Water conservancy flood control emergency regulation and control method and system based on digital twinning

Abstract

The invention discloses a digital twinning-based water conservancy flood control emergency regulation and control method and system, and relates to the technical field of digital twinning water conservancy flood control emergency regulation and control, and the method comprises the steps: building a uniquely numbered object file for a reservoir, a gate, a pump station, a river reach, a control section and a waterlogging-prone point, and solidifying a boundary constraint, a threshold value, a constraint statement and a collection binding relation; carrying out unit verification, quality marking and missing measurement completion, and merging into an observation frame and a forecast frame with a unified time step length; and outputting influence association, calculating and storing weight parameters of each associated edge in the sliding window, and calculating predicted water levels of the control section and the river reach in combination with the edge weight and the equivalent flow increment. According to the method, the emergency deduction fitting degree is improved based on sliding updating of the influence association edge weight and rapid prediction of the water level; the optimal regulation and control combination is output in a water level rigid rejection and evaluation preferential mode, a closed loop is formed through change rate, interval and linkage conflict check issuing and receipt writing back, and regulation and control safety, performability and timeliness are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of digital twin water conservancy flood control emergency control technology, specifically a water conservancy flood control emergency control method and system based on digital twins. Background Technology

[0002] With the development of IoT sensing, spatiotemporal data management, and hydrological and hydrodynamic models, flood control in water conservancy is gradually shifting from experience-based scheduling to a technology route that combines data-driven and model-assisted approaches. Online monitoring networks for rainfall, water levels, flow rates, and sluice gate and pump operations have been widely established in watershed and urban flood control systems at all levels, forming a data acquisition chain based on SCADA / telemetry, and overlaying digital management of rainfall forecasts, inflow forecasts, and scheduling procedures. Building on this foundation, the concept of digital twins has been introduced into the fields of water conservancy engineering and watershed flood control. Through the digital mapping of objects such as reservoirs, rivers, and sluice gates and pumps, the visualization and predictive simulation of flood conditions are achieved, thereby supporting collaborative decision-making and command issuance for emergency control.

[0003] Current flood control scheduling and digital twin applications still have significant shortcomings: First, although multi-source data has been integrated, the unified numbering of objects, boundary constraints, thresholds, and computational expression of procedural clauses are incomplete. This makes it difficult to form a verifiable closed-loop constraint system between data, procedures, and controlled objects, and control schemes often rely on manual interpretation and experience-based discretion. Second, existing forecasting and simulations mostly rely on offline hydrodynamic models or fixed-parameter empirical models, making it difficult to dynamically update the impact intensity in emergency scenarios based on data quality fluctuations, changes in operating conditions, and upstream control relationships. This results in unstable rapid predictions of water levels at key control sections and insufficient traceability. Third, the scheme generation and execution levels generally lack calculable verification mechanisms for action change rates, minimum action intervals, pump station power, and upstream-downstream linkage conflicts, easily leading to problems of being calculable but uncontrollable or controllable but unexecutable. Fourth, after the issuance of control commands, the feedback data is often not structured and written back to participate in the next state update, making it difficult to reflect actual execution deviations in subsequent decisions in a timely manner, and hindering the realization of rolling closed-loop control for emergency situations. Summary of the Invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the technical problem solved by this invention is that existing water conservancy flood control scheduling and digital twin-assisted decision-making methods have the following problems: the constraints of objects and procedures are difficult to be calculated and solidified; the parameters of influence correlation and prediction are difficult to be dynamically updated with the working conditions and have insufficient traceability; the control scheme lacks executable verification of action change rate, interval and linkage conflict and the feedback is difficult to write back in a closed loop; and how to achieve rolling prediction, constraint screening and closed-loop issuance of instructions under the constraints of multi-source data quality fluctuation and emergency timeliness.

[0006] To address the aforementioned technical problems, this invention provides the following technical solution: a water conservancy flood control emergency control method based on digital twins, comprising establishing uniquely numbered object files for reservoirs, gates, pumping stations, river sections, control sections, and flood-prone points, and solidifying boundary constraints, thresholds, constraint statements, and data acquisition binding relationships; accessing rainfall, water level, flow rate, and gate / pump operating conditions and forecast data according to the binding relationships; performing unit verification, quality marking, missing data supplementation, and merging into observation frames and forecast frames with a unified time step; and outputting data based on the upstream and downstream relationships and control relationships of the object files, combined with the observation frame sequence. The system influences the correlation and calculates and saves the weight parameters of each correlation edge within a sliding window. On the current observation frame, candidate control values ​​are generated according to the object file and effective constraints. The predicted water level of the control section and river segment is calculated by combining the edge weights and the equivalent flow increment. The candidate values ​​are merged into control combinations, and hard elimination is performed using the guaranteed water level. The evaluation value is then calculated using the predicted water level and the change in action, and the combination with the smallest value is selected. The selected combination is converted into an instruction data packet, and the change rate, interval, and linkage conflict are checked before being issued. The receipt is collected, and the actual execution result is written into the status record of the next moment.

[0007] As a preferred embodiment of the water conservancy flood control emergency control method based on digital twins described in this invention, the following steps are included: establishing uniquely numbered object files for reservoirs, gates, pumping stations, river sections, control sections, and flood-prone points, and solidifying boundary constraints, thresholds, constraint statements, and data acquisition binding relationships. This involves generating a unique number for each object and writing its name, governing unit, status (in service, under maintenance, out of service, faulty), and update time, while simultaneously registering its spatial positioning information. The spatial positioning information includes latitude and longitude, elevation, river name, and river mileage or section number. For reservoir objects, the water level and storage capacity relationship data and flood discharge facility parameters are registered. For gate objects, the number of gates, maximum opening degree, upper limit of opening and closing speed, and minimum action interval are registered. For pumping station objects, the number of units, rated flow rate per unit, minimum start-stop interval, and maximum allowable power are registered. The upper and lower limits and allowable change rates of each adjustable quantity are clearly defined and solidified.

[0008] As a preferred embodiment of the water conservancy flood control emergency control method based on digital twins described in this invention, the following steps are included: The process of accessing rainfall, water level, flow rate, and gate / pump operating conditions and forecast data according to binding relationships, performing unit verification, quality marking, missing data supplementation, and merging them into observation frames and forecast frames with a unified time step, includes solidifying the data acquisition binding relationships by associating the locations of rainfall stations, water level stations, flow rate stations, and gate / pump control systems with unique object numbers, and retaining both the sampling time and reception time for each acquisition record; during the data access process, consistency verification is performed according to the units solidified in the object file, and unit conversion is completed before writing; and when there are no direct measurement points, the gate flow rate and pump station outflow rate are converted using the actual gate opening and the water level above and below the gate, or the number of pump stations in operation and the rated discharge parameters, respectively. The conversion results are stored separately from the original acquired values ​​and do not overwrite the original data.

[0009] As a preferred embodiment of the water conservancy flood control emergency control method based on digital twins described in this invention, the quality markers include missing data, jump data, out-of-range data, and delay data markers. Missing data markers are determined according to the sampling period. Jump data markers are determined using a sliding window differential threshold and continuity test. Out-of-range data markers are determined using the legal value range fixed in the object file. When generating supplementary values ​​for missing data, if the missing data length does not exceed the preset duration, it is supplemented by time interpolation at the same station. If the missing data length exceeds the preset duration, it is supplemented by weighted estimation of neighboring stations or related objects upstream and downstream. The supplementary values ​​are written into an independent field and the original missing data markers are retained. Data are merged to generate observation frames according to a unified time step, and aggregation rules are fixed according to the index type. Rainfall is accumulated within a time window, water level is taken as the last value or average, flow rate is taken as the average, and opening degree and number of stations are taken as the last value.

[0010] As a preferred embodiment of the water conservancy flood control emergency control method based on digital twins described in this invention, the output of influence correlation and the calculation and storage of the weight parameters of each correlation edge within a sliding window include establishing the influence correlation as a directed relationship based on the upstream and downstream correlation and control relationship in the object file, and extracting the source object equivalent flow sequence and the target object water level sequence for each directed relationship within the most recent continuous historical window; the equivalent flow sequence is preferentially obtained by directly collecting the flow, and if it does not exist, it is obtained by converting the gate opening, the excess flow, or the number of pump stations and the discharge capacity; only the time when the source object equivalent flow and the target object water level are simultaneously valid are selected for calculation within the window, and the calculated edge weights are versioned and stored together with the window start and end time, the number of valid samples, and the object file version number. When there are insufficient valid samples or calculation abnormalities, the previous valid weight is used and a reuse mark is recorded.

[0011] As a preferred embodiment of the water conservancy flood control emergency control method based on digital twins described in this invention, the step of generating candidate control values ​​on the current observation frame according to object files and effective constraints, and calculating the predicted water level of the control section and river segment in combination with edge weights and equivalent flow increments includes the following steps when generating candidate control values ​​on the current observation frame: for gate objects, the target opening degree is used as the candidate value, while simultaneously satisfying the upper and lower limits of the opening degree, the maximum change in opening degree per unit time, and the minimum action interval constraints; for pump station objects, the target number of pumps in operation is used as the candidate value, while simultaneously satisfying the constraints that the number of pumps is an integer, the minimum start-stop interval, and the maximum allowable power constraints; for reservoir objects, the target discharge flow is used as the candidate value, while simultaneously satisfying the upper and lower limits of discharge and the maximum change in discharge per unit time constraints; the candidate values ​​are truncated or eliminated according to the effective constraint statements; and after completing the equivalent flow conversion, the predicted water level of the control section and river segment is calculated according to the edge weights and equivalent flow increments of the source objects stored in the influence association; the prediction results are associated with and stored in conjunction with the number of source objects participating in the summation and the version number used.

[0012] As a preferred embodiment of the water conservancy flood control emergency control method based on digital twins described in this invention, the following steps are included: merging candidate values ​​into control combinations, performing hard elimination using guaranteed water levels, and then calculating evaluation values ​​and selecting the smallest combination using predicted water levels and changes in action. This includes merging candidate values ​​into control combinations, first performing hard elimination of control combinations based on the guaranteed water levels of the control sections, calculating evaluation values ​​for the eliminated control combinations and sorting them from smallest to largest. If evaluation values ​​are tied, they are sorted from smallest to largest by the sum of the absolute values ​​of changes in action and the optimal combination is determined. When converting the optimal combination into an instruction data packet, an instruction item is generated for each object, containing a unique object number, instruction type, target value, allowable rate of change, instruction start time, and effective end time. Before issuance, the same object's repeated instructions, rate of change, interval, power, and linkage limit conflict checks are performed. After issuance, acknowledgment data containing reception status, execution status, actual execution value, and failure reason is received, and the acknowledgment data is written into the status record of the next moment according to the unique object number.

[0013] Another objective of this invention is to provide a water conservancy flood control emergency control system based on digital twins. This system can output influence correlations by combining upstream and downstream control relationships based on object archives with observation frame sequences, and calculate and save the weight parameters of each correlation edge within a sliding window. On the current observation frame, candidate control values ​​are generated according to the object archives and effective constraints. The predicted water levels of control sections and river segments are calculated by combining edge weights and equivalent flow increments. This solves the problem that current water conservancy flood control scheduling and digital twin-assisted decision-making methods have difficulty in dynamically updating influence correlations and prediction parameters according to operating conditions, and lack traceability.

[0014] As a preferred embodiment of the digital twin-based emergency control system for water conservancy and flood control described in this invention, the system includes: an object and rule solidification module, a data merging and association modeling module, and a control solution and closed-loop execution module. The object and rule solidification module is used to establish uniquely numbered object files for reservoirs, gates, pumping stations, river sections, control sections, and flood-prone points, and to solidify boundary constraints, thresholds, constraint statements, and the binding relationships of data collection points. The data merging and association modeling module is used to access rainfall, water level, flow rate, and gate / pump operating conditions and forecast data, complete unit verification, quality marking, missing data completion, and generate observation frames and forecast frames. Simultaneously, it forms influence associations based on upstream and downstream relationships and control relationships and calculates and saves edge weight parameters. The control solution and closed-loop execution module is used to generate candidate control values ​​on the current observation frame, calculate the predicted water level of the control section and river section, evaluate and select the optimal control combination, generate instruction data packets for issuance and retrieve receipts, and write the actual execution results into the status record for the next time step.

[0015] A computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement a water conservancy flood control emergency control method based on digital twins.

[0016] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of a digital twin-based emergency control method for flood control and water conservancy.

[0017] The beneficial effects of this invention are as follows: The water conservancy flood control emergency control method based on digital twins provided by this invention establishes uniquely numbered object files for reservoirs, gates, pumping stations, river sections, control sections, and flood-prone points, and solidifies boundary constraints, thresholds, constraint statements, and data acquisition binding relationships. This achieves a deterministic mapping between engineering objects, regulatory clauses, and controllable quantities, eliminating data ambiguity and constraint deficiencies caused by cross-system access and objects with the same name, thus providing unified standards and traceability for control calculations. Furthermore, by accessing rainfall, water level, flow rate, and gate / pump operating conditions and forecast data, and performing unit verification, quality marking, missing data completion, and unified step size merging, the method achieves temporal consistency and explicit quality of multi-source asynchronous data. This enables the stable formation of observation frames and forecast frames under conditions of delay, missing data, and jumps, thereby reducing the amplified impact of dirty data on inference and decision-making. Finally, by outputting influence correlations based on upstream and downstream and control relationships and sliding... The system calculates and saves edge weights within the window, enabling adaptive updates of influence intensity based on the latest observations. This addresses response drift caused by changes in operating conditions, thereby improving the alignment between the control section and the predicted water level of the river segment, as well as version traceability. Based on this, candidate control values ​​are generated according to object files and effective constraints, and rapid prediction is performed using edge weights and equivalent flow increments. This achieves a constraint-first, deductive, feasible domain contraction, used to preemptively eliminate ineffective actions, thus improving the efficiency and executability of emergency rolling solutions. Finally, candidate values ​​are merged into control combinations, with hard elimination based on guaranteed water level and selection based on predicted water level and action changes. Combined with changes in rate of change, intervals, and linkage conflict checks, a command data package is generated and issued, with the receipt written back to the status record. This achieves a closed-loop link from calculation to execution to feedback, explicitly reflecting execution deviations and driving the next round of updates, thereby improving the safety, stability, and continuous availability of emergency control. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 The first embodiment of the present invention provides an overall flowchart of a water conservancy flood control emergency control method based on digital twins. Detailed Implementation

[0020] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0021] Example 1, referring to Figure 1 As an embodiment of the present invention, a water conservancy flood control emergency control method based on digital twins is provided, comprising: S1: Establish uniquely numbered object files for reservoirs, gates, pumping stations, river sections, control sections, and flood-prone points, and solidify boundary constraints, thresholds, constraint statements, and data acquisition binding relationships. Access rainfall, water level, flow rate, gate and pump operation status, and forecast data according to the binding relationships, perform unit verification, quality marking, fill in missing measurements, and merge them into observation frames and forecast frames with a unified time step.

[0022] Furthermore, within the scope of flood control and regulation in a river basin or region, each entity involved in the regulation is first verified and registered. The registered entities are limited to six categories: reservoirs, sluice gates (including gate stations), pumping stations, river sections, control sections, and flood-prone areas. A unique number is generated for each entity, consisting of a management area identifier, an entity category identifier, and a sequence number, and is permanently written into the entity file. The entity name, affiliated unit, operational status (in service, under maintenance, out of service, faulty), and update time are also written into the entity file. After the entity file is established, all collected data and regulation commands are indexed using this unique number, without using the entity name for location, to avoid data concatenation errors caused by duplicate or renamed entities.

[0023] For each object, supplement its spatial positioning information. Spatial positioning information should include at least: latitude and longitude coordinates, elevation, river name, river mileage or station number, and sub-basin or zone number. For control sections and river segments, supplement the definition of the section location and the start and end points of the river segment. The section location is recorded as the river mileage paired with the section number, and the river segment location is recorded as the starting section number paired with the ending section number. For reservoir objects, supplement the registration of reservoir capacity curves or water level-capacity relationship data (stored in discrete point tables or piecewise linear form), and register the design parameters of flood discharge facilities such as spillways and flood discharge tunnels; for gate objects, register the number of gate openings, gate type, maximum opening degree, upper limit of opening and closing speed, and minimum operating interval; for pumping station objects, register the number of units, rated flow rate per unit, minimum start-stop interval, maximum allowable number of consecutive start-stop cycles, and power supply constraints (such as maximum allowable power or maximum load).

[0024] In the object file, the required operational quantities for each type of object are fixed. Only quantities that can be collected or directly calculated are selected. The operational quantities for reservoir objects include at least reservoir water level, inflow, and outflow; the operational quantities for river sections and control sections include at least water level and flow; the operational quantities for gate objects include at least the actual gate opening and the water levels above and below the gate; the operational quantities for pumping station objects include at least the number of pumps in operation, outflow, and operating power; the operational quantities for flood-prone points include at least the water level or water depth (if there is no water level gauge on site, the equivalent water level calculated by balancing the water level and drainage flow of adjacent drainage outlets is used instead). The unit and value range of each operational quantity are specified, for example, water level is measured in meters, flow is measured in cubic meters per second, opening is measured as a percentage, and the number of pumps is measured as an integer. The valid range of the operational quantity is written into the object file for verification.

[0025] It should be noted that a fixed adjustable quantity is set for each object, and the boundaries and action constraints of the adjustable quantity are written into it. For gates, the adjustable quantity is the target opening degree, with the lower limit, upper limit, maximum allowable change in opening degree per unit time, and minimum time interval between two actions written into the variable quantity. For pump stations, the adjustable quantity is the target number of pumps started or start / stop commands, with the lower limit, upper limit, minimum start / stop interval, and maximum allowable power written into the variable quantity. For reservoirs, the adjustable quantity is the target discharge flow rate or discharge process line, with the lower limit, upper limit, and maximum allowable change in discharge flow rate per unit time written into the variable quantity. No adjustable quantities are set for river sections, control sections, and flood-prone points; they are only used as constraint monitoring objects. All the above boundaries and constraints are stored using explicit numerical values, and the object file specifies that if the adjustable quantity exceeds the boundary or changes too rapidly, it should not be written into the variable quantity.

[0026] Key thresholds in dispatching procedures and emergency plans are solidified at the object-specific granularity. For control sections and river segments, warning water levels and guaranteed water levels (and, if necessary, restriction water levels) are written; for reservoirs, flood control restriction water levels and flood limit water levels are written; for flood-prone areas, water accumulation alarm thresholds and emergency response thresholds are written. When writing thresholds, the applicable conditions must be clearly defined, such as flood season / non-flood season, flood diversion area activation / deactivation, and applicable labels under a specific warning level. Multiple sets of threshold records are allowed for the same object, but each set must specify the applicable conditions and valid date range to avoid threshold conflicts. After the threshold records are written, numerical relationship verification is performed; for example, the warning water level must be less than the guaranteed water level, and the guaranteed water level must be less than or equal to the restriction water level. Threshold records that fail verification are not allowed to be saved.

[0027] The regulations are translated into directly calculable and verifiable constraint statements, which are then bound to specific objects. Constraint statements are written using a condition-based restriction structure: the condition part consists of collectable state quantities and applicable labels, such as "when the water level at a certain control section reaches the warning level," "when the reservoir water level is higher than the flood control limit level," or "when the warning level is Level II." The restriction part only allows upper and lower limits, rate of change limits, and action interval limits for controllable quantities, such as "the target gate opening must not exceed a certain upper limit," "the target discharge flow must not exceed a certain upper limit," or "the interval between two pump starts must not be less than a certain minute." Thresholds appearing in the constraint statements all refer to explicit values ​​from the aforementioned threshold records, not verbal descriptions. Each constraint statement is written with its source clause number and version number recorded, and the effective date range is retained; when the regulations are revised, a new version number is added, but historical versions are retained to ensure that the constraint set used at that time can be traced back to the same point in time.

[0028] Establish a visible one-to-one binding relationship between collected data and object operation quantities, and complete verification at the field point level. For water level stations, rain gauge stations, and flow stations, record the correspondence between their station number, installation location, and object number; for gates and pump stations, record the correspondence between the control system point name or measuring point name and the object number. After the binding relationship is written, perform a field test verification: read the station data at a certain moment and check whether it can be written into the corresponding object's operation quantity field; for abnormal situations (such as inconsistent units, incorrect station binding, unreasonable data fluctuation range), directly correct the binding relationship or supplement the unit conversion. Use clear proportional coefficients and offsets to record unit conversions and dimension conversions, such as the conversion between millimeters and meters, and the conversion between hourly rainfall and minute rainfall intensity; when it comes to the conversion from gate opening to flow rate, and the conversion from the number of pump stations to the discharge capacity, use engineering design curves or historical calibration curves for discrete point storage or piecewise linear storage, and record the object number and operating conditions to which the curve applies.

[0029] After the object files, threshold records, constraint statements, and data acquisition bindings are all established, a consistency check is performed and the baseline version is frozen. The consistency check includes at least: whether each object has spatial location information; whether all runtime variables are bound to a data source; whether controllable variables have boundary and action constraints; whether key thresholds are complete and satisfy numerical relationships; and whether constraint statements reference existing threshold values ​​and can be resolved to specific objects. After the check passes, the current object file and rule set are marked as the same baseline version, and the version number and effective time are saved. Subsequent changes to engineering parameters, thresholds, or clauses are written as new versions, and historical version records and their validity periods are retained to ensure that a unique set of object parameters and constraint data can be located at any given time. Furthermore, after the object files, spatial positioning information, operational and controllable quantities, boundary and action constraints, threshold records, constraint statements, and data binding relationships have all been established and the baseline version frozen, the next stage will be the access and processing of multi-source monitoring and forecast data. The accessed data sources are limited to hydrological monitoring data sources such as rain gauges, water level stations, and flow stations, as well as operational data sources from gate and pumping station control systems. Simultaneously, radar quantitative rainfall grid data and time-segmented rainfall forecast products provided by meteorological departments or existing interfaces will also be accessed. All of the above data sources will be uniquely identified by station number, location name, or product identifier, and will maintain a one-to-one correspondence with the object number; object names will not be used for location.

[0030] For each type of data source, the sampling period and arrival method are first determined, and the sampling period is written into the access configuration. Rain gauge data is sampled every minute or 5 minutes, water level and flow data are sampled every 1 to 15 minutes, and operational data such as actual gate opening, water level above and below the gate, number of pumps in operation, outflow and operating power are sampled according to the control system refresh cycle; rainfall forecast products are obtained according to their release frequency and their original timeliness identifiers are maintained. Each data record retains both the sampling time and the receiving time. The sampling time is used for time-series merging, and the receiving time is used to identify transmission delays and data replenishment. The time field is uniformly stored in the same time zone and the same format, and a monotonicity check is performed on the sampling time. If out-of-order arrival is found, the data is rearranged according to the sampling time but the original receiving time is retained.

[0031] The raw data received is uniformly converted into station / location, indicator, numerical, and time formats for storage, and unit consistency checks are performed before writing. Rainfall data is uniformly converted to millimeters or millimeters per hour, water level data is uniformly converted to meters, flow rate data is uniformly converted to cubic meters per second, gate opening is uniformly converted to a percentage, pump station count is an integer, and power is in kilowatts. When the original units are inconsistent with the units written in the object file, the conversion is performed according to the recorded proportional coefficient and offset before writing. When it comes to the conversion from gate opening to flow rate or from pump station count to discharge capacity, the registered engineering design curve or historical calibration curve is called for conversion. The conversion result is saved simultaneously with the original opening / count, and the original value is not overwritten by the conversion result.

[0032] Quality tags are attached to each piece of data written, and missing, jump, and out-of-range tags are generated according to fixed rules. Missing data tags are based on the sampling period: if no data is received for two consecutive sampling times, that sampling time is marked as missing. Jump tags use a sliding window difference rule: for the same indicator at the same station, the difference between adjacent time points is calculated; if the difference exceeds the rate of change of the indicator within the legal range registered in the object file or exceeds a preset jump threshold, it is marked as a jump. Out-of-range tags are based on the legal range registered in the object file; when water level, flow rate, opening degree, power, etc., exceed the range, they are marked as out of range. Quality tags are saved simultaneously with the original values; directly rewriting the original values ​​when generating quality tags is prohibited.

[0033] It should be noted that missing data is filled in, but the filled values ​​and original values ​​are stored in separate fields. The filling rules are limited to two categories: the first is intra-station time interpolation filling, where linear interpolation is used to generate filled values ​​when the missing data length does not exceed a preset duration (e.g., no more than 3 sampling periods); the second is neighboring station or related object filling, where the missing data length exceeds a preset duration, the station closest in the object archive spatial location information with a clear hydraulic relationship is selected as the reference station, and a weighted estimate is generated using a fixed weight. For missing water levels in river sections and control sections, a weighted estimate is performed by referring to the water levels of upstream or downstream control sections; for missing flow rates, an estimate is made by referring to the flow rates of adjacent sections or the flow rates converted from the water level-flow relationship curve; for missing gate openings, the last valid readout value of the control system is referenced, and the action interval constraint is used to determine whether it remains unchanged. All filled values ​​are written to the filling field, while the original fields retain the missing data status and a missing data marker.

[0034] After quality marking and data completion, the data are merged sequentially according to a unified time step to form observation frames with a fixed format. The unified time step is consistent with the least common multiple resolution of the sampling periods of each indicator in the object file, usually 1 minute, 5 minutes, or 15 minutes. For data with sampling periods shorter than the unified time step, aggregation is performed according to the unified time step, and the aggregation method is fixed according to the indicator type: rainfall is accumulated by time window, rainfall intensity is averaged by time window, water level is calculated by the last value or average of the time window (one is fixed in the configuration), flow rate is averaged by time window, pump opening is calculated by the last value of the time window, number of pump stations is calculated by the last value of the time window, and power is averaged by time window. For data with sampling periods longer than the unified time step, the most recent valid value is used for forward hold, and the source sampling period and hold count records are retained.

[0035] The observation frames are organized by object number as the index. Each frame contains the runtime values, imputation values ​​(if any), quality flags, and applicable label status for the corresponding threshold record of all objects within that time step. When multiple conflicting data sources occur for the same object at the same time step, the primary data source is selected according to the data source priority defined in the access configuration. The remaining sources are saved in backup fields and written with conflict flags. The conflict flag includes the conflicting data source identifier, conflict value, and difference magnitude. Each frame is written with the frame number and generation time, and the baseline version number of the object file and the constraint statement version number used are recorded to ensure that the observation frame is consistent with the frozen baseline version.

[0036] Rainfall forecasts and upstream inflow forecasts are processed using the same rules and then written into forecast frames. Forecast frames are stored with the release time, forecast time, grid or sub-basin number, and rainfall amount, and the product source and version are recorded. When the forecast is in gridded format, gridded rainfall is spatially aggregated according to the registered sub-basin or zoning number. The aggregation rule is fixed as area-weighted average or area-weighted cumulative, and the aggregation rule used is recorded. Forecast frames and observation frames are aligned on the timeline: for each observation frame, the most recently released and still valid forecast frame is associated, and its identifier is written into the association field of the observation frame to avoid mixing different release batches.

[0037] After generating observation and forecast frames, an access consistency check is performed, and an access operation log is generated. The check includes at least the following: whether the key operational quantities for each object have valid values ​​within a continuous time window; whether the missing measurement ratio exceeds the threshold; whether jump and out-of-range markers appear in clusters; whether the completion ratio is abnormal; and whether the data source latency exceeds the threshold. For any issues discovered during the check, each item is verified against the binding relationship and unit conversion records. If necessary, the correspondence between the site / location and the object number is corrected, or the conversion parameters are corrected. A new version is then written into the object file and the data collection binding relationship, retaining the original version's validity period and change records.

[0038] S2: Based on the upstream and downstream relationships and control relationships of the object archive and combined with the observation frame sequence, output the influence correlation and calculate and save the weight parameters of each correlation edge within the sliding window. Generate candidate control values ​​according to the object archive and effective constraints on the current observation frame, and calculate the predicted water level of the control section and river segment by combining the edge weight and equivalent flow increment.

[0039] Furthermore, based on the fact that observation frames and forecast frames have been generated and continuously written at a unified time step, the influence relationships between objects are organized into a set of directed relationships according to the upstream and downstream connections and control relationships in the object archive. Influence relationships are established only between six types of objects: directed relationships between upstream reservoir objects and downstream river sections or control sections; directed relationships between gate objects and their downstream control sections; directed relationships between river section objects and downstream control sections; and directed relationships between pumping station objects and their corresponding flood-prone points. Only one valid relationship record is retained for the same pair of source and target objects. If duplicate relationship records with overlapping validity periods exist in the object archive, the record with the newer update time is used, and the remaining records are deleted. For each target object, the set of all source objects pointing to that target object is aggregated to form the upstream influence set of that target object, and this set is kept consistent with the object's unique ID; object names are not used for matching.

[0040] For each directed relation in the influence association, prepare time series quantities that can be used for calculation. Prepare equivalent flow series for the source object side and water level series for the target object side. The equivalent flow series preferentially uses directly collected flow data: outflow for reservoir objects and cross-sectional flow for river sections or control sections; when there are no direct flow measurement points for gate objects, the gate flow is calculated by converting the actual gate opening and the water level above and below the gate according to the registered engineering design curve or historical calibration curve, and the converted gate flow is used as the equivalent flow; when there are no outflow measurement points for pumping station objects, the pumping station outflow is calculated by converting the number of pumps in operation and the rated flow or calibration discharge curve of a single pump, and the converted outflow is used as the equivalent flow. All the above conversions retain the original opening, number of pumps, and water level data, and the conversion results are saved as independent values ​​without overwriting the original collected values; when there are missing or jump marks in the water level above and below the gate or the number of pumps in operation required for the conversion, the equivalent flow inherits the quality mark simultaneously.

[0041] The water level sequence of the target object is uniformly taken from the collected water level or supplementary water level in the observation frame, and a fixed value selection rule is applied before calculation: if the collected water level exists and is not marked as missing or out of range, the collected water level is used; otherwise, the supplementary water level is used; if the supplementary water level also does not exist, that moment is not included in the parameter calculation. The equivalent flow sequence also has a fixed value selection rule: if the directly collected flow is available, the collected flow is used; if it is not available, the converted flow is used; if neither is available, that moment is not included in the parameter calculation. All time series involved in the calculation are uniformly aligned according to the time step of the observation frame. If there is a sampling gap between the source object and the target object at a certain moment, that moment is completely removed and not interpolated before being included in the parameter calculation.

[0042] It should be noted that the edge weights of each directed relation are calculated within the most recent continuous historical window. The length of the historical window is taken from the most recent 24 hours of valid data, and the number of frames included in the window is automatically determined according to a uniform time step: when the time step is... At the minute mark, the window frame count is taken as follows: Integer values; only moments that simultaneously satisfy the condition of valid equivalent flow of the source object and valid water level of the target object are included in the window set. The data within the window is first differentially calculated: the target object water level difference is the difference between water levels at adjacent moments, and the source object equivalent flow difference is the difference between equivalent flow rates at adjacent moments; the differential time is based on the observation frame time, and the differential results are not included in the summation when missing measurements, out-of-range conditions, or jump markers exist. The edge weights of each relationship are directly calculated from the window data using the least squares form, expressed as:

[0043] in, Indicates from the source object To the target object The influence of edge weights Indicates the sample number within the window. This indicates the length of the sliding window (the number of sample points contained within the window). Indicates the current calculation time. Indicates the first in the window A historical moment (looking back at the past) (time step) Represents the source object At any moment The equivalent flow increment, Represents the target object At any moment The increase in water level.

[0044] The calculated edge weights, along with the corresponding unique source object ID, unique target object ID, calculation window start and end times, number of valid samples, and the baseline version number of the object archive used, are written to the edge parameter record. The edge parameter record is saved using a versioned writing method: a new version record is generated for each calculation; the effective time of the new version record is the calculation completion time, and the expiration time is the effective time of the next version record; only one effective edge weight record is allowed for the same directed relation at any given time. The numerical range of the edge weights is validated: when an edge weight is non-numerical, infinite, or exceeds a reasonable order of magnitude (e.g., extreme values ​​caused by abnormal differences), the weight record is rejected, and the previous effective weight is used. Simultaneously, the reason for rejection and the corresponding source and target object IDs are recorded in the runtime log.

[0045] After the edge weight record is written, an upstream influence set for the target object is formed and a static snapshot of the upstream set is stored. The upstream influence set is indexed by the unique ID of the target object, and the set elements are the unique IDs of all source objects that point to the target object and are currently valid. When the upstream influence set of a target object is empty, the target object only retains its own water level sequence as a state variable and does not generate an upstream set record. When there are multiple upstream relationships for the same target object, all upstream relationships are kept active in parallel without manual merging. When the number of upstream relationships exceeds a preset limit (e.g., more than 20), they are sorted by the absolute value of the edge weight from largest to smallest, and only the first few are retained and added to the set. The remaining relationships are still kept in the edge parameter record but are not added to the set snapshot. The set snapshot records both the truncation threshold and the number of retained relationships.

[0046] The aforementioned impact associations, directed relationship sets, edge weight records, and upstream impact set snapshots are all bound to and saved with the baseline version number of the frozen object file. When a new version of the object file is added (e.g., a new object is added, relationship changes occur, or curve parameters change), the impact association set is refreshed synchronously with the object file version, and the edge weights are recalculated and a new version record is generated under the new version. Historical version records are retained without being overwritten.

[0047] Furthermore, under the condition that observation frames and forecast frames are continuously written, and the influence correlation and edge weight records have formed effective versions, the current state of each object is valued at a uniform time step, and the value range of candidate control quantities is organized accordingly. A fixed priority rule is adopted for the current state value at any given time: if an object has an unmarked and unmarked out-of-range acquisition value in the observation frame, the acquisition value is taken; otherwise, the supplementary value is taken; if neither the acquisition value nor the supplementary value exists, the object is not included in the calculation at that time, and the object's unique number and the name of the missing indicator are recorded in the operation log. Water level, flow rate, actual gate opening, number of pumping stations started, and operating power are all valued according to the above rules; the values ​​and corresponding quality marks are retained together, and the original observation frames are not written back to modify them.

[0048] For controllable objects, establish a set of controllable quantities for the current period and solidify value constraints. For gate objects, the controllable quantity is the target opening degree, whose allowable range is directly taken from the lower and upper limits of the opening degree registered in the object file. Simultaneously, action constraint verification is performed: the interval between two actions must not be less than the minimum action interval registered in the object file; under allowed action conditions, the absolute value of the difference between the target opening degree and the current actual opening degree must not exceed the product of the maximum change in opening degree per unit time and the time step registered in the object file. For pump station objects, the controllable quantity is the target number of operating units, limited to an integer value from 0 to the total number of units registered in the object file. Start-stop interval verification and power limit verification are performed: when the time since the last action is less than the minimum start-stop interval registered in the object file, the target number of pump station objects is fixed to the current actual number of units; when a change in the target number causes the estimated power to exceed the maximum allowable power registered in the object file, that target number value is not allowed to enter the candidate set. The controllable quantity of the reservoir object is the target discharge flow, and its allowable range is directly taken from the lower limit and upper limit of the discharge flow registered in the object file. At the same time, the rate of change constraint is implemented: the absolute value of the difference between the target discharge flow and the current outflow flow shall not exceed the product of the maximum discharge change per unit time and the time step registered in the object file.

[0049] The procedure thresholds and constraint statements are applied to the set of controllable quantities to form the current effective limits. For constraint statements related to control sections, the effectiveness condition is first determined. The effectiveness condition is determined only using the current water level, current flow rate, warning level label, and validity period label. When the effectiveness condition is met, the controllable quantity set of the corresponding object is truncated or removed according to the upper limit, lower limit, or rate of change limit explicitly written in the constraint statement. When multiple constraint statements for the same object are effective at the same time, they are merged according to the priority in the rule record. The merging method is fixed as follows: for upper limit constraints, the minimum upper limit is taken; for lower limit constraints, the maximum lower limit is taken; and for rate of change constraints, the minimum allowable rate of change is taken. If a conflict occurs after merging where the lower limit is greater than the upper limit, no candidate value is output for that object at that time, and the conflict information (unique object number, conflict constraint source clause number, conflict value) is recorded.

[0050] For each candidate value requiring equivalent flow rate calculation, the equivalent flow rate is calculated according to the conversion relationship registered in the object file, ensuring the original quantity is not overwritten. For reservoir objects, the equivalent flow rate is directly taken from the reservoir flow rate; the current equivalent flow rate is taken from the outflow flow rate of the current observation frame, and the candidate equivalent flow rate is taken from the outflow flow rate of the candidate target. For gate objects, the equivalent flow rate is taken from the gate passage flow rate; the current gate passage flow rate is preferentially taken from directly collected flow measurement points; if no direct measurement points exist, it is calculated using the actual gate opening, upstream water level, and downstream water level according to the engineering design curve or historical calibration curve registered in the object file. The conversion uses piecewise linear interpolation: when the curve is stored as several discrete points, linear interpolation is performed according to the interval where the opening and head are located to obtain the gate passage flow rate. For pump station objects, the equivalent flow rate is taken from the flow rate; the current outflow flow rate is preferentially taken from the directly collected outflow measurement points; if no direct measurement points exist, it is calculated using the current number of pumps in operation and the single-unit rated flow rate or calibration discharge rate curve registered in the object file; the candidate outflow flow rate is calculated using the same curve based on the number of candidate target pumps. The curve parameters required for the above conversion are all taken from the valid records of the frozen baseline version of the object file or its subsequent versions. The unique number and validity period of the applicable object of the curve must be consistent with the current object. If there is any inconsistency, the conversion will not be performed and an exception will be recorded.

[0051] After the candidate controllable quantities are selected and equivalent flow rates are converted, predicted water level records are generated for the target objects requiring rapid water level calculation in the influence association. The predicted objects are limited to control section objects and river segment objects; for each predicted object, its current water level is used as the base value, and the predicted water level is calculated using the edge weights of all source objects in its upstream influence set and the candidate equivalent flow rate increment. The predicted water level is expressed as:

[0052] in, Represents the target object The predicted water level value for the next time step; Represents the target object The water level at the current moment; Indicates a uniform time step; This represents the candidate control value combination; it consists of the target opening degree of the gate object, the target number of pumping stations, and the target discharge flow of the reservoir object, etc. The value range and action constraints are derived from the object file and the effective restrictions of the constraint statement. Represents the target object The collection of source objects; Represents the source object In candidate values The equivalent flow increment is expressed as:

[0053] in, The current equivalent flow rate of the source object is obtained from the observation frame according to a fixed priority rule (directly collected flow rate is preferred, and flow rate converted from curve is an alternative). The equivalent flow rate of the source object under the candidate values ​​is as follows: for the reservoir object, the discharge flow rate of the candidate target is taken; for the gate object, the flow rate through the gate is calculated according to the opening degree of the candidate target and combined with the water level above and below the gate in the current period, using piecewise linear interpolation of the object file registration curve; for the pumping station object, the flow rate is calculated according to the number of candidate targets and the calibrated discharge curve.

[0054] It should be noted that for each predicted water level record, the unique identifier of the target object, the current time, the candidate value combination identifier, the number of source objects participating in the summation, the edge weight version number used, the object archive baseline version number, and the observation frame number are simultaneously written. For any source object participating in the summation, if its If a source object is missing or its quality is marked as out of range, it is removed from the summation item of that record, and the reason for removal is recorded. When the number of source objects is 0 after removal, only the predicted water level record is retained. The results are then marked as having no valid upstream terms. The above predicted water level records are saved in time series format, without modifying the object files, threshold records, constraint statements, or the original content of the observation frames during the saving process.

[0055] S3: Merge the candidate values ​​into control combinations, use the guaranteed water level for hard elimination, then use the predicted water level and the change in action to calculate the evaluation value and select the combination with the smallest value. Convert the selected combination into an instruction data packet and verify the rate of change, interval and linkage conflict before issuing it. Collect the receipt and write the actual execution result into the status record of the next moment.

[0056] Furthermore, after candidate combinations of controllable values ​​have been formed and corresponding predicted water level records have been obtained for each control section and river segment, the candidate value combinations are systematically standardized. The candidate value combinations are indexed by the unique object number, merging the candidate values ​​of all controllable objects at the same time into a single control combination. During merging, duplicate candidate values ​​for the same object are not allowed. If multiple candidate value records exist for the same object at the same time, only candidate value records that satisfy the minimum action interval, rate of change constraint, boundary constraint, power constraint, and have the same source version number will be retained. The remaining records will be marked as conflicting and discarded. For each group... Generate a combination identifier and save the combination content, which includes: the target opening degree of the gate object, the target number of pump stations to be started, the target discharge flow of the reservoir object, and the current actual values ​​of each object (actual gate opening degree, actual number of pump stations, and current outflow of the reservoir) and the upper and lower limit parameters registered in the object file.

[0057] For each group A hard threshold verification is performed, using only the fixed, defined values ​​from the threshold records for judgment. The verification objects are limited to control section objects and river sections; the verification indicator is "predicted water level does not exceed the guaranteed water level." Specifically, for each control object... Read the control object in the group Predicted water level And read the guaranteed water level from the threshold record of the controlled object. When a valid At that time, if Then the group If a controlled object is directly deemed unqualified and removed, or if the record is not within the current applicable conditions and validity period, the controlled object will not participate in the hard threshold verification, but its unique number, missing threshold type, and currently applicable label will be recorded in the operation log. The reasons for removal (exceeding the limit object number, exceeding the limit range, and the threshold record version number used) will be retained for the candidate combinations and written to the combination log.

[0058] Candidate combinations that pass the hard threshold verification Calculate the evaluation value The evaluation value is calculated as follows:

[0059] in, Indicate candidate combinations Calculate the evaluation value. This represents the set of controlled objects participating in the evaluation. Represents the controlled object The warning water level, Represents the controlled object The guaranteed water level This represents the set of controllable objects involved in the calculation of action costs. and Representation Object The upper and lower limits of the adjustable amount, Indicate candidate combinations For objects The amount of change in action; Among them, when When it is a gate object, ,in The target opening degree of the gate in the candidate combinations. To observe the actual opening degree of the gate in the observation frame; when When the target is a pumping station, ,in The number of candidate target units, The actual number of units in operation in the observation frame; when When the object is a reservoir, ,in Discharge flow to candidate targets, This represents the current outflow rate in the observed frame.

[0060] Represents the controlled object In candidate combinations The predicted water level is expressed as:

[0061] When a In If no predicted water level record exists, the current water level shall be used directly. As and the Marked as no prediction.

[0062] For each group that passed verification Calculated Then, an evaluation record is generated and written to the evaluation table. The evaluation record must include at least: a combination identifier, the target value for each object, the predicted water level for each controlled object, the warning and guarantee threshold version number used for each controlled object, the unique number of the controlled object corresponding to the maximum value of the first item and its exceedance of the warning level, the sum of the action changes, the normalized denominator value of the action, and the final... The data includes numerical values, observation frame numbers, object archive baseline version numbers, constraint statement version numbers, and edge weight record version numbers. If duplicate evaluation records exist for the same combination of identifiers, only the record with the most recent generation time and the same version number will be retained; the remaining records will be marked as duplicates and removed.

[0063] After the evaluation record is written, the combination with the smallest evaluation value is selected from all valid candidate combinations. The selection rule is fixed as follows: first select the combination with the smallest evaluation value. Sort by smallest to largest. If there are ties for the smallest value, sort by the sum of the absolute values ​​of the changes in action. If they are still ties, select the combination with the smallest lexicographical order of the combination identifier. Selected combinations are written to the selected combination record, which contains the combination identifier and a complete list of values, and is associated with the evaluation record through the combination identifier. Evaluation records of unselected candidate combinations are retained without deletion, and only an unselected mark and ranking are written to them.

[0064] It should be noted that after the candidate combination with the smallest evaluation value has been determined and the selected combination record has been formed, the target values ​​of each object in the selected combination are converted into a set of directly issued instruction items. Instruction items are generated using the object's unique ID as an index. Each instruction item corresponds to only one object and contains only one type of control quantity. The instruction item content includes: object unique ID, object type, instruction type, target value, allowed rate of change, instruction generation time, instruction effective start time, instruction effective end time, and instruction version number. The instruction effective start time is fixed at the start of the next time step of the observation frame time corresponding to the selected combination; the instruction effective end time is fixed at the effective start time plus a preset effective window length (e.g., 30 minutes or 60 minutes). The effective window length is a specified value in the access configuration and remains unchanged. The allowed rate of change is directly taken as the maximum change per unit time registered in the object file; if a minimum action interval is also registered in the object file, this minimum action interval is written into the instruction item and saved as an execution verification field.

[0065] Generate a target opening instruction for the gate object. The target value is taken as the target opening of the gate object in the selected combination. The target value is a percentage value and is limited to the lower and upper limits of the opening registered in the object file. The instruction also writes the current actual gate opening as a reference field. The current actual gate opening is obtained from the observation frame according to a predetermined value priority rule (acquired value first, supplementary value as an alternative). When the absolute value of the difference between the target opening in the selected combination and the current actual opening exceeds the product of the maximum opening change per unit time and the time step registered in the object file, the target value is not truncated or rewritten. Instead, the instruction is marked as not meeting the rate of change and the generation of the gate's instruction is stopped. At the same time, the object's unique number, target value, actual value, and allowable change are recorded in the operation log. Instructions for other objects are still generated as usual.

[0066] Generate a target number of pump station units instruction for the pump station object. The target value is the target number of pump stations to be started in the selected combination, which is an integer and limited to the total number of units registered in the object file. The instruction item also writes the current actual number of units started and the current operating power as reference fields, which are obtained from the observation frame according to the established value priority rule. Perform start-stop interval verification on the target number of units: read the effective time of the last issued instruction item that is still valid for the pump station object. If the effective time is less than the start time of the current instruction, the pump station object will not generate an instruction item this time and the reason for rejection will be recorded. Perform power verification on the target number of units: when there is a power measurement point, the target power is estimated using the power per unit number of units; when there is no power measurement point, the target power is estimated using the rated power curve registered in the object file; if the estimated target power exceeds the maximum allowable power registered in the object file, the pump station object will not generate an instruction item this time and the reason for rejection will be recorded.

[0067] Generate a target discharge instruction for the reservoir object. The target value is taken as the target discharge flow rate of the reservoir object in the selected combination, which is a value in cubic meters per second and limited to the lower and upper limits of the discharge flow rate registered in the object file. The instruction item also writes the current outflow flow rate as a reference field. The current outflow flow rate is obtained from the observation frame according to the established value priority rule. Perform a change rate check on the target discharge: compare the absolute value of the difference between the target discharge and the current outflow flow rate with the product of the maximum discharge change per unit time and the time step registered in the object file. If it exceeds, no instruction item is generated for the reservoir object this time and the reason for rejection is recorded. If the reservoir object has a target value in the form of a discharge process line, the process line is discretized into a segmented target value sequence with the same time step, and a sub-instruction item is generated for each segment. The effective start time of each sub-instruction item is written in ascending order of segment. The change rate check of each segment is executed segment by segment. If any segment fails, the generation of instruction items for the entire process line is stopped and the reason for rejection is recorded.

[0068] After each object instruction item is generated, conflict and consistency checks are performed on the instruction item set. Conflict checks include: whether there are multiple instruction items for the same object within the same valid window; if so, only the instruction with the highest version number and the most recent generation time is retained, and the rest are marked as conflicting and removed. For objects with linkage constraints (such as upstream reservoir discharge and downstream gate opening, pumping station drainage and flood-prone point water level control constraints), the fixed linkage limit values ​​in the constraint statements are read, and it is checked whether the target values ​​of the corresponding objects in the instruction item set simultaneously meet the constraints; if not, the instruction item label values ​​are not automatically rewritten, but the instruction items involving the objects are marked as linkage conflicts and the generation and issuance of records are stopped, while the conflict source clause number, conflict object unique number, and conflict value are recorded.

[0069] The verified set of instructions is encapsulated into an instruction data package. The instruction data package includes: data package number, generation time, object archive baseline version number, threshold record version number, constraint statement version number, edge weight record version number, corresponding observation frame number, and a list of instructions. The data package number is generated using a date combined with a batch number, with the batch number monotonically increasing within the same day. The instruction item list is sorted by object type: reservoir instructions first, gate instructions second, and pump station instructions last; within the same type, instructions are sorted lexicographically by the object's unique ID. Instruction data packages cannot be modified after being written to storage; new versions are only allowed by adding new data packages. Any change to an instruction item is implemented by generating a new data package; historical data packages are retained but not overwritten.

[0070] When sending instruction data packets to the control system or dispatching platform, field mapping is performed using the registered data source and object number binding relationship. The target opening degree of a gate object is mapped to the target opening degree point of the gate control system; the target number of pump stations is mapped to the pump station start / stop or number setting point; and the target discharge of a reservoir object is mapped to the discharge setting point of the reservoir dispatching system. Secondary unit conversion is not allowed during mapping; unit conversion is only permitted during the data access and object file curve conversion stages. During transmission, the sending time, reception confirmation status, and transmission channel identifier are recorded, and this record is associated with and saved with the instruction data packet number. If reception confirmation fails or times out, the fixed number of retries and retry interval are executed according to the access configuration. If the transmission still fails after exceeding the retry count, a failure status is written, and transmission is stopped.

[0071] The system receives receipt data from the control system or dispatch platform and matches it with the unique object number and instruction data packet number. Receipt data includes at least: unique object number, receipt time, reception status, execution status, actual execution value, and failure reason. For gate objects, the actual execution value is the actual opening readback value; for pump station objects, the actual execution value is the actual number of pumps in operation and the outflow readback value; for reservoir objects, the actual execution value is the actual discharge flow readback value. The receipt data is written to the receipt record table, retaining the original failure reason text without merging or rewriting. When the receipt shows execution failure or the deviation between the actual execution value and the target value exceeds the allowable deviation threshold registered in the object file, the object's operating status in the object file is updated from "in service" to "fault" or "pending status requiring manual confirmation for maintenance." Boundary parameters and threshold parameters are not automatically rewritten; only deviation information and the receipt reason are recorded, and this change is presented in the operating status field in the observation frame of the next time step. The aforementioned receipt records, issuance records, and instruction data packets are all associated with and saved with the corresponding observation frame number, object file baseline version number, and rule version number, maintaining a consistent index relationship for decision and execution data in the same batch.

[0072] Example 2, one embodiment of the present invention, provides a water conservancy flood control emergency control method based on digital twins. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0073] First, a verification was conducted using a small-to-medium-sized river basin-urban integrated flood control scenario. The basin included one medium-sized reservoir (with adjustable discharge capacity), two sluice gates (adjustable opening degree), two pumping stations (adjustable number of operating units), three key control sections (downstream constraint points), and two flood-prone areas. Three independent flood events were used as comparisons: E1 was a short-duration heavy rainfall event, E2 was a continuous moderate-to-heavy rainfall event, and E3 was an extreme event with even higher peak intensity. All three events used the same monitoring network and the same set of engineering capacity parameters to ensure comparability. First, a unique file was established for all objects, registering their spatial location (latitude and longitude, elevation, river mileage / section number), operational status fields (water level, flow rate, opening degree, number of pumping stations, power, etc.), adjustable quantities and boundary parameters (opening degree / number of pumping stations / upper and lower limits of discharge, allowable rate of change, minimum action interval, maximum power, etc.). The warning and guarantee thresholds for the control sections, as well as the applicable linkage constraint clauses during the flood season, were clearly defined and fixed numerically. Subsequently, the locations of rain gauges, water level stations, flow stations, and gate pump control points were bound and verified, and the sampling and reception times were retained. Data with inconsistent units were converted before being written. For gate flow and pump station outflow in the absence of direct measurement points, existing calibration curves or rated discharge parameters were used for conversion, and the conversion results were saved in parallel with the original opening degree / number of pumps. After the data entered, missing measurements, jumps, out-of-range, and delays were marked. For short missing measurements, interpolation at the same station was used to fill in the gaps; for long missing measurements, weighted filling was used with relevant upstream and downstream stations, and the filled values ​​were stored separately. Then, observation frames were generated by merging at a uniform time step, and rainfall forecasts were aligned to each observation frame according to the release batch. Next, the influence correlations were sorted out based on upstream and downstream and control relationships, and the source-side equivalent flow sequence and target-side water level sequence were extracted. Within a 24-hour sliding window, the weight of each correlation edge was updated with effective samples and saved in a versioned manner. At each time step, under effective constraints, a set of candidate values ​​for the target gate opening, the target number of pumping stations, and the target reservoir discharge is generated. After completing the equivalent flow increment conversion, the predicted water level for the next time step is calculated for the control section and river segment. Subsequently, the candidate values ​​of each object are merged into control combinations. First, combinations that guarantee water level execution are hard-rejected. Then, a comprehensive evaluation is performed on the eliminated combinations, and the smallest value is selected. The selected combination is converted into a command data packet, and the rate of change, minimum interval, maximum power, and upstream and downstream linkage conflicts are checked. Commands that pass the check are issued according to the predetermined mapping. The receipt includes the reception status, execution status, actual execution value, and failure reason. The receipt is written into the status record of the next time step to reflect the execution deviation and changes in equipment availability. The control group adopts the traditional joint commissioning process: relying on fixed experience coefficients and manual adjustments, without edge weight sliding updates and command-level conflict checks, and only performing routine threshold alarms.

[0074] Table 1 Experimental Data

[0075] As shown in Table 1, the indicators exhibit a consistent trend of improvement, and the sources of improvement clearly correspond to the key technical points of each stage. Firstly, regarding predictive capability, the peak prediction errors for E1, E2, and E3 decreased from 0.36 / 0.29 / 0.44 m to 0.12 / 0.10 / 0.15 m, indicating that the short-term prediction of the water level at the control section is more closely aligned with actual flood conditions. This difference is not simply due to a more complex model, but rather consistent with the approach of updating the influence-related boundary weights within a sliding window and estimating the response intensity based on effective samples for equivalent flow increments and water level increments. When rainfall intensity and engineering control cause a drift in the hydraulic response, traditional schemes using fixed empirical coefficients are prone to systematic biases, while adjusting the boundary weights according to the latest observations can absorb the drift as parameter changes, thus demonstrating a lower peak error. Secondly, regarding the safety constraints, the duration of exceeding the guaranteed water level decreased from 78 min to 18 min in E1, from 55 min to 12 min in E2, and from 132 min to 35 min in E3. Simultaneously, the peak water level at the control section also decreased by 0.18 / 0.14 / 0.28 m, respectively, indicating that the control combination more effectively suppressed the peak and shortened the duration of exceeding limits within the constraints. This demonstrates the role of the hard elimination of guaranteed water level schemes and the optimal selection based on comprehensive evaluation: hard elimination prevented schemes that exceeded the limit from entering the candidate set, while the optimal selection based on evaluation allowed for the selection of a better compromise between risk level and action intensity while meeting the bottom line, thus significantly shortening the duration of exceeding limits in all three processes. Thirdly, regarding executability, traditional joint commissioning experienced execution failures due to exceeding the rate of change / minimum interval and execution failures due to linkage conflicts in all three processes (e.g., 6 times and 3 times in E3, respectively), while the corresponding indicators for this scheme were all 0, and the success rate of issuance was increased to approximately 99.4%–99.7%. The results are highly consistent with the verification of change rate, interval, power, and linkage conflict before the generation of instruction data packets: although traditional processes may provide theoretically feasible target values, the lack of instruction-level verification can easily trigger equipment action limitations or upstream and downstream linkage conflicts at the execution level, leading to failure and repeated adjustments. The verification mechanism eliminates these unexecutable situations before issuance, significantly improving the issuance success rate. Furthermore, the average execution deviation decreased from 7.8% / 6.5% / 9.2% to 2.1% / 1.9% / 2.7%, indicating that the feedback and actual execution are closer to the target setting; combined with the closed-loop mechanism of "writing the feedback into the next moment's status record," long-term accumulation of deviations can be avoided and repeated trial-and-error adjustments can be reduced. Finally, in terms of emergency response timeliness and operational costs, the time for a single solution deployment decreased from 145-190s to 26-32s, the number of actions decreased by about 35%-46%, and the pump station energy consumption also decreased by about 4%-7% in three processes, demonstrating that fewer and more effective action combinations are formed in a shorter decision-making time, reducing execution frequency and energy consumption burden.In summary, Table 1 shows that the present invention demonstrates repeatable improvements in predictive fit, threshold constraint satisfaction, instruction executability, and closed-loop stability. It provides a structured improvement path to address the shortcomings of traditional technologies in terms of fixed parameters, lack of instruction-level verification, and insufficient acknowledgment closed-loop, reflecting the comprehensive advantages brought by novelty and inventiveness.

[0076] Example 3, an embodiment of the present invention, provides a water conservancy flood control emergency control system based on digital twin, including an object and rule solidification module, a data merging and association modeling module, and a control solution and closed-loop execution module.

[0077] The object and rule solidification module is used to establish uniquely numbered object files for reservoirs, gates, pumping stations, river sections, control sections, and flood-prone points, and solidify boundary constraints, thresholds, constraint statements, and binding relationships of data collection points. The data merging and association modeling module is used to access rainfall, water level, flow rate, gate and pump operation status and forecast data, complete unit verification, quality marking, missing data completion, and generate observation frames and forecast frames. At the same time, it forms influence associations based on upstream and downstream and control relationships and calculates and saves edge weight parameters. The regulation solution and closed-loop execution module is used to generate candidate regulation values ​​on the current observation frame, calculate the predicted water level of control sections and river sections, evaluate and select the optimal regulation combination, generate instruction data packets for issuance and collect receipts, and write the actual execution results into the status record of the next time step.

[0078] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0079] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-including system, or other system that can fetch and execute instructions from, an instruction execution system, apparatus, or device). For the purposes of this specification, "computer-readable medium" can be any means that can contain, store, communicate, propagate, or transmit programs for use by, or in conjunction with, an instruction execution system, apparatus, or device.

[0080] More specific examples (a non-exhaustive list) of computer-readable media include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which programs can be printed, because programs can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0081] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc. It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

[0082] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A water conservancy flood control emergency control method based on digital twins, characterized in that, include: Establish uniquely numbered object files for reservoirs, sluice gates, pumping stations, river sections, control sections, and flood-prone points, and solidify boundary constraints, thresholds, constraint statements, and data acquisition binding relationships. Access rainfall, water level, flow rate, sluice gate and pump operation status, and forecast data according to the binding relationships, perform unit verification, quality marking, fill in missing data, and merge them into observation frames and forecast frames with a unified time step. Based on the upstream and downstream relationships and control relationships of the object archive and combined with the observation frame sequence, the influence correlation is output and the weight parameters of each correlation edge are calculated and saved within the sliding window. On the current observation frame, candidate control values ​​are generated according to the object archive and effective constraints, and the predicted water level of the control section and river segment is calculated by combining the edge weights and equivalent flow increments. Candidate values ​​are merged into control combinations, and hard elimination is performed using the guaranteed water level. Then, the evaluation value is calculated using the predicted water level and the change in action, and the combination with the smallest value is selected. The selected combination is converted into an instruction data packet, and the change rate, interval and linkage conflict are checked before being issued. The receipt is retrieved and the actual execution result is written into the status record of the next moment.

2. The water conservancy flood control emergency control method based on digital twin as described in claim 1, characterized in that: The process of establishing uniquely numbered object files for reservoirs, sluice gates, pumping stations, river sections, control sections, and flood-prone points, and solidifying boundary constraints, thresholds, constraint statements, and data acquisition binding relationships, includes generating a unique number for each object and writing its name, governing unit, status (in service, under maintenance, out of service, faulty operation), and update time, while also registering its spatial location information. Spatial positioning information includes latitude and longitude, elevation, river name, and river mileage or section number; For reservoirs, register the water level and storage capacity relationship data and flood discharge facility parameters; for gates, register the number of gates, maximum opening degree, upper limit of opening and closing speed and minimum action interval; for pumping stations, register the number of units, rated flow of a single unit, minimum start-stop interval and maximum allowable power, and specify and solidify the upper and lower limits and allowable change rates of each adjustable quantity.

3. The water conservancy flood control emergency control method based on digital twin as described in claim 2, characterized in that: The process of accessing rainfall, water level, flow rate, and gate pump operation status and forecast data according to the binding relationship, performing unit verification, quality marking, missing data supplementation, and merging into observation frames and forecast frames with a unified time step includes solidifying the acquisition binding relationship by associating the locations of rainfall stations, water level stations, flow rate stations, and gate pump control systems with unique object numbers, and retaining both the sampling time and the receiving time for each acquisition record. During the data access process, consistency verification is performed according to the units fixed in the object file, and unit conversion is completed before writing. In the absence of direct measurement points, the gate flow rate and pump station outflow rate are converted by the actual gate opening and the water level above and below the gate or the number of pump stations in operation and the rated discharge parameters, respectively. The conversion results are stored separately from the original collected values ​​and do not overwrite the original data.

4. The water conservancy flood control emergency control method based on digital twin as described in claim 3, characterized in that: The quality markers include missing, jump, out-of-range, and delayed markers. Missing markers are determined according to the sampling period, jump markers are determined using a sliding window differential threshold and continuity test, and out-of-range markers are determined using the valid value range fixed in the object file. When generating supplementary values ​​for missing data, if the missing data length does not exceed the preset duration, interpolation based on the same station time is used for supplementation; if the missing data length exceeds the preset duration, weighted estimation based on neighboring stations or upstream and downstream related objects is used for supplementation. The supplementary values ​​are written into an independent field and the original missing data marker is retained. Data are merged to generate observation frames at a uniform time step, and aggregation rules are fixed according to index type. Rainfall is accumulated over a time window, water level is taken as the last value or average, flow rate is taken as the average, and opening degree and number of stations are taken as the last value.

5. The water conservancy flood control emergency control method based on digital twin as described in claim 4, characterized in that: The output influences the correlation and calculates and saves the weight parameters of each correlation edge within the sliding window. The influence correlation is established as a directed relationship based on the upstream and downstream correlation and control relationship in the object file. For each directed relationship, the equivalent flow sequence of the source object and the water level sequence of the target object are extracted within the most recent continuous historical window. Equivalent flow sequences are preferentially obtained by directly collecting flow rates. If they do not exist, they are obtained by converting gate opening, flow rate, or number of pump stations and discharge capacity. Within the window, only the moment when the equivalent flow of the source object and the water level of the target object are simultaneously valid is selected for calculation. The calculated edge weights are saved together with the window start and end times, the number of valid samples, and the object file version number. When there are insufficient valid samples or calculation errors, the previous valid weight is used and a reuse mark is recorded.

6. The water conservancy flood control emergency control method based on digital twin as described in claim 5, characterized in that: The process of generating candidate control values ​​based on object files and effective constraints on the current observation frame, and calculating the predicted water level of the control section and river segment in combination with edge weights and equivalent flow increments, includes the following when generating candidate control values ​​on the current observation frame: for gate objects, the target opening degree is used as the candidate value and simultaneously satisfies the upper and lower limits of the opening degree, the maximum change in opening degree per unit time, and the minimum action interval constraints; for pump station objects, the target number of pumps in operation is used as the candidate value and simultaneously satisfies the constraints that the number of pumps is an integer, the minimum start-stop interval, and the maximum allowable power constraints; for reservoir objects, the target discharge flow rate is used as the candidate value and simultaneously satisfies the upper and lower limits of discharge and the maximum change in discharge per unit time constraints. Candidate values ​​are truncated or eliminated based on the effective constraint statements. After the equivalent flow conversion is completed, the predicted water level of the control section and river segment is calculated according to the edge weights saved in the influence association and the equivalent flow increment of the source object. The prediction results are associated with and saved with the number of source objects participating in the summation and the version number used.

7. The water conservancy flood control emergency control method based on digital twin as described in claim 6, characterized in that: The process of merging candidate values ​​into control combinations, using guaranteed water levels for hard elimination, and then using predicted water levels and changes in action to calculate evaluation values ​​and select the smallest combination includes merging candidate values ​​into control combinations, firstly using guaranteed water levels of control sections for hard elimination of control combinations, calculating evaluation values ​​for the eliminated control combinations and sorting them from smallest to largest evaluation values, and if evaluation values ​​are tied, sorting them from smallest to largest by the sum of absolute values ​​of changes in action and determining the optimal combination. When converting the optimal combination into an instruction data packet, an instruction item is generated for each object, containing the object's unique number, instruction type, target value, allowed rate of change, instruction start time, and effective end time. Before issuing the instruction, the duplicate instructions, rate of change, interval, power, and linkage limit conflict checks for the same object are performed. After the data is sent, receive the receipt data containing the receipt status, execution status, actual execution value and failure reason, and write the receipt data into the status record of the next moment according to the object's unique number.

8. A system employing the digital twin-based emergency control method for flood control and water conservancy as described in any one of claims 1 to 7, characterized in that: It includes a module for object and rule solidification, a module for data merging and association modeling, and a module for control, solution and closed-loop execution; The object and rule solidification module is used to establish uniquely numbered object files for reservoirs, gates, pumping stations, river sections, control sections and flood-prone points, and solidify boundary constraints, thresholds, constraint statements and data collection point binding relationships. The data merging and association modeling module is used to access rainfall, water level, flow rate, sluice gate and pump operation status and forecast data, complete unit verification, quality marking, missing measurement filling and generate observation frames and forecast frames, and at the same time form influence associations based on upstream and downstream and control relationships and calculate and save edge weight parameters. The regulation solution and closed-loop execution module is used to generate candidate regulation values, calculate the predicted water level of the control section and river section on the current observation frame, evaluate and screen the optimal regulation combination, generate instruction data packets for issuance and retrieve receipts, and write the actual execution results into the status record of the next time moment.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the water conservancy flood control emergency control method based on any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the steps of the water conservancy flood control emergency control method based on any one of claims 1 to 7.

Images