Water conservancy project safety monitoring method and system based on artificial intelligence

Abstract

The invention relates to the field of hydraulic engineering safety monitoring, in particular to a hydraulic engineering safety monitoring method and system based on artificial intelligence. The method comprises the following steps: acquiring historical monitoring data, classifying the historical monitoring data into a plurality of water regimen categories, and respectively establishing a proportional relation graph; respectively processing the received real-time monitoring data, and judging the affiliation state of the corresponding real-time monitoring data in the proportional relation graph; based on a preset atlas discreteness rule, identifying whether the corresponding monitoring section has a proportional drift anomaly at present; identifying a synchronous offset section cluster according to the abnormal condition of the proportional drift; analyzing the synchronous offset section clusters, determining a disturbance conduction relation between the synchronous offset section clusters, and constructing a disturbance linkage path; and constructing a monitoring situation map of the water conservancy project area according to the proportion map model, the proportion drift anomaly identification result and the disturbance linkage path. According to the invention, the monitoring precision and timeliness of the water conservancy project can be improved.

Description

Technical Field

[0001] This invention relates to the field of water conservancy project safety monitoring, specifically to a water conservancy project safety monitoring method and system based on artificial intelligence. Background Technology

[0002] With the continuous increase in the number of large-scale water conservancy projects in my country, their operational safety has become a key issue in hydrological scheduling, water resource optimization and allocation, and flood control and disaster reduction. In order to effectively manage the operational status of water conservancy projects, it is often necessary to set up multiple monitoring sections in key areas to collect multi-source hydrological monitoring data, including flow velocity, flow rate, water level, and rainfall. More and more water conservancy monitoring systems are introducing artificial intelligence technology and attempting to process massive amounts of monitoring data in depth through intelligent clustering, feature recognition and graph modeling.

[0003] Chinese patent publication number CN119150434A discloses a water conservancy project dam safety monitoring device and monitoring method, including: learning the historical water level influence characteristics and the historical water levels corresponding to the historical water level influence characteristics by using a deep learning algorithm to obtain a water level prediction model that can predict the water level, and then identifying the real-time water level influence characteristics through the water level prediction model, and finally realizing the prediction of the water level.

[0004] In existing technologies, current water conservancy project safety monitoring methods lack dynamic analysis under different hydrological scenarios and are difficult to adapt to changing hydrological classification conditions. In terms of data anomaly identification, they often ignore the relationship between flow velocity and flow rate and cannot build synchronous offset analysis and linkage path reasoning between multiple cross sections. Existing technologies have failed to effectively integrate artificial intelligence algorithms to process the complex distribution characteristics of monitoring data and cannot achieve data analysis between historical hydrological data and real-time data. These are the problems we need to solve. Summary of the Invention

[0005] The purpose of this invention is to address the problems existing in the background technology by proposing an artificial intelligence-based method and system for monitoring the safety of water conservancy projects.

[0006] The technical solution of this invention: A method for safety monitoring of water conservancy projects based on artificial intelligence, comprising the following steps: S1. Obtain historical monitoring data from multiple monitoring sections in water conservancy projects, classify the historical monitoring data into multiple water condition categories according to hydrological characteristics, and establish a proportional relationship map of each monitoring section under the corresponding water condition category. S2. Process the real-time monitoring data received from multiple monitoring sections respectively, determine the status of the corresponding real-time monitoring data in the proportional relationship map, and identify whether there is a proportional drift anomaly in the corresponding monitoring section based on the preset map discreteness rules. S3. Identify synchronous offset section clusters based on the proportional drift anomalies in the proportional relationship map; analyze the synchronous offset section clusters, determine the disturbance transmission relationship between synchronous offset section clusters based on the proportional offset consistency, and construct the disturbance linkage path. S4. Based on the scale map model, the scale drift anomaly identification results, and the disturbance linkage path, construct the monitoring status map of the water conservancy project area.

[0007] Preferably, the process of acquiring historical monitoring data from multiple monitoring sections in a water conservancy project, classifying the historical monitoring data into multiple hydrological categories according to hydrological characteristics, and establishing a proportional relationship map of each monitoring section under the corresponding hydrological category includes: Historical monitoring data from multiple monitoring sections deployed within the water conservancy project area are acquired, and the monitoring sections are numbered to obtain monitoring section numbers. Historical monitoring data includes historical flow velocity data, historical flow rate data, historical hydrological data, and historical monitoring periods. Historical hydrological data includes water level data, rainfall data, and upstream-downstream differences. The collected flow velocity data and flow rate data are clustered based on their distribution characteristics in a two-dimensional feature space formed within the historical monitoring period to classify water condition categories, construct water condition classification labels, and number the water condition classification labels to obtain water condition category numbers. Using the monitoring section number and the hydrological category number as a joint index, the velocity and flow data of each monitoring section corresponding to each hydrological category are analyzed to construct a corresponding proportional relationship map. The proportional relationship map expresses the quantitative relationship between velocity and flow in the form of a scatter cloud or a fitted curve, and the following three structural intervals are divided accordingly.

[0008] Preferably, the process of processing the real-time monitoring data received from multiple monitoring sections and determining the corresponding real-time monitoring data's place in the proportional relationship map includes: Real-time monitoring data received from multiple monitoring sections in water conservancy projects are collected, and state attribution analysis and anomaly identification of proportional drift are performed on the proportional relationship graphs. The real-time monitoring data includes real-time flow velocity data, real-time flow rate data, and real-time monitoring time. Based on the monitoring section number and real-time monitoring time of the real-time monitoring data, the water condition category structure corresponding to the real-time monitoring data in the proportional relationship map is matched; the real-time monitoring data is normalized in the two-dimensional feature space to ensure that the real-time monitoring data and the proportional coordinate system in the proportional relationship map are consistent in dimensions, and mapped into the proportional relationship map under the corresponding water condition category. Based on the three types of structural intervals in the proportional relationship graph, the attribution status of real-time monitoring data on the proportional relationship graph is determined: the attribution status includes stable state, transitional state and abnormal state.

[0009] Preferably, the process of identifying whether there is an abnormal proportional drift at the corresponding monitoring section based on the preset spectral discreteness rules is as follows: The map discreteness rules are set, including map discreteness rule one, map discreteness rule two, and map discreteness rule three. Map discreteness rule one states that if the normalized Euclidean distance between the real-time monitored data point and the boundary of the stable interval exceeds a preset tolerance threshold, it indicates a significant proportional deviation trend. Map discreteness rule two states that if, within n consecutive sampling periods, at least m data points in the same monitoring section continuously fall into the map discrete interval and there is no regression behavior of falling back into the stable interval, it indicates a proportional structure drift event. Map discreteness rule three states that if, within period T, data points in the same monitoring section jump into multiple discrete intervals or move away from the main dense area of ​​the map, combined with the point cloud density, it is determined to be a structural proportional instability phenomenon caused by non-random disturbance. When any two simultaneously satisfy the discreteness rule of the map, it is determined that there is a proportional drift anomaly in the corresponding monitoring section. The result of the proportional drift anomaly will be recorded in real time and the monitoring section number and its status will be output.

[0010] Preferably, the process of identifying synchronous migration section clusters based on proportional drift anomalies in the proportional relationship map includes: In water conservancy projects, all monitoring sections with proportional drift anomalies are acquired, and the proportional change feature sequence within the historical period U is extracted. The proportional change feature is the sequence corresponding to the ratio of flow velocity to flow rate per unit time of the monitoring section. Based on the normalized coordinate difference, coordinate change direction, and spectral interval jump sequence, a proportional change vector is constructed. The proportional change vectors of multiple monitoring sections with proportional drift anomalies are compared and analyzed. Using cosine similarity based on the angle, a set of sections with similar offset features is identified and denoted as the synchronous offset section cluster.

[0011] Preferably, the process of analyzing the synchronous offset section cluster, determining the disturbance transmission relationship between the synchronous offset section clusters based on the proportional offset consistency, and constructing the disturbance linkage path includes: The characteristic sequence of the proportion change of the key monitoring section at the upstream water intake of the water conservancy project is obtained and denoted as the reference source sequence. The key monitoring section refers to the monitoring section located at the main upstream water intake, which is the reference monitoring position for identifying the starting point of the disturbance. The characteristic sequence of the proportion change of each monitoring section in the synchronous offset section cluster is compared and a proportion offset rule is set. If the proportion offset rule is satisfied, there is a disturbance transmission relationship. Based on the disturbance transmission relationship identified by the proportional offset rule, the structural path connecting the upstream inlet point and each abnormal section is constructed in the spatial topology map of the water conservancy project, and a disturbance linkage path is formed with the proportional relationship map to form a joint index relationship.

[0012] Preferably, the process of constructing a monitoring status map of the water conservancy project area based on the scale map model, the scale drift anomaly identification results, and the disturbance linkage path includes: Based on the distribution of the attribution status of each monitoring section in the proportional relationship map during the current observation period, offset status identifiers are generated; combined with the offset status identifiers of the starting point of the disturbance linkage path, the location of potential disturbance sources is constructed and highlighted in the proportional relationship map to construct a monitoring situation map.

[0013] This invention also discloses an artificial intelligence-based water conservancy project safety monitoring system, including a management center, which is communicatively connected to a water conservancy monitoring module, a status identification module, an anomaly disturbance module, and a water conservancy situation module. The water conservancy monitoring module is used to acquire historical monitoring data from multiple monitoring sections in water conservancy projects, classify the historical monitoring data into multiple water condition categories according to hydrological characteristics, and establish a proportional relationship map of each monitoring section under the corresponding water condition category. The status recognition module is used to process the real-time monitoring data received from multiple monitoring sections, determine the status of the corresponding real-time monitoring data in the proportional relationship map, and identify whether there is a proportional drift anomaly in the corresponding monitoring section based on the preset map discreteness rules. The abnormal disturbance module is used to identify synchronous offset section clusters based on the proportional drift anomalies in the proportional relationship map; analyze the synchronous offset section clusters, determine the disturbance transmission relationship between synchronous offset section clusters based on the proportional offset consistency, and construct the disturbance linkage path; The water conservancy situation module is used to construct a monitoring situation map of the water conservancy project area based on the scale map model, the scale drift anomaly identification results, and the disturbance linkage path.

[0014] Compared with the prior art, the above-mentioned technical solution of the present invention has the following beneficial technical effects: by constructing a proportional relationship map adapted to different water conditions, and combining the attribution status of real-time monitoring data with the discreteness rules of the map, intelligent discrimination of the proportional stability status of monitoring sections is realized, effectively improving the anomaly identification accuracy and response timeliness of multi-section monitoring data of water conservancy projects; by using feature clustering and similarity measurement methods in artificial intelligence, synchronous offset section clusters and their disturbance linkage paths are identified, and a monitoring situation map is constructed, which can realize the visualization highlighting and dynamic positioning of disturbance source locations, enhancing the intelligent perception capability and regional risk early warning capability of complex disturbance propagation. Attached Figure Description

[0015] Figure 1 This is a flowchart of one embodiment of the present invention. Detailed Implementation

[0016] Example 1, as Figure 1 As shown, the present invention proposes an artificial intelligence-based method for safety monitoring of water conservancy projects, comprising the following steps: S1. Obtain historical monitoring data from multiple monitoring sections in water conservancy projects, classify the historical monitoring data into multiple water condition categories according to hydrological characteristics, and establish a proportional relationship map of each monitoring section under the corresponding water condition category. S2. Process the real-time monitoring data received from multiple monitoring sections respectively, determine the status of the corresponding real-time monitoring data in the proportional relationship map, and identify whether there is a proportional drift anomaly in the corresponding monitoring section based on the preset map discreteness rules. S3. Identify synchronous offset section clusters based on the proportional drift anomalies in the proportional relationship map; analyze the synchronous offset section clusters, determine the disturbance transmission relationship between synchronous offset section clusters based on the proportional offset consistency, and construct the disturbance linkage path. S4. Based on the scale map model, the scale drift anomaly identification results, and the disturbance linkage path, construct the monitoring status map of the water conservancy project area.

[0017] It should be further explained that, in the specific implementation process, the process of acquiring historical monitoring data from multiple monitoring sections in the water conservancy project, classifying the historical monitoring data into multiple hydrological categories according to hydrological characteristics, and establishing a proportional relationship map of each monitoring section under the corresponding hydrological category is as follows: Historical monitoring data from multiple monitoring sections deployed within the water conservancy project area are acquired. These monitoring sections refer to cross-sectional locations set along hydraulic structures such as rivers, open channels, or water conveyance tunnels. They are used for long-term real-time monitoring of hydrological and hydraulic parameters such as water velocity, flow rate, and water level. These sections are deployed at key control points, flood-prone areas, or upstream and downstream nodes of the project, and are numbered to obtain monitoring section numbers. The historical monitoring data is uploaded in real-time to the water conservancy monitoring center via a telemetry terminal using sensor devices such as electromagnetic velocity meters, ultrasonic flow meters, and water level gauges. The historical monitoring data includes historical flow velocity data, historical flow rate data, historical hydrological data, and historical monitoring cycles. The historical flow velocity data represents the average flow velocity of water at the cross-section per unit historical time. The historical flow rate data represents the volume of water passing through the cross-section per unit historical time, calculated based on the measured water level combined with the cross-sectional area and flow velocity. The historical hydrological data is used to enhance subsequent hydrological assessments and includes water level data, rainfall data, and upstream-downstream differences. The upstream-downstream differences refer to the difference in flow rate or water level at a certain moment between upstream and downstream cross-sections set at different locations in the same river segment, used to characterize hydrodynamic fluctuations or the possibility of siltation. The collected flow velocity and flow rate data are clustered based on their distribution characteristics in a two-dimensional feature space formed during historical monitoring periods to classify water conditions into categories, construct water condition classification labels, and assign numbers to the water condition classification labels to obtain water condition category numbers.

[0018] It should be noted that the specific process of clustering based on the distribution characteristics in the two-dimensional feature space formed within the historical monitoring period is as follows: Historical monitoring data from each monitoring section is cleaned and standardized to remove physically invalid values ​​and outliers. The K-Means++ clustering algorithm, an artificial intelligence algorithm, is used to cluster the historical monitoring data of each monitoring section. The clustering dimension is the two-dimensional feature space formed by velocity and flow data. The inter-cluster separation and intra-cluster compactness of each candidate cluster number K are calculated based on the silhouette coefficient. The cluster number K corresponding to the largest silhouette coefficient is selected as the optimal cluster number. The clustering results are used as the basis for classifying the hydrological response characteristics of the corresponding monitoring sections at different times. Each cluster is defined as a hydrological category, which includes dry and stable type, early rising type, flood peak response type, and receding disturbance type. The classification criteria for the hydrological categories can refer to the industry standards for water conservancy information management or common types of measured response processes. Each hydrological category is assigned a unique hydrological category number, a hydrological feature center value, and a corresponding representative time period. The hydrological feature center value is the cluster center point, and the representative time period is the time period with the most observations in the corresponding category. Hydrological classification labels are constructed. Using the monitoring section number and the water condition category number as a joint index, the flow velocity and flow rate data of each monitoring section corresponding to each water condition category are analyzed to construct a corresponding proportional relationship map. The proportional relationship map records the relationship curve or point cloud density distribution between flow velocity and flow rate in the monitoring section under different water condition categories in two-dimensional coordinate form, and extracts the structural labels of stable intervals, discrete intervals and transition intervals for subsequent real-time data status identification. Specifically, for each hydrological category, the corresponding velocity and flow rate data are normalized to be distributed in a two-dimensional spatial coordinate system with unified dimensions. Based on the density distribution of data points in the two-dimensional plane and the main fitting direction, a proportional relationship map is constructed. The proportional relationship map expresses the quantitative relationship between velocity and flow rate in the form of a scatter cloud or a fitted curve, and is divided into the following three structural intervals: a stable interval, a discrete interval, and a transition interval. The stable interval is the region with the densest data distribution and the smallest fluctuation, reflecting the normal proportional relationship under this category. The discrete interval is the region that deviates significantly from the stable interval and is outside the distribution boundary, often related to equipment disturbance, measurement error, or upstream hydraulic abrupt change. The transition interval is the region between the stable and discrete intervals, used to accommodate some proportional shifts caused by normal disturbances.

[0019] It should be further explained that, in the specific implementation process, the real-time monitoring data received from multiple monitoring sections are processed separately to determine the corresponding real-time monitoring data's belonging status in the proportional relationship map; and based on the preset map dispersion rules, the process of identifying whether there is a proportional drift anomaly at the corresponding monitoring section is as follows: Real-time monitoring data received from multiple monitoring sections in water conservancy projects are collected, and the state attribution analysis of the proportional relationship spectrum and the identification of proportional drift anomalies are performed respectively, so as to realize the function of water situation determination and anomaly early warning under asynchronous monitoring conditions. The real-time monitoring data includes real-time flow velocity data, real-time flow rate data, and real-time monitoring time. The real-time flow velocity data represents the average flow velocity of the water body at the monitoring section per unit time at the current observation time, and the real-time flow rate data represents the volume of water passing through the monitoring section per unit time. Based on the monitoring section number and real-time monitoring time of the real-time monitoring data, the water condition category structure corresponding to the real-time monitoring data in the proportional relationship map is matched; the real-time monitoring data is normalized in the two-dimensional feature space to ensure that the scale coordinate system of the real-time monitoring data and the proportional relationship map are consistent, and mapped into the proportional relationship map under the corresponding water condition category.

[0020] Based on the three structural intervals in the proportional relationship map, the attribution status of real-time monitoring data on the proportional relationship map is determined: if the real-time monitoring data point is located in the stable interval of the proportional relationship map, it indicates that the current proportional relationship is consistent with the structural relationship of the corresponding water condition category, and the attribution status is recorded as "stable state"; if the real-time monitoring data point is located in the transition interval of the proportional relationship map, it indicates that there is a certain degree of proportional deviation, but it is still within the allowable range, and the attribution status is recorded as "transitional state"; if the real-time monitoring data point is located in the discrete interval of the proportional relationship map, it indicates that the current proportional relationship between flow velocity and flow rate has significantly deviated from the structural distribution under the same historical water condition, and the attribution status is recorded as "abnormal state". The map discreteness rules are set, including map discreteness rule one, map discreteness rule two, and map discreteness rule three. Map discreteness rule one states that if the normalized Euclidean distance between the real-time monitored data point and the boundary of the stable interval exceeds a preset tolerance threshold, it indicates a significant proportional deviation trend. Map discreteness rule two states that if, within n consecutive sampling periods, at least m data points in the same monitoring section continuously fall into the map discrete interval and there is no regression behavior of falling back into the stable interval, it indicates a proportional structure drift event. Map discreteness rule three states that if, within period T, data points in the same monitoring section jump into multiple discrete intervals or move away from the main dense area of ​​the map, combined with the point cloud density, it is determined to be a structural proportional instability phenomenon caused by non-random disturbance. When any two simultaneously satisfy the discreteness rule of the map, it is determined that there is a proportional drift anomaly in the corresponding monitoring section. The result of the proportional drift anomaly will be recorded in real time and the monitoring section number and its status will be output.

[0021] It should be further explained that, in the specific implementation process, based on the anomalies in the proportional drift in the proportional relationship map, clusters of synchronous offset sections are identified; the synchronous offset section clusters are analyzed, and the disturbance transmission relationship between the synchronous offset section clusters is determined based on the consistency of proportional offset. The process of constructing the disturbance linkage path is as follows: In water conservancy projects, all monitoring sections with proportional drift anomalies are acquired, and the proportional change feature sequence within the historical period U is extracted. The proportional change feature is the sequence corresponding to the ratio of flow velocity to flow rate per unit time of the monitoring section. Based on the normalized coordinate difference, coordinate change direction, and spectral interval jump sequence, a proportional change vector is constructed. The proportional change vectors of multiple monitoring sections with proportional drift anomalies are compared and analyzed. Using cosine similarity based on the angle, a set of sections with similar offset features is identified and denoted as the synchronous offset section cluster.

[0022] A sequence of proportional changes in key monitoring sections at the upstream intake of a water conservancy project is obtained and denoted as the reference source sequence. The key monitoring sections refer to those located at the main upstream intake, serving as reference monitoring locations for identifying the starting point of disturbances. The proportional change sequence of each monitoring section in the synchronous offset section cluster is compared, and proportional offset rules are set. These rules include three rules: Rule 1, Rule 2, and Rule 3. Rule 1 requires that the cosine similarity between the proportional change vectors of upstream and downstream sections exceeds a set similarity threshold. Rule 2 requires that the coordinate change directions are the same and the anomaly occurrence times satisfy a temporal sequence. Rule 3 requires that the reference monitoring location and the abnormal monitoring section maintain a stable or slow transition state. If the proportional offset rules are satisfied, a disturbance propagation relationship exists. Based on the disturbance transmission relationship identified by the proportional offset rule, a structural path is constructed in the spatial topology map of the hydraulic engineering to connect the upstream inlet point and each abnormal section, thereby constructing a disturbance linkage path. The disturbance linkage path improves the ability to identify and respond to abnormal transmission under asynchronous monitoring conditions, and can serve as the structural basis for subsequent intervention and regulation, source tracing analysis and hydraulic control logic, and forms a joint index relationship with the proportional relationship map.

[0023] It should be further explained that, in the specific implementation process, the process of constructing a monitoring status map of the water conservancy project area based on the scale map model, the results of scale drift anomaly identification, and the disturbance linkage path is as follows: Based on the distribution of the status of each monitoring section in the proportional relationship map during the current observation period, a shift status identifier is generated. The shift status identifier graphically represents the current proportional stability level of the monitoring section. The proportional stability level is a stable level, a slight drift level, a moderate drift level, and a severe drift level, which correspond to the state of the monitoring section falling into the stable interval, approaching the transition interval, the transition interval, and the discrete interval in the proportional map, respectively. By combining the offset status indicators of the starting points of disturbance linkage paths, potential disturbance source locations are constructed. These potential disturbance source locations represent the most likely starting disturbance source section or regional node in the current disturbance propagation path. The determination criteria include, but are not limited to, prioritizing the monitoring section with the lowest proportional stability level and the earliest time of the first occurrence of the anomaly among the first nodes in all linkage paths as a candidate disturbance source point. If multiple disturbance source candidate points exist, they are selected based on the priority of the path topology in terms of the large impact coverage and the wide propagation range, generating potential disturbance source locations, which are then highlighted in the map to construct a monitoring situation map. The monitoring situation map includes proportional stability level, key disturbance linkage paths, and potential disturbance source locations. The structure of the key disturbance linkage paths represents the associated propagation paths during the proportional instability process. The potential disturbance source locations are used to provide precise target support for subsequent anomaly localization, emergency response, and intervention control.

[0024] Example 2: The artificial intelligence-based water conservancy project safety monitoring system proposed in this invention is applied to the artificial intelligence-based water conservancy project safety monitoring method described in Example 1. Specifically, it includes a management center, which is communicatively connected to a water conservancy monitoring module, a status identification module, an anomaly disturbance module, and a water conservancy situation module. The water conservancy monitoring module is used to acquire historical monitoring data from multiple monitoring sections in water conservancy projects, classify the historical monitoring data into multiple water condition categories according to hydrological characteristics, and establish a proportional relationship map of each monitoring section under the corresponding water condition category. The status recognition module is used to process the real-time monitoring data received from multiple monitoring sections, determine the status of the corresponding real-time monitoring data in the proportional relationship map, and identify whether there is a proportional drift anomaly in the corresponding monitoring section based on the preset map discreteness rules. The abnormal disturbance module is used to identify synchronous offset section clusters based on the proportional drift anomalies in the proportional relationship map; analyze the synchronous offset section clusters, determine the disturbance transmission relationship between synchronous offset section clusters based on the proportional offset consistency, and construct the disturbance linkage path; The water conservancy situation module is used to construct a monitoring situation map of the water conservancy project area based on the scale map model, the scale drift anomaly identification results, and the disturbance linkage path.

[0025] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited thereto. Various changes can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of the present invention.

Claims

1. A method for safety monitoring of water conservancy projects based on artificial intelligence, characterized in that, Includes the following steps: S1. Obtain historical monitoring data from multiple monitoring sections in water conservancy projects, classify the historical monitoring data into multiple water condition categories according to hydrological characteristics, and establish a proportional relationship map of each monitoring section under the corresponding water condition category. S2. Process the real-time monitoring data received from multiple monitoring sections respectively, determine the belonging status of the corresponding real-time monitoring data in the proportional relationship map; and identify whether there is a proportional drift anomaly in the corresponding monitoring section based on the preset map discreteness rules. S3. Identify synchronous offset section clusters based on the proportional drift anomalies in the proportional relationship map; analyze the synchronous offset section clusters, determine the disturbance transmission relationship between synchronous offset section clusters based on the proportional offset consistency, and construct the disturbance linkage path. S4. Based on the scale map model, the scale drift anomaly identification results, and the disturbance linkage path, construct the monitoring status map of the water conservancy project area.

2. The method for monitoring the safety of water conservancy projects based on artificial intelligence according to claim 1, characterized in that, The process of acquiring historical monitoring data from multiple monitoring sections in a water conservancy project, classifying the historical monitoring data into multiple hydrological categories according to hydrological characteristics, and establishing a proportional relationship map of each monitoring section under the corresponding hydrological category includes: Historical monitoring data from multiple monitoring sections deployed within the water conservancy project area are acquired, and the monitoring sections are numbered to obtain monitoring section numbers. Historical monitoring data includes historical flow velocity data, historical flow rate data, historical hydrological data, and historical monitoring periods. Historical hydrological data includes water level data, rainfall data, and upstream-downstream differences. The collected flow velocity data and flow rate data are clustered based on their distribution characteristics in a two-dimensional feature space formed within the historical monitoring period to classify water condition categories, construct water condition classification labels, and number the water condition classification labels to obtain water condition category numbers. Using the monitoring section number and the hydrological category number as a joint index, the velocity and flow data of each monitoring section corresponding to each hydrological category are analyzed to construct a corresponding proportional relationship map. The proportional relationship map expresses the quantitative relationship between velocity and flow in the form of a scatter cloud or a fitted curve, and the following three structural intervals are divided accordingly.

3. The method for monitoring the safety of water conservancy projects based on artificial intelligence according to claim 2, characterized in that, The process of processing real-time monitoring data received from multiple monitoring sections and determining the corresponding real-time monitoring data's place in the proportional relationship map includes: Real-time monitoring data received from multiple monitoring sections in water conservancy projects are collected, and state attribution analysis and scale drift anomaly identification are performed on the scale relationship maps. The real-time monitoring data includes real-time flow velocity data, real-time flow rate data, and real-time monitoring time. Based on the monitoring section number and real-time monitoring time to which the real-time monitoring data belongs, the water situation category structure corresponding to the real-time monitoring data in the scale relationship map is matched. The real-time monitoring data is normalized in a two-dimensional feature space to ensure that the scale coordinate system of the real-time monitoring data and the scale relationship map are consistent in dimensions, and mapped into the scale relationship map under the corresponding water situation category. Based on the three types of structural intervals in the proportional relationship graph, the attribution status of real-time monitoring data on the proportional relationship graph is determined: the attribution status includes stable state, transitional state and abnormal state.

4. The method for monitoring the safety of water conservancy projects based on artificial intelligence according to claim 3, characterized in that, Based on the preset spectral discreteness rules, the process of identifying whether there is a proportional drift anomaly at the corresponding monitoring section is as follows: The map discreteness rules are set, including map discreteness rule one, map discreteness rule two, and map discreteness rule three. Map discreteness rule one states that if the normalized Euclidean distance between the real-time monitored data point and the boundary of the stable interval exceeds a preset tolerance threshold, it indicates a significant proportional deviation trend. Map discreteness rule two states that if, within n consecutive sampling periods, at least m data points in the same monitoring section continuously fall into the map discrete interval and there is no regression behavior of falling back into the stable interval, it indicates a proportional structure drift event. Map discreteness rule three states that if, within period T, data points in the same monitoring section jump into multiple discrete intervals or move away from the main dense area of ​​the map, combined with the point cloud density, it is determined to be a structural proportional instability phenomenon caused by non-random disturbance. When any two simultaneously satisfy the discreteness rule of the map, it is determined that there is a proportional drift anomaly in the corresponding monitoring section. The result of the proportional drift anomaly will be recorded in real time and the monitoring section number and its status will be output.

5. The method for monitoring the safety of water conservancy projects based on artificial intelligence according to claim 4, characterized in that, The process of identifying synchronous migration section clusters based on the proportional drift anomalies in the proportional relationship map includes: In water conservancy projects, all monitoring sections with proportional drift anomalies are acquired, and the proportional change feature sequence within the historical period U is extracted. The proportional change feature is the sequence corresponding to the ratio of flow velocity to flow rate per unit time of the monitoring section. Based on the normalized coordinate difference, coordinate change direction, and spectral interval jump sequence, a proportional change vector is constructed. The proportional change vectors of multiple monitoring sections with proportional drift anomalies are compared and analyzed. Using cosine similarity based on the angle, a set of sections with similar offset features is identified and denoted as the synchronous offset section cluster.

6. The method for monitoring the safety of water conservancy projects based on artificial intelligence according to claim 5, characterized in that, The process of analyzing synchronous migration section clusters, determining the disturbance transmission relationship between synchronous migration section clusters based on proportional migration consistency, and constructing disturbance linkage paths includes: The characteristic sequence of the proportion change of the key monitoring section at the upstream water intake of the water conservancy project is obtained and denoted as the reference source sequence. The key monitoring section refers to the monitoring section located at the main upstream water intake, which is the reference monitoring position for identifying the starting point of the disturbance. The characteristic sequence of the proportion change of each monitoring section in the synchronous offset section cluster is compared and a proportion offset rule is set. If the proportion offset rule is satisfied, there is a disturbance transmission relationship. Based on the disturbance transmission relationship identified by the proportional offset rule, the structural path connecting the upstream inlet point and each abnormal section is constructed in the spatial topology map of the water conservancy project, and a disturbance linkage path is formed with the proportional relationship map to form a joint index relationship.

7. The method for monitoring the safety of water conservancy projects based on artificial intelligence according to claim 6, characterized in that, The process of constructing a monitoring status map of the water conservancy project area based on the scale map model, the results of scale drift anomaly identification, and the disturbance linkage path includes: Based on the distribution of the attribution status of each monitoring section in the proportional relationship map during the current observation period, offset status identifiers are generated; combined with the offset status identifiers of the starting point of the disturbance linkage path, the location of potential disturbance sources is constructed and highlighted in the proportional relationship map to construct a monitoring situation map.

8. An artificial intelligence-based water conservancy project safety monitoring system, specifically applied to the artificial intelligence-based water conservancy project safety monitoring method described in any one of claims 1 to 7, comprising a management center, characterized in that, The management center's communication connections include a water resources monitoring module, a status identification module, an anomaly disturbance module, and a water resources situation module. The water conservancy monitoring module is used to acquire historical monitoring data from multiple monitoring sections in water conservancy projects, classify the historical monitoring data into multiple water condition categories according to hydrological characteristics, and establish a proportional relationship map of each monitoring section under the corresponding water condition category. The status recognition module is used to process the real-time monitoring data received from multiple monitoring sections, determine the status of the corresponding real-time monitoring data in the proportional relationship map, and identify whether there is a proportional drift anomaly in the corresponding monitoring section based on the preset map discreteness rules. The abnormal disturbance module is used to identify synchronous offset section clusters based on the proportional drift anomalies in the proportional relationship map; analyze the synchronous offset section clusters, determine the disturbance transmission relationship between synchronous offset section clusters based on the proportional offset consistency, and construct the disturbance linkage path; The water conservancy situation module is used to construct a monitoring situation map of the water conservancy project area based on the scale map model, the scale drift anomaly identification results, and the disturbance linkage path.

Images