Concrete gravity dam operation period safety risk value calculation method and system

Abstract

The invention relates to the technical field of data processing, in particular to a concrete gravity dam operation period safety risk value calculation method and system.The method comprises the steps that structure monitoring data, environment data and operation data of a dam body in the operation period are continuously collected through a safety monitoring system; carrying out preprocessing and standardization, coordinate conversion and risk feature extraction on the collected data, and obtaining fusion state data through data fusion processing; constructing a dam body operation environment parameter set according to the environment data, and calculating a safety risk value by using a deep belief network model in combination with the environment parameter set and the fusion state data; and dividing risk levels according to the security risk values and outputting early warning information. According to the method, dynamic quantitative evaluation of the safety risk of the concrete gravity dam in the operation period is realized through multi-source data fusion and a deep learning technology, and a scientific basis is provided for dam body safety management and risk early warning.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method and system for calculating the safety risk value of a concrete gravity dam during its operation. Background Technology

[0002] As a key control structure in water conservancy projects, the operational safety of concrete gravity dams directly affects the safety of people's lives and property downstream, regional economic development, and ecological environment stability. With increasing service time, changes in hydrological and meteorological conditions, material deterioration, and geological evolution, the dam structure is prone to cumulative damage, and safety risks gradually become prominent. Traditional safety assessment methods often rely on single monitoring data (such as displacement and stress) or periodic manual inspections, lacking quantitative descriptions of risk levels and tiered control criteria, making it difficult to support refined decision-making.

[0003] The above content is only used to help understand the technical solution of the present invention and does not represent an admission that the above content is prior art. Summary of the Invention

[0004] The main objective of this invention is to provide a method and system for calculating the safety risk value during the operation of a concrete gravity dam. This aims to solve the technical problem that traditional concrete gravity dam safety assessment methods rely heavily on single monitoring data (such as displacement and stress) or periodic manual inspections, lacking quantitative descriptions of risk levels and tiered control criteria, making it difficult to support refined decision-making.

[0005] To achieve the above objectives, the present invention provides a method for calculating the safety risk value of a concrete gravity dam during its operation, the method comprising: The structural monitoring data, environmental data, and operational data of the concrete gravity dam during its operation are continuously collected through the concrete gravity dam safety monitoring system. The structural monitoring data, environmental data, and operational data are processed to obtain fused status data, wherein the data processing includes data preprocessing and standardization, coordinate transformation and extraction of risk features, and data fusion processing. A set of dam operating environment parameters is constructed based on environmental data, and a preset deep belief network model is used to calculate the safety risk value based on the set of dam operating environment parameters and the fused state data. Risk levels are classified based on safety risk values, and early warning information is output.

[0006] Optionally, the data preprocessing and standardization includes: Align the timestamps of the data streams of structural monitoring data, environmental data, and operational data to ensure that data from different monitoring sources remain consistent over time. Use the isolated forest algorithm or wavelet thresholding to remove noise and outliers from structural monitoring data, environmental data, and operational data. The processed structural monitoring data, environmental data, and operational data are standardized by using min-max normalization or Z-score standardization to unify the data dimensions.

[0007] Optionally, the coordinate transformation and risk feature extraction includes: Structural monitoring data from different monitoring points of a concrete gravity dam are converted to a unified three-dimensional coordinate system for the dam body; Structural risk characteristics of concrete gravity dams are extracted from the converted structural monitoring data, and environmental-operational risk characteristics of concrete gravity dams are extracted from environmental and operational data. The structural risk characteristics include dam displacement rate, stress-strain amplitude, and crack propagation parameters. The environmental-operational risk characteristics include water level fluctuation, seepage rate, temperature gradient, and gate opening and closing frequency.

[0008] Optionally, the data fusion process includes: Key risk factors for concrete gravity dams were identified based on structural risk characteristics and environmental-operational risk characteristics, and a state assessment model was initialized. The improved Hungarian algorithm is used to continuously match the key risk factors with real-time monitoring features; Based on the real-time monitoring characteristics, the evolution trend of the key risk factors is predicted using a state assessment model to obtain prediction results; The original state data is merged with the prediction results, and the parameters of the state evaluation model are updated in real time through an adaptive filter. The state assessment model is a dynamic fusion model based on an adaptive filter, which is a Kalman-Bayes combined filter used to correct the prediction error of key risk factors in real time.

[0009] Optionally, the improved Hungarian algorithm is implemented through the following steps: S101. Given a cost matrix, where the elements represent the Euclidean distance cost between the key risk factors and real-time monitoring features of a concrete gravity dam, the key risk factors include the cumulative displacement threshold of the dam body and the seepage threshold, and the real-time monitoring features include the current displacement value and the seepage flow. S102. Subtract the minimum value from each row and each column of the cost matrix to obtain the standardized cost matrix; S103. Find a specified position in the standardized cost matrix such that each row and each column has at least one marked 0 element; if the specified position cannot be found, proceed to step S105; otherwise, mark the specified position and set the other 0 elements in its row and column to an unmarked state. S104. Find unpaired 0 elements in the marked rows and connect them with 0 elements in the marked columns to form matching edges; if no unpaired 0 elements exist, proceed to the next step. S105. Modify the cost matrix according to the new matching edges, while retaining the existing matching edges; if all key risk factors have been matched with the monitoring features, the process ends; otherwise, return to step S103.

[0010] Optionally, the preset deep belief network model is a multilayer restricted Boltzmann machine, which includes an input layer, an output layer, and multiple hidden layers; The output of the hidden layer is represented as follows: ,in , , Use the Sigmoid activation function; In the formula, H is the hidden layer output, and X is the hidden layer input, which is the concatenated vector of environmental parameters and fused state data. and Here, represents the weights and biases of the hidden layer, m represents the number of input features, and k represents the number of neurons in the hidden layer. The output of the output layer is represented as follows: ,in , ; In the formula, This represents a safety risk value, ranging from 0 to 1. and ...

[0011] Optionally, the step of classifying risk levels based on security risk values ​​and outputting early warning information includes: When the safety risk value is less than or equal to the first preset threshold, it is determined to be a low risk level, and the normal operation information of the concrete gravity dam is output. When the safety risk value is greater than the first preset threshold and less than or equal to the second preset threshold, it is determined to be a medium risk level, a yellow warning signal is output, and the monitoring frequency of the concrete gravity dam is increased. When the safety risk value is greater than the second preset threshold, it is determined to be an extremely high risk level, a red warning signal is output and the emergency response mechanism for concrete gravity dams is activated, wherein the first preset threshold is less than the second preset threshold.

[0012] Furthermore, to achieve the above objectives, the present invention also provides a system for calculating the safety risk value during the operation period of a concrete gravity dam, the system comprising: The data acquisition module is used to continuously collect structural monitoring data, environmental data, and operational data of the concrete gravity dam during its operation through the concrete gravity dam safety monitoring system. The data processing module is used to process the structural monitoring data, environmental data, and operational data to obtain fused status data. The data processing includes data preprocessing and standardization, coordinate transformation and risk feature extraction, and data fusion processing. The risk calculation module is used to construct a set of dam operating environment parameters based on environmental data, and to calculate the safety risk value using a preset deep belief network model based on the set of dam operating environment parameters and the fused state data. The risk classification module is used to classify risk levels based on safety risk values ​​and output early warning information.

[0013] Furthermore, to achieve the above objectives, the present invention also provides a device for calculating the safety risk value during the operation period of a concrete gravity dam. The device includes: a memory, a processor, and a program for calculating the safety risk value during the operation period of a concrete gravity dam stored in the memory and executable on the processor. The program for calculating the safety risk value during the operation period of a concrete gravity dam is configured to implement the steps of the method for calculating the safety risk value during the operation period of a concrete gravity dam as described above.

[0014] In addition, to achieve the above objectives, the present invention also provides a storage medium storing a program for calculating the safety risk value of a concrete gravity dam during its operation period. When the program is executed by a processor, it implements the steps of the method for calculating the safety risk value of a concrete gravity dam during its operation period as described above.

[0015] This invention provides a method for calculating the safety risk value of a concrete gravity dam during its operation. The method integrates structural monitoring data, environmental data, and operational data, and through preprocessing, coordinate transformation, and feature extraction, achieves deep fusion of multi-source data, avoiding the limitations of a single data source and significantly improving the comprehensiveness and accuracy of risk assessment. Utilizing a deep belief network model combined with environmental parameter sets and fused state data, it can dynamically calculate the safety risk value, capturing the evolution trend of the dam's safety status in real time, providing a scientific basis for risk early warning. By classifying risk levels according to the safety risk value and outputting early warning information, it realizes a shift from a binary judgment of safety or unsafety to multi-level risk early warning, providing refined decision support for dam operation management. Through automated data processing and the application of deep learning models, it reduces manual intervention, improves the efficiency and reliability of risk assessment, and reduces the risk of human error. The data preprocessing and standardization methods used in the scheme can effectively remove noise and outliers, enhance the system's anti-interference ability in complex environments, and ensure the stability of the assessment results. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the structure of the safety risk value calculation device for the operation period of a concrete gravity dam in the hardware operating environment involved in the embodiments of the present invention; Figure 2 This is a flowchart illustrating an embodiment of the method for calculating the safety risk value of a concrete gravity dam during operation according to the present invention. Figure 3 This is a structural block diagram of an embodiment of the safety risk value calculation system for the operation period of a concrete gravity dam according to the present invention.

[0017] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0018] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0019] Reference Figure 1 , Figure 1 This is a schematic diagram of the structure of a concrete gravity dam operation period safety risk value calculation device in the hardware operating environment involved in the embodiments of the present invention.

[0020] like Figure 1 As shown, the equipment for calculating the safety risk value during the operation of a concrete gravity dam may include: a processor 1001, such as a central processing unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. The communication bus 1002 is used to enable communication between these components. The user interface 1003 may include a display screen, and optionally, it may also include a standard wired interface or a wireless interface. In this invention, the wired interface of the user interface 1003 may be a USB interface. The network interface 1004 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface). The memory 1005 may be a high-speed random access memory (RAM) or a non-volatile memory (NVM), such as a disk storage device. The memory 1005 may also optionally be a storage device independent of the aforementioned processor 1001.

[0021] Those skilled in the art will understand that Figure 1 The structure shown does not constitute a limitation on the equipment for calculating the safety risk value during the operation of a concrete gravity dam. It may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0022] like Figure 1 As shown, the memory 1005, which serves as a computer storage medium, may include an operating system, a network communication module, a user interface module, and a program for calculating the safety risk value during the operation of a concrete gravity dam.

[0023] exist Figure 1 In the concrete gravity dam operation period safety risk value calculation device shown, the network interface 1004 is mainly used to connect to the backend server and communicate data with the backend server; the user interface 1003 is mainly used to connect to peripheral devices; the concrete gravity dam operation period safety risk value calculation device calls the concrete gravity dam operation period safety risk value calculation program stored in the memory 1005 through the processor 1001 and executes the concrete gravity dam operation period safety risk value calculation method provided in the embodiment of the present invention.

[0024] Based on the above hardware structure, an embodiment of the method for calculating the safety risk value of concrete gravity dams during operation is proposed.

[0025] Reference Figure 2 , Figure 2 This is a flowchart illustrating an embodiment of the method for calculating the safety risk value of a concrete gravity dam during its operation, as presented in this invention.

[0026] In one embodiment, the method for calculating the safety risk value during the operation of the concrete gravity dam includes the following steps: Step S100: Continuously collect structural monitoring data, environmental data, and operational data of the concrete gravity dam during its operation period through the concrete gravity dam safety monitoring system.

[0027] The structural monitoring data can be real-time monitoring information reflecting the structural state of a concrete gravity dam, and can be used to assess the stability and degree of damage to the dam structure. In this embodiment, the structural monitoring data can be collected through a sensor network deployed on the dam body to obtain measured values ​​of physical quantities such as displacement, stress, strain, and seepage pressure. Furthermore, the structural monitoring data can participate in data fusion processing with environmental data and operational data to form fused state data. For example, the structural monitoring data may include, but is not limited to, one or more of displacement data, stress data, and seepage pressure data.

[0028] Environmental data can be external natural condition parameters affecting the operational state of a concrete gravity dam, and can be used as an important external stimulus input for analyzing changes in the dam's response. In an exemplary embodiment, environmental data can be collected through meteorological stations, hydrological stations, and geological monitoring equipment to obtain information such as temperature, rainfall, and upstream and downstream water levels. Furthermore, environmental data can be used to construct a set of dam operating environment parameters and input together with fused state data into a deep belief network model. Exemplarily, environmental data can include, but is not limited to, one or more of temperature data, water level data, and rainfall data.

[0029] Operational data can be records of operations and conditions generated during the scheduling and operation of a concrete gravity dam, reflecting the actual operational load borne by the dam body and its fluctuations. In one specific embodiment, operational data can be obtained from automated systems such as reservoir scheduling systems and gate control systems, collecting information such as discharge flow, flood discharge frequency, and reservoir capacity changes. Furthermore, operational data can be used in conjunction with structural monitoring data and environmental data to generate fused status data. For example, operational data may include, but is not limited to, one or more of the following: flood discharge flow data, reservoir water level change data, and gate opening and closing records.

[0030] Structural monitoring data, environmental data, and operational data can be collected by periodically or continuously acquiring these three types of data streams through a deployed data acquisition system. Furthermore, this operation can achieve real-time acquisition via wired or wireless sensor networks, and combine this with edge computing nodes for preliminary filtering and compression during transmission, thereby ensuring the integrity and timeliness of the raw data and supporting subsequent analysis.

[0031] Step S200 involves processing the structural monitoring data, environmental data, and operational data to obtain fused status data. The data processing includes data preprocessing and standardization, coordinate transformation and risk feature extraction, and data fusion processing.

[0032] The fused state data can be a multi-source data set comprehensively representing the dam's state after preprocessing and feature extraction. It can be used to provide a unified and representative input feature vector for calculating safety risk values. In this embodiment, the fused state data is obtained by preprocessing, coordinate transformation, feature extraction, and fusion of structural monitoring data, environmental data, and operational data. Furthermore, the fused state data, along with the dam's operational environment parameter set, can be used as input to a deep belief network model.

[0033] Data processing can involve sequentially performing data preprocessing, standardization, coordinate transformation, feature extraction, and fusion operations. Furthermore, data processing can employ sliding window denoising, Z-score standardization, principal component analysis for dimensionality reduction, and weighted fusion strategies to improve data quality and expressive power, thereby eliminating noise and dimensional differences, extracting key features, and enhancing the effectiveness of model input.

[0034] Step S300: Construct a set of dam operating environment parameters based on environmental data, and use a preset deep belief network model to calculate the safety risk value based on the set of dam operating environment parameters and the fused state data.

[0035] The dam operating environment parameter set can be a set of parameters extracted from environmental data and organized into a model-usable format, which can be used to characterize the intensity of the impact of the external environment on dam safety. In an exemplary embodiment, the dam operating environment parameter set can be filtered, normalized, and dimensionally integrated based on the time series of environmental data to form structured input variables. Furthermore, the dam operating environment parameter set can be input together with fused state data into a deep belief network model to calculate safety risk values.

[0036] A deep belief network model can be a deep generative model based on stacked restricted Boltzmann machines, which can be used for nonlinear feature learning and function approximation. In this embodiment, the deep belief network model is obtained through offline training using historical monitoring data and has multi-layer nonlinear mapping capabilities. Furthermore, the deep belief network model can receive fused state data and a set of dam operating environment parameters as input and output a safety risk value.

[0037] Calculating the safety risk value can be achieved by inputting fused state data and a set of dam operating environment parameters into a trained deep belief network model, dynamically outputting a risk score. Furthermore, this operation can incorporate an online fine-tuning mechanism, using the latest data to update the model weights to adapt to long-term evolution trends, thereby enabling a dynamic quantitative assessment of the dam's safety status.

[0038] Step S400: Classify risk levels based on safety risk values ​​and output early warning information.

[0039] The safety risk value can be a numerical indicator that quantifies the safety level of a concrete gravity dam under its current operating state, and can be used to support subsequent risk level classification and early warning decisions. In one specific embodiment, the safety risk value is dynamically output by a deep belief network model, reflecting the overall safety status of the dam. Furthermore, the safety risk value can serve as the basis for classifying risk levels and triggering the output of early warning information.

[0040] Risk levels can be different grades based on security risk values, representing different degrees of security threats. In this embodiment, risk levels map continuous security risk values ​​to discrete grades by setting a threshold range. Furthermore, risk levels can be derived based on security risk values ​​and used to generate corresponding warning information. For example, risk levels can include, but are not limited to, one or more of low risk, medium risk, and high risk.

[0041] Early warning information can be an output indicating abnormal safety status or potential risks in the dam body, and can be used to provide timely risk alerts and handling suggestions to operation and management personnel. In one exemplary embodiment, the early warning information automatically generates warning content in the form of text, sound and light, or remote push signals based on the risk level. Furthermore, the content and urgency of the early warning information can be determined by the risk level. For example, the early warning information may include, but is not limited to, one or more of the following: yellow warning, orange warning, and red warning.

[0042] Risk classification and early warning information output can be achieved by mapping safety risk values ​​to risk levels according to preset rules and generating corresponding early warning signals. Furthermore, this operation can configure multi-level response plans and automatically push them to monitoring platforms or responsible personnel's terminals, thereby realizing closed-loop management from quantitative assessment to qualitative response.

[0043] For example, in the scenario of dynamic monitoring of the safety of gravity dams during the flood season, the method for calculating the safety risk value of concrete gravity dams during operation in this embodiment can be as follows: During the main flood season, the system continuously collects data such as the deformation displacement of the dam, the rate of rise of the reservoir water level, temperature fluctuations, and the opening and closing frequency of the flood discharge gates; after data cleaning and feature fusion, a fused state data containing structural response and environmental excitation is formed; at the same time, rainfall intensity, upstream and downstream water level differences, etc., constitute the dam's operating environment parameter set; both are input into a trained deep belief network model, and the current safety risk value is output in real time; when the value exceeds a preset threshold, the system determines that it has entered a high-risk level, automatically triggers a red alert, and issues an alarm via SMS and the central control console, prompting the management unit to activate the emergency inspection and flood discharge scheduling plan.

[0044] This embodiment provides a method for calculating the safety risk value during the operation of a concrete gravity dam. By continuously collecting structural monitoring data, environmental data, and operational data, the method preprocesses, extracts features, and fuses multi-source data to generate fused state data. Combined with a set of dam operation environment parameters constructed from environmental data, the method uses an offline-trained deep belief network model to perform nonlinear mapping calculation of the safety risk value. Based on this risk value, the method performs hierarchical judgment and issues early warning information. This enables the system to comprehensively reflect the coupling effect of the dam structure state, external environmental excitation, and actual operating load, achieving continuous, dynamic, and refined assessment of the operational safety of concrete gravity dams. This reduces reliance on manual intervention, improves the accuracy of risk identification and the level of automation in response, and achieves the technical effect of improving the scientific nature and reliability of dam safety management.

[0045] Furthermore, the data preprocessing and standardization include: Aligning the timestamps of data streams from structural monitoring data, environmental data, and operational data ensures that data from different monitoring sources remain consistent over time.

[0046] Data stream timestamp alignment can be achieved by synchronizing structural monitoring data, environmental data, and operational data from different monitoring sources according to a unified time base, ensuring the comparability of each data point at the same time. In this embodiment, data stream timestamp alignment can be achieved by interpolation to fill in missing time points or by using time series resampling technology to achieve equal-interval alignment. For asynchronous acquisition systems, timestamp matching algorithms are used for precise alignment, thereby eliminating time misalignment caused by different acquisition frequencies or communication delays, and ensuring the accuracy of subsequent fusion processing.

[0047] Use the isolated forest algorithm or wavelet thresholding to remove noise and outliers from structural monitoring data, environmental data, and operational data.

[0048] Noise and outlier removal can be achieved by applying the Isolation Forest algorithm to identify and remove outliers from the data, or by using wavelet thresholding to decompose the signal and filter out high-frequency noise components. Furthermore, for environmental data with strong periodicity, wavelet transform combined with soft thresholding can be used for denoising. For monitoring data with non-Gaussian distribution, Isolation Forest can be used for anomaly detection, marking or correcting data from anomalous periods, thereby improving data quality and reducing the interference of false signals on feature extraction and model judgment.

[0049] The processed structural monitoring data, environmental data, and operational data are standardized by using min-max normalization or Z-score standardization to unify the data dimensions.

[0050] Standardization can involve unifying the dimensions of the denoised data, using min-max normalization to map the data to a specified interval, or using Z-score standardization to make the data conform to a zero-mean, unit-variance distribution. In an exemplary embodiment, a suitable standardization method can be selected based on the data distribution characteristics. For example, Z-score can be preferentially used for data that follows a normal distribution, while min-max normalization can be used for data with well-defined boundaries. Furthermore, the standardization parameters should be kept consistent before inputting them into the model, thereby avoiding the problem of numerical dominance between different physical magnitudes and enhancing the model training stability and feature fusion effect.

[0051] Taking the integration and processing of multi-source asynchronous monitoring data as an example, the method for calculating the safety risk value during the operation period of a concrete gravity dam in this embodiment can be as follows: A concrete gravity dam is equipped with a variety of monitoring devices, of which the deformation sensor uploads displacement data every 10 minutes, the meteorological station records temperature and rainfall every hour, and the gate control system only generates operation logs during operation; the system first timestamps all data in 5-minute increments, and supplements the environmental parameters at intermediate moments through linear interpolation. After detecting a set of abrupt seepage pressure readings in the structural monitoring data, the isolated forest algorithm is used to determine that it is an outlier and remove it; then, Z-score standardization (applicable to stress and temperature) and min-max normalization (applicable to reservoir water level and flow rate) are applied to all variables respectively, and finally a clean dataset in a unified format is output for feature extraction and fusion.

[0052] Furthermore, the coordinate transformation and risk feature extraction includes: transforming the structural monitoring data of different monitoring points of the concrete gravity dam into a unified three-dimensional coordinate system of the dam body; The dam's three-dimensional coordinate system serves as a unified spatial reference for concrete gravity dams. It maps structural monitoring data from different monitoring points onto the same spatial framework for geometric alignment, enabling the alignment and comparison of data from different monitoring points within this unified framework. In an exemplary embodiment, the dam's three-dimensional coordinate system can be constructed by calibrating the X, Y, and Z axes and the origin position using surveying methods such as a total station or GNSS, based on dam design drawings and measurement control points. Furthermore, the dam's three-dimensional coordinate system can serve as a target reference system for coordinate transformation of structural monitoring data.

[0053] Coordinate transformation can be the operation of mapping structural monitoring data distributed in different spatial locations to a unified three-dimensional coordinate system of the dam body based on their physical installation coordinates. For example, coordinate transformation can be achieved by using coordinate transformation matrices to perform translation, rotation, and scaling operations, eliminating data deviations caused by differences in sensor deployment orientation. This ensures spatial comparability of multi-point monitoring data and improves the consistency of subsequent feature extraction.

[0054] Structural risk characteristics of concrete gravity dams were extracted from the converted structural monitoring data, and environmental-operational risk characteristics of concrete gravity dams were extracted from environmental and operational data. Structural risk characteristics include dam displacement rate, stress-strain amplitude, and crack propagation parameters. Environmental-operational risk characteristics include water level fluctuation, seepage rate, temperature gradient, and gate opening and closing frequency.

[0055] Structural risk characteristics can be a set of key parameters characterizing the evolution trend of the concrete gravity dam's structural state, reflecting the intensity of changes in the internal mechanical behavior of the dam structure and the development of potential damage. In this embodiment, structural risk characteristics can be obtained from structural monitoring data transformed to a unified coordinate system through differential operations, extreme value identification, or curve fitting methods. Furthermore, structural risk characteristics can be generated from the transformed structural monitoring data and incorporated into the fused state data. For example, structural risk characteristics may include, but are not limited to, one or more of the following: dam displacement rate, stress-strain amplitude, crack propagation parameters, etc.

[0056] Environmental-operational risk characteristics can be dynamic indicators that comprehensively reflect the impact of external environmental changes and operational actions on the dam body. They can be used to characterize the severity of external stimuli and the safety responses they may induce. In one specific embodiment, environmental-operational risk characteristics can be obtained from environmental and operational data by extracting statistical features such as change rates, gradients, or frequencies of occurrence. Furthermore, environmental-operational risk characteristics can form a core component of fused state data together with structural risk characteristics. For example, environmental-operational risk characteristics may include, but are not limited to, one or more of the following: water level fluctuations, seepage rate increases, temperature gradients, and gate opening and closing frequencies.

[0057] Extracting risk features can be an operation that identifies and calculates quantitative indicators with safety implications from structural, environmental, and operational data. For example, risk features can be extracted by differentiating the displacement sequence to obtain the displacement rate, performing sliding window growth rate analysis on seepage data, calculating the temperature gradient using a temperature difference algorithm, and counting the number of gate actions per unit time. This can transform raw observations into low-dimensional, high-information-density features that have greater physical meaning and predictive value.

[0058] For example, in the scenario of monitoring non-uniform expansion of the dam body during the spring snowmelt season, the method for calculating the safety risk value of the concrete gravity dam during operation in this embodiment can be as follows: As the temperature rises, a significant temperature difference occurs between the upstream and downstream surfaces of the dam body, while the reservoir water level rises slowly due to snowmelt replenishment; the system collects temperature and deformation data from each measuring point, converts the readings of the dispersed displacement gauges to the three-dimensional coordinate system of the dam body, calculates the vertical and horizontal displacement rates, and extracts the temperature gradient as an environmental-operational risk feature; combined with the seepage rate increase and historical crack monitoring records, an abnormal expansion trend is identified in a certain dam section; after this feature combination is input into the deep confidence network model, the output safety risk value increases, triggering a medium-level warning, prompting maintenance personnel to pay close attention to the thermal expansion effect and the structural coordination deformation capacity of the area.

[0059] This embodiment transforms structural monitoring data from different monitoring points of a concrete gravity dam into a unified three-dimensional coordinate system of the dam body. It then extracts structural risk characteristics of the concrete gravity dam from the transformed structural monitoring data, as well as environmental and operational risk characteristics from environmental and operational data. By eliminating the spatial heterogeneity of multi-source monitoring data through a unified spatial reference frame, and transforming the original observation data into low-dimensional, high-information-density features with clear engineering physical meaning, this embodiment achieves the technical effects of improving the accuracy of multi-point monitoring data fusion, enhancing the sensitivity and interpretability of state characterization, and providing high-quality input for safety assessment models to improve the rationality and timeliness of risk identification.

[0060] In one embodiment, the data fusion process includes: Key risk factors for concrete gravity dams were identified based on structural risk characteristics and environmental-operational risk characteristics, and a state assessment model was initialized. Key risk factors can be core parameters that significantly influence the safety state of concrete gravity dams, identified from structural risk characteristics and environmental-operational risk characteristics. These parameters can characterize the key drivers of safety evolution. In this embodiment, key risk factors can be obtained through correlation analysis, principal component analysis, or by screening characteristic variables strongly correlated with the dam's abnormal response based on prior knowledge. Furthermore, key risk factors are jointly identified by structural risk characteristics and environmental-operational risk characteristics and participate in the prediction process of the state assessment model. For example, key risk factors may include, but are not limited to, one or more of the following: dam segment displacement rate anomaly factor, seepage pressure growth factor, and temperature-stress coupling factor.

[0061] Identifying key risk factors can be based on the correlation between structural risk characteristics and environmental-operational risk characteristics, as well as engineering experience, to screen a subset of features that drive safety changes. Furthermore, identifying key risk factors can be achieved by using mutual information calculation, sensitivity analysis, or expert rule base matching methods to extract key variables from high-dimensional features. This reduces redundant input to the model, focuses on core risk sources, and improves assessment efficiency and interpretability.

[0062] A state assessment model can be a mathematical modeling tool used to evaluate the evolution trend of key risk factors in concrete gravity dams. It can output predicted values ​​of the future states of key risk factors, supporting early risk identification. In an exemplary embodiment, the state assessment model is built based on an adaptive filter, continuously receiving matching results and performing recursive predictions after initialization. Furthermore, the state assessment model receives matching results of key risk factors and real-time monitoring features as input, and its parameters are updated by a Kalman-Bayes combined filter.

[0063] The improved Hungarian algorithm is used to continuously match key risk factors with real-time monitoring features; The improved Hungarian algorithm, an optimized task allocation algorithm, is used to solve multi-objective matching problems. In this scheme, it achieves the optimal association between risk factors and real-time monitoring features, enabling efficient and accurate matching between key risk factors and the latest collected monitoring features, thus avoiding mismatches that lead to evaluation bias. In one specific embodiment, the improved Hungarian algorithm introduces a dynamic weight adjustment mechanism based on the traditional Hungarian algorithm to adapt to the time-series changes in monitoring features. Furthermore, the improved Hungarian algorithm connects key risk factors and real-time monitoring features, providing the correct input mapping for the state assessment model.

[0064] Matching key risk factors with real-time monitoring features can be achieved by using an improved Hungarian algorithm to establish the optimal correspondence between key risk factors and current monitoring features. Furthermore, matching key risk factors with real-time monitoring features can be achieved by setting the feature similarity within a time sliding window as the matching cost function, dynamically adjusting the allocation strategy to cope with sensor drift or the addition of new measurement points. This ensures a consistent mapping between historical models and current observations, preventing misjudgments due to feature misalignment.

[0065] Based on real-time monitoring characteristics, a state assessment model is used to predict the evolution trend of key risk factors and obtain prediction results. Predicting the evolution trend of key risk factors can be based on a state assessment model, combining historical states with the latest matching results to generate a predictive output of future short-term evolution. Furthermore, predicting the evolution trend of key risk factors can employ a recursive prediction structure, updating the input and rolling out multi-step prediction results each period, thereby enabling proactive perception of potential risk development and supporting early warning.

[0066] The original state data is merged with the prediction results, and the parameters of the state evaluation model are updated in real time through an adaptive filter. The Kalman-Bayes combined filter can be an adaptive filtering method that integrates the advantages of Kalman filtering and Bayesian inference. It is used to dynamically estimate and correct system states, and can be used for online estimation and compensation of prediction errors for key risk factors, improving prediction accuracy and stability. For example, the Kalman-Bayes combined filter combines the state prediction-update framework of Kalman filtering with the probability density correction capability of Bayesian methods to adjust model uncertainty in real time. Furthermore, as a specific implementation of an adaptive filter, the Kalman-Bayes combined filter is used to update the parameters of the state assessment model in real time and fuse the original state data with the prediction results.

[0067] Merging existing state data with prediction results and updating model parameters can be achieved by fusing historical state data with current predictions and using a Kalman-Bayes combined filter to correct the model's internal parameters. Furthermore, this operation can be triggered by setting a prediction residual covariance threshold to establish a parameter readjustment mechanism, enabling rapid convergence to the true state after abrupt events. This enhances the model's adaptability to non-stationary conditions and maintains its long-term evaluation accuracy.

[0068] Taking the dynamic tracking of seepage risk during heavy rainfall as an example, the method for calculating the safety risk value of a concrete gravity dam during operation in this embodiment can be as follows: During continuous heavy rain, the system identifies the seepage flow rate increase and downstream seepage pressure anomaly as key risk factors; the improved Hungarian algorithm is dynamically matched with the real-time piezometer readings distributed in different dam sections, maintaining correct correlation even if some sensor signals are delayed; the state assessment model predicts that the seepage pressure in a certain area may exceed the limit within the next two hours based on the matching result; at the same time, the Kalman-Bayes combined filter corrects the model parameters in real time according to the deviation between the measured value and the predicted value, and merges the corrected prediction result with the original state data, finally outputting a continuously updated safety risk value, providing a reliable basis for flood control scheduling.

[0069] This embodiment identifies key risk factors of concrete gravity dams based on structural risk characteristics and environmental-operational risk characteristics, initializes a state assessment model, and continuously matches key risk factors with real-time monitoring features using an improved Hungarian algorithm. Based on the real-time monitoring features, the state assessment model predicts the evolution trend of key risk factors to obtain prediction results. The original state data and prediction results are merged, and the parameters of the state assessment model are updated in real time through an adaptive filter. By identifying key risk factors to focus on core risk sources and reduce the dimensionality of model input, the improved Hungarian algorithm achieves dynamic and accurate mapping between multi-source monitoring features and risk factors. The state assessment model enables recursive prediction of the evolution path of key factors, and the Kalman-Bayes combined filter is used to compensate for prediction errors online and adaptively adjust model parameters. This significantly improves the robustness and timeliness of data fusion processing, enhances the dynamic adaptability and prediction reliability of safety risk assessment under complex working conditions, and achieves the technical effect of improving the reliability of prediction.

[0070] In one embodiment, the improved Hungarian algorithm is implemented through the following steps: S101. Given a cost matrix, where the elements represent the Euclidean distance cost between the key risk factors and real-time monitoring features of a concrete gravity dam, the key risk factors include the cumulative displacement threshold of the dam body and the seepage threshold, and the real-time monitoring features include the current displacement value and the seepage flow. The cost matrix can be a cost matrix with key risk factors as rows and real-time monitoring features as columns, where each element is the Euclidean distance between the two. This matrix quantifies the cost of different matching combinations. In this embodiment, the cost matrix is ​​generated by calculating the Euclidean distance between each key risk factor and all real-time monitoring features; a smaller distance indicates a higher matching probability. Furthermore, the cost matrix is ​​jointly constructed from key risk factors and real-time monitoring features for subsequent standardization and matching operations. The Euclidean distance cost can be a measure of the geometric distance between two points calculated based on the threshold of a key risk factor and the corresponding real-time monitoring feature. This reflects the degree of deviation between the actual observed value and the preset risk benchmark, serving as a basis for matching priority. In an exemplary embodiment, the Euclidean distance cost is calculated based on the dam's cumulative displacement threshold and the current displacement value, and the seepage flow threshold and the seepage flow rate, respectively. The cost matrix can be a two-dimensional matrix constructed based on key risk factors and real-time monitoring features, where each element is the corresponding Euclidean distance, thus reflecting the ease or difficulty of matching different combinations. Furthermore, the construction cost matrix can be accelerated by using vectorized computation to solve many-to-many distances and supports dynamic addition and deletion of factors and feature dimensions, thus enabling the technical effect of providing structured input for the matching process.

[0071] S102. Subtract the minimum value from each row and each column of the cost matrix to obtain the standardized cost matrix; The standardized cost matrix can be obtained by subtracting the minimum value from each row and column of the original cost matrix, and is used for subsequent labeling and matching operations. For example, the standardized cost matrix is ​​generated by subtracting the minimum element value from each row and column of the original cost matrix, maintaining non-negativity while achieving normalization. Furthermore, the standardized cost matrix is ​​derived from the cost matrix and forms the basis for subsequent labeling and matching operations. Generating the standardized cost matrix can be achieved by subtracting the minimum value of each row or column of the cost matrix, resulting in a non-negative standardized cost matrix. In a specific embodiment, a threshold protection mechanism can be introduced to generate the standardized cost matrix to prevent negative values ​​due to floating-point errors, thereby improving the algorithm's adaptability to data of different scales and eliminating the influence of dimensions and offsets.

[0072] S103. Find a specified position in the standardized cost matrix such that each row and each column has at least one marked 0 element; if the specified position cannot be found, proceed to step S105; otherwise, mark the specified position and set the other 0 elements in its row and column to an unmarked state. Marking a zero element at a specified position can be achieved by finding a set of zero elements located in different rows and columns, ensuring that each row and column is covered by at least one zero element. In this embodiment, marking a zero element at a specified position can be achieved by combining row and column scanning with a greedy strategy to prioritize isolated zero points, reducing the probability of collisions. Furthermore, this operation can determine a preliminary feasible matching set, providing candidates for forming a complete matching edge. For example, when a row or column has a unique zero element, that position is marked first, and the marking status of other zero elements in the same row and column is suppressed to ensure that each row and column has at most one valid mark.

[0073] S104. Find unpaired 0 elements in the marked rows and connect them with 0 elements in the marked columns to form matching edges; if no unpaired 0 elements exist, proceed to the next step. The matching edge can be a one-to-one correspondence between key risk factors and monitoring features established by marking zero elements in a standardized cost matrix. This ensures accurate information mapping by achieving a one-to-one allocation of key risk factors and monitoring features. In an exemplary embodiment, the matching edge is generated by finding the zero element that is not involved in pairing in the marked row and connecting it with the zero element marked in the same column to form a unique matching edge. Furthermore, the operation of forming matching edges can progressively build effective matching relationships and advance the overall matching process. For example, a weighted backtracking mechanism can be introduced to form matching edges, selecting the connection with higher historical consistency when multiple alternative paths exist, thereby improving matching stability.

[0074] S105. Modify the cost matrix according to the new matching edges, while retaining the existing matching edges; if all key risk factors have been matched with the monitoring features, the process ends; otherwise, return to step S103.

[0075] Modifying the cost matrix and iterative matching can be done when a complete match is not achieved. Based on the current matching edges, the cost matrix is ​​modified, and the process returns to the standardization steps to continue iterating. In this embodiment, modifying the cost matrix and iterative matching can construct new zero points by increasing the minimum value of uncovered rows and decreasing the value of intersection regions. Furthermore, this operation ensures that the algorithm can converge in complex situations, improving the matching success rate. For example, when some key risk factors have not yet been paired, the system adjusts the cost matrix to generate a new distribution of zero elements, and then re-executes the labeling and matching process until all factors are bound.

[0076] For example, in the scenario of identifying seepage anomalies in multiple dam sections during the flood season, the method for calculating the safety risk value of a concrete gravity dam during operation in this embodiment can be as follows: During heavy rainfall, the system detects an increase in seepage flow in multiple dam sections, and it is necessary to accurately match key risk factors such as seepage flow thresholds with real-time seepage flow data distributed at different locations; the system first calculates the Euclidean distance between each factor and the monitoring value to form a cost matrix, and performs row and column standardization processing; then, it attempts to establish an initial match by marking independent zero points. If a certain seepage pressure anomaly point is not covered, the matrix adjustment mechanism is triggered and iterative optimization is performed, ultimately achieving accurate binding of all key risk factors with corresponding measuring points, ensuring the correctness of the input to the state assessment model.

[0077] This embodiment constructs a cost matrix based on Euclidean distance and performs row and column standardization to eliminate the influence of dimensional differences and offsets between data. It determines the initial matching position by marking zero elements in the standardized matrix that satisfy row and column uniqueness, and connects these marked elements to form one-to-one matching edges. When the matching is incomplete, the cost matrix is ​​adjusted according to the existing matching state, and the marking and matching process is repeated until all key risk factors and real-time monitoring features are paired. Through the synergistic effect of these steps, the accuracy and robustness of the matching between key risk factors and monitoring features can be enhanced. Especially in scenarios with multi-source concurrency and signal overlap, it can maintain high reliability, providing an accurate data alignment foundation for subsequent state prediction and risk assessment.

[0078] In one embodiment, the preset deep belief network model is a multilayer restricted Boltzmann machine, which includes an input layer, an output layer, and multiple hidden layers. The deep neural network structure can be a deep neural network structure composed of multiple stacked restricted Boltzmann machines. It is used to perform layer-by-layer nonlinear transformations on the concatenated vector of environmental parameters and fused state data to extract high-order feature representations. It can serve as a specific implementation of a deep belief network model and undertake the calculation of risk assessment indicators. In this embodiment, the deep neural network structure can initialize the hidden layer parameters through unsupervised layer-by-layer pre-training and combine this with supervised fine-tuning to optimize the output layer weights, thereby achieving the modeling of complex nonlinear relationships in the input data. Furthermore, the deep neural network structure can receive the concatenated vector of environmental parameters and fused state data as input, and after layer-by-layer transformation through multiple hidden layers, output a risk assessment indicator in the range of 0 to 1.

[0079] The output of the hidden layer is represented as follows: ,in , , Use the Sigmoid activation function; In the formula, H is the hidden layer output, and X is the hidden layer input, which is the concatenated vector of environmental parameters and fused state data. and Here, represents the weights and biases of the hidden layer, m represents the number of input features, and k represents the number of neurons in the hidden layer. The hidden layer forward propagation computation can be achieved by weighting and summing the input vector X (a concatenation of environmental parameters and fused state data) with the hidden layer's weight matrix W and bias vector b, and then mapping it to the hidden layer output H through a sigmoid activation function. In an exemplary embodiment, the hidden layer forward propagation computation can accelerate convergence by batch normalizing the input data or using a dropout mechanism to prevent overfitting, thereby achieving a nonlinear transformation of the input features and extracting deep abstract representations.

[0080] The output of the output layer is represented as follows: ,in , ; In the formula, This represents a safety risk value, ranging from 0 to 1. and where represents the weights and biases of the output layer, and q represents the number of outputs of the output layer.

[0081] Specifically, the risk value generation of the output layer can be achieved by weighted summing of the output of the last hidden layer and the weights and biases of the output layer, and then compressing it to the 0-1 range using the Sigmoid function to obtain a risk assessment index with a value range of 0-1. For example, the generation of output layer risk values ​​can incorporate a temperature coefficient to adjust the output sensitivity, or combine a moving average strategy to smooth out instantaneous fluctuations, so that the output has a quantitative indicator with clear physical meaning, which facilitates subsequent risk level classification.

[0082] For example, in the scenario of dynamic risk assessment under extreme rainstorm conditions, the method for calculating the safety risk value of the concrete gravity dam during operation in this embodiment can be as follows: During continuous heavy rainfall, the system collects abnormal signals in real time, such as rapid rise in reservoir water level, increased seepage pressure in dam foundation, and drastic changes in temperature; after data fusion processing, a high-dimensional state vector is formed, which is then spliced ​​with environmental parameters and input into a deep belief network model composed of a deep neural network structure; each hidden layer extracts composite risk features step by step, and the final output layer outputs the current risk assessment index as 0.87. Based on this, the system determines it to be a high-risk level, automatically triggers a red alert, and suggests that the dispatch center release floodwater in advance to reduce the load.

[0083] This embodiment provides a method for calculating the safety risk value of a concrete gravity dam during its operation. It employs a deep neural network structure composed of multiple stacked restricted Boltzmann machines to perform layer-by-layer nonlinear transformation on the concatenated vector of environmental parameters and fused state data. The method utilizes forward propagation calculation in the hidden layers to achieve nonlinear mapping of input features to extract deep abstract representations. The final feature representation is transformed into a continuous risk assessment index within the range of 0 to 1 through risk value generation in the output layer. The model parameters are optimized through a combination of unsupervised pre-training and supervised fine-tuning. The Sigmoid activation function ensures the interpretability and continuity of the output. This method enhances the model's ability to capture complex coupling relationships between variables, improving the accuracy, stability, and adaptability to operating conditions during dynamic risk assessment.

[0084] Furthermore, the step of classifying risk levels based on safety risk values ​​and outputting early warning information includes: When the safety risk value is less than or equal to the first preset threshold, it is determined to be a low-risk level, and the normal operation information of the concrete gravity dam is output.

[0085] The first preset threshold can be a benchmark value used to distinguish between low-risk and medium-risk levels, and can be used as a basis for determining whether a concrete gravity dam is in a low-risk state. In this embodiment, the first preset threshold is set based on historical monitoring data statistical analysis, model verification, and engineering experience, and can be adjusted through iterative optimization. Furthermore, the first preset threshold is compared with a safety risk value to determine the risk level, participating in the classification logic of medium and low risk. Determining a low-risk level and outputting normal operation information can be achieved by comparing the safety risk value with the first preset threshold; if it is less than or equal to the threshold, the low-risk logic branch is executed. In an exemplary embodiment, this operation can be achieved by writing the determination result to a log system, synchronously updating the status indicator light on the monitoring interface to green, and generating periodic operation reports, thereby enabling automated confirmation and feedback of the normal operating status and reducing unnecessary intervention.

[0086] When the safety risk value is greater than the first preset threshold and less than or equal to the second preset threshold, it is determined to be at a medium risk level, a yellow warning signal is output, and the monitoring frequency of the concrete gravity dam is increased.

[0087] The second preset threshold can be a benchmark value used to distinguish between medium and extremely high risk levels, and can serve as a key criterion for determining whether a concrete gravity dam has entered an extremely high risk state. For example, the second preset threshold is set in conjunction with structural safety margin, extreme condition simulation results, and operation and management requirements, and is typically higher than the first preset threshold. Furthermore, the second preset threshold is compared with the safety risk value to participate in the determination of the extremely high risk level and trigger the emergency response mechanism. The yellow warning signal can be a form of warning information output indicating a moderate level of safety risk, which can be used to prompt operation and management personnel to increase monitoring frequency, conduct special inspections, or prepare emergency plans. In this embodiment, the yellow warning signal is automatically generated when the safety risk value exceeds the first preset threshold but does not exceed the second preset threshold. Furthermore, the yellow warning signal is triggered by the medium risk level and has a mapping relationship with the warning information. Determining a medium risk level and outputting a yellow warning signal can be achieved by activating the medium risk response logic when the safety risk value is between the first and second preset thresholds. In one specific embodiment, this operation can be achieved by automatically increasing the sensor sampling frequency, pushing reminders to the operation and maintenance terminal, and activating the trend prediction module to assist in the analysis, thereby enabling early risk identification and monitoring enhancement, and improving the ability to deal with potential problems.

[0088] When the safety risk value is greater than the second preset threshold, it is determined to be an extremely high risk level, a red warning signal is output, and the emergency response mechanism for concrete gravity dams is activated, wherein the first preset threshold is less than the second preset threshold.

[0089] A red alert signal can be an emergency warning message indicating an extremely high safety risk. It can be used to trigger an emergency response mechanism, notifying relevant units to take emergency dispatch, evacuation, or rescue measures. In one exemplary embodiment, the red alert signal is automatically activated when the safety risk value exceeds a second preset threshold. Furthermore, the red alert signal is triggered by an extremely high risk level, directly linking to the activation of the emergency response mechanism. Determining an extremely high risk level and outputting a red alert signal can be achieved by immediately executing the highest-level alarm procedure when the safety risk value exceeds the second preset threshold. For example, this operation can be implemented by linking a broadcast system to issue audible and visual alarms, sending emergency messages to the command center, and automatically invoking GIS positioning and impact range simulation tools, thereby ensuring a comprehensive response is initiated immediately upon the occurrence of a serious risk. The emergency response mechanism can be a pre-defined and configured rapid response procedure and linkage measures for major safety threats. It can be used to quickly control the situation and reduce potential losses when an extremely high risk is detected. In this embodiment, the emergency response mechanism achieves multi-departmental collaborative response through preset procedures and management protocols, including actions such as flood discharge operations, personnel evacuation, and equipment shutdown. Furthermore, the emergency response mechanism is triggered by a red alert signal and relies on a safety risk value reaching a specific threshold. Activating the emergency response mechanism can simultaneously initiate pre-set emergency response procedures upon issuing a red alert signal. In one specific embodiment, this operation can be achieved by calling the scheduling system via an interface to execute flood discharge operations, connecting to the communication platform to notify responsible personnel, and activating video surveillance of key areas, thereby forming a closed-loop control from risk identification to action execution and enhancing overall safety management capabilities.

[0090] For example, in the scenario of risk response during a rapid rise in reservoir water level caused by heavy rainfall, the method for calculating the safety risk value of a concrete gravity dam during operation in this embodiment can be as follows: During continuous heavy rain, the safety risk value of a concrete gravity dam gradually increases due to increased seepage pressure, intensified micro-deformation of the dam body, and the reservoir capacity approaching the flood limit level. After the system detects that the safety risk value exceeds the first preset threshold, it automatically determines it to be at a medium risk level, issues a yellow warning signal, and prompts the dispatching department to increase the frequency of patrols and open additional monitoring channels. As the rainfall continues, the safety risk value further exceeds the second preset threshold, and the system immediately determines it to be at an extremely high risk level, issues a red warning signal, and automatically activates the emergency response mechanism, closes unnecessary water intakes, opens the spillway gates for flood discharge, and pushes a hazard report to the flood control headquarters, completing the automatic linkage of the entire process from monitoring to response.

[0091] This embodiment provides a method for calculating the safety risk value during the operation of a concrete gravity dam. By comparing the safety risk value with a first preset threshold and a second preset threshold, it sequentially determines the risk levels as low, medium, and extremely high. It then outputs normal operation information, a yellow warning signal and enhanced monitoring prompts, and a red warning signal and emergency response mechanism activation instructions, respectively. By constructing an orderly judgment logic using graded thresholds, it achieves hierarchical identification and differentiated response to risk states. This allows the system to automatically match corresponding management actions based on the degree of risk. The yellow warning supports pre-intervention, and the red warning links to emergency response procedures, significantly improving the response speed and control capability to sudden high-risk events. This achieves the technical effect of enhancing the safety of dam operation and the level of management automation.

[0092] Furthermore, this embodiment of the invention also proposes a storage medium storing a program for calculating the safety risk value of a concrete gravity dam during its operation period. When the program is executed by a processor, it implements the steps of the method for calculating the safety risk value of a concrete gravity dam during its operation period as described above.

[0093] In addition, refer to Figure 3 This invention also proposes a system for calculating the safety risk value during the operation of a concrete gravity dam. The system includes: Data acquisition module 10 is used to continuously collect structural monitoring data, environmental data and operational data of concrete gravity dam during its operation through the concrete gravity dam safety monitoring system; Data processing module 20 is used to process the structural monitoring data, environmental data and operational data to obtain fused status data, wherein the data processing includes data preprocessing and standardization, coordinate transformation and risk feature extraction, and data fusion processing; Risk calculation module 30 is used to construct a set of dam operating environment parameters based on environmental data, and to calculate the safety risk value using a preset deep belief network model based on the set of dam operating environment parameters and the fused state data. The risk classification module 40 is used to classify risk levels based on safety risk values ​​and output early warning information.

[0094] Other embodiments or specific implementations of the concrete gravity dam operation safety risk value calculation system described in this invention can be referred to the above-mentioned method embodiments, and will not be repeated here.

[0095] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A method for calculating the safety risk value during the operation period of a concrete gravity dam, characterized in that, The method includes: The structural monitoring data, environmental data, and operational data of the concrete gravity dam during its operation are continuously collected through the concrete gravity dam safety monitoring system. The structural monitoring data, environmental data, and operational data are processed to obtain fused status data, wherein the data processing includes data preprocessing and standardization, coordinate transformation and extraction of risk features, and data fusion processing. A set of dam operating environment parameters is constructed based on environmental data, and a preset deep belief network model is used to calculate the safety risk value based on the set of dam operating environment parameters and the fused state data. Risk levels are classified based on safety risk values, and early warning information is output.

2. The method for calculating the safety risk value during the operation period of a concrete gravity dam as described in claim 1, characterized in that, The data preprocessing and standardization include: Align the timestamps of the data streams of structural monitoring data, environmental data, and operational data to ensure that data from different monitoring sources remain consistent over time. Use the isolated forest algorithm or wavelet thresholding to remove noise and outliers from structural monitoring data, environmental data, and operational data. The processed structural monitoring data, environmental data, and operational data are standardized by using min-max normalization or Z-score standardization to unify the data dimensions.

3. The method for calculating the safety risk value during the operation period of a concrete gravity dam as described in claim 1, characterized in that, The coordinate transformation and risk feature extraction include: Structural monitoring data from different monitoring points of a concrete gravity dam are converted to a unified three-dimensional coordinate system for the dam body; Structural risk characteristics of concrete gravity dams are extracted from the converted structural monitoring data, and environmental-operational risk characteristics of concrete gravity dams are extracted from environmental and operational data. The structural risk characteristics include dam displacement rate, stress-strain amplitude, and crack propagation parameters. The environmental-operational risk characteristics include water level fluctuation, seepage rate, temperature gradient, and gate opening and closing frequency.

4. The method for calculating the safety risk value during the operation period of a concrete gravity dam as described in claim 3, characterized in that, The data fusion process includes: Key risk factors for concrete gravity dams were identified based on structural risk characteristics and environmental-operational risk characteristics, and a state assessment model was initialized. The improved Hungarian algorithm is used to continuously match the key risk factors with real-time monitoring features; Based on the real-time monitoring characteristics, the evolution trend of the key risk factors is predicted using a state assessment model to obtain prediction results; The original state data is merged with the prediction results, and the parameters of the state evaluation model are updated in real time through an adaptive filter. The state assessment model is a dynamic fusion model based on an adaptive filter, which is a Kalman-Bayes combined filter used to correct the prediction error of key risk factors in real time.

5. The method for calculating the safety risk value during the operation period of a concrete gravity dam as described in claim 4, characterized in that, The improved Hungarian algorithm is implemented through the following steps: S101. Given a cost matrix, where the elements represent the Euclidean distance cost between the key risk factors and real-time monitoring features of a concrete gravity dam, the key risk factors include the cumulative displacement threshold of the dam body and the seepage threshold, and the real-time monitoring features include the current displacement value and the seepage flow. S102. Subtract the minimum value from each row and each column of the cost matrix to obtain the standardized cost matrix; S103. Find a specified position in the standardized cost matrix such that each row and each column has at least one marked 0 element; if the specified position cannot be found, proceed to step S105; otherwise, mark the specified position and set the other 0 elements in its row and column to an unmarked state. S104. Find unpaired 0 elements in the marked rows and connect them with 0 elements in the marked columns to form matching edges; if no unpaired 0 elements exist, proceed to the next step. S105. Modify the cost matrix according to the new matching edges, while retaining the existing matching edges; if all key risk factors have been matched with the monitoring features, the process ends; otherwise, return to step S103.

6. The method for calculating the safety risk value during the operation period of a concrete gravity dam as described in claim 1, characterized in that, The preset deep belief network model is a multilayer restricted Boltzmann machine, which includes an input layer, an output layer, and multiple hidden layers. The output of the hidden layer is represented as follows: ,in , , Use the Sigmoid activation function; In the formula, H is the hidden layer output, and X is the hidden layer input, which is the concatenated vector of environmental parameters and fused state data. and Here, represents the weights and biases of the hidden layer, m represents the number of input features, and k represents the number of neurons in the hidden layer. The output of the output layer is represented as follows: ,in , ; In the formula, This is a safety risk value, ranging from 0 to 1. and ...

7. The method for calculating the safety risk value during the operation period of a concrete gravity dam as described in claim 1, characterized in that, The process of classifying risk levels based on security risk values ​​and outputting early warning information includes: When the safety risk value is less than or equal to the first preset threshold, it is determined to be a low risk level, and the normal operation information of the concrete gravity dam is output. When the safety risk value is greater than the first preset threshold and less than or equal to the second preset threshold, it is determined to be a medium risk level, a yellow warning signal is output, and the monitoring frequency of the concrete gravity dam is increased. When the safety risk value is greater than the second preset threshold, it is determined to be an extremely high risk level, a red warning signal is output and the emergency response mechanism for concrete gravity dams is activated, wherein the first preset threshold is less than the second preset threshold.

8. A system for calculating the safety risk value during the operation of a concrete gravity dam, characterized in that, The system for calculating the safety risk value during the operation of the concrete gravity dam includes: The data acquisition module is used to continuously collect structural monitoring data, environmental data, and operational data of the concrete gravity dam during its operation through the concrete gravity dam safety monitoring system. The data processing module is used to process the structural monitoring data, environmental data, and operational data to obtain fused status data. The data processing includes data preprocessing and standardization, coordinate transformation and risk feature extraction, and data fusion processing. The risk calculation module is used to construct a set of dam operating environment parameters based on environmental data, and to calculate the safety risk value using a preset deep belief network model based on the set of dam operating environment parameters and the fused state data. The risk classification module is used to classify risk levels based on safety risk values ​​and output early warning information.

9. A device for calculating the safety risk value during the operation of a concrete gravity dam, characterized in that, The device includes: a memory, a processor, and a concrete gravity dam operation period safety risk value calculation program stored in the memory and executable on the processor, the concrete gravity dam operation period safety risk value calculation program being configured to implement the steps of the concrete gravity dam operation period safety risk value calculation method as described in any one of claims 1 to 7.

10. A storage medium, characterized in that, The storage medium stores a program for calculating the safety risk value of a concrete gravity dam during its operation period. When the processor executes the program, it implements the steps of the method for calculating the safety risk value of a concrete gravity dam during its operation period as described in any one of claims 1 to 7.

Images