A method for analyzing the influence of a horizontal joint of a panel on the overall stability of a dam body

Abstract

The application provides a method for analyzing influence of panel horizontal joint setting on overall stability of a dam body, comprising receiving dam body design parameters, collecting panel horizontal joint setting, monitoring and environmental load data; analyzing panel stress, dam body stability risk and dynamic response, generating deformation influence information in combination with panel deformation monitoring data; cooperatively analyzing stress, risk and response information, generating cooperative results; based on the cooperative results and the deformation influence information, proposing dam body reinforcement suggestions and sending to a design device. The application can improve dam body safety and reliability.

Description

Technical Field

[0001] This invention relates to the field of dam stability analysis technology in water conservancy and hydropower engineering, and more specifically, to a method for analyzing the impact of horizontal joint settings on the overall stability of the dam body. Background Technology

[0002] In water conservancy and hydropower projects, panel dams are an important dam type, and their stability directly affects the safe operation of the project. The stability of panel dams is influenced by various factors, including the structural design of the panels, the load conditions on the dam body, and environmental factors. Traditional stability analysis methods for panel dams mainly focus on the analysis of single factors, such as stress analysis of the panels or displacement monitoring of the dam body. While these methods can provide stability assessments to some extent, they lack a comprehensive consideration of the synergistic effects of multiple factors. Furthermore, traditional analysis methods often rely on empirical formulas when processing panel deformation monitoring data, lacking in-depth data mining and accurate assessment of deformation.

[0003] In implementing the embodiments of the present invention, the prior art has at least the following problems or defects: the prior art fails to fully consider the impact of the horizontal joint setting of the panel on the overall stability of the dam body, lacks collaborative analysis of panel stress distribution, dam dynamic response and panel deformation monitoring data, resulting in an incomplete and inaccurate assessment of dam stability, and makes it difficult to provide scientific and effective reinforcement suggestions. Summary of the Invention

[0004] This invention provides a method for analyzing the impact of horizontal joint settings on the overall stability of a dam, including: Receive dam design parameters as input and obtain the dam design parameters as target parameters; Based on the target parameters, the following stability impact analysis is performed: Collect panel horizontal joint setting data, dam body monitoring data, and dam body environmental load data corresponding to the target parameters; Based on the panel horizontal seam setting data, panel stress analysis is performed to generate panel stress distribution information; Based on the dam monitoring data, a dam stability risk assessment is conducted to obtain at least one stability risk information. Based on the dam's environmental load data, dynamic response analysis of the dam is performed to generate dynamic response information of the dam. Based on the at least one stability risk information, the panel stress distribution information, and the dam body dynamic response information, a collaborative analysis of dam body stability is performed to generate collaborative analysis results. In the stability impact analysis process, in response to receiving panel deformation monitoring data sent by the panel deformation monitoring device, panel deformation impact information is generated based on the panel deformation monitoring data; Based on the collaborative analysis results and the panel deformation impact information, dam reinforcement recommendations are generated. The dam reinforcement recommendation information is sent to the dam reinforcement design equipment.

[0005] Furthermore, the panel horizontal seam setting data includes setting data for each panel unit. Each panel unit setting data includes a unit number and a horizontal seam parameter set. The horizontal seam parameter set includes panel thickness parameters and panel elastic modulus parameters. The panel stress analysis based on the panel horizontal seam setting data and the generation of panel stress distribution information includes: for each panel unit setting data included in the panel horizontal seam setting data, performing the following steps: calculating the stress value of the corresponding panel unit based on the panel thickness parameters and panel elastic modulus parameters included in the panel unit setting data. In response to determining that the stress value is greater than a preset stress threshold, the panel unit setting data and information characterizing the risk of panel stress anomalies are determined as panel stress distribution sub-information; The determined sub-information of each panel stress distribution is used to define the panel stress distribution information.

[0006] Further, the step of calculating the stress value of the corresponding panel unit based on the panel thickness parameter and panel elastic modulus parameter included in the panel unit setting data includes: Obtain the dam load data corresponding to the panel unit, the dam load data including water pressure value and rockfill pressure value; Based on the panel thickness parameter, panel elastic modulus parameter, water pressure value, and rockfill pressure value, the stress value is calculated according to the stress calculation formula, wherein the stress calculation formula is: ,in, Indicates the stress value. This represents the stress adjustment factor, and E represents the panel's elastic modulus parameter. Indicates the water pressure value. t represents the pressure value of the rockfill, and t represents the panel thickness parameter.

[0007] Furthermore, each of the at least one stability risk information corresponds to a dam body location identifier, and the dam body stability collaborative analysis based on the at least one stability risk information, the panel stress distribution information, and the dam body dynamic response information, generating collaborative analysis results, includes: sending the panel stress distribution information, the at least one stability risk information, and the dam body dynamic response information to at least one analysis terminal; For each of the at least one stability risk information, the following analysis steps are performed: the dam body part identifier corresponding to the stability risk information is determined as the risk part identifier; The stability risk information is sent to at least one analysis terminal corresponding to the risk location identifier; The collaborative analysis results are generated based on the panel stress distribution information and the dam dynamic response information.

[0008] Further, the step of generating panel deformation impact information based on the panel deformation monitoring data includes: for each panel deformation value in the panel deformation value sequence included in the panel deformation monitoring data, performing the following deformation impact analysis processing: inputting the panel deformation value into a pre-trained panel deformation impact assessment model to obtain panel deformation impact assessment information corresponding to the panel deformation value; In response to determining that the panel deformation impact assessment information characterizes the dam body part corresponding to the detected panel deformation value has a stability risk, the panel measuring point location information corresponding to the panel deformation value is determined as the risk location information; The location information of each identified risk point is used as the panel deformation impact information.

[0009] Furthermore, the panel deformation impact assessment model is trained through the following steps: Historical panel deformation value sequences and historical panel deformation impact assessment information were collected as initial training samples; The initial training samples are cleaned to remove outliers, resulting in cleaned training samples. The cleaned training samples are normalized to obtain normalized training samples. Based on the normalized training samples, a neural network model is trained, wherein the neural network model includes an input layer, a hidden layer, and an output layer, and the hidden layer uses an activation function to perform a nonlinear transformation; The trained neural network model is validated using a validation set, and the model accuracy is calculated. In response to the model accuracy falling below a preset accuracy threshold, the hyperparameters of the neural network model are adjusted and retrained until the model accuracy meets the preset accuracy threshold, thus obtaining the panel deformation impact assessment model.

[0010] Furthermore, after receiving panel deformation monitoring data sent by the panel deformation monitoring device, the method further includes: for each panel deformation value in the panel deformation value sequence included in the panel deformation monitoring data, performing the following panel crack detection processing: based on the dam load analysis corresponding to the target parameter, normalizing the panel deformation value to obtain a normalized panel deformation value; The normalized panel deformation value is input into a pre-trained panel crack recognition model to obtain the panel crack risk value. Based on the panel crack risk value, a panel crack early warning information is generated; The panel crack warning information is sent to at least one panel maintenance terminal.

[0011] Furthermore, the panel crack recognition model is trained through the following steps: Historical panel deformation value sequences and historical panel crack data were collected as initial training samples. The initial training samples are subjected to feature extraction processing to extract panel deformation features and crack features to obtain feature training samples; The feature training samples are standardized to obtain standardized training samples; Based on the standardized training samples, a support vector machine model is trained, wherein the support vector machine model uses a kernel function for non-linear classification; Cross-validation was used to optimize the trained support vector machine model and adjust the model parameters. Based on the optimized support vector machine model, the model performance index is calculated. In response to the model performance index meeting the preset performance threshold, the panel crack recognition model is obtained.

[0012] Furthermore, the method also includes: In response to receiving dam seepage monitoring data sent by the dam seepage monitoring equipment, the dam seepage monitoring data is identified as dam seepage data to be analyzed. Based on the seepage data to be analyzed in the dam body, the seepage velocity sequence and seepage pressure value sequence are extracted; Based on the seepage velocity sequence and seepage pressure sequence, a seepage stability index is calculated, wherein the seepage stability index is calculated using a seepage stability formula, which is: Where S represents the seepage stability index, This represents the seepage adjustment coefficient. This represents the average value of the seepage velocity sequence. This represents the maximum value in the osmotic pressure value sequence; In response to determining that the seepage stability index is greater than a preset seepage threshold, seepage stability information of the dam body is generated; The seepage stability information of the dam body is sent to the dam body reinforcement design equipment.

[0013] Furthermore, based on the dam monitoring data, the dam stability risk assessment is performed to obtain at least one stability risk information, including: For each dam displacement monitoring data in the dam displacement monitoring data sequence included in the dam monitoring data, the following risk identification steps are performed: the dam displacement monitoring data is filtered to obtain filtered displacement data; The filtered displacement data is input into a pre-trained stability risk identification model to obtain the stability risk value; Based on the stability risk value, time series analysis is performed to determine the stability risk trend; In response to determining that the stability risk value is greater than a preset risk threshold or that the stability risk trend indicates an increase in risk, the dam displacement monitoring data and information characterizing the existence of stability risk are determined as stability risk information. Each identified stability risk is designated as at least one stability risk.

[0014] The embodiments of the present invention have at least the following beneficial effects: 1. By comprehensively collecting data on the horizontal joint settings of the panel, dam monitoring data, and dam environmental load data, and conducting panel stress analysis, dam stability risk assessment, and dam dynamic response analysis, a multi-dimensional and comprehensive assessment of dam stability is achieved. This solves the problem of one-sided assessment results caused by relying on only a single factor analysis in existing technologies, and improves the comprehensiveness and accuracy of dam stability assessment.

[0015] 2. Based on panel deformation monitoring data, panel deformation impact information is generated, and combined with collaborative analysis results, dam reinforcement recommendations are generated. This realizes the direct conversion from monitoring data to reinforcement measures, solving the problems of insufficient utilization of deformation data and lack of scientific basis for reinforcement recommendations in existing technologies, and enhancing the pertinence and effectiveness of dam reinforcement measures.

[0016] 3. By adopting a pre-trained panel deformation impact assessment model and panel crack identification model, the panel deformation value is analyzed in depth and crack risk is warned. This solves the problem of low assessment accuracy caused by relying on empirical formulas to process deformation data in existing technologies, and improves the ability to identify potential risks of the dam body and the accuracy of early warning. Attached Figure Description

[0017] The above and other objects, features, and advantages of exemplary embodiments of the present invention will become readily apparent from the following detailed description taken in conjunction with the accompanying drawings. Several embodiments of the invention are illustrated in the drawings by way of example, not limitation, in which: Figure 1 This is a flowchart illustrating the method for analyzing the impact of horizontal joint settings on the overall stability of a dam, as provided in an embodiment of the present invention. Detailed Implementation

[0018] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0019] Traditional stability analyses of panel dams primarily rely on independent processing of single-dimensional data, such as isolated assessments of panel stress or dam displacement. Because a correlation analysis mechanism between panel horizontal joint parameters and dam dynamic response is lacking, the coupling effect between abnormal stress distribution and dynamic load response cannot be quantified and identified. Furthermore, panel deformation monitoring data and the dam stability assessment process are disconnected; deformation characteristics are not effectively integrated into the overall analysis framework, resulting in insufficient accuracy in locating risk areas.

[0020] For example, in the design of concrete-faced rockfill dams, when using traditional methods to handle panel thickness and elastic modulus parameters, local stress values ​​are calculated solely based on static mechanical models. The displacement trends in dam monitoring data are not synchronously correlated with the dynamic fluctuations of environmental loads, making it impossible to verify the spatiotemporal correspondence between stress concentration areas and dynamic response peak intervals. Furthermore, millimeter-level deformation data collected by panel deformation monitoring equipment is used only for independent early warning threshold determination, failing to form a closed-loop feedback with dam reinforcement decisions, resulting in reinforcement measures lagging behind actual deformation development.

[0021] If the above problems are not addressed, the mismatch between the panel structure parameters and the dynamic response of the dam body will lead to the continuous accumulation of damage in areas of abnormal stress. The disconnect between deformation monitoring data and stability assessment will delay the reinforcement of critical components. Under dynamic loads, the stress redistribution at the interface between the panel and the dam body may trigger a chain reaction of failures, while unidentified synergistic mechanisms will exacerbate the expansion of seepage paths within the dam body, ultimately reducing the safety redundancy of the dam throughout its entire life cycle.

[0022] like Figure 1 As shown, this application proposes a method for analyzing the impact of horizontal joint settings on the overall stability of a dam body, including: S1. Receive the dam design parameters input and obtain the dam design parameters as target parameters; S2. Based on the target parameters, perform the following stability impact analysis: S3. Collect panel horizontal joint setting data, dam monitoring data and dam environmental load data corresponding to the target parameters; S4. Based on the panel horizontal seam setting data, perform panel stress analysis and generate panel stress distribution information; S5. Based on the dam body monitoring data, conduct a dam body stability risk assessment to obtain at least one stability risk information. S6. Based on the dam body environmental load data, perform dynamic response analysis of the dam body and generate dynamic response information of the dam body; S7. Based on the at least one stability risk information, the panel stress distribution information, and the dam body dynamic response information, perform a collaborative analysis of dam body stability and generate collaborative analysis results; S8. In the stability impact analysis and processing, in response to receiving panel deformation monitoring data sent by the panel deformation monitoring device, panel deformation impact information is generated based on the panel deformation monitoring data. S9. Based on the collaborative analysis results and the panel deformation impact information, generate dam reinforcement recommendation information; S10. Send the dam reinforcement suggestion information to the dam reinforcement design equipment.

[0023] Panel horizontal joint setting data refers to the set of structural parameters of the panel in the horizontal direction. Specifically, this can be achieved by using sensors to collect parameters such as panel unit thickness and elastic modulus, reflecting the physical characteristics of the panel in the horizontal joint. Dam body monitoring data refers to real-time dam body status data obtained through monitoring equipment such as displacement gauges and stress gauges. Specifically, an automated data acquisition system can periodically record displacement and stress changes to assess the current stability of the dam body. Dam body environmental load data refers to the dynamic load parameters applied to the dam body by the external environment. Specifically, water pressure data can be obtained from hydrological monitoring stations, and the rockfill pressure value can be calculated by combining it with geological exploration data to analyze the dynamic impact of environmental factors on the dam body.

[0024] Panel stress analysis refers to calculating the internal stress distribution of a panel based on the principles of materials mechanics. Specifically, it can use finite element simulation or pre-defined stress calculation formulas to process thickness, elastic modulus, and load data to identify areas of localized stress anomalies. Dam stability risk assessment quantifies the probability of dam instability using a risk model. Specifically, machine learning models can be used to analyze the time-series characteristics of monitoring data to pinpoint potential risk areas. Dam dynamic response analysis simulates the deformation and vibration characteristics of the dam under load. Specifically, it can use dynamic equations combined with numerical calculation methods to assess the overall structural response behavior of the dam. Collaborative analysis results refer to comprehensive assessment conclusions integrating stress, risk, and dynamic response data. Specifically, multi-source data fusion algorithms can be used to generate quantitative indicators to reveal the correlation and influence mechanisms between various factors.

[0025] Panel deformation impact information refers to the trend of dam stability changes inferred from deformation monitoring data. Specifically, neural network models can be used to predict the relationship between deformation and stability, supplementing the real-time nature of collaborative analysis results. Dam reinforcement recommendation information refers to targeted engineering measures generated based on the analysis results. Specifically, an expert system can be used to match a pre-set reinforcement strategy library to guide dam structure optimization.

[0026] The core innovation of this application lies in the dynamic integration of panel horizontal joint parameters, real-time monitoring data, environmental load and deformation monitoring through a multi-source data collaborative analysis mechanism. This constructs a comprehensive analysis framework covering panel stress, dam risk and dynamic response, breaking through the limitations of traditional single-factor analysis. This improves the comprehensiveness and accuracy of dam stability assessment and provides a scientific basis for reinforcement design.

[0027] In the stability impact analysis, panel deformation monitoring data sent by the panel deformation monitoring equipment is also received, and panel deformation impact information is generated based on this data. Finally, based on the collaborative analysis results and the panel deformation impact information, dam reinforcement recommendation information is generated and sent to the dam reinforcement design equipment.

[0028] This method comprehensively analyzes the overall stability of the dam by considering multiple factors, including the setting of horizontal joints in the dam panel, dam monitoring, environmental loads, and panel deformation. Panel stress analysis identifies stress concentration areas, dam stability risk assessment reveals potential risk points, and dynamic response analysis reveals the dam's behavior under different loads. Synergistic analysis of these results yields a more comprehensive and accurate stability assessment.

[0029] As a preferred embodiment, in a concrete-faced rockfill dam project, the design parameters of the dam body are first received, including key parameters such as dam height, dam crest width, and upstream and downstream slope ratios. These parameters are set as target parameters for subsequent analysis.

[0030] Based on the target parameters, a stability impact analysis was performed. Data on the horizontal joint settings of the panel were collected, including panel thickness and elastic modulus; dam monitoring data were collected, including displacement, stress, and seepage data; and environmental load data for the dam were collected, including water level changes and seismic loads.

[0031] When performing panel stress analysis, the panel structure model is established using the finite element method, the panel horizontal joint setting data is input, the stress distribution under different load conditions is simulated, and a panel stress distribution cloud map is generated.

[0032] The dam stability risk assessment adopts a multi-factor comprehensive scoring method, which calculates the scores of various indicators based on dam monitoring data, comprehensively assesses the dam stability risk level, and generates a risk area distribution map.

[0033] The dynamic response analysis of the dam body uses the time history analysis method. The environmental load data of the dam body is input, the acceleration and displacement response of the dam body under dynamic load are calculated, and the dynamic response curve is generated.

[0034] In the dam stability collaborative analysis stage, the stress distribution of the panel, the stability risk assessment results, and the dynamic response characteristics are overlaid and analyzed to identify high-risk areas and generate a collaborative analysis report.

[0035] During the analysis, deformation monitoring data transmitted from the panel deformation monitoring equipment is received in real time. Data mining algorithms are used to analyze deformation trends, assess the impact of deformation on stability, and generate a panel deformation impact report.

[0036] Based on the collaborative analysis report and the panel deformation impact report, an expert system is used to generate dam reinforcement recommendations, including reinforcement locations, methods, and priorities. These recommendations are then sent to the dam reinforcement design equipment via a data interface for designers to reference.

[0037] Furthermore, it also includes: collecting panel horizontal seam setting data, wherein the panel horizontal seam setting data includes setting data for each panel unit, and each panel unit setting data includes a unit number and a horizontal seam parameter set, the horizontal seam parameter set including panel thickness parameters and panel elastic modulus parameters; for each panel unit setting data, calculating the stress value of the corresponding panel unit based on the panel thickness parameters and panel elastic modulus parameters; when the stress value exceeds a preset stress threshold, combining the panel unit setting data with information characterizing the risk of abnormal panel stress into panel stress distribution sub-information; and summarizing all panel stress distribution sub-information into panel stress distribution information.

[0038] The panel unit settings data uses unit numbers to independently identify each panel unit, and the horizontal seam parameter set defines the physical properties of the units using panel thickness and elastic modulus parameters. During stress calculation, a preset stress threshold is introduced as a criterion to filter out units with abnormal risks. For example, the panel thickness parameter can be set to 0.5 meters to 1.2 meters, the panel elastic modulus parameter can be set to 20 GPa to 35 GPa, and the preset stress threshold is set to 15 MPa based on the material's tensile strength.

[0039] In some embodiments, during the panel stress analysis phase, each panel element is calculated independently based on its thickness and modulus of elasticity. The specific location is determined by the element number, and the stress calculation formula is applied in conjunction with the physical property parameters from the horizontal seam parameter set. The system derives stress values. When the calculation results exceed a preset threshold, it automatically associates the element number with abnormal risk information, generating distribution sub-information including location identifiers. All sub-information is integrated to form a global stress distribution map, visually displaying the stress state of each element. This process, by refining stress assessment at the element level, accurately identifies high-risk areas, providing data support for subsequent reinforcement design and solving the assessment bias problem caused by the lack of differentiation between elements in traditional methods.

[0040] When performing panel stress analysis, for each panel unit setting data included in the panel horizontal joint setting data, the following steps are performed: Based on the panel thickness parameter and panel elastic modulus parameter included in the panel unit setting data, calculate the stress value of the corresponding panel unit. For example, for panel unit number A1, its panel thickness parameter is 0.5 meters, and its panel elastic modulus parameter is 30 GPa. The stress value calculated using the stress calculation formula is 5 MPa.

[0041] Furthermore, in response to determining that the stress value is greater than a preset stress threshold, the panel element configuration data and information characterizing the risk of panel stress anomalies are identified as panel stress distribution sub-information. Specifically, assuming the preset stress threshold is 4 MPa, since the calculated stress value of 5 MPa is greater than the preset stress threshold, the panel element configuration data of element number A1 and the information characterizing the risk of panel stress anomalies are identified as panel stress distribution sub-information.

[0042] Therefore, the determined sub-information of each panel stress distribution is defined as the panel stress distribution information. In a preferred embodiment, the panel stress distribution information may include multiple sub-information pieces of panel stress distribution, each corresponding to a panel unit with a risk of stress anomaly.

[0043] Furthermore, it also includes: calculating the stress value based on the panel thickness parameter, panel elastic modulus parameter, water pressure value, and rockfill pressure value, according to the stress calculation formula, wherein the stress calculation formula is: ,in, Indicates the stress value. This represents the stress adjustment factor, and E represents the panel's elastic modulus parameter. Indicates the water pressure value. t represents the pressure value of the rockfill, and t represents the panel thickness parameter.

[0044] The acquisition of dam load data corresponding to the panel unit includes water pressure and rockfill pressure. Water pressure is calculated by collecting water depth data at the corresponding location of the panel unit using pressure sensors. Rockfill pressure is measured by earth pressure gauges embedded within the rockfill. The stress adjustment coefficient α is preset based on the panel material type and connection method, ranging from 0.85 to 1.15. Panel thickness parameter t is obtained from design drawings or on-site measurements, and panel elastic modulus parameter E is determined through laboratory material testing. In the stress calculation formula, the water pressure value... With the pressure value of the rockfill Divide each value by the panel thickness parameter t, multiply by the elastic modulus parameter E, and correct using the stress adjustment coefficient α to finally obtain the stress value σ of the panel unit.

[0045] In the panel stress analysis process, the water pressure and rockfill pressure values ​​at the location of the panel unit are first extracted from monitoring equipment or databases. These values ​​are then combined with the panel design parameters, such as thickness and elastic modulus, and substituted into the stress calculation formula for iterative calculation. For example, when the panel thickness parameter t is 0.8 meters, the elastic modulus parameter E is 30 GPa, and the water pressure value... The pressure value of the rockfill body is 200 kPa. When the stress is 150 kPa and the stress adjustment factor α is 1.0, the calculated stress value σ is: 1.0×30×(200 / 0.8+150 / 0.8)=1.0×30×(250+187.5)=13125MPa.

[0046] The calculation results comprehensively consider the interaction between external loads and panel structure parameters. By introducing a stress adjustment coefficient α to eliminate the influence of nonlinear material deformation, the stress analysis results are closer to the actual working conditions, providing an accurate data basis for subsequent stability risk assessment.

[0047] Furthermore, it also includes: sending panel stress distribution information, at least one stability risk information, and dam dynamic response information to at least one analysis terminal; for each stability risk information, performing the following analysis steps: identifying the dam part identifier corresponding to the stability risk information as the risk part identifier; sending the stability risk information to at least one analysis terminal corresponding to the risk part identifier; and generating collaborative analysis results based on the panel stress distribution information and dam dynamic response information.

[0048] The risk location identifier is used to uniquely mark specific areas within the dam body where stability risks exist, such as the upstream panel area or the dam foundation area. Analysis terminals are assigned based on the risk location identifier, with each terminal corresponding to one or more dam body locations. The panel stress distribution information sent to the analysis terminals includes stress distribution data for each panel element, and the dam body dynamic response information includes displacement or vibration data of the dam body under different loads. In a preferred embodiment, the analysis terminals employ a distributed computing architecture, with each terminal configured with an independent data processing module to receive and process the risk data for its corresponding location. For example, when the risk location identifier is upstream panel area A, the corresponding analysis terminal will receive the stability risk information for that area and simultaneously acquire the panel stress distribution data and dynamic response data for that area.

[0049] In generating collaborative analysis results, panel stress distribution information, stability risk information, and dam dynamic response information are first distributed to multiple analysis terminals via a data transmission interface. After the dam location identifier carried by each stability risk information is extracted, the system determines the analysis terminal responsible for processing that location based on a preset mapping relationship. For example, when stability risk information corresponding to dam displacement monitoring data carries a dam center identifier, this risk information is directed to analysis terminal numbered 03. This terminal simultaneously acquires the panel stress distribution data for that location and, combined with the dynamic response data, performs local stability calculations. After all analysis terminals complete their assigned location analyses, the local results are aggregated to the main control module, which then generates the overall collaborative analysis results through a data fusion algorithm. During this process, due to the correspondence between risk location identifiers and analysis terminals, stability risks for different locations can be processed in parallel. Furthermore, by combining the correlation data between panel stress and dynamic response, computational delays when a single terminal processes global data are avoided, improving analysis efficiency. For example, when an analysis terminal detects that the stability risk value of downstream panel B exceeds the threshold, it can immediately call up the panel stress distribution data of that area and combine it with the water pressure change data in the dynamic response to quickly determine whether the risk has increased due to stress concentration, thereby generating an accurate assessment for that area in the collaborative analysis results.

[0050] In some embodiments, the panel stress distribution information includes the stress value and location information of each panel element. At least one stability risk information includes the risk level, risk type, and dam body location identifier. The dam body dynamic response information includes displacement, strain, and acceleration data at each monitoring point on the dam body. This information is transmitted via a data transmission network to analysis terminals distributed across different areas of the dam body.

[0051] Furthermore, for each stability risk information, the system automatically matches its dam body location identifier with a preset risk location correspondence table to determine the corresponding risk location identifier. For example, if the dam body location identifier in the stability risk information is the middle of the dam crest, then the risk location identifier is determined to be area A.

[0052] Therefore, the system sends this stability risk information to the analysis terminal identified as area A. After receiving the information, the analysis terminal performs a comprehensive analysis by combining the panel stress distribution information and the dam's dynamic response information.

[0053] In the specific analysis process, the analysis terminal performs spatiotemporal correlation analysis between the panel stress distribution and the dam's dynamic response data to identify the overlap between stress concentration areas and displacement anomaly areas. Through data fusion algorithms, collaborative analysis results reflecting the overall stability of the dam are generated, including risk level assessment, deformation trend prediction, and reinforcement recommendations.

[0054] Furthermore, it also includes: for each panel deformation value in the panel deformation value sequence included in the panel deformation monitoring data, performing the following deformation impact analysis processing: inputting the panel deformation value into a pre-trained panel deformation impact assessment model to obtain panel deformation impact assessment information for the corresponding panel deformation value; in response to determining that the panel deformation impact assessment information characterizes the dam body part corresponding to the detected panel deformation value as having a stability risk, determining the panel measuring point location information of the corresponding panel deformation value as the risk location information; and determining the determined risk location information as panel deformation impact information.

[0055] The panel deformation impact assessment model establishes a correlation between deformation values ​​and stability risks through a nonlinear mapping relationship. Deformation impact analysis employs a point-by-point traversal method to iterate through the deformation sequence, ensuring that data from each measuring point is independently evaluated. Risk location information is spatially located through the mapping relationship between measuring point coordinates and the dam's 3D model, enabling visual identification of risk areas.

[0056] Furthermore, the method also includes: collecting historical panel deformation value sequences and historical panel deformation impact assessment information as initial training samples; cleaning the initial training samples to remove outliers, resulting in cleaned training samples; normalizing the cleaned training samples to obtain normalized training samples; training a neural network model based on the normalized training samples, wherein the neural network model includes an input layer, a hidden layer, and an output layer, and the hidden layer uses an activation function for nonlinear transformation; validating the trained neural network model using a validation set and calculating the model accuracy; and adjusting the hyperparameters of the neural network model and retraining it in response to the model accuracy falling below a preset accuracy threshold, until the model accuracy meets the preset accuracy threshold, thus obtaining the panel deformation impact assessment model.

[0057] The data cleaning process involves identifying and removing data points that exceed a set threshold for data fluctuation range, for example, using the three-standard-deviation method to filter outliers. Normalization uses a maximum-minimum method to map the data to the 0-1 interval, eliminating the impact of dimensional differences on model training. The hidden layer activation function of the neural network model is the ReLU function, and the output layer uses the Sigmoid function to achieve binary classification probability output. Hyperparameter tuning includes gradually decreasing the learning rate from 0.1 to 0.001, setting the batch size to 32 or 64, and dynamically adjusting the number of hidden layer neurons from 8 to 16 based on the input feature dimension.

[0058] In some embodiments, after data cleaning, outliers are removed from the initial training samples to ensure that the training data conforms to the actual working conditions. When the normalized data is input into the neural network model, the convergence speed of the gradient descent process is improved by more than 20%. The hidden layer introduces nonlinear transformation capability through the ReLU function, enabling the model to capture the complex relationship between panel deformation values ​​and stability risk. During the validation phase, the model accuracy is calculated using an independent validation set. When the accuracy is below 90%, retraining is performed by reducing the learning rate or increasing the number of neurons in the hidden layer until the accuracy reaches a preset threshold. The final model achieves an accuracy of over 92% on the test set, effectively identifying the impact of panel deformation on dam stability.

[0059] As a preferred embodiment, the solution of this application is specifically implemented as follows: The panel deformation impact assessment model is trained through the following steps: Historical panel deformation value sequences and historical panel deformation impact assessment information are collected as initial training samples. For example, panel deformation monitoring data from the past 5 years are collected, including daily measured deformation values ​​and corresponding engineer assessment results.

[0060] The initial training samples are cleaned to remove outliers, resulting in cleaned training samples. Specifically, the 3σ principle is used to identify and remove outlier data points to ensure data quality.

[0061] The cleaned training samples are normalized to obtain normalized training samples. Furthermore, the min-max normalization method is used to map the data to the [0,1] interval.

[0062] A neural network model is trained based on normalized training samples. This model consists of an input layer, hidden layers, and an output layer. The hidden layers use an activation function for nonlinear transformation. Thus, a three-layer feedforward neural network is constructed, with the number of nodes in the input layer equal to the number of features in the deformed value. The hidden layers use the ReLU activation function, and the output layer corresponds to the evaluation result.

[0063] The trained neural network model is validated using a validation set, and the model accuracy is calculated. For example, a 10-fold cross-validation method is used to evaluate model performance.

[0064] In response to a model accuracy falling below a preset accuracy threshold, the hyperparameters of the neural network model are adjusted, and retraining is performed until the model accuracy meets the preset accuracy threshold, resulting in a panel distortion impact assessment model. Specifically, hyperparameters such as the learning rate and batch size are optimized using a grid search method, and iterative training is conducted until the model accuracy reaches above 95%.

[0065] Furthermore, it also includes: after receiving the panel deformation monitoring data, for each panel deformation value in the panel deformation value sequence, performing panel crack detection processing: based on the dam load analysis corresponding to the target parameters, normalizing the panel deformation value to obtain a normalized panel deformation value; inputting the normalized panel deformation value into a pre-trained panel crack identification model to obtain a panel crack risk value; generating panel crack early warning information based on the panel crack risk value; and sending the panel crack early warning information to at least one panel maintenance terminal.

[0066] When performing normalization based on dam load analysis, the theoretical deformation range under the current dam load state is obtained, and the actual deformation values ​​are scaled proportionally to a uniform dimension range to eliminate the influence of different load conditions on the magnitude differences of deformation values. The panel crack identification model uses a support vector machine (SVM) model, which maps low-dimensional, nonlinear, inseparable deformation features to a high-dimensional space through a kernel function, achieving crack risk classification. During model training, features are extracted from historical panel deformation value sequences, including multi-dimensional features such as deformation rate, deformation fluctuation amplitude, and deformation trend. These features are then labeled using historical crack data to form standardized training samples.

[0067] In some embodiments, after the panel deformation monitoring data is input, a theoretical deformation threshold is first calculated based on the current water pressure and rockfill pressure values ​​of the dam body. The actual deformation value is then divided by this threshold to obtain a normalized panel deformation value, ensuring comparability of deformation data under different load conditions. The normalized data is then input into a support vector machine model. The model uses a pre-learned classification hyperplane to determine whether the current deformation value belongs to a crack risk mode. When the panel crack risk value exceeds a set threshold, a panel crack early warning message containing the risk level and measurement point location is generated and directly pushed to the maintenance terminal. For example, when the normalized deformation value of a measurement point is 1.2 and the model outputs a risk value of 0.85, a level one early warning is triggered, indicating a cracking risk at that location. This method enables proactive maintenance by identifying potential risks based on the nonlinear characteristics of deformation data before visible cracks appear in the panel.

[0068] Furthermore, the method also includes: collecting historical panel deformation value sequences and historical panel crack data as initial training samples; performing feature extraction processing on the initial training samples to extract panel deformation features and crack features to obtain feature training samples; performing standardization processing on the feature training samples to obtain standardized training samples; training a support vector machine model based on the standardized training samples, with the support vector machine model using a kernel function for nonlinear classification; optimizing the trained support vector machine model using cross-validation to adjust the model parameters; calculating the model performance index based on the optimized support vector machine model, and obtaining a panel crack recognition model when the model performance index meets a preset performance threshold.

[0069] Feature extraction utilizes time-domain and frequency-domain analysis to extract panel deformation features, such as deformation rate and deformation fluctuation amplitude, while also incorporating crack length and width as crack features. Standardization employs the Z-score method to eliminate dimensional differences and ensure consistent data distribution. The support vector machine model uses radial basis functions as the kernel function, handling nonlinear classification problems by mapping to a high-dimensional space. Cross-validation uses the K-fold method to divide the training and validation sets, adjusting the penalty factor and kernel parameters to optimize the classification boundary. Model performance metrics include accuracy, recall, and F1 score; an F1 score of 0.85 or higher is considered to meet the performance threshold.

[0070] In some embodiments, after feature extraction, the deformation rate and fluctuation amplitude of the historical panel deformation value sequence are quantized into feature vectors, while crack length and width are used as label data. Standardization normalizes the feature vectors to the [-1,1] interval to avoid interference from different units of measurement during model training. The support vector machine model maps low-dimensional nonlinear data to a high-dimensional space using a kernel function, constructing an optimal hyperplane to effectively separate cracked and non-cracked states. During cross-validation, K-fold partitioning ensures that each validation set does not overlap, and parameter combinations are adjusted through grid search to improve the model's generalization ability. The optimized model obtains performance metrics by calculating the confusion matrix. When the F1 score exceeds a preset threshold, it indicates that the model possesses high-precision recognition capabilities and can be deployed to actual monitoring scenarios for crack early warning.

[0071] As a preferred embodiment, the specific implementation of this application is as follows: During the training process of the panel crack identification model, historical panel deformation value sequences under different working conditions, along with their corresponding crack widths and crack densities, are first collected as initial training samples. Time-frequency domain features are extracted from the deformation value sequences using wavelet transform as panel deformation features, while texture features are extracted from crack images using the gray-level co-occurrence matrix algorithm as crack features, forming feature training samples. These feature training samples are input into the Z-Score normalization module for data normalization to eliminate dimensional differences. The normalized data is then input into a support vector machine model based on radial basis function kernels for training, with the penalty factor and kernel function parameters adjusted using a grid search method. The model is optimized using a five-fold cross-validation method, and the F1 score of each fold validation set is calculated as a performance indicator. Training stops when the average F1 score reaches 0.92, ultimately obtaining a panel crack identification model that accurately correlates panel deformation features with crack risk.

[0072] This application further proposes a panel crack recognition model trained through the following steps: collecting historical panel deformation value sequences and historical panel crack data as initial training samples; performing feature extraction processing on the initial training samples to extract panel deformation features and crack features, obtaining feature training samples; standardizing the feature training samples to obtain standardized training samples; training a support vector machine model based on the standardized training samples, with the support vector machine model using a kernel function for nonlinear classification; optimizing the trained support vector machine model using cross-validation to adjust model parameters; calculating the model performance index based on the optimized support vector machine model, and obtaining the panel crack recognition model when the model performance index meets a preset performance threshold.

[0073] Feature extraction separates crack-related fluctuation and trend features from the panel deformation value sequence through mathematical transformations, such as using wavelet transform to decompose deformation components in different frequency bands. Standardization maps feature values ​​to a uniform dimension range, eliminating the impact of magnitude differences on model training. A radial basis function is chosen as the kernel function, transforming low-dimensional inseparable data into a high-dimensional separable space through nonlinear mapping. Cross-validation employs a K-fold partitioning method, dividing the dataset into training and validation subsets, and iteratively adjusting the penalty factor and kernel function parameter combination to balance model complexity and generalization ability.

[0074] In some embodiments, after feature extraction, the initial training samples include panel deformation features such as deformation rate extrema, cumulative deformation, and spectral energy distribution; and crack features such as crack length, width, and propagation direction. Standardization is performed using the Z-score method to convert the feature data into a distribution with a mean of zero and a standard deviation of one. The support vector machine model achieves classification by solving for the maximum margin hyperplane, and the kernel function parameters are determined optimally using a grid search method. During cross-validation, a subset is retained as validation data in each iteration, while the remainder is used for training. This process is repeated K times, and the average accuracy is used as the model performance metric. When the accuracy reaches 95% or higher, the model stops optimization and outputs the final parameters, ensuring that the crack identification model maintains high classification accuracy under complex loading conditions.

[0075] As a preferred embodiment, the solution of this application is implemented as follows: After receiving the seepage monitoring data transmitted by the dam seepage monitoring equipment, the data is parsed into a seepage velocity sequence and a seepage pressure value sequence. The seepage velocity sequence is obtained by calculating the average flow velocity within a time window, and the seepage pressure value sequence is taken as the peak pressure value within the monitoring period. The seepage stability formula is then used... Calculate the seepage stability index, where β is taken as an empirical coefficient of 0.85. This is the moving average of the seepage velocity sequence. This represents the maximum value in the seepage pressure sequence. When the calculated S value exceeds a preset threshold of 2.5, seepage stability information containing the location of abnormal seepage areas is generated. This information is transmitted to the dam reinforcement design system via an encrypted protocol, triggering the automatic generation of a seepage barrier reinforcement scheme.

[0076] Furthermore, it also includes: for each dam displacement monitoring data in the dam displacement monitoring data sequence included in the dam monitoring data, performing a risk identification step: filtering the dam displacement monitoring data to obtain filtered displacement data; inputting the filtered displacement data into a pre-trained stability risk identification model to obtain a stability risk value; performing time series analysis based on the stability risk value to determine the stability risk trend; in response to determining that the stability risk value is greater than a preset risk threshold or that the stability risk trend indicates an increase in risk, identifying the dam displacement monitoring data and information characterizing the existence of stability risk as stability risk information; and identifying each identified stability risk information as at least one stability risk information.

[0077] Filtering employs either a moving average algorithm or a wavelet denoising algorithm to eliminate high-frequency noise interference in the displacement data. The stability risk identification model is trained using historical displacement data and corresponding risk labels, and a random forest algorithm is used to construct the classification model. The input dimensions include displacement amount, displacement rate, and displacement direction angle. Time series analysis uses an autoregressive integral moving average model to fit the trend of risk values ​​within a continuous time window, calculating the slope of risk change as the basis for trend judgment. The preset risk threshold is dynamically adjusted based on the safety factor in the dam design parameters, specifically calculated by multiplying the safety factor by the allowable displacement value.

[0078] Preferably, in the process of processing the dam displacement monitoring data sequence, wavelet transform is first used to filter the original displacement monitoring data to eliminate environmental noise interference. The filtered displacement data is then input into a stability risk identification model trained based on the random forest algorithm. This model is trained by mapping historical displacement data to stability states and outputs the corresponding stability risk values. The risk value sequence is further analyzed using an ARIMA time series model to calculate the moving average and standard deviation of the risk values. When the risk value exceeds a preset threshold or the trend analysis shows that the risk growth rate exceeds the critical slope, the system automatically triggers an early warning mechanism, generating risk information containing the coordinates of the displacement monitoring points and the risk level, and transmitting it to the central monitoring platform via an encrypted protocol.

[0079] Through the above technical solution, this application effectively solves the problem of insufficient feature extraction from displacement monitoring data using traditional methods. Through multi-level data processing and intelligent analysis, it achieves dynamic identification and trend prediction of dam stability risks. The combination of machine learning models and time series analysis significantly improves the sensitivity and accuracy of risk identification, enabling timely detection of potential instability signs and providing a reliable basis for engineering maintenance decisions.

[0080] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A method for analyzing the impact of horizontal joint settings on the overall stability of a dam body, comprising: Receive dam design parameters as input and obtain the dam design parameters as target parameters; Based on the target parameters, the following stability impact analysis is performed: Collect panel horizontal joint setting data, dam body monitoring data, and dam body environmental load data corresponding to the target parameters; Based on the panel horizontal seam setting data, panel stress analysis is performed to generate panel stress distribution information; Based on the dam monitoring data, a dam stability risk assessment is conducted to obtain at least one stability risk information. Based on the dam's environmental load data, dynamic response analysis of the dam is performed to generate dynamic response information of the dam. Based on the at least one stability risk information, the panel stress distribution information, and the dam body dynamic response information, a collaborative analysis of dam body stability is performed to generate collaborative analysis results. In the stability impact analysis process, in response to receiving panel deformation monitoring data sent by the panel deformation monitoring device, panel deformation impact information is generated based on the panel deformation monitoring data; Based on the collaborative analysis results and the panel deformation impact information, dam reinforcement recommendations are generated. The dam reinforcement recommendation information is sent to the dam reinforcement design equipment.

2. The method according to claim 1, characterized in that, The panel horizontal seam setting data includes setting data for each panel unit. Each panel unit setting data includes a unit number and a horizontal seam parameter set. The horizontal seam parameter set includes panel thickness parameters and panel elastic modulus parameters. Based on the panel horizontal seam setting data, panel stress analysis is performed to generate panel stress distribution information, including: for each panel unit setting data included in the panel horizontal seam setting data, the following steps are performed: based on the panel thickness parameters and panel elastic modulus parameters included in the panel unit setting data, the stress value of the corresponding panel unit is calculated. In response to determining that the stress value is greater than a preset stress threshold, the panel unit setting data and information characterizing the risk of panel stress anomalies are determined as panel stress distribution sub-information; The determined sub-information of each panel stress distribution is used to define the panel stress distribution information.

3. The method according to claim 2, characterized in that, The calculation of the stress value of the corresponding panel unit based on the panel thickness parameter and panel elastic modulus parameter included in the panel unit setting data includes: Obtain the dam load data corresponding to the panel unit, the dam load data including water pressure value and rockfill pressure value; Based on the panel thickness parameter, panel elastic modulus parameter, water pressure value, and rockfill pressure value, the stress value is calculated according to the stress calculation formula, wherein the stress calculation formula is: ,in, Indicates the stress value. This represents the stress adjustment factor, and E represents the panel's elastic modulus parameter. Indicates the water pressure value. t represents the pressure value of the rockfill, and t represents the panel thickness parameter.

4. The method according to claim 1, characterized in that, Each of the at least one stability risk information corresponds to a dam body location identifier, and the method of performing a collaborative analysis of dam body stability based on the at least one stability risk information, the panel stress distribution information, and the dam body dynamic response information, and generating a collaborative analysis result, includes: sending the panel stress distribution information, the at least one stability risk information, and the dam body dynamic response information to at least one analysis terminal; For each of the at least one stability risk information, the following analysis steps are performed: the dam body part identifier corresponding to the stability risk information is determined as the risk part identifier; The stability risk information is sent to at least one analysis terminal corresponding to the risk location identifier; The collaborative analysis results are generated based on the panel stress distribution information and the dam dynamic response information.

5. The method according to claim 1, characterized in that, The step of generating panel deformation impact information based on the panel deformation monitoring data includes: for each panel deformation value in the panel deformation value sequence included in the panel deformation monitoring data, performing the following deformation impact analysis processing: inputting the panel deformation value into a pre-trained panel deformation impact assessment model to obtain panel deformation impact assessment information corresponding to the panel deformation value; In response to determining that the panel deformation impact assessment information characterizes the dam body part corresponding to the detected panel deformation value has a stability risk, the panel measuring point location information corresponding to the panel deformation value is determined as the risk location information; The location information of each identified risk point is used as the panel deformation impact information.

6. The method according to claim 5, characterized in that, The panel deformation impact assessment model is trained through the following steps: Historical panel deformation value sequences and historical panel deformation impact assessment information were collected as initial training samples; The initial training samples are cleaned to remove outliers, resulting in cleaned training samples. The cleaned training samples are normalized to obtain normalized training samples. Based on the normalized training samples, a neural network model is trained, wherein the neural network model includes an input layer, a hidden layer, and an output layer, and the hidden layer uses an activation function to perform a nonlinear transformation; The trained neural network model is validated using a validation set, and the model accuracy is calculated. In response to the model accuracy falling below a preset accuracy threshold, the hyperparameters of the neural network model are adjusted and retrained until the model accuracy meets the preset accuracy threshold, thus obtaining the panel deformation impact assessment model.

7. The method according to claim 1, characterized in that, After receiving panel deformation monitoring data sent by the panel deformation monitoring device, the method further includes: for each panel deformation value in the panel deformation value sequence included in the panel deformation monitoring data, performing the following panel crack detection processing: based on the dam load analysis corresponding to the target parameter, normalizing the panel deformation value to obtain a normalized panel deformation value; The normalized panel deformation value is input into a pre-trained panel crack recognition model to obtain the panel crack risk value. Based on the panel crack risk value, a panel crack early warning information is generated; The panel crack warning information is sent to at least one panel maintenance terminal.

8. The method according to claim 7, characterized in that, The panel crack identification model is trained through the following steps: Historical panel deformation value sequences and historical panel crack data were collected as initial training samples. The initial training samples are subjected to feature extraction processing to extract panel deformation features and crack features to obtain feature training samples; The feature training samples are standardized to obtain standardized training samples; Based on the standardized training samples, a support vector machine model is trained, wherein the support vector machine model uses a kernel function for non-linear classification; Cross-validation was used to optimize the trained support vector machine model and adjust the model parameters. Based on the optimized support vector machine model, the model performance index is calculated. In response to the model performance index meeting the preset performance threshold, the panel crack recognition model is obtained.

9. The method according to claim 1, characterized in that, The method further includes: In response to receiving dam seepage monitoring data sent by the dam seepage monitoring equipment, the dam seepage monitoring data is identified as dam seepage data to be analyzed. Based on the seepage data to be analyzed in the dam body, the seepage velocity sequence and seepage pressure value sequence are extracted; Based on the seepage velocity sequence and seepage pressure sequence, a seepage stability index is calculated, wherein the seepage stability index is calculated using a seepage stability formula, which is: Where S represents the seepage stability index, This represents the seepage adjustment coefficient. This represents the average value of the seepage velocity sequence. This represents the maximum value in the osmotic pressure value sequence; In response to determining that the seepage stability index is greater than a preset seepage threshold, seepage stability information of the dam body is generated; The seepage stability information of the dam body is sent to the dam body reinforcement design equipment.

10. The method according to claim 1, characterized in that, The process of conducting a dam stability risk assessment based on the dam monitoring data yields at least one stability risk information item, including: For each dam displacement monitoring data in the dam displacement monitoring data sequence included in the dam monitoring data, the following risk identification steps are performed: the dam displacement monitoring data is filtered to obtain filtered displacement data; The filtered displacement data is input into a pre-trained stability risk identification model to obtain the stability risk value; Based on the stability risk value, time series analysis is performed to determine the stability risk trend; In response to determining that the stability risk value is greater than a preset risk threshold or that the stability risk trend indicates an increase in risk, the dam displacement monitoring data and information characterizing the existence of stability risk are determined as stability risk information. Each identified stability risk is designated as at least one stability risk.

Images