Multi-source hydrological data monitoring and safety early warning system for intelligent water conservancy

Abstract

The invention belongs to the technical field of water conservancy informatization and automatic control, and discloses an intelligent water conservancy-oriented multi-source hydrological data monitoring and safety early warning system, which comprises a monitoring link for acquiring water level, rainfall, flow velocity, flow and meteorological elements, and acquiring environment information in combination with videos and remote sensing images; the communication processing link supports multi-link transmission and generates an integrity mark for the data; in the data processing link, field mapping is realized based on a unified semantic model, conflict judgment is carried out through a timestamp, an integrity mark and an adjacent monitoring point weighting sequence, and an abnormal mark is generated at the same time; in the digital modeling link, three-dimensional twin models of a river channel, a reservoir and a dam are established, and abnormal factors are written in; in the risk prediction link, a risk assessment result is output in a multi-factor weighting mode; the early warning control link determines a release object based on the three-dimensional permission matrix and executes a boundary rule; and the linkage link issues a scheduling instruction after manual reexamination and confirmation, and feeds back an audit record to the model to form closed-loop correction.

Description

Technical Field

[0001] This invention belongs to the field of water conservancy informatization and automation control technology, specifically a multi-source hydrological data monitoring and safety early warning system for smart water conservancy. Background Technology

[0002] The safe operation of water conservancy projects depends on continuous monitoring and timely early warning of key components such as rivers, reservoirs and dams. Existing hydrological monitoring systems are mostly based on single elements, such as water level stations collecting water levels, rain gauge stations recording rainfall, or local cross-sections observing flow velocity. Although these systems can provide basic data, they often lack the comprehensive utilization of meteorological conditions, video images and remote sensing information, resulting in limited monitoring data coverage, insufficient information dimensions, and difficulty in supporting comprehensive judgment in complex environments.

[0003] In the communication process, traditional systems generally rely on a single transmission method. Extreme weather or complex geographical environments can easily cause link interruptions. The lack of multi-channel redundancy and data integrity verification makes it difficult to detect data loss or errors during transmission in a timely manner. Existing data processing methods usually use simple averaging or fixed priority to resolve multi-source data conflicts, failing to comprehensively consider factors such as time sequence, data reliability, or spatial proximity. Therefore, deviations are prone to occur when facing abnormal or contradictory data.

[0004] In terms of modeling and prediction, although existing research has attempted to apply digital twin technology to water conservancy projects, it has focused on the simulation of physical states and lacks an inheritance mechanism for anomalies in monitoring data. Risk factors often cannot be carried over in the model, resulting in a disconnect between prediction and actual risk. Existing early warning methods mainly rely on hierarchical triggering, with fixed information transmission paths, making it difficult to take into account the multidimensional constraints between watershed level, administrative division and engineering level, and lacking cross-regional and cross-level coordination capabilities.

[0005] Furthermore, existing digital twins mostly adopt a state update method driven by a single physical quantity, lacking explicit coding and cross-cycle inheritance of monitored anomalies, resulting in the inability to reflect the accumulated risk of anomalies in the model's state vector. At the same time, prediction models usually calculate risks directly based on single points or single types of elements, without incorporating data reliability and spatial neighborhood consistency into the same prediction framework, making it difficult to make stable judgments and distributions of water conservancy risks across regions and levels.

[0006] Furthermore, in the emergency dispatch phase, existing systems mainly rely on automated commands or single manual confirmations, lacking review mechanisms and execution log feedback. This results in a lack of a closed loop between dispatch execution and subsequent risk prediction, making it difficult for the system to correct prediction errors in a timely manner. Consequently, existing technologies have shortcomings in multi-source monitoring, data fusion, twin modeling, risk prediction, and hierarchical early warning. Summary of the Invention

[0007] The purpose of this invention is to provide a multi-source hydrological data monitoring and safety early warning system for smart water conservancy, so as to solve the problems mentioned in the background art.

[0008] To achieve the above objectives, the present invention provides the following technical solution: a multi-source hydrological data monitoring and security early warning system for smart water conservancy, comprising: This system conducts multi-source monitoring of typical water conservancy objects such as rivers, reservoirs, and dams. It acquires structured hydrological elements such as water level, rainfall, flow velocity, flow rate, and meteorological data, and supplements environmental situation information by combining video monitoring and remote sensing imagery, so that subsequent analysis is no longer limited to a single data dimension. The collected data is accessed by the communication processing unit, which supports adaptive switching between wired, wireless, and satellite links. When data enters, each data entry is bound with an integrity tag to characterize the reliability of the corresponding link and message, so that the data value and data credibility can be considered simultaneously during subsequent fusion. On this basis, the data fusion processing unit completes field standardization and consistency verification based on a unified semantic model. When there are differences in the same parameter from multiple sources, the system makes a decision based on timestamp, integrity tag weight, and spatial correction order of neighboring monitoring points, reducing the disturbance of conflicting data to the downstream model from the source. In the continuous detection window, deviation trends are identified and anomaly tags corresponding to the parameters are generated one by one. The digital twin modeling unit uses formatted data after adjudication and anomaly markers as input to construct digital twins of rivers, reservoirs, and dams, and updates water levels, flow rates, weather conditions, and structural status in real time. Anomaly markers are written into the twin state factors, enabling the twins to not only reflect physical operational status but also continuously carry accumulated information on abnormal risks. The risk prediction unit extracts predictive factors such as flow rate changes, accumulated rainfall, structural anomalies, and integrity markers based on the twin state factors, and adaptively adjusts the weights of each factor using historical feedback, outputting corresponding risk assessment results. The early warning issuance unit then issues warnings based on the risk assessment results. When the threshold is exceeded, an early warning is triggered, and the target audience and scope of the warning are determined by a three-dimensional permission matrix consisting of watershed level, administrative division, and engineering level. When the warning involves cross-regional or cross-level scenarios, horizontal isolation and vertical transmission rules are executed according to the permission boundaries to ensure that the warning is traceable and can be implemented within the management system. After receiving a high-level warning, the dispatching and linkage unit generates a dispatching instruction, which is then manually reviewed and confirmed before being sent to the execution terminal. The execution logs and manual intervention information of the linkage process form an audit record and are sent back to the twin model to correct subsequent risk prediction parameters, thereby keeping the prediction, early warning and dispatching in a closed loop.

[0009] Preferably, the monitoring process includes multiple types of sensors and terminals: water level sensors deployed at river cross-sections for continuous water level recording; rainfall sensors installed at watershed rain gauge stations for hourly rainfall collection; and flow velocity meters deployed at the monitoring cross-sections for acquiring flow velocity and, in conjunction with cross-sectional geometric parameters, calculating flow rate using a flow calculation unit. The flow rate can be calculated using the following formula: ; In the formula: : Indicates the cross-sectional flow rate, with the unit being cubic meters per second (m's); : Represents the average flow velocity of the cross section, with the unit being meters per second (m / s). Its value can be obtained by weighted averaging of the instantaneous flow velocities collected by the flow velocity measuring instrument at different measuring points in the cross section. : Represents the cross-sectional area through which water flows, in square meters (m²), and its value can be calculated from the water depth obtained by the water level sensor combined with the cross-sectional geometric parameters; Meteorological monitoring units are set up near reservoirs and dams to measure temperature, humidity, air pressure, and wind speed. In addition, cameras and remote sensing terminals are deployed at key hydraulic structures and important water areas to acquire video images and remote sensing images to supplement environmental information. Unlike the acquisition of single hydrological elements in existing technologies, this invention acquires data simultaneously from multiple dimensions such as hydrology, meteorology, and images, providing complete input for subsequent fusion and modeling.

[0010] Preferably, the communication processing stage includes: an interface unit for transmitting collected data via a wired network; a communication unit for wireless transmission; and a unit for transmitting data via a satellite link when the first two types of networks are unavailable. Upon data access, the system generates an integrity tag for each data item. This integrity tag can be calculated based on a message checksum, and the calculation formula is as follows: ; In the formula, For integrity marking, For the first in the message One data unit, The total number of message data units. Preset modulus; The integrity markers generated in the above manner can reflect the consistency and integrity of the message during transmission. Integrity markers can also be generated by timestamp comparison or link confirmation signals.

[0011] Preferably, the data processing stage includes a field mapping unit, a consistency verification unit, and an anomaly marking unit. The field mapping unit standardizes data from different sources based on a unified semantic model. The consistency verification unit compares multi-source data for the same parameter. When conflicts exist, the decision is made in the following order: latest timestamp, higher integrity marker value, and weighted average of neighboring monitoring points. The weighted average can be calculated using the following formula: ; In the formula, These are the corrected parameter values. For the first Measurements from neighboring monitoring points, The weights assigned to this monitoring point, The number of neighboring points involved in the calculation; An anomaly marker unit generates an anomaly marker when the same parameter exceeds a threshold in three consecutive detection windows. The threshold can be set based on historical hydrological statistics or industry standards. The detection window length can be 5 minutes, 10 minutes, or set according to operational needs. Anomaly determination can be described by the following formula: ; in, Indicates an anomaly marker. Indicates the consecutive number The parameter values ​​within each detection window This refers to the expected value or historical average of this parameter. A preset threshold is set. The anomaly markers are stored in association with formatted data for use by the digital twin model.

[0012] Preferably, the digital modeling stage includes a data interface, a modeling unit, a state update unit, and a tag processing unit. The data interface is used to receive formatted data and anomaly tags; the modeling unit generates three-dimensional structural models of rivers, reservoirs, and dams. The status update unit dynamically updates parameters such as water level, flow rate, and weather conditions during operation. This update can be described by the following formula: ; In the formula, For twin models at time State parameters, For actual measured data in the monitoring process These are the state parameters from the previous moment. To update the weight coefficients; The marking processing unit is used to write the anomaly marker into the twin state factor and correct it. The correction can be achieved through the following formula: ; in, The corrected state factor value. Uncorrected state parameters An exception flag (value 0 or 1), This is the anomaly correction factor.

[0013] Preferably, the risk prediction stage includes an interface unit, a factor management unit, a weight adjustment unit, and a result output unit. The interface unit receives the twin state factors output by the modeling stage. The factor management unit sets flow change values, cumulative rainfall values, structural anomaly state parameters, and integrity markers as prediction factors; The weight adjustment unit adjusts the factor weights based on historical feedback in each prediction period. The adjustment method can be error inverse distribution, moving average, or statistical regression. The risk prediction can be calculated using the following formula: ; In the formula, Based on the risk assessment results, For the first The predictive factors include flow change values, cumulative rainfall values, structural anomaly state parameters, and integrity markers. The weights of the corresponding factors, The total number of predictor factors, and their weights during operation. The system will make dynamic adjustments based on the deviation between historical forecasts and actual feedback to ensure the reliability of risk assessment results. After completing the calculation, the result output unit will transmit the risk assessment results to the early warning and control stage.

[0014] Preferably, the early warning control unit includes an input unit, a threshold determination unit, an access control unit, and a boundary processing unit. The input unit receives the risk assessment results output by the risk prediction unit. The threshold determination unit generates an early warning signal when the risk assessment results exceed a set threshold, which can be determined based on historical statistical patterns, industry management standards, or after calibration by a risk prediction model. The access control unit determines the early warning issuance targets based on a three-dimensional access control matrix composed of watershed level, administrative division, and engineering level. The matrix can be stored using a three-dimensional array or table structure, and each entry contains a level identifier, administrative code, and engineering level. Its retrieval relationship can be represented as follows: ; in, For the set of objects to which the early warning is issued, It is a three-dimensional permission matrix. For watershed level identification, For administrative division codes, For engineering grade identification; In cross-regional or cross-level situations, the boundary processing unit performs horizontal isolation or vertical transmission according to the matrix retrieval results to ensure the integrity and accuracy of the transmission of early warning information between different spatial and management levels.

[0015] Preferably, the linkage mechanism includes an instruction generation unit, a review and issuance unit, a log recording unit, and a feedback unit; The instruction generation unit generates a scheduling instruction upon receiving a high-level warning signal; the review and issuance unit manually reviews the scheduling instruction and issues it to the execution terminal after confirmation; the log recording unit records the execution time, operation steps, and manual intervention information during the execution process to form an audit record; the feedback unit writes the audit record to the state update unit of the twin model through the database interface, and uses it as an additional input variable in the next prediction cycle to correct the prediction.

[0016] The beneficial effects of this invention are as follows: 1. This invention achieves comprehensive monitoring of hydrology, meteorology, and environment from multiple dimensions by deploying water level, rainfall, flow velocity, flow rate, and meteorological sensors in river, reservoir, and dam scenarios, combined with video surveillance and remote sensing terminals. Unlike the single-element collection in existing technologies, this invention can form a complete dataset within the same time frame, avoiding judgment bias caused by missing information. At the same time, the communication link introduces a multi-channel switching mechanism of wired, wireless, and satellite links, and generates a bound integrity tag when data is accessed, ensuring that the data always has traceability and reliability during transmission and processing. As a result, this system can maintain stable data acquisition and transmission in extreme weather and complex environments, significantly improving the integrity and continuity of monitoring information.

[0017] 2. This invention establishes a triple conflict resolution mechanism in the data processing stage, which prioritizes timestamps, assigns weight to integrity markers, and uses a weighted average of neighboring monitoring points. This mechanism enables deterministic processing results when there are discrepancies in multi-source data, avoiding the distortion caused by simple averaging in existing schemes. Simultaneously, anomaly markers from continuous detection windows are written into the state factors of the digital twin model, allowing the model to not only reflect physical states but also inherit abnormal states, achieving a cumulative expression of potential risk factors. In the risk prediction stage, the system sets flow changes, rainfall accumulation, structural anomalies, and integrity markers as prediction factors and dynamically adjusts the weights based on historical feedback, making the prediction results more consistent with the actual hydrological evolution. This combination of anomaly inheritance and weight adjustment significantly enhances the accuracy and adaptability of the prediction stage.

[0018] 3. This invention establishes an information dissemination path corresponding to the management system through a three-dimensional permission matrix composed of watershed level, administrative division, and engineering level. It performs horizontal isolation or vertical transmission in cross-regional or cross-level situations, ensuring that early warning information can be accurately transmitted to the target unit according to the institutional boundaries, avoiding omissions or overstepping levels caused by single-path transmission in the prior art. At the same time, in the linkage link, the dispatching instructions must be manually reviewed and confirmed before they can be issued, and an audit record containing operation time, steps and intervention status is generated during the execution process. This record is fed back to the twin model to form a closed-loop correction mechanism between prediction and execution. Through this early warning and dispatch linkage design, this system can achieve cross-level coordination in emergencies and continuously optimize subsequent risk prediction, realizing dynamic protection for the operation of water conservancy projects. Attached Figure Description

[0019] Figure 1 This is a flowchart of the multi-source hydrological data monitoring and safety early warning system for smart water conservancy, as described in this invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] like Figure 1 As shown, this embodiment of the invention provides a multi-source hydrological data monitoring and security early warning system for smart water conservancy, deployed in the monitoring network of typical water conservancy objects such as rivers, reservoirs, and dams within a watershed. The system operates collaboratively according to the following link: multi-source sensing access - reliable transmission marking - semantic fusion adjudication - twin modeling inheritance - risk prediction and assessment - access control early warning release - review and linkage reinjection. The system's monitoring components are set up at river cross-sections, reservoir areas, and key parts of dams. It acquires corresponding hydrological and meteorological elements through water level, rainfall, flow velocity, flow rate, and meteorological sensors. Cameras and remote sensing terminals are deployed at important water areas and hydraulic structures to simultaneously collect on-site video images and remote sensing images as supplementary information on the environmental situation. All types of monitoring data are packaged and output using monitoring point number, parameter type, and collection timestamp as indexes to ensure time alignment and spatial positioning basis for subsequent processing. Monitoring data is accessed by the communication processing stage, which is equipped with wired, wireless, and satellite communication interfaces. It performs adaptive switching according to preset link priorities and real-time link availability. When switching occurs, the original timestamp of the monitoring data at the same moment remains unchanged, and only the link identifier field is updated. The communication processing stage generates a corresponding integrity tag when each data entry arrives, and binds the tag to the data for storage and transmission with the packet. The integrity tag is used to characterize the message consistency and transmission reliability level of the data under the current link conditions, providing a reliable basis for subsequent conflict resolution and risk prediction. After receiving multi-source monitoring data carrying integrity tags, the data processing stage first performs field mapping based on a unified semantic model, converting data fields from different devices and protocols into standard fields in the unified semantic space. Then, it performs consistency checks on the multi-source values ​​of the same parameter. When discrepancies exist, the data processing stage prioritizes data with more recent timestamps. If timestamps are indistinguishable or discrepancies persist, it further compares the reliability levels corresponding to the integrity tags and prioritizes data with higher reliability levels. If the above two steps still cannot eliminate the discrepancies, it introduces data from several monitoring points spatially adjacent to the target monitoring point for weighted correction to obtain the adjudicated parameter value. Within a continuous detection window, the data processing stage performs trend detection on the adjudicated parameter value. When it finds that the same parameter continuously deviates from its historical average or management threshold and meets preset continuity conditions, it generates an anomaly tag corresponding to that parameter. The anomaly tag is associated with and stored with the corresponding formatted data and output to the digital modeling stage. The digital modeling process uses formatted data and anomaly markers as input to construct digital twin models of the river channel, reservoir, and dam. The structural layer of the twin model establishes a three-dimensional structural framework based on the river cross-section geometry, reservoir capacity curve, and dam design parameters, and establishes a mapping relationship between monitoring point numbers and twin nodes. During operation, the digital modeling process updates the twin's water level, flow rate, accumulated rainfall, meteorological elements, and structural operating status in real time based on the formatted data, ensuring that the twin's state remains synchronized with the physical object's state. Anomaly markers are written into the twin's state factors and inherited across cycles with state updates, allowing the twin to explicitly retain accumulated information on anomaly risks while expressing the physical state, forming a set of twin state factors for risk prediction. After receiving the twin state factor set, the risk prediction stage extracts flow change trends, cumulative rainfall, structural anomaly parameters, and reliability parameters corresponding to integrity markers as prediction factors. The risk prediction stage then constructs a risk assessment model based on these prediction factors and adaptively adjusts the weights of each prediction factor by incorporating historical feedback records, ensuring that the risk assessment results dynamically adjust with hydrological evolution and data reliability. The risk prediction stage outputs an assessment message containing the risk assessment results and their corresponding risk levels, which is then transmitted to the early warning and control stage. After receiving the risk assessment results, the early warning control mechanism triggers an early warning based on the correspondence between the assessment results and the threshold system. This threshold system is associated with at least the risk type, engineering level, and flood season correction parameters to ensure consistent and traceable early warning triggering conditions for different objects and seasons. The early warning control mechanism further determines the early warning release targets and scope based on a three-dimensional permission matrix composed of watershed level, administrative division, and engineering level. The permission matrix stores the set of release targets and their receiving priorities using a three-element index of watershed level identifier, administrative division code, and engineering level identifier. When an early warning involves cross-administrative divisions or cross-watershed levels, the early warning control mechanism executes horizontal isolation or vertical transmission rules under the permission boundary conditions: for cross-district scenarios within the same watershed, horizontal isolation rules are applied to units in adjacent districts with engineering levels no lower than the source object; for cross-level scenarios within the same district, vertical transmission rules are applied to synchronize with the superior watershed management unit; for scenarios involving both cross-district and cross-level scenarios, a two-level release list is generated in a vertical-then-horizontal order to determine the final receiving unit and its release path.

[0022] Upon receiving a high-level early warning from the early warning and control stage, the linkage mechanism generates a dispatch instruction. Before being issued, the dispatch instruction undergoes manual review and confirmation. Once approved, it is sent to the execution terminal via a pre-defined interface. During dispatch execution, the linkage mechanism records the instruction generation time, review results, execution steps, and any manual interventions to form an audit log. This audit log is transmitted back to the digital modeling stage via a data interface and written into the twin model's operation log area. It serves as the basis for weight correction and parameter calibration during subsequent risk prediction, ensuring a consistent closed-loop system from prediction to early warning to dispatch.

[0023] The monitoring process includes various types of sensors and terminals: water level sensors deployed at river cross-sections for continuous water level recording; rainfall sensors installed at watershed rain gauges for hourly rainfall collection; and flow velocity meters deployed at monitoring cross-sections to acquire flow velocity, which is then converted into flow rate by a flow calculation unit based on cross-sectional geometric parameters. The flow rate can be calculated using the following formula: ; In the formula: : Indicates the cross-sectional flow rate, with the unit being cubic meters per second (m's); : Represents the average flow velocity of the cross section, with the unit being meters per second (m / s). Its value can be obtained by weighted averaging of the instantaneous flow velocities collected by the flow velocity measuring instrument at different measuring points in the cross section. : Represents the cross-sectional area through which water flows, in square meters (m²), and its value can be calculated from the water depth obtained by the water level sensor combined with the cross-sectional geometric parameters; Meteorological monitoring units are set up near reservoirs and dams to measure temperature, humidity, air pressure, and wind speed. In addition, cameras and remote sensing terminals are deployed at key hydraulic structures and important water areas to acquire video images and remote sensing images to supplement environmental information. Unlike the acquisition of single hydrological elements in existing technologies, this invention acquires data simultaneously from multiple dimensions such as hydrology, meteorology, and images, providing complete input for subsequent fusion and modeling.

[0024] The communication processing stage includes: an interface unit for transmitting collected data via a wired network; a communication unit for wireless transmission; and a unit for transmitting data via a satellite link when the first two types of networks are unavailable. During data access, the system generates an integrity tag for each data item. The integrity tag can be calculated based on a message checksum, and the calculation formula is as follows: ; In the formula, For integrity marking, For the first in the message One data unit, The total number of message data units. Preset modulus; The integrity markers generated in the above manner can reflect the consistency and integrity of the message during transmission. The integrity markers can also be generated by timestamp comparison or link confirmation signals. This invention does not limit this. The integrity markers maintain a corresponding relationship during data transmission, verification and prediction, and are used to trace data reliability.

[0025] The data processing stage includes a field mapping unit, a consistency verification unit, and an anomaly marking unit. The field mapping unit standardizes data from different sources based on a unified semantic model. The consistency verification unit compares multi-source data for the same parameter. When conflicts exist, the resolution is made in the following order: latest timestamp, higher integrity flag value, and weighted average of nearby monitoring points. The weighted average can be calculated using the following formula: ; In the formula, These are the corrected parameter values. For the first Measurements from neighboring monitoring points, The weights assigned to this monitoring point, The number of neighboring points involved in the calculation; An anomaly marker unit generates an anomaly marker when the same parameter exceeds a threshold in three consecutive detection windows. The threshold can be set based on historical hydrological statistics or industry standards. The detection window length can be 5 minutes, 10 minutes, or set according to operational needs. Anomaly determination can be described by the following formula: ; in, Indicates an anomaly marker. Indicates the consecutive number The parameter values ​​within each detection window This refers to the expected value or historical average of this parameter. A preset threshold is set. The anomaly markers are stored in association with formatted data for use by the digital twin model.

[0026] The digital modeling process includes a data interface, a modeling unit, a state update unit, and a tagging processing unit. The data interface is used to receive formatted hydrological data, meteorological data and corresponding anomaly markers from the data processing stage, and to build a ternary index according to the monitoring point number, parameter type and timestamp; The modeling unit establishes a three-dimensional twin model based on the river cross-section geometry, reservoir capacity curve, and dam structural design parameters. The river model uses a spatial grid divided according to the cross-section station number, the reservoir model uses a voxel grid with the reservoir area contour lines as the boundary, and the dam model uses a finite element structure grid layered according to dam sections and elevation. Each grid node is bound to a corresponding monitoring point number to achieve a one-to-one mapping between monitoring points and twin nodes. The state update unit uses a twin state vector This indicates that Li Shengti is at all times The set of state parameters, wherein the state vector includes at least water level, flow rate, accumulated rainfall, structural displacement or seepage pressure, and meteorological elements; when the time is received... Actual measured data When merging and updating, use the following formula: ; in, This is the twin state vector from the previous time step. This is the measured data vector after field mapping. To update the weight coefficients, the value range is 0 < ≤1, its size is determined by the reliability weight corresponding to the integrity tag; The marking processing unit marks the exception. Write twin state factors And form a risk state vector with anomaly inheritance. Its update method is as follows: ; ; in, The risk status factor for the previous cycle, This is the abnormal inheritance coefficient, and its value range is 0 < ≤1; This is a set of twin state factors for use in the risk prediction process; through the above-mentioned abnormal inheritance writing method, the monitored abnormalities are accumulated and expressed across cycles in the Li Shengti, providing traceable risk input for subsequent predictions; The risk prediction process includes an interface unit, a factor management unit, a weight adjustment unit, and a result output unit. The interface unit receives the twin state factors output from the digital modeling stage. and according to the predicted cycle Forming a sliding sample sequence ; The factor management unit extracts a set of predictive factors from the twin state factors. The predictive factors include at least: flow rate of change. Rainfall accumulation value Structural abnormal state parameters and the reliability parameters corresponding to the integrity tag. The rate of change of flow rate is calculated using the following formula: ; In the formula, For a moment The cross-sectional flow rate, This refers to the cross-sectional flow rate of the previous cycle; The weight adjustment unit constructs the prediction error based on historical feedback records. The weight vector is updated using an error-reverse assignment method. The update rules are as follows: ; in, This is the learning rate coefficient, with a value range of 0 < ≤1, For the first One predictive factor; each weight is normalized and used for risk calculation in this period. The output unit calculates the risk assessment result according to the weighted risk model. ; ; The risk assessment results, along with the corresponding risk level labels, will be output to the early warning and control system.

[0027] The early warning and control unit consists of an input unit, a threshold determination unit, an access matching unit, and a boundary processing unit. Each unit works collaboratively in the order of risk input, threshold triggering, object matching, and boundary release. The input unit is connected to the risk prediction stage via a data bus and is used to receive the risk assessment results output by the risk prediction stage. and their corresponding risk source identifiers The risk source identifier includes at least the watershed unit number corresponding to the current risk. Administrative division codes and engineering grade markings This is to facilitate the determination of subsequent permission indexes and publishing scope; The threshold determination unit is used to classify and determine the risk assessment results; preset thresholds are used. Instead of using a fixed single value, the threshold is determined according to a combination rule of risk type, engineering level, and seasonal flood limit; firstly, benchmark thresholds are set for different risk types (such as flood risk, leakage risk, and displacement risk). ; Secondly, apply a grade coefficient to the benchmark threshold according to the engineering grade. The higher the engineering level, the stricter the threshold. Third, during the flood season, a seasonal correction factor is used. Further adjust the threshold; Therefore, the threshold can be expressed as: ; in, Derived from historical statistical quantiles or industry management standards, and Pre-configured by the water administration department or engineering operation and maintenance procedures; the threshold determination logic is: when ≥ Timely generation of early warning signals and provide an early warning level. The warning level was changed from The correspondence between the threshold intervals and the multi-level threshold intervals is determined; The permission matching unit is based on a three-dimensional permission matrix. Select the target for issuing the alert; the three-dimensional permission matrix uses a ternary index (...). , , The matrix is ​​composed of elements corresponding to watershed levels, administrative divisions, and engineering levels, respectively. It can be stored using a three-dimensional array or a relational table, with each entry representing an access record, containing at least: a level identifier. Administrative Code Engineering level List of Published Objects and the receiving priority of the object. The matrix retrieval relationship is as follows: ; in, To match the set of early warning release objects with the ternary index of risk sources, and to avoid index non-uniqueness caused by cross-source projects, the permission matching unit uses... For the first index, For the second index, Perform a step-by-step search for the third index; when no exact match is found in the third index, automatically fall back to a higher-level engineering entry under the same watershed and administrative division to ensure that the target of the publication is identified; The boundary processing unit is used to resolve release conflicts across regions or levels. It first determines the source of the risk: the ternary index and the set. The relationship between the indexes of various objects in the middle; If there exists an object that satisfies ≠ and = In cases where the situation is cross-level but within the same zoning, the vertical transmission rule applies: the early warning is simultaneously sent to the next higher-level river basin management unit and its corresponding command node within the same zoning. If there exists an object that satisfies ≠ and = This is considered a case spanning multiple administrative divisions but within the same river basin, and the horizontal isolation rule applies: only adjacent administrative divisions with engineering standards no lower than [specific level] are included. The target sends an alert, and marks the main horizontal boundary source field in the published message: When they occur simultaneously ≠ and ≠ In such cases, a two-level release list is generated in a sequence of first vertical and then horizontal: first, it is transmitted vertically to the next higher-level watershed unit, and then the higher-level unit distributes it to cross-regional objects according to the horizontal isolation rules, thereby avoiding the overreach or omission of early warnings.

[0028] The aforementioned boundary rules are jointly completed by the object index output by the permission matching unit and the condition judgment of the boundary processing unit, so that the scope of early warning release is consistent with the three-dimensional constraints of watershed-administration-engineering in the water conservancy management system; Early warning messages should include at least: risk assessment results. Threshold Warning Level Source identification ( , , List of published objects and boundary identifier fields, warning signals output by the warning control link. The list of objects to which the data is published will be transmitted to the coordination mechanism for subsequent scheduling decisions.

[0029] The linkage mechanism includes an instruction generation unit, a review and issuance unit, a log recording unit, and a feedback unit. The instruction generation unit generates a scheduling instruction upon receiving a high-level warning signal; the review and issuance unit manually reviews the scheduling instruction and issues it to the execution terminal after confirmation; the log recording unit records the execution time, operation steps, and manual intervention information during the execution process to form an audit record; the feedback unit writes the audit record to the state update unit of the twin model through the database interface, and uses it as an additional input variable in the next prediction cycle to correct the prediction.

[0030] Unlike existing automated or one-way confirmation modes of scheduling, this invention achieves closed-loop linkage between scheduling execution and prediction correction through manual review and log backfeeding.

[0031] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0032] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A multi-source hydrological data monitoring and safety warning system for smart water conservancy, characterized in that: The application relates to a river, reservoir and dam monitoring system, which comprises the following parts: a monitoring device for collecting water level, rainfall, flow velocity, flow rate and meteorological elements in river, reservoir and dam scenes, and obtaining environmental information by combining video monitoring and remote sensing images; a communication processing module for receiving data output by the monitoring device, the communication processing module being capable of switching between wired communication, wireless communication and satellite communication, and generating an integrity mark for each piece of access data, the integrity mark being in one-to-one correspondence with the corresponding data and being transmitted together with the data; a data processing device for field mapping and consistency checking of access data based on a unified semantic model, when there are multiple sources of conflict for the same parameter, the data processing device makes a decision in the order of time stamp priority, integrity mark weight and adjacent monitoring point weighted correction, and generates an abnormal mark when a parameter deviation trend is found in a continuous detection window; a digital modeling device for constructing a digital twin model of a river, reservoir and dam based on the formatted data and abnormal marks output by the data processing device, and updating the structure state and operation state in the twin body in real time, the abnormal mark being written into the twin state factor; a risk prediction device for taking the twin state factor as input, taking flow change trend, rainfall cumulative value, structure abnormal state and integrity mark as prediction factors, and adaptively adjusting the weight of the prediction factors according to historical feedback records to generate a risk assessment result; a warning control module for triggering a warning when the risk assessment result exceeds a set threshold, generating a corresponding warning level, determining the release range of the warning through a three-dimensional permission matrix composed of a river basin level, an administrative division and an engineering level, and executing horizontal isolation and vertical transmission rules under the permission boundary condition; a linkage device for generating a scheduling instruction when a high-level warning of the warning control module is received, the scheduling instruction needing to be confirmed by manual review before being issued, and storing the execution log and manual intervention into an audit record, the audit record being fed back to the digital modeling device for correcting subsequent risk prediction parameters.

2. The multi-source hydrological data monitoring and safety warning system for smart water conservancy according to claim 1, characterized in that: The monitoring device comprises: a water level sensor arranged at a river section for obtaining real-time water level data; a rainfall sensor arranged at a river basin rainfall station for obtaining hourly rainfall data; a flow velocity measuring instrument and a flow calculation unit arranged at a monitoring section for measuring flow velocity and calculating flow rate respectively; a meteorological monitoring unit arranged near a reservoir and a dam for obtaining temperature, humidity, air pressure and wind speed data; and a camera and a remote sensing terminal arranged at key hydraulic structures and important water areas for obtaining video images and remote sensing images.

3. The multi-source hydrological data monitoring and safety warning system for smart water conservancy according to claim 2, characterized in that: The communication processing module comprises a wired communication interface unit, a wireless communication unit, a satellite communication unit and an integrity mark generation unit for generating corresponding integrity marks when data is accessed, the integrity marks being stored in association with the collected data and maintaining the corresponding relationship in the data transmission and subsequent processing process.

4. The multi-source hydrological data monitoring and safety warning system for smart water conservancy according to claim 3, characterized in that: The data processing device comprises: a field mapping unit for converting data fields of different sources to a unified semantic space; a consistency checking unit for checking multi-source data of the same parameter and sequentially performing conflict resolution according to the order of the latest timestamp, higher integrity label value, and weighted average of adjacent monitoring points; an abnormal label generation unit for generating an abnormal label when the difference between the same parameter value and the corresponding historical average value exceeds the preset threshold value in three consecutive detection windows, and storing the abnormal label in association with the resolved data.

5. The multi-source hydrological data monitoring and safety warning system for smart water conservancy according to claim 4, characterized in that: The digital modeling device comprises: a data interface unit; a modeling unit; a state updating unit; a label processing unit; wherein the state updating unit is used to dynamically update the water level, flow and meteorological state parameters of the twin model based on the formatted data; the label processing unit is used to write the abnormal label into the state factor storage area of the twin model, and the state factor is used as an input variable in the risk prediction device for calculation.

6. The multi-source hydrological data monitoring and safety warning system for smart water conservancy according to claim 5, characterized in that: The risk prediction device comprises: a twin state factor input interface unit; a factor management unit; a weight adjustment unit; a result output unit; the factor management unit is used to set the flow change value, rainfall accumulation value, structural abnormal state parameter and integrity label as prediction factors; the weight adjustment unit is used to modify the weight of each prediction factor according to the historical feedback record in each prediction period, and the risk assessment result is output by the result output unit.

7. The multi-source hydrological data monitoring and safety warning system for smart water conservancy according to claim 6, characterized in that: The early warning control module comprises: an input unit; a threshold determination unit; a permission matching unit; a boundary processing unit; the threshold determination unit is used to generate a warning signal and determine the warning level when the risk assessment result exceeds the preset threshold value; the permission matching unit is used to determine the warning release object based on the three-dimensional permission matrix constructed based on the basin level, administrative division and engineering level; the boundary processing unit is used to perform horizontal isolation or vertical transmission rules according to the matrix retrieval result in the case of cross-region or cross-level.

8. The multi-source hydrological data monitoring and safety warning system for smart water conservancy according to claim 7, characterized in that: The linkage device comprises: an instruction generation unit; a review and issuance unit; a log recording unit; an audit feedback unit; the instruction generation unit is used to generate a dispatching instruction when the warning level is high; the review and issuance unit is used to manually review and confirm the dispatching instruction and issue it after confirmation; the log recording unit is used to record the execution time, execution steps and manual intervention information and form an audit record; the audit feedback unit is used to feed back the audit record to the digital modeling device, so that it is used as a correction parameter in subsequent prediction calculation.

Images