction, and operation and maintenance strategy generation.

[0026] S2: Receive the initial engineering response status, detect local response gaps caused by missing or abnormal data in the initial engineering response status; map the local response gaps to the structural topology network and historical monitoring trajectory of the water conservancy project, and complete the response data of the gap area through residual inference and neighborhood response tensor propagation algorithm, and output the reconstructed structural response status. S2 specifically includes: S21: After receiving the initial engineering response status, the engineering response status includes the distribution results of the multidimensional response variables of the structure under the current working conditions over time. The distribution results are mapped to a time-space response field in the structural domain, and anomaly detection is performed on the response field to identify regions where there may be missing responses or abnormal data. In the response status data, the time-space distribution curves of all response quantities are extracted, and continuous spatial regions that exceed the preset response threshold range or experience gradient abrupt changes are marked as potential local response void regions, represented as: ; in, This serves as a structural node index, representing the analyzed location within the hydraulic engineering structure, corresponding to the node coordinates or number in the structural topology network. It is the domain of response of a hydraulic engineering structure, that is, the spatial extent of the entire response field. It is the distribution of the response quantity with respect to time and space, nodes At any moment The response value range depends on the response type. For displacement, the range is 0.1-100; for stress, the range is 0.1-50; and for pressure-related responses, the range is 1-500. This range must cover both normal operation and potential abnormal conditions. This is the upper limit of the set response threshold. The value range depends on the response type. For displacement, the value range is 80%-90% of the structure's allowable deformation limit. For stress, the value range is 60%-70% of the material's yield strength. This is used to identify abnormal response regions that deviate significantly from the normal state. It is set slightly below the failure threshold to provide early warning. This is the spatial gradient operator, representing the rate of change of the response value within the structure. Abrupt gradient changes may indicate local structural anomalies, damage, or monitoring interruptions. As threshold abrupt change standards, the displacement gradient ranges from 5 to 15, the stress gradient from 5 to 20, and the permeability gradient from 20 to 100. Values ​​exceeding these ranges typically indicate discontinuous responses, potentially suggesting the presence of voids or abnormal disturbances. This represents the set of spatial locations of local response void regions, indicating data regions in the structural topology network where response anomalies or missing data occur. This step designs a void detection method based on response distribution characteristics. Its core objective is to identify regions where local structural response information is missing or distorted due to data gaps, abnormal disturbances, or simulation instability—that is, local response voids. This method uses multiple structural response quantities included in the initial engineering response state as the analysis object, extracting their temporal and spatial distributions. By analyzing the distribution trend of the response quantities within the structural domain, it focuses on two key aspects: whether the response values ​​exceed the preset normal threshold range for the engineering project, and whether the response quantities exhibit drastic spatial abrupt changes.

[0027] If the response value in a certain spatial region continuously exceeds the warning threshold, or if its spatial gradient (i.e., the rate of change of response between neighboring points) changes sharply, it can be determined that there are abnormal response characteristics in the region, and it can be preliminarily identified as a potential local response void. Such voids may be caused by various reasons such as sensor failure, simulation boundary non-closure, and failure to accurately map external disturbances. In subsequent steps, the response data needs to be reconstructed and supplemented.

[0028] S22: The detected local response void regions mentioned above Mapped to the structural topology network of the water conservancy project, the structural topology network is constructed from the geometric connections and physical coupling paths between all engineering units. Simultaneously, historical operating conditions similar to the current disturbance conditions (such as water level and rainfall) are retrieved from the historical monitoring trajectory database. The response evolution trajectory of the target cavity area under normal conditions is extracted and constructed as the desired reference model. The structural topology network is represented as: ; in, It is a topological network of hydraulic engineering structures, with 100-10000 nodes and 2-6 edges per node. In a network, a structural node represents a monitoring or simulation unit. The connecting edges between nodes represent the mechanical transmission path or the seepage coupling path; S23: Calculate residuals by utilizing the difference between historical reference patterns and the current situation, and propagate this calculation within the structured topology network. The residuals are represented as follows: ; Use the residual value as the source term It is propagated along the structural topology network to the hole neighborhood, and diffused to adjacent nodes through weighted averaging or graph convolution, as shown below: ; in, This refers to the observed response values ​​in the neighborhood of the cavity region in the current initial response situation. It is used to identify the response deviation in the cavity region, and its accuracy affects the residual calculation results. These are residual values, the difference between the current response and the historical reference pattern. Displacement residuals range from 1 to 30, and stress residuals range from 1 to 20. Larger residual values ​​indicate a greater difference between the current state and the normal historical state, suggesting a higher level of potential risk. For nodes The neighborhood set, It is the topological weight between nodes. It is a node The neighborhood residual aggregation result represents the reconstruction driving value of the hole neighborhood; After identifying the local response void regions, it is necessary to further estimate the reasonable structural response values ​​that the void locations should exhibit. Since these regions lack effective observation or simulation data and cannot be directly judged, this step adopts a residual inference combined with topological diffusion method. By comparing the deviation between historical data patterns and the current structural response, the basis for repairing these void regions is indirectly constructed.

[0029] First, the historical monitoring database is retrieved, and typical operating states similar to the current working conditions are selected. The response change trajectory of the void area under normal conditions is extracted. These historical reference values ​​are compared with the observed values ​​of the void's neighborhood in the current response situation to calculate the residuals. These residuals reflect the difference between the expected response and the actual observed response, and are an important basis for judging whether there is an abnormal response in the void. Next, the residuals are used as initial driving information and transmitted into the structural topology network for spatial diffusion into the void area. During the diffusion process, different propagation weights are set according to factors such as the connection relationship of the structure, the stiffness of the components, and the degree of coupling, so that the residual information propagates along the path with strong structural coupling first, avoiding misjudgment. At the same time, in order to improve the stability of propagation and engineering reliability, the diffusion of residuals also considers the consistency of neighborhood nodes, and uses weighted averaging or graph convolution to smooth the results.

[0030] In summary, this step, by introducing residuals as a criterion for judging physical rationality and combining them with topological structure for regional diffusion, achieves indirect reconstruction of local response void regions, laying the foundation for subsequent more refined tensor field completion and risk identification.

[0031] S24: After obtaining the preliminary residual diffusion results, a neighborhood response tensor propagation algorithm is introduced. Utilizing the constitutive properties of the structural material and network boundary conditions, a tensor field propagation model is constructed. The structural response state after the response tensor field is stably output and completed is then determined. The tensor field propagation model is expressed as: ; in, It is the first Nodes in the next iteration The response tensor estimate, It is the first Nodes in the next iteration The response tensor estimate is a core variable in the propagation iteration calculation and is ultimately used to complete the result value of the process. Is with Adjacent topological nodes, This represents the current iteration step, with a value ranging from 1 to 50. It is a normalization factor used to ensure propagation balance. For nodes With adjacent nodes The topological weights between nodes, ranging from 0 to 1, represent the relative influence between nodes. They are constructed by combining the stiffness ratio between nodes, the proportion of coupling area, and the correlation of historical collaborative responses, and are then normalized. Initial conditions... That is, initialized to the residual value after propagation, the iteration process continues until the response tensor field is stable, that is, satisfies At that time, the tensor propagation results are backfilled into the void regions to form a complete response data matrix, and the completed structural response state is output. This is used for subsequent risk path identification and control strategy generation. The set convergence threshold has a range of values. The smaller the value of the control propagation termination condition, the higher the precision requirement. The node in two adjacent iterations The magnitude of the tensor change.

[0032] S3: Based on the response status of the reconstructed structure, extract multiple potential risk transmission paths; calculate the dominant causality and transmission potential of each path, and select the key dominant risk paths accordingly; generate specific operation and maintenance control strategies for the control key points on the dominant risk paths, and output them to the operation and maintenance execution module to control the water conservancy facilities. S3 specifically includes: S31: Nodes whose response values ​​exceed the warning threshold are identified from the reconstructed structural response status. These nodes represent potential risk sources, indicating locations within the engineering structure that are currently in a high-response, high-risk state. They are the starting points for potential risk transmission and are represented as follows: ; Combining the structural mechanics model and hydraulic transmission path model of water conservancy projects, a system is constructed based on each risk source node. Starting with key engineering facilities Directed graph of risk transmission with endpoint The direction of the edges in the risk transmission directed graph is determined by physical mechanisms (such as stress transmission paths and water pressure transmission paths) to simulate how potential risks gradually affect core facilities through structural coupling. in, This is a set of risk source nodes, representing structural locations that are in a high-risk state under the current response situation. If the proportion is too high, the system may issue false alarms; if it is too low, it may miss detections. It is the first in the network A structural node is the basic unit of position used in structural topology modeling. This is the set of all structural nodes in the topology network, with a value ranging from 100 to 10000, depending on the complexity of the project and the required monitoring accuracy. For nodes The response value, These serve as early warning thresholds for response variables: displacement 50-80, stress 25-40, and seepage pressure 300-400. They are used to identify potential anomalies in advance, rather than responding only after failure. It is a directed graph of risk transmission, used to simulate potential risk propagation paths. It is the basic structure for risk path analysis, and the direction of the edges is determined by the physical transmission mechanism. This is the set of nodes in the risk transmission diagram, with values ​​ranging from 100 to 10000. These nodes constitute the vertices of the risk diagram, representing the locations of various responses. It is the set of edges in the risk transmission graph, representing the physical connection or risk transmission path between structural nodes, with an average of 2-6 edges connected to each node; S32: Using a path search algorithm, starting from each risk source node, search for all possible paths to critical facilities. Each path represents a potential risk evolution channel, reflecting the way in which risk may gradually propagate from peripheral areas to critical nodes. Extract the path from each risk source node. To key infrastructure nodes The set of all feasible paths is represented as: ; in, Represents a node To key infrastructure nodes The set of all candidate paths, A single risk transmission path consists of several nodes arranged in a directed order, ranging from 3 to 10 nodes. Too few nodes make it difficult to reflect the transmission logic, while too many nodes may cross too wide a domain, reducing controllability. This range balances physical accessibility and evolutionary level depth. Path search algorithms are methods used to find the optimal path from the starting point to the ending point in a graph structure. In this invention, the risk transmission graph consists of structural nodes and their physical or hydraulic connections. The search objective is to find all possible transmission paths from the risk source node to the critical engineering facility. The system uses Dijkstra's algorithm, which can enumerate or filter multiple effective risk transmission paths while ensuring efficiency, for subsequent causal analysis and risk potential assessment. The path search design considers the actual coupling of the structural topology. Edges in the graph can be weighted according to connection strength, physical distance, or risk relevance to guide the algorithm to prioritize more reliable transmission paths. At the same time, by limiting parameters such as path length and node visit count, invalid paths or loops can be avoided from interfering with the search results, improving the accuracy and robustness of the overall analysis.

[0033] S33: To quantify the impact of each risk path, two key indicators are introduced to measure the logical coherence and risk intensity along the path, respectively. The key indicators include dominant causality and transmission potential. (1) Dominant Causality Strength: Dominant causality strength is used to measure whether there is a clear transmission relationship between the response changes of adjacent nodes on the path. It is obtained by calculating the transfer entropy or Granger causality strength between response variables and multiplying them to obtain the causal coherence index of the entire path. The higher the causality strength, the clearer and more traceable the transmission mechanism on the path, and the better the causality of the path. Each pair of adjacent nodes Calculate the transition entropy or Granger causality strength between its response variables. The dominant causality of a path is represented as: ; in, It is the first One risk transmission path, Representing a path A pair of adjacent nodes is a pair of nodes in a structural physical network that has a direct transmission relationship. This represents the causal strength of the responses between nodes, ranging from 0 to 1. It can be calculated from transfer entropy, Granger causality, or correlation coefficient. A larger value indicates a stronger causal relationship, while values ​​less than 0.8 typically indicate weaker causality. It is a path The dominant causal degree, (2) Propagation Potential: The propagation potential is used to measure the risk level of each node on the path and its influence on the structure. The risk level of a node is determined by the extent to which its response value exceeds the warning threshold and its importance in the structure. For a path, its overall propagation potential is the weighted product of the risk levels of all nodes on the path, reflecting the accumulated risk intensity of the path. Each node on the path The risk level and the transmission potential of the path are expressed as: ; ; in, This is the response value of the node, representing the current response strength of the node. The value is directly derived from the structural response status data. It corresponds to the early warning threshold of the response variable, reflecting the critical boundary of engineering safety, identifying potential risks in advance, and avoiding waiting until the point of failure to take action. This is the importance coefficient of a node in the structure (such as connectivity, betweenness centrality, etc.), with a value ranging from 0.1 to 1.0. It can be calculated based on betweenness centrality, connectivity, or load ratio, and represents the control effect of the node on the overall structure. The node risk level ranges from 0.05 to 3.0. A low risk level indicates a node response that just exceeds the warning threshold and has a relatively small structural effect; a high risk level indicates a node response that significantly exceeds the threshold and is a critical node. It is the path transmission potential value, which represents the intensity of risk accumulated by the path. The higher the value, the more risk impact the path may accumulate and the stronger the risk evolution potential. It is one of the core indicators of path scoring. S34: To select the most representative dominant risk path from multiple candidate paths, a comprehensive scoring function is introduced. This function weights and sums the dominant causality and transmission potential according to set weights. The comprehensive scoring function can be adjusted to favor causality or intensity. Finally, the path with the highest comprehensive score is selected as the key risk path most worthy of attention and intervention at the current stage, expressed as: ; ; in, A comprehensive path score, ranging from 0 to 10, is used to characterize the importance of the path and to select critical paths. These are the weighting coefficients for causality and transmission potential, all fixed at 0.5 to ensure the scoring function is normalized and facilitates sorting and comparison. It is the complete set of all candidate paths, with values ​​ranging from 5 to 20, and selects the most representative dominant path from multiple possible paths; S35: In the dominant risk path In the above, the betweenness centrality of each node or edge is calculated, and the key control nodes with significant control effects are selected, represented as: ; in, As a key control node with significant control effect, This represents a structural node on the dominant risk path. It is a node The intermediation centrality of a node, ranging from 0 to 1, represents the node's ability to "transmit control" across multiple paths; the higher the value, the more likely it is to be a central point for risk diffusion. For critical control nodes The system simulates different control strategies in a digital twin system and outputs operation and maintenance control strategies, specifically including: (1) Opening the floodgates to release floodwater: specifying the flow rate and timing; (2) Reinforcement and support: specified location, material type and quantity of work.

[0034] This invention encompasses any substitutions, modifications, equivalent methods, and solutions made within the spirit and scope of this invention. To provide the public with a thorough understanding of this invention, specific details are described in detail in the following preferred embodiments; however, those skilled in the art will fully understand the invention even without these details. Furthermore, to avoid unnecessary misunderstanding of the essence of this invention, well-known methods, processes, procedures, components, and circuits are not described in detail.

[0035] The above description is only a preferred embodiment of the present invention. It should be noted that for those skilled in the art, several improvements and modifications can be made without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A water conservancy project operation and maintenance management system based on digital twins, characterized in that, This includes executing the following: S1: Acquire structural monitoring data and environmental disturbance data of the water conservancy project; perform time rhythm analysis on the structural monitoring data and environmental disturbance data to identify asynchronous intervals where their update cycles do not overlap; based on interpolation prediction and time window alignment methods, fill and align the data in the asynchronous intervals to generate a spatiotemporally synchronized driving sequence, and input the driving sequence into the digital twin system. The digital twin system simulates and generates the initial engineering response state based on the spatiotemporally synchronized driving sequence. S2: Receive the initial engineering response status, detect local response gaps caused by missing or abnormal data in the initial engineering response status; map the local response gaps to the structural topology network and historical monitoring trajectory of the water conservancy project, and complete the response data of the gap region through residual inference algorithm and neighborhood response tensor propagation algorithm, outputting the reconstructed structural response status; specifically including: The residual inference algorithm calculates the residual between the observed value of the current cavity region and the expected value obtained by inference based on historical data patterns, and distributes it to adjacent network nodes. The residual inference algorithm constructs a reference response pattern similar to the current environmental disturbance conditions based on historical monitoring data, compares the reference response pattern with the actual observed response value in the neighborhood of the cavity region to obtain the residual, and distributes the residual to the adjacent nodes of the cavity region according to the structural topology network. Based on the neighborhood response tensor propagation algorithm, the response data of the void region is completed according to the constitutive relation and boundary conditions of the structural material, and the reconstructed structural response state is output. The neighborhood response tensor propagation algorithm uses local residuals as source terms to construct tensor propagation paths in the structural topology network. It performs weighted diffusion of the response quantities between adjacent nodes according to the constitutive relation and boundary conditions of the structural material, and iterates through multiple rounds until the tensor field converges to form a continuous and balanced response tensor distribution. S3: Based on the reconstructed structural response status, extract multiple potential risk transmission paths; calculate the dominant causality and transmission potential of each path, and select the key dominant risk paths accordingly; generate specific operation and maintenance control strategies for the control key points on the dominant risk paths, and output them to the operation and maintenance execution module to control the water conservancy facilities. The dominant causality is quantified by calculating the transfer entropy or Granger causality strength of response variables between adjacent nodes on the risk transmission path, and then multiplying them along the risk transmission path. The transmission potential is obtained by calculating the weighted product of the risk levels of each node on the risk transmission path, wherein the node risk level is determined by the magnitude of the node's response value exceeding the threshold and its importance in the structure.

2. The water conservancy project operation and maintenance management system based on digital twins according to claim 1, characterized in that, S1 includes: S11: Obtain structural monitoring data through a sensor network deployed on the hydraulic engineering structure, the structural monitoring data including at least displacement, stress, seepage pressure and vibration data; at the same time, obtain environmental disturbance data through meteorological and hydrological monitoring stations and remote data interfaces, the environmental disturbance data including at least rainfall, water level, flow velocity and wind speed data; S12: Perform spectral analysis or periodic verification on the structural monitoring data and environmental disturbance data respectively to identify the first dominant update cycle of the structural monitoring data and the second dominant update cycle of the environmental disturbance data; compare the first dominant update cycle and the second dominant update cycle, and mark the time period in which the two cycles are out of phase and the data points are missing as the asynchronous interval. S13: Within the asynchronous interval, interpolation and alignment processing strategies are adopted for data sequences with different update frequencies to generate a spatiotemporally synchronized driving sequence. S14: Input the spatiotemporally synchronized driving sequence into the digital twin system, which has a built-in finite element analysis model of the water conservancy project, and simulate and generate the initial engineering response situation by running the finite element analysis model.

3. The water conservancy project operation and maintenance management system based on digital twins according to claim 2, characterized in that, S13 includes: S131: For data sequences with low update frequency, a data completion process based on interpolation prediction is used. By constructing a smooth change curve between adjacent known data points, missing data points are filled in, and the temporal continuity of data points is restored. S132: For data sequences with high update frequency, a time window-based moving average method is used for data alignment within the asynchronous interval. By weighting and averaging multiple historical data points within a fixed time window, short-term fluctuations are suppressed, and the stability and rhythm consistency of the data are improved.

4. A water conservancy project operation and maintenance management system based on digital twins according to claim 2, characterized in that, S14 includes: S141: The spatiotemporal synchronization drive sequence after interpolation and time alignment processing is input into the digital twin system. The digital twin system has an embedded structural response modeling module for water conservancy projects. The modeling module calls a preset finite element analysis model or physical mechanism model based on the topological characteristics and stress boundary conditions of the engineering structure to simulate the dynamic response process of the water conservancy project structure under the influence of external environmental disturbances. S142: Based on the multidimensional monitoring data in the spatiotemporal synchronous driving sequence, perform time-series deduction of structural response evolution and output initial response status information, which includes structural displacement distribution, stress state, vibration characteristics and seepage pressure response.

5. A water conservancy project operation and maintenance management system based on digital twins according to claim 1, characterized in that, S2 further includes: S21: Receive the initial engineering response status and mark the continuous regions in the status data where the response value exceeds the preset response threshold range or the gradient change changes abruptly as potential local response holes. S22: Map the local response void to the structural topology network of the water conservancy project, which is constructed by the geometric connection relationship and mechanical transmission path of the engineering structure; at the same time, retrieve the normal operation data pattern of the void area under historical and similar environmental disturbances in the historical monitoring trajectory database.

6. A water conservancy project operation and maintenance management system based on digital twins according to claim 1, characterized in that, S3 includes: S31: Based on the reconstructed structural response status, identify multiple risk source nodes whose response values ​​exceed the warning threshold; based on the structural mechanics model and hydraulic transmission model of the water conservancy project, construct a risk transmission directed graph with the risk source nodes as the starting point and the key engineering facilities as the ending point; S32: In the directed graph of risk transmission, a path search algorithm is used to extract multiple potential risk transmission paths from each risk source node to the key engineering facilities; S33: Calculate the dominant causality and transmission potential of each path; S34: Based on the preset screening strategy, select at least one path with the highest comprehensive score of dominant causality and transmission potential as the key dominant risk path. S35: Identify nodes or connecting edges with high intermediation centrality on the dominant risk path and determine them as key control points; based on the simulation results of applying control measures to the key control points using the digital twin system, generate specific operation and maintenance control strategies and output them to the operation and maintenance execution module.

7. A water conservancy project operation and maintenance management system based on digital twins according to claim 6, characterized in that, The transfer entropy is used to measure the direction and intensity of information flow in the response sequence between nodes. The Granger causality strength is obtained by performing causality tests on the historical response sequences of adjacent nodes in the path and extracting a significant statistic as the causal strength.

Images