Dam safety monitoring method and system based on big data

By capturing the spatiotemporal correlation and constructing a causal relationship diagram in dam safety monitoring, combining the LSTM model and local anomaly factor method, the problem of incomplete monitoring results is solved, efficient abnormality detection and emergency response of the dam are achieved, and monitoring is improved Accuracy and risk identification capabilities.

CN119025946AInactive Publication Date: 2024-11-26SHENZHEN HONGHEDA ELECTRONICS CO LTD
View PDF 0 Cites 20 Cited by

Patent Information

Application Number
CN202411033258.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-30
Publication Date
2024-11-26
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing dam safety monitoring technology fails to fully utilize the spatial and temporal correlation when processing complex environmental data, and the dynamic relationship capture ability between monitoring points is weak, resulting in incomplete monitoring results and insufficient abnormal detection and response.

Method used

By collecting data from dam monitoring points, capturing spatiotemporal correlation characteristics for clustering and partitioning, building a causal relationship diagram, using LSTM model to predict environmental changes, and using local anomaly factor methods for abnormal detection and classification, the cause of abnormality is analyzed based on the causal relationship diagram to formulate Emergency response.

Benefits of technology

Improve the accuracy and prediction capabilities of dam monitoring, enhance the early identification and rapid response capabilities of potential risks, and ensure the safety and environmental stability of the dam.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119025946A_ABST
    Figure CN119025946A_ABST
Patent Text Reader

Abstract

The invention discloses a dam safety monitoring method and system based on big data, and relates to the technical field of dam safety monitoring. Capturing the space-time correlation characteristics of the monitoring points, carrying out clustering and partitioning on the monitoring points, and constructing a causal relationship graph according to the layout and the mutual relationship of the dam monitoring points; and extracting and fusing data features of each clustering partition, constructing an environment change prediction model by using an LSTM model to calculate a residual value of each monitoring point, and performing anomaly detection by using a local anomaly factor method. The method can accurately track and predict the potential risk area through determining the causal relationship among the monitoring points, can identify the data deviating from the normal mode in real time through constructing the environment change prediction model and combining the LOF method, effectively maintains the dam safety and the environment stability, remarkably improves the accuracy and prediction capability of dam monitoring, and improves the safety of the dam. And the early recognition and quick response capabilities of potential risks are enhanced.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of dam safety monitoring, and in particular to a dam safety monitoring method and system based on big data. Background Art

[0002] Dam safety monitoring refers to the use of various technologies and methods to continuously or regularly monitor the structural performance of the dam and the surrounding environment to ensure the safety and reliability of the dam under design and operating conditions. Dam safety monitoring is a key measure to prevent dam failures and catastrophic dam failures. With the rapid development of big data and Internet of Things technologies, dam safety monitoring technology has also made significant progress. With the introduction of advanced computing power and machine learning technology, real-time monitoring and prediction technologies have begun to be used in the maintenance of such infrastructure, which can provide more continuous and comprehensive monitoring data. However, although the application of these technologies has improved the real-time monitoring and the degree of automation of data processing, the existing technologies still face many challenges in terms of deep data integration, prediction accuracy and timeliness of anomaly detection. When processing complex environmental data, the existing technologies often fail to make full use of the spatiotemporal correlation in the data, and the ability to capture the dynamic relationship between the various monitoring points of the dam is weak, resulting in the monitoring results not being able to fully reflect the safety status of the dam and the need for further improvement in the detection and response of anomalies in the dam. Summary of the invention

[0003] In view of the problems existing in the above-mentioned existing dam safety monitoring methods and systems based on big data, the present invention is proposed.

[0004] Therefore, the problem to be solved by the present invention is that the existing technology often fails to fully utilize the spatiotemporal correlation in the data when processing complex environmental data, and the ability to capture the dynamic relationship between the various monitoring points of the dam is weak, resulting in the monitoring results cannot fully reflect the safety status of the dam and there is still room for improvement in the abnormal detection and response of the dam.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions: a dam safety monitoring method based on big data, which includes collecting environmental data from dam monitoring points and preprocessing them, capturing the spatiotemporal correlation characteristics of the monitoring points to cluster and partition the monitoring points, and constructing a cause-effect relationship diagram according to the layout and mutual relationship of the dam monitoring points;

[0006] Extract the data features of each cluster partition and fuse them, use the LSTM model to build an environmental change prediction model to calculate the residual value of each monitoring point, and use the local anomaly factor method to detect anomalies;

[0007] The detected dam abnormalities are classified and the causes of the abnormalities are analyzed through the cause-effect diagram, and emergency responses are carried out according to the abnormality type and cause.

[0008] As a preferred solution of the dam safety monitoring method based on big data described in the present invention, the collecting of environmental data from dam monitoring points and preprocessing thereof refers to forming a comprehensive monitoring network based on the structural diagram of the dam and the deployment of sensors at the monitoring points, collecting environmental data of each monitoring point in real time through the monitoring network, performing outlier detection and data standardization on the collected data, marking a timestamp for each data point and synchronizing the data to a central database.

[0009] As a preferred solution of the dam safety monitoring method based on big data of the present invention, wherein: the capturing of the spatiotemporal correlation characteristics of the monitoring points to cluster and partition the monitoring points and constructing a causal relationship diagram according to the layout and mutual relationship of the dam monitoring points refers to selecting the input variables for clustering based on the collected dam environmental data and structural data, constructing a data matrix, each row represents a monitoring point, and the column represents the selected clustering variable;

[0010] Use Ward's method for hierarchical clustering and use Euclidean distance to calculate the similarity between clusters:

[0011]

[0012] Where D ij is the Euclidean distance between monitoring point i and monitoring point j, x iu and x ju is the value of the u-th variable at the i-th and j-th monitoring points, and n is the total number of variables;

[0013] Analyze the clustering structure, determine the groups of each monitoring point, select the time series data of the highly dynamic monitoring points from the groups of each monitoring point, and use the DTW algorithm to calculate the distance between the time series after preprocessing the selected time series data:

[0014]

[0015] In the formula, X and Y are two time series data, dist(x' I ,y' J ) is the sequence point x' I and y' J The distance between

[0016] Initialize the cluster center, select a random time series as the initial cluster center and iterate. During the iteration process, assign each time series to the nearest cluster center based on the DTW distance. Update the cluster center by calculating the average DTW distance of all time series in each cluster until the cluster center reaches the maximum number of iterations. Cluster the monitoring points based on the similarity of the monitoring point data to divide the monitoring points into regions.

[0017] Construct a VAR model, use the unit root test to check the stationarity of each time series and use the information criterion to determine the optimal lag order of the VAR model and determine the VAR model parameters. Perform residual diagnosis and stability tests on the model to obtain a trained VAR model. Use the VAR model to perform Granger causality test on the time series data of each pair of monitoring points:

[0018]

[0019] In the formula, Z t and W t are the time series variables of two monitoring points, p and q are the lag periods, β a and μ b is the model coefficient, ε t is the error term, Z t-a is the value of the target variable Z at the previous a time point, W t-b is the value of the target variable W at the previous b time points, α is a constant term, and t is time;

[0020] Analyze and record the causal relationship between each pair of monitoring points, including the causal direction and the strength of the correlation. Construct a comprehensive causal relationship graph based on the results of the Granger causality test. The nodes in the graph represent the monitoring points and the edges represent the causal relationships. Store the clustering results and causal relationship graph in the central database.

[0021] As a preferred solution of the dam safety monitoring method based on big data described in the present invention, the extraction of data features of each cluster partition and the fusion thereof refers to selecting a feature extraction method according to the monitoring point data in each cluster partition, calculating the statistical features of each monitoring point, including the mean and variance, and performing FFT to extract the frequency components, combining the features extracted from each monitoring point and the dependency relationship in the causal graph, and using a typical correlation analysis method to fuse the data features.

[0022] As a preferred solution of the dam safety monitoring method based on big data described in the present invention, wherein: the use of the LSTM model to build an environmental change prediction model to calculate the residual value of each monitoring point, and the sampling of the local anomaly factor method for anomaly detection refers to the use of the LSTM model to build an environmental change prediction model;

[0023] Add LSTM layer, set input dimension, number of hidden layer units;

[0024] Add a fully connected layer and an output layer after the last LSTM layer, and set the output layer dimension to 1;

[0025] Using mean square error as the loss function, the historical environmental data training data set is input into the model, and the model parameters are iteratively optimized based on the Adam optimizer until the loss is minimized and then the iteration is stopped. The iterated model parameters are input into the environmental change prediction model to obtain the environmental change prediction model.

[0026] Input the real-time environmental data characteristics of each cluster partition into the environmental change prediction model to obtain the environmental changes of each cluster partition of the dam in the future period;

[0027] For each time point, calculate the difference R between the actual observed value and the predicted value t :

[0028] R t =T1-T2,

[0029] In the formula, T1 is the actual observed environmental value, T2 is the predicted environmental value, and t is the time;

[0030] Evaluate the abnormality of each data point based on the residual data of the monitoring point:

[0031]

[0032] Where H is the anomaly score of the data point, N(c) is the neighborhood of data point c, r(c,k) is the distance from data point c to data point k, l(k) is the local reachability density of data point k, and l(c) is the local reachability density of data point c.

[0033] Use the LOF algorithm to calculate the environmental data of each monitoring point, record the LOF scores of all data points, set the abnormal judgment threshold Q, and compare the abnormal score H with the judgment threshold Q to determine the degree of abnormality of the data point. If the calculated abnormal score H is greater than the abnormal judgment threshold Q, the data point is abnormal.

[0034] As a preferred solution of the dam safety monitoring method based on big data described in the present invention, wherein: the classification of detected dam abnormalities and the analysis of abnormal causes through a causal relationship diagram refers to building an abnormal classification model using an RBF kernel function, setting a candidate value range and an initial candidate value of a regularization parameter C and a kernel function parameter γ, training fault classification using historical feature encoding as a training set, and defining a fault classification optimization objective function;

[0035] The candidate values ​​of the regularization parameter C and the kernel function parameter γ are combined, each combination is substituted into the fault classification model for model training to obtain the optimized objective function output, and the regularization parameter C and the kernel function parameter γ combination of the minimum optimized objective function is selected as the model parameter output to the fault classification model;

[0036] Collect historical abnormal data and input them into the abnormal classification model to obtain the dam abnormality type;

[0037] The abnormal monitoring point is located in the causal relationship diagram, and the breadth-first search BFS algorithm is used to start from the abnormal monitoring point, trace all causal paths, identify other monitoring points directly related to the abnormal monitoring point, record all related causal paths and abnormal monitoring points involved, collect historical and real-time data of the abnormal monitoring point and all abnormal monitoring points on the causal path, analyze the data change pattern of the abnormal monitoring point and the external environmental factors to obtain the cause of the dam abnormality.

[0038] As a preferred solution of the dam safety monitoring method based on big data described in the present invention, the emergency response based on the abnormality type and cause refers to formulating corresponding emergency response measures based on the data abnormality type of the dam abnormality monitoring point and the abnormality cause obtained by analysis, determining the required resources and personnel to be deployed to relevant locations, continuously monitoring the execution effect of the emergency measures and collecting data during the implementation process, evaluating the actual effect of the emergency measures on dam safety maintenance and making real-time adjustments.

[0039] Another object of the present invention is to provide a dam safety monitoring system based on big data, which comprises:

[0040] Data collection module, used to deploy a comprehensive monitoring network to collect environmental data from various monitoring points of the dam in real time and perform pre-processing;

[0041] A module is constructed to partition the various monitoring points of the dam using a clustering algorithm and to construct and update a causal relationship diagram to analyze the dynamic relationship between the monitoring points;

[0042] The data fusion module is used to extract and fuse the features of the monitoring point data in each cluster partition;

[0043] Anomaly detection module, which is used to build an environmental change prediction model to predict future environmental change trends and perform real-time dam anomaly detection based on residual values;

[0044] The response management module is used to classify the detected anomalies and analyze the causes of dam anomalies based on the cause-effect diagram, formulate corresponding measures according to the anomaly type and cause, and adjust them in real time.

[0045] A computer device comprises: a memory and a processor; the memory stores a computer program, and the processor implements the steps of a dam safety monitoring method based on big data when executing the computer program.

[0046] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of a dam safety monitoring method based on big data.

[0047] The beneficial effects of the present invention are as follows: by clarifying the causal relationship between each monitoring point, the present invention can accurately track and predict potential risk areas, and construct an environmental change prediction model combined with the LOF method to identify data that deviates from the normal mode in real time, effectively maintaining dam safety and environmental stability, which not only significantly improves the accuracy and prediction ability of dam monitoring, but also enhances the early identification and rapid response capabilities to potential risks. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0049] Figure 1 Schematic diagram of the process of dam safety monitoring method based on big data.

[0050] Figure 2 This is a schematic diagram of the structure of the dam safety monitoring system based on big data. DETAILED DESCRIPTION

[0051] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings.

[0052] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0053] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor is it a separate or selective embodiment that is mutually exclusive with other embodiments.

[0054] Example 1, reference Figure 1 and Figure 2 , which is the first embodiment of the present invention, and which provides a dam safety monitoring method based on big data. The dam safety monitoring method based on big data includes:

[0055] S1. Collect environmental data from dam monitoring points and pre-process them, capture the spatiotemporal correlation characteristics of the monitoring points, cluster and partition the monitoring points, and construct a causal relationship diagram based on the layout and mutual relationship of the dam monitoring points;

[0056] Specifically, collecting environmental data from dam monitoring points and preprocessing it means deploying sensors based on the dam's structural diagram and monitoring points to form a comprehensive monitoring network, collecting environmental data from each monitoring point in real time through the monitoring network, performing outlier detection and data standardization on the collected data, marking a timestamp for each data point, and synchronizing the data to a central database.

[0057] Through the precise layout of the monitoring network, the key parts of the dam and potential risk areas can be fully covered to ensure the comprehensiveness and representativeness of the data. The effective layout of monitoring points can maximize monitoring efficiency, reduce blind spots, and improve the timeliness and accuracy of risk warnings. Through advanced data preprocessing technology, noise and errors in the data can be effectively screened out to improve data quality. This step ensures the accuracy and reliability of data analysis, and provides a solid foundation for subsequent data analysis and decision-making. Marking timestamps for each data point helps to conduct time series analysis, track data change trends, and identify potential abnormal development patterns. Timestamps also support data synchronization processing, especially in distributed database environments, to maintain data consistency and timeliness.

[0058] Furthermore, the spatiotemporal correlation characteristics of the monitoring points are captured to cluster and partition the monitoring points, and a causal relationship diagram is constructed based on the layout and mutual relationship of the dam monitoring points. The input variables for clustering are selected based on the collected dam environmental data and structural data, and a data matrix is ​​constructed, in which each row represents a monitoring point and the column represents the selected clustering variable;

[0059] Use Ward's method for hierarchical clustering and use Euclidean distance to calculate the similarity between clusters:

[0060]

[0061] Where D ij is the Euclidean distance between monitoring point i and monitoring point j, x iu and x ju is the value of the u-th variable at the i-th and j-th monitoring points, and n is the total number of variables;

[0062] Analyze the clustering structure, determine the groups of each monitoring point, select the time series data of the highly dynamic monitoring points from the groups of each monitoring point, and use the DTW algorithm to calculate the distance between the time series after preprocessing the selected time series data:

[0063]

[0064] In the formula, X and Y are two time series data, dist(x' I ,y' J ) is the sequence point x' I and y' J The distance between

[0065] Initialize the cluster center, select a random time series as the initial cluster center and iterate. During the iteration process, assign each time series to the nearest cluster center based on the DTW distance. Update the cluster center by calculating the average DTW distance of all time series in each cluster until the cluster center reaches the maximum number of iterations. Cluster the monitoring points based on the similarity of the monitoring point data to divide the monitoring points into regions.

[0066] Construct a VAR model, use the unit root test to check the stationarity of each time series and use the information criterion to determine the optimal lag order of the VAR model and determine the VAR model parameters. Perform residual diagnosis and stability tests on the model to obtain a trained VAR model. Use the VAR model to perform Granger causality test on the time series data of each pair of monitoring points:

[0067]

[0068] In the formula, Z t and W t are the time series variables of two monitoring points, p and q are the lag periods, β a and μ b is the model coefficient, ε t is the error term, Z t-a is the value of the target variable Z at the previous a time point, W t-b is the value of the target variable W at the previous b time points, α is a constant term representing the intercept of the model, and t is time;

[0069] Analyze and record the causal relationship between each pair of monitoring points, including the causal direction and the strength of the correlation. According to the results of the Granger causality test, a comprehensive causal relationship graph is constructed. The nodes in the graph represent the monitoring points and the edges represent the causal relationship. The clustering results and causal relationship graph are stored in the central database.

[0070] Considering the structural complexity of the dam and the variability of the environment, clustering of monitoring points can effectively group monitoring points according to geographical location and environmental factors (such as water flow velocity, soil moisture, etc.). For example, monitoring points upstream and downstream of the dam may show different data characteristics due to differences in water flow conditions. Cluster analysis can help identify these patterns and group monitoring points with similar characteristics. By clustering monitoring points with similar environmental responses, environmental change trends and potential risks in specific areas can be analyzed more accurately. This approach improves the accuracy of data analysis, making predictions and responses more targeted and timely. For example, if monitoring points in a cluster area continuously report a sharp rise in water levels, detailed inspections and emergency plans for the area can be quickly initiated without having to conduct a comprehensive emergency response for the entire dam, thereby saving resources and reducing unnecessary operations. During the operation of the dam, monitoring points in different parts may have causal relationships due to factors such as structural connectivity or water flow paths. For example, changes in readings at a monitoring point on the dam foundation may affect the stability of the downstream sidewall. Causal graphs can reveal these associations based on historical and real-time data. The constructed causal graphs not only help engineers understand the interactions between monitoring points, but also predict the potential impact of specific events on other parts of the dam. In actual operation, this information is crucial, especially when dealing with emergencies such as foundation leakage or sidewall cracks. It can quickly determine the scope of the problem and deploy inspection and maintenance teams in a targeted manner. The LSTM model is used to predict possible environmental changes at each monitoring point of the dam, and the local anomaly factor method is combined for anomaly detection, which can monitor and predict the environmental and structural health status of the dam in real time. For example, by analyzing the water level and flow rate data upstream of the dam during the rainy season, the LSTM model can predict the trend of water level changes in the next few hours, which enables the monitoring team to intervene before the problem becomes a potential safety risk, such as adjusting the amount of water released from the reservoir or strengthening the monitoring of a part of the dam, thereby effectively avoiding or mitigating disastrous consequences.

[0071] S2, extract the data features of each cluster partition and fuse them, use the LSTM model to build an environmental change prediction model to calculate the residual value of each monitoring point, and use the local anomaly factor method to detect anomalies;

[0072] Specifically, extracting the data features of each cluster partition and fusing them means selecting a feature extraction method based on the monitoring point data in each cluster partition, calculating the statistical features of each monitoring point, including the mean and variance, and performing FFT to extract the frequency components, combining the features extracted from each monitoring point and the dependency relationship in the causal graph, and using the canonical correlation analysis method to fuse the data features.

[0073] Statistical features and frequency components are calculated from the monitoring point data of each cluster partition. These features are used to describe the current state and historical behavior of the monitoring points. Subsequently, these features are fused using canonical correlation analysis to explore the correlation between different monitoring points and between them and the overall safety of the dam, making the information extracted from a large amount of data more accurate and representative, and improving the quality and efficiency of data analysis. Through feature fusion, the interactions between the monitoring points and their impact on dam safety can be better understood, thus providing a scientific basis for risk assessment and decision-making. Statistical features help monitoring point managers quickly grasp the normality and range of changes in the monitoring points, while FFT analysis reveals the frequency characteristics of the data, which is crucial for identifying periodic changes and potential structural problems. The constructed causal relationship diagram not only provides an intuitive representation of the mutual influence between monitoring points, but also guides targeted monitoring and maintenance activities, especially when predicting and dealing with potential risks.

[0074] Furthermore, the LSTM model is used to construct an environmental change prediction model to calculate the residual value of each monitoring point, and the local anomaly factor method is sampled for anomaly detection, which refers to the use of the LSTM model to construct an environmental change prediction model;

[0075] Add LSTM layer, set input dimension, number of hidden layer units;

[0076] Add a fully connected layer and an output layer after the last LSTM layer, and set the output layer dimension to 1;

[0077] Using mean square error as the loss function, the historical environmental data training data set is input into the model, and the model parameters are iteratively optimized based on the Adam optimizer until the loss is minimized and then the iteration is stopped. The iterated model parameters are input into the environmental change prediction model to obtain the environmental change prediction model.

[0078] Input the real-time environmental data characteristics of each cluster partition into the environmental change prediction model to obtain the environmental changes of each cluster partition of the dam in the future period;

[0079] For each time point, calculate the difference R between the actual observed value and the predicted value t :

[0080] R t =T1-T2,

[0081] In the formula, T1 is the actual observed environmental value, T2 is the predicted environmental value, and t is the time;

[0082] Evaluate the abnormality of each data point based on the residual data of the monitoring point:

[0083]

[0084] Where H is the anomaly score of the data point, N(c) is the neighborhood of data point c, which consists of the nearest k points, r(c,k) is the distance from data point c to data point k, l(k) is the local reachability density, which is the reciprocal mean of the reachability distance of data point k, and l(c) is the local reachability density of data point c;

[0085] The LOF algorithm is used to calculate the environmental data of each monitoring point, and the LOF scores of all data points are recorded. The abnormality judgment threshold Q is set based on the statistical distribution of the LOF score. The abnormality score H is compared with the judgment threshold Q to determine the degree of abnormality of the data point. If the calculated abnormality score H is greater than the abnormality judgment threshold Q, the data point is abnormal.

[0086] In practical applications, the LSTM model first learns the dynamic change pattern of the dam environment based on historical monitoring data (such as water level, pressure, temperature, etc.). These data are collected from multiple monitoring points distributed at key structural locations of the dam, covering various cluster partitions of the dam. Subsequently, the model will predict the environmental status of each monitoring point in the next few days or weeks, such as predicting the pressure changes that the dam may experience in the upcoming rainy season. The prediction model can warn of potential risks before the actual situation becomes dangerous, such as predicting that the rising water level due to continuous rainfall may put pressure on the dam structure. Managers can take measures in advance based on these predictions, such as adjusting the reservoir water level to prevent flooding. This predictive capability makes dam management more proactive rather than reactive, greatly increasing the time window for dealing with potential problems and reducing the risk of disasters. The LOF algorithm is used to monitor the data of each monitoring point in real time. When the actual environmental data monitored (such as sudden temperature rise or pressure anomaly) differs significantly from the prediction of the LSTM model, the system automatically marks these deviations as potential anomalies. Real-time anomaly detection allows the management team to quickly respond to possible safety issues, such as the development of cracks or early signs of leakage. Rapidly identifying and addressing these issues can avoid greater structural damage and potential environmental disasters. The application of the LOF algorithm improves the sensitivity and accuracy of anomaly detection, reducing unnecessary maintenance costs caused by false alarms and risks caused by missed alarms.

[0087] S3, classify the detected dam abnormalities and analyze the causes of the abnormalities through the cause-effect diagram, and make emergency responses according to the abnormality types and causes;

[0088] Specifically, classifying the detected dam abnormalities and analyzing the abnormal causes through the cause-effect diagram refers to building an abnormal classification model using the RBF kernel function, setting the candidate value range and initial candidate value of the regularization parameter C and the kernel function parameter γ, training the fault classification using the historical feature encoding as a training set, and defining the fault classification optimization objective function;

[0089] The candidate values ​​of the regularization parameter C and the kernel function parameter γ are combined, each combination is substituted into the fault classification model for model training to obtain the optimized objective function output, and the regularization parameter C and the kernel function parameter γ combination of the minimum optimized objective function is selected as the model parameter output to the fault classification model;

[0090] Collect historical abnormal data and input them into the abnormal classification model to obtain the dam abnormality type;

[0091] The abnormal monitoring point is located in the causal relationship diagram, and the breadth-first search BFS algorithm is used to start from the abnormal monitoring point, trace all causal paths, identify other monitoring points directly related to the abnormal monitoring point, record all related causal paths and abnormal monitoring points involved, collect historical and real-time data of the abnormal monitoring point and all abnormal monitoring points on the causal path, analyze the data change pattern of the abnormal monitoring point and the external environmental factors to obtain the cause of the dam abnormality.

[0092] During the operation of the dam, various types of abnormal situations may be encountered, such as structural changes caused by natural disasters (floods, earthquakes), or safety problems caused by improper human operation and equipment failure. The SVM model constructed using the RBF kernel function in the present invention can effectively classify these abnormal situations. For example, in the flood season, the dam monitoring point may record abnormally high water level or pressure data. Through the pre-trained classification model, the system can quickly identify these data as natural disaster-related anomalies, thereby activating the corresponding early warning system, implementing emergency drainage measures, and ensuring the safety of the dam. Constructing a causal relationship diagram can help engineers clearly understand the dependencies and influence relationships between the monitoring points. For example, if the monitoring data shows that the pressure on one side of the dam is abnormally increased, through the causal relationship diagram, other monitoring points connected to the point can be quickly determined to check whether there are similar abnormal performances, thereby helping to locate the source of the problem and determine whether the local structure is damaged or the overall pressure is too large. This analysis is crucial for formulating specific maintenance or reinforcement measures to ensure timely and effective response to potential structural safety issues. In the dam monitoring system, the monitored data usually has complex time series characteristics, such as periodic fluctuations and sudden changes. By analyzing the changing patterns of these data, especially by combining historical and real-time data, it is possible to more accurately predict the risks that the dam may face. For example, by comparing historical data from the same period, the system may identify the gradual rise in water levels due to continuous rainfall, so that water release measures can be taken in advance to avoid excessive water levels and put pressure on the dam structure.

[0093] Furthermore, emergency response based on the type and cause of anomalies refers to formulating corresponding emergency response measures based on the data anomaly type of the dam abnormal monitoring point and the analyzed cause of the anomaly, determining the required resources and personnel to be deployed to relevant locations, continuously monitoring the implementation effect of emergency measures and collecting data during the implementation process, evaluating the actual effect of emergency measures on dam safety maintenance and making real-time adjustments.

[0094] Based on the type and analysis results of abnormal monitoring data, specific emergency response measures are formulated. Subsequently, the required resources and personnel are mobilized to ensure that these resources can be deployed to key locations in a timely manner. A systematic solution is provided to quickly and effectively respond to various abnormal events detected by the dam monitoring system. Through pre-established measures, potential damage or risks can be minimized, and continuous monitoring ensures that emergency measures can be adjusted and optimized according to actual conditions. Data collection not only helps evaluate the effectiveness of current measures, but also provides valuable experience and data support for future emergency responses. The actual effect of emergency response is evaluated based on the collected implementation data, and emergency measures are adjusted in real time according to the evaluation results, making emergency response measures more accurate and effective, and able to flexibly respond to complex and changing emergencies.

[0095] Example 2, reference Figure 2 , which is the second embodiment of the present invention, and which is different from the previous embodiment, provides a dam safety monitoring system based on big data, which includes:

[0096] Data collection module, used to deploy a comprehensive monitoring network to collect environmental data from various monitoring points of the dam in real time and perform pre-processing;

[0097] A module is constructed to partition the various monitoring points of the dam using a clustering algorithm and to construct and update a causal relationship diagram to analyze the dynamic relationship between the monitoring points;

[0098] The data fusion module is used to extract and fuse the features of the monitoring point data in each cluster partition;

[0099] Anomaly detection module, which is used to build an environmental change prediction model to predict future environmental change trends and perform real-time dam anomaly detection based on residual values;

[0100] The response management module is used to classify the detected anomalies and analyze the causes of dam anomalies based on the cause-effect diagram, formulate corresponding measures according to the anomaly type and cause, and adjust them in real time.

[0101] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.

[0102] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0103] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0104] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used to implement: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

Claims

1. A dam safety monitoring method based on big data, characterized in that: include, Collect environmental data from dam monitoring points and pre-process them, capture the spatiotemporal correlation characteristics of the monitoring points, cluster and partition the monitoring points, and construct a causal relationship diagram based on the layout and interrelationships of the dam monitoring points; Extract the data features of each cluster partition and fuse them, use the LSTM model to build an environmental change prediction model to calculate the residual value of each monitoring point, and use the local anomaly factor method to detect anomalies; The detected dam abnormalities are classified and the causes of the abnormalities are analyzed through the cause-effect diagram, and emergency responses are carried out according to the abnormality type and cause.

2. The dam safety monitoring method based on big data according to claim 1, characterized in that: The collecting of environmental data from the dam monitoring points and preprocessing thereof refers to the deployment of sensors according to the structural diagram of the dam and the monitoring points to form a comprehensive monitoring network, collecting the environmental data of each monitoring point in real time through the monitoring network, performing outlier detection and data standardization on the collected data, marking a timestamp for each data point and synchronizing the data to a central database.

3. The dam safety monitoring method based on big data according to claim 2, characterized in that: The capturing of the spatiotemporal correlation characteristics of the monitoring points to cluster and partition the monitoring points and constructing a causal relationship diagram according to the layout and mutual relationship of the dam monitoring points refers to selecting the input variables for clustering based on the collected dam environmental data and structural data, constructing a data matrix, each row represents a monitoring point, and the column represents the selected clustering variable; Use Ward's method for hierarchical clustering and use Euclidean distance to calculate the similarity between clusters: Where D ij is the Euclidean distance between monitoring point i and monitoring point j, x iu and x ju is the value of the u-th variable at the i-th and j-th monitoring points, and n is the total number of variables; Analyze the clustering structure, determine the groups of each monitoring point, select the time series data of the highly dynamic monitoring points from the groups of each monitoring point, and use the DTW algorithm to calculate the distance between the time series after preprocessing the selected time series data: In the formula, X and Y are two time series data, dist(x' I ,y' J ) is the sequence point x' I and y' J The distance between Initialize the cluster center, select a random time series as the initial cluster center and iterate. During the iteration process, assign each time series to the nearest cluster center based on the DTW distance. Update the cluster center by calculating the average DTW distance of all time series in each cluster until the cluster center reaches the maximum number of iterations. Cluster the monitoring points based on the similarity of the monitoring point data to divide the monitoring points into regions. Construct a VAR model, use the unit root test to check the stationarity of each time series and use the information criterion to determine the optimal lag order of the VAR model and determine the VAR model parameters. Perform residual diagnosis and stability tests on the model to obtain a trained VAR model. Use the VAR model to perform Granger causality test on the time series data of each pair of monitoring points: In the formula, Z t and W t are the time series variables of two monitoring points, p and q are the lag periods, β a and μ b is the model coefficient, ε t is the error term, Z t-a is the value of the target variable Z at the previous a time point, W t-b is the value of the target variable W at the previous b time points, α is a constant term, and t is time; Analyze and record the causal relationship between each pair of monitoring points, including the causal direction and the strength of the correlation. Construct a comprehensive causal relationship graph based on the results of the Granger causality test. The nodes in the graph represent the monitoring points and the edges represent the causal relationships. Store the clustering results and causal relationship graph in the central database.

4. The dam safety monitoring method based on big data according to claim 3 is characterized in that: The extracting and fusing of data features of each cluster partition refers to selecting a feature extraction method based on the monitoring point data in each cluster partition, calculating the statistical features of each monitoring point, including the mean and variance, and performing FFT to extract the frequency components, combining the features extracted from each monitoring point and the dependency relationship in the causal graph, and using a typical correlation analysis method to fuse the data features.

5. The dam safety monitoring method based on big data according to claim 4 is characterized in that: The use of the LSTM model to construct an environmental change prediction model to calculate the residual value of each monitoring point and to perform anomaly detection by sampling a local anomaly factor method refers to the use of the LSTM model to construct an environmental change prediction model; Add LSTM layer, set input dimension, number of hidden layer units; Add a fully connected layer and an output layer after the last LSTM layer, and set the output layer dimension to 1; Using mean square error as the loss function, the historical environmental data training data set is input into the model, and the model parameters are iteratively optimized based on the Adam optimizer until the loss is minimized and then the iteration is stopped. The iterated model parameters are input into the environmental change prediction model to obtain the environmental change prediction model. Input the real-time environmental data characteristics of each cluster partition into the environmental change prediction model to obtain the environmental changes of each cluster partition of the dam in the future period; For each time point, calculate the difference R between the actual observed value and the predicted value t : R t =T1-T2, In the formula, T1 is the actual observed environmental value, T2 is the predicted environmental value, and t is the time; Evaluate the abnormality of each data point based on the residual data of the monitoring point: Where H is the anomaly score of the data point, N(c) is the neighborhood of data point c, r(c,k) is the distance from data point c to data point k, l(k) is the local reachability density of data point k, and l(c) is the local reachability density of data point c. Use the LOF algorithm to calculate the environmental data of each monitoring point, record the LOF scores of all data points, set the abnormal judgment threshold Q, and compare the abnormal score H with the judgment threshold Q to determine the degree of abnormality of the data point. If the calculated abnormal score H is greater than the abnormal judgment threshold Q, the data point is abnormal.

6. The dam safety monitoring method based on big data according to claim 5, characterized in that: The classifying of the detected dam abnormalities and analyzing the abnormal causes through the cause-effect diagram refers to building an abnormal classification model using the RBF kernel function, setting the candidate value range and initial candidate value of the regularization parameter C and the kernel function parameter γ, training the fault classification using the historical feature encoding as a training set, and defining the fault classification optimization objective function; The candidate values ​​of the regularization parameter C and the kernel function parameter γ are combined, each combination is substituted into the fault classification model for model training to obtain the optimized objective function output, and the regularization parameter C and the kernel function parameter γ combination of the minimum optimized objective function is selected as the model parameter output to the fault classification model; Collect historical abnormal data and input them into the abnormal classification model to obtain the dam abnormality type; The abnormal monitoring point is located in the causal relationship diagram, and the breadth-first search BFS algorithm is used to start from the abnormal monitoring point, trace all causal paths, identify other monitoring points directly related to the abnormal monitoring point, record all related causal paths and abnormal monitoring points involved, collect historical and real-time data of the abnormal monitoring point and all abnormal monitoring points on the causal path, analyze the data change pattern of the abnormal monitoring point and the external environmental factors to obtain the cause of the dam abnormality.

7. The dam safety monitoring method based on big data according to claim 6 is characterized in that: The emergency response based on the abnormality type and cause refers to formulating corresponding emergency response measures according to the data abnormality type of the dam abnormality monitoring point and the abnormality cause obtained by analysis, determining the required resources and personnel to be deployed to relevant locations, continuously monitoring the execution effect of the emergency measures and collecting data during the implementation process, evaluating the actual effect of the emergency measures on the safety maintenance of the dam and making real-time adjustments.

8. A dam safety monitoring system based on big data based on the dam safety monitoring method based on big data according to any one of claims 1 to 7, characterized in that: include, Data collection module, used to deploy a comprehensive monitoring network to collect environmental data from various monitoring points of the dam in real time and perform pre-processing; A module is constructed to partition the various monitoring points of the dam using a clustering algorithm and to construct and update a causal relationship diagram to analyze the dynamic relationship between the monitoring points; The data fusion module is used to extract and fuse the features of the monitoring point data in each cluster partition; Anomaly detection module, which is used to build an environmental change prediction model to predict future environmental change trends and perform real-time dam anomaly detection based on residual values; The response management module is used to classify the detected anomalies and analyze the causes of dam anomalies based on the cause-effect diagram, formulate corresponding measures according to the anomaly type and cause, and adjust them in real time.

9. A computer device comprising: A memory and a processor; the memory stores a computer program, characterized in that: when the processor executes the computer program, the steps of the dam safety monitoring method based on big data described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the dam safety monitoring method based on big data described in any one of claims 1 to 7 are implemented.

Citation Information

Cited By

  • Data monitoring and processing method based on dam stability

    CN119293704A

  • Roadbed surface temperature control method and system for roadbed anti-freezing experiment

    CN119396223A

  • Hydraulic structure safety monitoring data abnormity self-diagnosis method and system

    CN119783007A

  • Smart park management method and system based on multi-system integration

    CN119848746A

  • Dangerous waste intelligent supervision system and method based on Internet of Things

    CN119885018A