Natural resource engineering data intelligent management method

Abstract

The invention provides an intelligent management method for natural resource engineering data, and relates to the technical field of data processing, and the method comprises the steps: inputting a time-space coupling model into real-time weather forecast data, simulating water distribution schemes under different rainfall scenes, automatically recognizing a river reach position of which the ecological flow is lower than a design threshold value, and carrying out the calculation of the river reach position. In combination with a historical water replenishing strategy case library, generating a multi-objective optimization scheduling scheme including a water replenishing time sequence, a water replenishing path and a water replenishing amount; three monitoring points are arranged at a downstream section of an ecological protection area, a flood control section and a canal head of an irrigation area in a drainage basin, and ecological flow, flood control section water level and canal water level data are monitored in real time respectively; and dynamically correcting the water supplementing time sequence, the water supplementing path and the water supplementing amount in the multi-objective optimization scheduling scheme to obtain a corrected scheduling scheme. According to the method, refinement and toughness of drainage basin water resource management can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly to an intelligent management method for natural resource engineering data. Background Art

[0002] Watershed management needs to integrate multi-source data such as satellite remote sensing (e.g., NDVI vegetation index, surface temperature inversion data), hydrological stations (CSV format time series data of water level, flow rate, and water quality monitoring), and irrigation area Internet of Things sensors (channel water level LoRa transmission data, gate opening Modbus protocol data). Due to the lack of a unified semantic parsing framework, the following problems occur: Spatio-temporal reference conflicts. For example, the water level data of a certain reservoir uses the elevation benchmark of the dam top, while adjacent stations use the national 85 elevation benchmark (such as the Huanghai elevation). Incorrect unit conversion may cause a calculation deviation of the water storage volume of more than 0.5m. There is also the problem of semantic ambiguity. For example, the "runoff" field represents the daily average value in some databases, while it is the instantaneous value in other systems. Direct fusion will lead to chaotic input time series of the model.

[0003] In addition, in the scenario of sudden heavy rain, if the reservoir storage capacity is set only according to the historical average value, the reservoir may approach the flood control limit water level within 12 hours and be forced to discharge emergency flood, resulting in a sudden drop in the ecological flow downstream. Existing models often assume instantaneous transfer of water volume between the main stream and tributaries, ignoring the water flow propagation time (for example, the water flow delay from a certain main stream section to the ecological protection area reaches 8 hours), resulting in a misalignment between the water supply instruction and the flow peak, and the actual water supply efficiency is less than 60% of the expected value. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide an intelligent management method for natural resource engineering data, which can improve the refinement and resilience of watershed water resource management.

[0005] To solve the above technical problem, the technical solution of the present invention is as follows: An intelligent management method for natural resource engineering data, the method comprising: Step 1, construct a semantic parsing model based on a knowledge graph, perform structured conversion on heterogeneous data, parse the semantic ambiguity of key fields through a preset watershed ontology library, establish a standardized data pool under a unified spatio-temporal coordinate system, and process missing values using a spatio-temporal correlation interpolation algorithm for upstream and downstream stations to obtain a standardized data pool; Step 2, based on the standardized data pool output in Step 1, combined with the scheduling rule constraint conditions in the historical scheduling case library, establish a spatio-temporal coupling model including a dynamic balance equation of reservoir water storage, a continuity equation of river channel runoff, and an ecological flow threshold constraint; Step 3: Input the spatiotemporal coupling model constructed in Step 2 into real-time meteorological forecast data, simulate the water volume allocation scheme under different rainfall scenarios, automatically identify the river section positions where the ecological flow is lower than the design threshold, and combine with the historical water replenishment strategy case library to generate a multi-objective optimal scheduling scheme including water replenishment timing, water replenishment path, and water replenishment volume. Step 4: Set up three monitoring points at the downstream section of the ecological protection area, the flood control section, and the head of the irrigation canal in the basin, and respectively monitor the ecological flow, the water level of the flood control section, and the water level of the canal in real time. According to the spatial position characteristics of the monitoring points and the real-time monitoring data, dynamically correct the water replenishment timing, water replenishment path, and water replenishment volume in the multi-objective optimal scheduling scheme generated in Step 3 to obtain a corrected scheduling scheme. Step 5: Encode the scheduling scheme corrected in Step 4 to generate an executable scheduling instruction.

[0006] The above scheme of the present invention has at least the following beneficial effects.

[0007] By extracting rules from the historical scheduling case library (such as the ecological water demand weight in the dry season ≥ 0.6), converting empirical knowledge into linear programming constraint conditions, the adaptability of the scheduling scheme of the model in the sudden drought scenario is increased by 35%; the dynamic balance equation of the reservoir water storage introduces real-time inflow prediction (such as the ARIMA model), and combines with the water volume transfer matrix of the river channel runoff continuity equation (such as the main stream propagation time correction coefficient), reducing the water volume allocation error of the joint operation of the reservoir group from 15% of the traditional model to 5%; set the ecological flow threshold interval according to the flood season / non-flood season (such as ≥ 10 m 3 / s in the non-flood season, ≥ 15 m 3 / s in the flood season), avoiding the problems of over-protection or under-protection of the ecology caused by fixed thresholds. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 is a schematic flowchart of a method for intelligent management of natural resource engineering data provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0009] Hereinafter, exemplary embodiments of the present disclosure will be described in more detail with reference to the drawings. Although the exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present disclosure can be more thoroughly understood and the scope of the present disclosure can be fully conveyed to those skilled in the art.

[0010] As Figure 1 shown, an embodiment of the present invention provides a method for intelligent management of natural resource engineering data, and the method includes the following steps: Step 1: Construct a semantic parsing model based on a knowledge graph to perform structured transformation on heterogeneous data. Parse the semantic ambiguity of keyword fields through a preset basin ontology library, establish a standardized data pool under a unified spatio-temporal coordinate system, and process missing values using an upstream-downstream station spatio-temporal correlation interpolation algorithm to obtain the standardized data pool. Step 2: Based on the standardized data pool output in Step 1, combined with the scheduling rule constraint conditions in the historical scheduling case library, establish a spatio-temporal coupling model that includes a dynamic balance equation for reservoir water storage, a river runoff continuity equation, and an ecological flow threshold constraint. Step 3: Input the spatio-temporal coupling model constructed in Step 2 into real-time meteorological forecast data, simulate water volume allocation schemes under different rainfall scenarios, automatically identify the river section positions where the ecological flow is lower than the design threshold, and generate a multi-objective optimal scheduling scheme that includes the water replenishment time sequence, water replenishment path, and water replenishment volume in combination with the historical water replenishment strategy case library. Step 4: Set up three monitoring points at the downstream section of the ecological protection area, the flood control control section, and the canal head of the irrigation area in the basin to respectively monitor the ecological flow, the water level of the flood control section, and the water level of the canal in real time. According to the spatial position characteristics of the monitoring points and the real-time monitoring data, dynamically correct the water replenishment time sequence, water replenishment path, and water replenishment volume in the multi-objective optimal scheduling scheme generated in Step 3 to obtain the corrected scheduling scheme. Step 5: Perform instruction encoding on the scheduling scheme corrected in Step 4 to generate executable scheduling instructions.

[0011] In the embodiment of the present invention, through the basin ontology library (such as defining entity relationships such as "reservoir water level - storage capacity relationship" and "irrigation area water use priority"), the format conflicts of heterogeneous data such as satellite remote sensing, hydrological stations, and Internet of Things sensors are resolved (such as unifying the Huanghai elevation datum), the semantic ambiguity of fields such as "water level elevation" is eliminated (such as distinguishing absolute / relative reference values), and the data standardization processing efficiency is increased by more than 40%; the upstream-downstream station spatio-temporal correlation algorithm (such as interpolation based on the hydraulic diffusion model) is used to reduce the interpolation error of missing values from ±20% of the traditional linear method to ±5%. Especially in the case of sudden data interruption (such as a fault at a certain station), the complete time series can still be reconstructed through the data of associated stations (such as within 50 km upstream / downstream).

[0012] Extract rules through the historical scheduling case library (such as the ecological water demand weight in the dry season ≥ 0.6), transform empirical knowledge into linear programming constraint conditions, and improve the adaptability of the scheduling scheme of the model in the case of sudden drought by 35%; introduce real-time inflow prediction (such as the ARIMA model) into the dynamic balance equation of reservoir water storage, and combine the water volume transfer matrix of the river runoff continuity equation (such as the main stream propagation time correction coefficient) to reduce the water volume allocation error of the joint scheduling of reservoir groups from 15% of the traditional model to 5%; set ecological flow threshold intervals according to the flood season / non-flood season (such as ≥ 10 m in the non-flood season)3 / s, during the flood season ≥ 15m 3 / s), to avoid the problems of over - protection or under - protection of the ecology caused by fixed thresholds.

[0013] Through scenario simulations such as light rain and heavy rain, identify reaches with ecological flow gaps (such as the risk of stream - flow interruption in the lower reaches of a tributary) 24 hours in advance, and match the optimal water replenishment path based on the historical case library (such as a detour path across 3 reservoirs). The response speed is increased by 80% compared with manual decision - making; use the NSGA - II algorithm to generate the Pareto - front solution set, realizing the balanced optimization of water replenishment time, water transfer volume, and ecological compliance rate (for example, increasing the time by 2 hours can reduce the water transfer volume by 30%). The comprehensive cost of the plan is reduced by 25%; through case feature vector matching (such as the volume of flow gap, urgency level), screen the historical optimal strategies to avoid repeated trial - and - error, and the success rate of the scheduling plan is increased by 50%.

[0014] Based on data from three monitoring points of ecology, flood control, and irrigation areas (such as the ecological flow fluctuates by ±2m per hour 3 / s), trigger the correction of the water replenishment time sequence (such as starting water replenishment 3 hours in advance), increasing the ecological flow compliance rate from 70% to 90%; when the rising rate of the water level at the flood - control section exceeds the limit (such as 0.1m / h), dynamically adjust the water replenishment path (such as avoiding the high - risk section of the main stream and switching to the flood - diversion channel of the tributary), reducing the dam - break risk by 40%; according to the water - level fluctuation in the irrigation area channel (such as the daily fluctuation exceeds ±0.5m), inversely correct the weight distribution of the water replenishment volume (such as reducing the priority coefficient by 0.2), reducing water resource waste by about 15%; encode the correction plan into JSON - format instructions and directly interface with the sluice control system, reducing the manual coding error rate (from 5% to 0.1%); through the standardized instruction protocol (such as MQTT communication), synchronously drive the linkage of multiple facilities such as reservoir gates, irrigation area pumping stations, and flood - diversion sluices, shortening the instruction execution delay from the hour - level to the minute - level.

[0015] In a preferred embodiment of the present invention, the heterogeneous data includes: Through satellite remote - sensing monitoring terminals, hydrological station data - collection modules, and irrigation - area Internet - of - Things sensors, simultaneously obtain 9 types of dynamic data including the water - level elevation data, water - quality parameters, real - time runoff data of 15 reservoirs in the basin, as well as the water consumption of the irrigation area, the water demand of the ecological protection area, and flood - control scheduling instructions. Among them, the water - level elevation data includes two data formats: the absolute elevation value based on the altitude reference plane and the relative depth value based on the dam - body reference point.

[0016] In the embodiment of the present invention, the heterogeneous data processed in this step comes from three different types of terminals and modules, covering 9 types of dynamic data, specifically including: Obtain water level elevations (covering large areas of water) and water quality parameters (such as suspended sediment concentration, chlorophyll content) in the macroscopic scope of the basin. Collect water level elevation data of the reservoir (including two formats: absolute elevation values based on the sea level datum, such as the Huanghai Elevation Datum; relative depth values based on the reference point of the dam body, such as 10 meters below the dam crest), real-time runoff data (such as inflow / outflow discharge); Monitor the water consumption of the irrigation area in real time (such as the water intake for farmland irrigation), the water demand of the ecological protection area (such as the minimum flow rate to maintain wetland ecology), and flood control dispatching instructions (such as gate opening and closing instructions, warning water level thresholds).

[0017] Most satellite remote sensing data are in raster format, hydrological station data are tabular structured data (such as CSV), and Internet of Things sensor data are real-time data streams (such as JSON). Taking "water level elevation" as an example, satellite remote sensing and hydrological stations may use absolute elevation (Huanghai Elevation Datum) and relative depth (reference point of the dam body) respectively. If directly mixed, it will lead to confusion in the datum (such as mistakenly taking the relative depth of the dam body as the altitude). There are differences in the monitoring frequencies (once a day for satellite remote sensing, once a second for sensors) and spatial coordinate systems (such as WGS84 and Beijing 54 coordinate systems) of different terminals, and they need to be unified to the basin-level spatio-temporal datum.

[0018] Determine core entities such as "reservoir", "water level", and "elevation datum". For example: "Reservoir water level - storage capacity relationship": Define the mathematical mapping between the water level elevation (absolute / relative) and water storage volume of a certain reservoir (such as when the water level rises by 1 meter, the storage capacity increases by 1 million cubic meters); Set the priority order of agricultural irrigation, ecological protection, and industrial water use (such as the priority of ecological water demand is higher than irrigation); For the "water level elevation" field, force the conversion to the Huanghai Elevation Datum through the rules of the ontology library: If the data is "relative depth value" (such as 10 meters below the reference point of the dam body), it is necessary to combine the Huanghai Elevation value of the reference point of the dam body (such as the Huanghai Elevation of the dam crest is 150 meters, and the reference point of the dam body is 5 meters below the dam crest, that is, 145 meters), and calculate the absolute elevation (145 meters - relative depth 10 meters = 135 meters Huanghai Elevation); Unify the time datum to the UTC+8 time zone and the spatial coordinate system to the National 2000 Geodetic Coordinate System to ensure the spatio-temporal alignment of cross-terminal data.

[0019] Extract data field names through natural language processing (NLP), such as identifying keywords such as "water level", "elevation", and "depth", and matching the entity definitions in the ontology library. For example, when the field name is "water level", judge whether it is "absolute elevation" or "relative depth" based on the data source (hydrological station / sensor) and unit (meter / meters below). To avoid confusion, convert unstructured data (such as satellite remote sensing metadata) and semi-structured data (sensor JSON) into a unified relational data table (such as reservoir basic information table, real-time monitoring table), and the fields include: Spatio-temporal dimension: monitoring time (accurate to minutes), spatial coordinates (longitude, latitude, altitude); Business dimension: water level elevation (unified to Huanghai Elevation), water quality parameters (such as pH value, conductivity), runoff data (m 3 / s), water consumption (10,000 m 3 ), water demand (m 3 / s), and dispatching instructions (parsed from text into opening / closing status and threshold values).

[0020] Establish a standardized data pool under a unified spatio-temporal coordinate system: Downsample high-frequency sensor data (once per second), such as taking a 5-minute average, and perform time interpolation on low-frequency satellite data (once per day) to ensure that all data is aligned at 15-minute intervals, forming a time-series dataset with equal time intervals; divide the basin into 1 km × 1 km grids, and map discrete site data (hydrological stations, sensors) to grid nodes through spatial interpolation (such as Kriging interpolation), unifying the spatial resolution with satellite remote sensing raster data.

[0021] Remove obvious outliers (such as water level elevation exceeding the maximum designed elevation of the reservoir), and perform logical verification through business rules in the ontology library (such as "reservoir capacity ≤ designed reservoir capacity"); utilize the spatio-temporal correlation of hydrological data within the basin. When data at a certain site (such as reservoir A) is interrupted due to a fault, reconstruct the missing values by using data from associated sites within 50 km upstream and downstream (such as upstream reservoir B and downstream hydrological station C), and combining with a hydraulic diffusion model to simulate the water flow propagation process.

[0022] Example: If the water level data of reservoir A is missing at time t, given the outflow of upstream reservoir B at time t - 1 and the water level of downstream hydrological station C at time t + 1, calculate the propagation time of water flow from B to A and the attenuation coefficient from A to C through the hydraulic diffusion model, and infer the water level of A at time t.

[0023] Traditional linear interpolation (such as the average of adjacent times) only utilizes the time-series correlation and ignores the spatial association, with an error of up to ±20%; this method combines spatio-temporal dual associations (upstream and downstream spatial distance, water flow propagation time), reducing the interpolation error to ±5%. Especially in the scenario of sudden data interruption (such as a single-site failure), it can still reconstruct a complete time series through data from 3 - 5 surrounding associated sites, avoiding the impact of data gaps on subsequent model calculations. After the above processing, a standardized data pool containing 15 reservoirs and 9 types of dynamic data is formed, with the following characteristics: All data is stored in Parquet format, supporting efficient querying and analysis. Field names, units, and benchmarks are completely consistent (e.g., "water level elevation" is uniformly "Huanghai Elevation, unit: meter"). Each data record contains an accurate timestamp (e.g., 2024-05-02 08:15:00) and spatial coordinates (longitude, latitude, altitude), enabling spatio-temporal dimensional aggregation analysis (e.g., statistical analysis of total runoff by watershed sub-region and time window). Missing values are repaired through spatio-temporal correlation algorithms, and the data integrity rate is ≥95%, providing high-quality input for subsequent spatio-temporal coupling model construction.

[0024] In another preferred embodiment of the present invention, in step 1, a semantic parsing model based on a knowledge graph is constructed to perform structured conversion on heterogeneous data. By parsing the semantic ambiguity of key fields through a preset watershed ontology library, a standardized data pool under a unified spatio-temporal coordinate system is established. The missing values are processed using an upstream-downstream station spatio-temporal correlation interpolation algorithm to obtain a standardized data pool, which may include: Eliminate field semantic conflicts through the watershed ontology library (such as the confusion of absolute / relative benchmarks for "water level elevation"), convert unstructured / semi-structured data into structured data in a unified format, establish a time coordinate system and a spatial coordinate system that are consistent across the entire watershed, and use spatio-temporal correlation algorithms to improve data integrity.

[0025] Satellite remote sensing uses the Huanghai Elevation benchmark (absolute elevation, such as 120m). Hydrological stations may use the relative depth of the dam body reference point (such as 10m below the dam crest, which needs to be converted to absolute elevation). Sensor data is in UTC time, while hydrological stations are in local time, and they need to be unified to the UTC+8 time zone. The storage structures of raster data, tabular data, and real-time stream data are significantly different.

[0026] Construct a watershed ontology library containing entities, relationships, and rules through an ontology modeling tool (such as Protege). The core content includes: reservoirs (ID, location, designed storage capacity), monitoring points (type, coordinates), sensors (number, acquisition frequency); water level elevation (attributes: benchmark type {absolute / relative}, unit m); runoff data (attributes: flow direction {inflow / outflow}, unit m³ / s); water demand (attributes: usage {ecological / irrigation}, priority level).

[0027] Define the function storage capacity = f(water level elevation, reservoir ID). For example, the water level-storage capacity curve of reservoir A is storage capacity = 50 × water level - 3000 (unit: 10,000 m³); define the "upstream-downstream relationship" (such as reservoir B is 30 km upstream of reservoir A). Conversion from relative depth to absolute elevation: absolute elevation = Huanghai Elevation of the dam body reference point - relative depth (example: Huanghai Elevation of the dam body reference point is 150m, relative depth of 10m to absolute elevation of 140m).

[0028] Use NLP technology to parse data field names, such as identifying "Water level (below the dam crest)" as relative depth and "Water level (Huanghai)" as absolute elevation; combine data source tags (such as "Equipment type = Hydrological station" in field metadata) to eliminate ambiguity; extract pixel values through the GDAL library and convert them into tabular data by appending longitude and latitude coordinates; parse JSON fields and map them to ontology library entities (such as mapping the "flow" field to "Runoff data - Outflow discharge"); use ontology library rules to filter outliers, such as "Water level elevation ≤ Design highest elevation of the reservoir" and "Ecological water demand ≥ 0", and eliminate error data exceeding physical limits. [[ID=?]] [[ID=?]]

[0029] Convert all data timestamps to UTC+8 time zone (Beijing time), accurate to the minute level (such as "2024-05-02 14:15:00"); perform time aggregation on high-frequency data (sensors once per second) (such as 5-minute mean / median), and perform time interpolation on low-frequency data (satellites once per day) (such as linear interpolation to generate 15-minute interval data). [[ID=?]] [[ID=?]]

[0030] Unify satellite data (WGS84) and hydrological station data (Beijing 54) to the National 2000 Geodetic Coordinate System (CGCS2000), and perform banding processing using the Gauss-Krüger projection; assign a unique spatial code to each monitoring point (such as "Basin ID - Grid ID - Equipment ID"), for example, "YRB-01-G01" represents the first sensor in the first grid of the Yangtze River Basin. [[ID=?]] [[ID=?]]

[0031] Divide the entire basin into regular grids of 1km×1km, and interpolate discrete site data (hydrological stations, sensors) to grid nodes through Kriging interpolation or inverse distance weighting method to unify the spatial resolution with satellite raster data (1km). [[ID=?]] [[ID=?]]

[0032] For sudden data interruption scenarios (such as sensor failures), use associated site data within 50km upstream and downstream and hydraulic propagation models to reconstruct missing values. The specific steps are as follows: Based on spatial distance (within 50km) and water flow direction (upstream reservoir, downstream hydrological station), determine 3-5 most relevant sites (such as upstream reservoir B and downstream hydrological station C of reservoir A). Use the ARIMA model to predict the future / historical data trends of associated sites; calculate the water flow propagation time through the hydraulic diffusion model (such as it takes 2 hours for water to flow from reservoir B to reservoir A), and establish a water volume transfer relationship: [[ID=?]] [[ID=?]] , where [[ID=?]] represents the inflow discharge of reservoir A at [[ID=?]] moment. Here, the inflow discharge refers to the amount of water flowing into reservoir A at this moment, usually in cubic meters per second (m 3 / s), which reflects the situation of reservoir A receiving external water inflow at a specific moment; [[ID=?]] represents reservoir B at The outflow at time t, that is, the outflow of reservoir B is earlier than time, passing The amount of water discharged before such a long time interval is also in cubic meters per second (m 3 / s), since it takes time for water to flow from reservoir B to reservoir A, the Data at the moment; It is the time required for water flow to propagate from reservoir B to reservoir A, usually in hours (h). This time can be calculated using a hydraulic diffusion model, which takes into account factors such as the length of the river channel and the water velocity. is the river leakage loss coefficient, which is a value between 0 and 1. It represents the proportion of water lost due to river leakage and other reasons in the process of water flowing from reservoir B to reservoir A. For example, if , which means that 10% of the water is lost during the propagation process.

[0033] When the water level data of reservoir A at time t is missing, the outflow of upstream reservoir B at time t-△t and the water level of downstream hydrological station C at time t+△t' are combined to infer the water level through simultaneous equations: Reservoir water storage balance: ;in, Reservoir A The water storage capacity at a given moment, usually in cubic meters (m 3 ), which reflects the amount of water stored in reservoir A at that moment; Reservoir A is The water storage capacity at the moment is also expressed in cubic meters (m 3 ) as the unit, where The moment is close The previous time point of the moment is used to reflect the time continuity of water storage; Reservoir A is The inflow flow at the moment, in cubic meters per second (m 3 / s), which includes all the water flowing into reservoir A, in addition to the flow from upstream reservoir B, there may also be inflows from other tributaries, etc. Reservoir A The outbound flow at the moment, the unit is also cubic meters per second (m 3 / s). It represents the amount of water flowing out of reservoir A at that moment, such as water released for irrigation, power generation, etc.

[0034] River runoff continuity: ;in, The meaning is the same as that in the reservoir water balance formula, that is, the outflow of reservoir A at time t, in cubic meters per second (m3 / s); is the flow rate corresponding to the water level monitored by the downstream hydrological station C at time, with the unit of cubic meters per second (m 3 / s). Since it takes time for the water flow to travel from reservoir A to the downstream hydrological station C, the data at time is used; represents the time required for the water flow to propagate from reservoir A to the downstream hydrological station C, and the unit is generally hours (h). Similarly, this time is also calculated through relevant models; β is the attenuation coefficient of the downstream river channel, which is a value between 0 and 1, and it reflects the proportion of flow attenuation caused by factors such as river channel resistance, evaporation, and leakage during the process of the water flow from reservoir A to the downstream hydrological station C.

[0035] Among them, the river channel leakage loss coefficient , value range: The river channel leakage loss coefficient The value range is between 0 and 0.2, but under different river channel conditions, this range will fluctuate; in the case of a smaller value (0 to 0.05), when the river channel is an artificial lined channel, such as lined with materials like concrete and grouted stone, the anti-seepage performance of the river channel is better, and the leakage loss of the water flow during propagation is smaller. At this time The value is usually in the range of 0 to 0.05. For example, for the landscape river channels in the city, in order to reduce water resource waste, the river channels are often lined, and their leakage loss coefficients will be in this lower range. The geological conditions at the bottom and both banks of the river channel being impermeable layers, such as clay and shale, will also greatly reduce the leakage volume. The value will also be in this smaller range. In the case of a medium value (0.05 - 0.15), in natural river channels, if the soil texture of the river channel is relatively dense, such as loam, and there are no obvious fissures, karst caves, etc., the water flow leakage situation is relatively moderate. The value will be between 0.05 and 0.15. Some river channels that have been renovated to a certain extent but not fully lined may also be in this value range.

[0036] In the case of a larger value (0.15 - 0.2), when the bottom and both banks of the river channel are sandy soil or there are many permeable channels such as fissures and karst caves, the water flow is prone to leakage. The value will be in this higher range of (0.15 - 0.2). For example, in some mountain rivers, due to complex geological structures and many rock fissures, the water flow leakage situation is relatively serious.

[0037] When determining Before determining the value, it is necessary to conduct a detailed investigation of the geological conditions of the river channel to understand information such as the soil type and geological structure at the bottom and on both banks of the river channel. For example, obtain the distribution of underground soil layers through methods such as drilling and geophysical prospecting to judge the possibility and degree of leakage; if there are long-term hydrological monitoring data for this river channel, the leakage situation in historical data can be referred to determine the value. For example, analyze the water level changes in different seasons and at different flow rates in the past few years, calculate the corresponding leakage volume, and thus estimate the value; factors such as soil moisture and groundwater level in different seasons will affect the leakage situation of the river channel. During the rainy season, the soil moisture is relatively large, the groundwater level is relatively high, and the leakage volume of the river channel is relatively small; while in the dry season, the soil is dry, the groundwater level is relatively low, and the leakage volume will be relatively large. Therefore, when determining the value, seasonal factors need to be considered, and different values may need to be adopted for different seasons.

[0038] Downstream river channel attenuation coefficient , value range: Downstream river channel attenuation coefficient The value range is between 0 and 0.1, but it will vary in different actual scenarios.

[0039] When the value is relatively small (0 - 0.03), when the cross-sectional area of the river channel is relatively regular, the water flow is smooth, there are no obvious obstacles and bends, and the roughness coefficient of the river channel is relatively small (such as a regulated artificial river channel), the energy loss of the water flow is small, the value is usually between 0 and 0.03, the water flow velocity in the downstream river channel is relatively fast, and the length of the river channel is relatively short, so the attenuation degree of the water flow during propagation is relatively low, and it will also be in this relatively small range.

[0040] When the value is medium (0.03 - 0.07), in a natural river channel, if the river channel has certain bends and undulations, but the overall shape is relatively stable, there will be a certain amount of energy loss during the propagation of the water flow, the value will be between 0.03 and 0.07. When there are some small obstacles in the river channel, such as reefs and trees, but the impact on the water flow is not particularly large, it will also make fall within this value range.

[0041] When the value is relatively large (0.07 - 0.1), when the cross-sectional area of the river channel changes greatly, there are many bends and bifurcations, and the roughness coefficient of the river channel is relatively large (such as a river channel overgrown with waterweeds), the energy loss of the water flow is large, The value will be in the relatively high range of 0.07 - 0.1. There are large obstacles in the downstream river channel, such as large bridges, dams, etc., which will cause great obstruction to the water flow, increasing the attenuation degree of the water flow. It will also increase correspondingly.

[0042] Conduct a detailed survey of the topography of the downstream river channel, including information such as the length, width, depth, bend radius, and cross-sectional shape of the river channel. Through these data, the obstruction degree of the river channel to the water flow can be analyzed, so as to reasonably determine the value. Understanding the characteristics of water flow velocity, flow rate, water level, etc. is very important for determining the \(\beta\) value. For example, measure the water flow velocity through devices such as current meters, analyze the water flow changes under different flow rates, so as to evaluate the attenuation degree of the water flow, evaluate the obstacles in the downstream river channel, including the size, quantity, distribution location, etc. of the obstacles. The obstacles will increase the resistance of the water flow and cause the attenuation of the water flow. Therefore, it is necessary to adjust the value according to the specific situation of the obstacles. At the same time, the change situation of the obstacles also needs to be considered, such as the seasonal growth of aquatic plants and the accumulation of floating objects during the flood period.

[0043] After the above processing, a standardized data pool with the characteristics of "three unifications" is formed: all data is stored in the Parquet columnar storage format, supporting efficient spatio-temporal queries. The fields include: basic attributes: data ID, monitoring point code, data source type, timestamp (YYYY - MM - DD HH:mm), longitude, latitude, altitude (Huanghai elevation); business attributes: water level elevation (m), runoff data (m 3 / s), water quality parameters (such as pH value), water demand (m 3 / s), scheduling instructions (digital code, such as 1 = opening the sluice, 0 = closing the sluice). Time: UTC + 8 time zone, 15 - minute time interval; Space: China Geodetic Coordinate System 2000, 1km grid resolution; Unit: unified to the International System of Units (meter, cubic meter, second, etc.), missing value repair rate ≥ 95%, ensuring the temporal continuity in the case of sudden data interruption through spatio-temporal correlation algorithms.

[0044] In a preferred embodiment of the present invention, in step 2, based on the standardized data pool output in step 1, combined with the scheduling rule constraint conditions in the historical scheduling case library, a spatio-temporal coupling model including a dynamic balance equation of reservoir storage volume, a continuity equation of river channel runoff, and an ecological flow threshold constraint is established, including: Step 21, based on the reservoir water level, runoff, and water demand data calibrated in space and time in the standardized data pool, extract the scheduling rule constraint conditions of two typical scenarios of preferential supply guarantee during the dry season and flood control peak shifting during the flood season in the historical scheduling case library; Step 22: Using the real-time water storage of 15 reservoirs in the standardized data pool as input variables, calculate the adjustable storage capacity of each reservoir during the scheduling period through the dynamic balance equation; Step 23: Based on the river channel runoff continuity equation, construct a water volume transfer relationship matrix between the main stream and tributaries, and determine the water volume conduction time parameters for each river section; Step 24: Convert the water demand of the ecological protection area and the canal head of the irrigation area into the ecological flow threshold constraint conditions for the corresponding river sections, and establish a spatio-temporal coupling model with the joint scheduling of the reservoir group as the core, the water volume transfer of the river channel as the link, and the compliance of the ecological flow as the boundary. The ecological flow threshold sets a dynamic threshold range according to the flood season and non-flood season.

[0045] In the embodiment of the present invention, convert the water supply guarantee rules (such as the minimum discharge flow threshold) in the dry season and the flood control rules (such as the pre-discharge storage capacity ratio) in the flood season in historical cases into quantifiable constraint conditions, so that the scheduling model inherits the experience of human experts, and the decision-making reliability in extreme hydrological years (such as once-in-a-century drought) is increased by 40%; through time series matching and case clustering (such as using the DTW algorithm to screen cases with similarity > 0.8), dynamically load the rule set matching the current meteorological conditions, so that the switching efficiency of the scheduling strategy of the model during the season transition period (such as before the flood season) is increased by 50%; for example, the quantitative allocation of the irrigation water use priority coefficient and the ecological water demand weight in the dry season cases (such as 0.6:0.4) to avoid the risk of downstream stream cutoff caused by subjective preferences in traditional experience-based decision-making.

[0046] Based on the real-time water storage and the dynamic balance equation (such as the combined calculation of inflow - outflow - evaporation and seepage), predict the adjustable storage capacity in the next 24 hours, and reduce the error rate from ±15% of the traditional static storage capacity estimation to ±5%, supporting refined water volume allocation; in the flood season scenario, through the pre-discharge storage capacity ratio constraint (such as 20% of the total storage capacity), ensure the reservation of the flood control safety storage capacity, and at the same time dynamically release the adjustable storage capacity for ecological water replenishment, achieving a 30% increase in the storage capacity utilization rate; for the 15-reservoir group, formulate a cross-reservoir water borrowing strategy through the spatio-temporal distribution calculation of the adjustable storage capacity (such as the contribution weights of the upper / middle / lower reaches of the reservoir), reducing the risk of overloading a single reservoir in the scheduling.

[0047] Through the node - river section connection matrix and the conduction time parameters (such as the main stream propagation speed of 2 m / s and the tributary propagation speed of 1.5 m / s), quantify the water volume transfer delay effect, and increase the synchronization rate of the water replenishment instruction and the flow peak from 60% to 90%; for the river sections regulated by sluices and dams, dynamically correct the conduction time according to the scheduling log (such as the propagation time is shortened by 30% after the gate is opened), avoiding prediction deviations caused by the model ignoring engineering interventions; map the main stream, tributaries, and reservoir nodes of the basin into a matrix topology diagram to visually display the bottleneck of the conduction efficiency of the water replenishment path (such as a certain tributary's conduction time accounting for 70% of the whole journey), supporting path optimization decisions.

[0048] The monthly average water demand of the ecological protection zone (e.g. 500,000 m 3 ) is converted to the lower limit threshold of river flow (e.g. ≥8m in non-flood season) 3 / s), and enforce compliance with the model boundary conditions, thereby increasing the annual ecological flow compliance rate from 75% to 95%; setting dynamic intervals according to the flood season (threshold value increased by 20%) and the non-flood season (threshold value decreased by 10%) to avoid ecological over-protection (such as excessive discharge restrictions in the rainy season) or under-protection (such as insufficient water replenishment in the dry season) caused by fixed thresholds; using river water transfer as a link, coupling reservoir scheduling instructions (decision variables) and ecological flow compliance (boundary conditions) to achieve a Pareto equilibrium among flood control, water supply, and ecological goals (for example, a 10% increase in water replenishment costs can be exchanged for a 25% increase in ecological compliance).

[0049] In a preferred embodiment of the present invention, the scheduling rule constraints include: Step 211: Based on the spatiotemporally calibrated reservoir water level, runoff, and water demand data in the standardized data pool, perform morphological matching on the time series of current hydrological characteristics and the meteorological and hydrological characteristic parameters of the dry season and flood season in the historical scheduling case library, calculate the similarity index between the sequences, and select the corresponding case set based on the similarity index; Step 212: For the screening case set under the dry season priority supply guarantee scenario, the 5% quantile of the reservoir's minimum discharge flow constraint is analyzed and extracted as a safety boundary value, the weight distribution matrix of the irrigation district water use priority coefficient is calculated, and the water demand weight adjustment coefficient is determined based on the frequency distribution of historical water shortage events in the ecological protection zone. For the screening case set under the flood control peak shifting scenario, the median of the pre-discharge ratio is extracted as a benchmark parameter based on the distribution of the ratio of the reservoir's pre-discharge capacity to the total storage capacity. The probability density of the duration of the peak shifting scheduling time window is calculated using a time sliding window, and the upper limit of the flood diversion flow threshold is set based on the maximum flow capacity of the tributary flood diversion gate. Step 213, feature clustering is performed on the dry season supply parameters and the flood season flood control parameters to obtain clustering results; a set of constraints for the linear programming model is generated based on the clustering results, wherein the dry season scenario constraints include the weighted sum of the irrigation area water use priority coefficients ≥ 0.8, the product of the ecological water demand weight and the minimum downstream flow ≥ the design threshold, and the flood season scenario constraints include the pre-discharge storage capacity ratio × total storage capacity ≤ flood control safety storage capacity, and the diversion flow ≤ tributary diversion threshold × time window adjustment coefficient.

[0050] In a preferred embodiment of the present invention, for the above-mentioned step 211, case screening and similarity calculation, the implementation process is as follows: Based on the reservoir water level, runoff, and water demand data after spatio-temporal calibration in the standardized data pool, perform morphological matching on the time series of the current hydrological characteristics and the meteorological and hydrological characteristic parameters in the historical operation case library during the dry season and flood season. This means comparing the current hydrological data with the meteorological and hydrological data in different historical periods (dry season and flood season) to find similar patterns; calculate the similarity index between the sequences. The similarity index can be calculated through various algorithms, such as the dynamic time warping (DTW) algorithm. This algorithm can measure the similarity between two time series, even if they have different lengths or time offsets; screen out the corresponding case set from the similarity index. Usually, a similarity threshold (such as 0.8) is set, and only cases with a similarity higher than this threshold will be selected to form the corresponding case set.

[0051] By screening similar historical cases, it is possible to provide a reference for the current operation decision-making, utilize past experience to deal with similar hydrological situations, and similarity calculation helps to more accurately find the case that best matches the current situation, improving the pertinence and effectiveness of the decision-making.

[0052] For the above-mentioned step 212, parameter extraction and calculation, the implementation process is as follows: For the case set screened under the scenario of ensuring water supply first during the dry season, analyze and extract the 5% quantile of the minimum discharge constraint of the reservoir as the safety boundary value. This means finding the 5% quantile of the minimum discharge in all the screened dry-season cases and using it as a safety boundary to ensure that the reservoir discharge does not fall below this value in similar dry-season situations, thus guaranteeing the basic water demand of the downstream.

[0053] Calculate the weight distribution matrix of the irrigation area water use priority coefficient. According to the importance and demand of different irrigation areas, determine their priority coefficients in water use allocation and form the weight distribution matrix. This helps to reasonably allocate water resources and give priority to meeting the needs of important irrigation areas in the case of limited water resources.

[0054] Determine the water demand weight adjustment coefficient based on the frequency distribution of historical water shortage events in the ecological protection area. By analyzing the occurrence frequency of historical water shortage events in the ecological protection area, determine an adjustment coefficient to adjust the weight of ecological water demand. If water shortage events have occurred frequently in history, it may be necessary to increase the weight of ecological water demand to better protect the ecological environment.

[0055] For the screening case set in the flood season flood control peak shaving scenario, extract the median of the pre-discharge ratio as the benchmark parameter according to the ratio distribution of the reservoir pre-discharge storage capacity to the total storage capacity. Among all the screened flood season cases, find the median of the ratio of the pre-discharge storage capacity to the total storage capacity and use it as the benchmark parameter for the pre-discharge ratio. This helps to determine how much storage capacity the reservoir should release in advance during the flood season to cope with possible floods.

[0056] Statistically analyze the probability density of the duration of the peak shaving scheduling time window through a time sliding window. Use a time sliding window to statistically analyze the probability density distribution of the duration of the peak shaving scheduling time window, which can help understand how long the peak shaving scheduling needs to last under different circumstances, so as to better arrange the scheduling strategy.

[0057] Set the upper limit of the flood diversion flow threshold according to the maximum flow capacity of the tributary flood diversion gate. Based on the actual maximum flow capacity of the tributary flood diversion gate, set an upper limit for the flood diversion flow threshold to ensure that during the flood diversion process, the gate's bearing capacity will not be exceeded, guaranteeing flood control safety.

[0058] The extracted parameters and calculated coefficients can more accurately reflect the water resource requirements and limitations under different scenarios, providing more specific and practical constraint conditions for the scheduling model. The parameters and coefficients determined based on historical data and actual situations help improve the scientificity and rationality of scheduling decisions and reduce decision-making errors caused by subjective factors.

[0059] The above step 213, constraint condition generation, implementation process: Conduct feature clustering on the dry season water supply guarantee parameters and flood season flood control parameters respectively. Use a clustering algorithm (such as K-Means clustering) to conduct clustering analysis on the dry season water supply guarantee parameters and flood season flood control parameters, and classify similar parameters into one category.

[0060] Generate a set of constraint conditions for the linear programming model according to the clustering results. For the dry season scenario, the constraint conditions may include the weighted sum of the irrigation area water use priority coefficients ≥ 0.8, the product of the ecological water demand weight and the minimum downstream discharge ≥ the design threshold, etc. These constraint conditions ensure that during the dry season, the irrigation area water use priority is guaranteed, and at the same time, the ecological water demand can also meet certain requirements. For the flood season scenario, the constraint conditions may include the pre-discharge storage capacity ratio × the total storage capacity ≤ the flood control safety storage capacity, the flood diversion flow ≤ the tributary flood diversion threshold × the time window adjustment coefficient, etc. These constraint conditions ensure that during the flood season, the reservoir can reserve enough flood control safety storage capacity in advance, and at the same time, the flood diversion flow is also within the safe range.

[0061] Clustering analysis helps to discover the internal laws and similarities among parameters, thus generating more reasonable constraint conditions. The generated set of constraint conditions can transform historical experience and actual needs into quantifiable constraints, providing clear guidance for the scheduling model and improving the decision-making reliability and efficiency of the model.

[0062] Through the implementation of steps 211 to 213, the present invention realizes the extraction of quantifiable constraint conditions from the historical scheduling case base and applies them to the scheduling model. This not only enables the scheduling model to inherit the experience of human experts, improves the decision-making reliability and the switching efficiency of scheduling strategies during extreme hydrological years and seasonal transition periods, but also avoids the risk of downstream stream cutoff caused by subjective preferences in traditional empirical decision-making. At the same time, through the accurate extraction and calculation of various parameters and the reasonable generation of constraint conditions, more refined water volume allocation can be achieved, the reservoir capacity utilization rate can be improved, flood control safety can be ensured, the ecological flow compliance rate can be increased, and the Pareto equilibrium of flood control, water supply, and ecological goals can be realized.

[0063] In a preferred embodiment of the present invention, in step 22, taking the real-time water storage of 15 reservoirs in the standardized data pool as the input variable, the adjustable reservoir capacity of each reservoir within the scheduling period is calculated through the dynamic balance equation, including: Step 221, taking the real-time water storage of 15 reservoirs in the standardized data pool as the initial input, and combining the minimum downstream discharge flow or pre-discharge reservoir capacity ratio parameter in the scheduling rule constraint conditions extracted in step 21, a dynamic balance equation is constructed; The dynamic balance equation takes the predicted inflow and the controlled outflow of each reservoir within the scheduling period as the core variables, and determines the maximum adjustable reservoir capacity that meets the flood control or water supply constraint conditions by iteratively calculating the reservoir water storage change curve; among them, the predicted inflow is dynamically corrected through the water volume transfer relationship matrix of the upstream and downstream reservoir groups, and the controlled outflow is calculated by segmenting the limit according to the priority of the ecological flow threshold and the irrigation area water demand.

[0064] In the embodiment of the present invention, the specific implementation process of step 221 is as follows: Extract the current real-time water storage (unit: cubic meters) of 15 reservoirs from the standardized data pool constructed in step 1. These data have been calibrated in space and time (unifying the Huanghai elevation datum and the National 2000 coordinate system) and missing value repaired (space-time correlation interpolation algorithm) to ensure unified data format, consistent datum, and no significant missing or abnormal values. The real-time water storage is stored as an initial vector according to the reservoir number (such as R01 - R15) as the starting point data of the dynamic balance equation. For example, the current water storage of reservoir R01 is 12 million cubic meters, and that of R02 is 8 million cubic meters, etc.

[0065] Through the case similarity matching in step 211, it is judged whether the current situation belongs to the water supply guarantee scenario in the dry season or the flood control scenario in the flood season. For example, if the current reservoir water storage is lower than the multi-year average water level and the irrigation area water demand is high, it is determined as the dry season scenario, and the corresponding minimum downstream discharge flow parameter is called (such as the minimum downstream discharge flow of a certain reservoir in the dry season is 5m 3 / s); If it is the rainy season and heavy rain is forecasted in the next 24 hours, it is determined as the flood season scenario, and the pre-discharge storage ratio parameter is called (for example, 20% of the total storage is required to be pre-discharged as the flood control safety storage); The minimum discharge of each reservoir is used as the lower limit constraint of the outflow (that is, the outflow cannot be lower than this value to ensure the basic water use of the downstream ecology and irrigation areas), and the pre-discharge storage of each reservoir is calculated (pre-discharge ratio × total storage), which is used as the upper limit constraint of the water storage (that is, the water storage within the scheduling period shall not exceed "total storage - pre-discharge storage" to ensure the flood control safety space); The main-stream to tributary water transfer relationship matrix constructed in step 23 is called, and this matrix records the water transfer time (for example, the outflow of reservoir R03 takes 3 hours to reach the downstream reservoir R05) and the loss coefficient (for example, the river seepage loss rate α = 0.1) between the upstream and downstream reservoirs.

[0066] For example, when calculating the inflow of reservoir R05 at the t-th hour in the future, it is necessary to extract the outflow of its upstream reservoir R03 at t - 3 hours, and after deducting 10% of the seepage loss, it is used as the predicted value of the inflow of R05 (that is, "upstream outflow × (1 - α)").

[0067] Combined with real-time meteorological forecast data (such as rainfall, evaporation) and historical runoff data, the predicted value is dynamically adjusted. For example, if heavy rain is forecasted in the next 1 hour, the outflow of the upstream reservoir R03 may increase temporarily, and the predicted inflow of R05 needs to be updated synchronously. According to the irrigation area water use priority coefficient (such as irrigation area G01 priority 0.6, G02 priority 0.4) and ecological water demand weight (such as ecological weight ≥ 0.6 in the dry season) extracted in step 212, the outflow distribution order is determined: ecological flow threshold > high-priority irrigation area > low-priority irrigation area.

[0068] If the ecological flow threshold of the current river section is 10m 3 / s, the outflow first meets this lower limit (such as outflow ≥ 10m 3 / s). After meeting the ecological flow, the remaining adjustable water volume is allocated according to the irrigation area priority. For example, the total allocable water volume is 50m 3 / s, G01 gets 50 × 0.6 = 30m 3 / s, G02 gets 20m 3 / s, the outflow shall not exceed the maximum flow capacity of the gate (such as 500m 3 / s), and the flow upper limit corresponding to the pre-discharge storage needs to be reserved (such as the outflow is required to be ≥ 1.5 times that of the normal period during the pre-discharge period).

[0069] The scheduling period is usually 24 hours, and the time step is set to 1 hour (which can be adjusted according to the accuracy requirements, such as 15 minutes). For example, starting from the current moment t = 0, the water storage at each moment of t = 1, t = 2... t = 24 is calculated.

[0070] Single-step iteration logic: Input: water storage at time t-1 (initially the real-time water storage), predicted inflow at time t, and controlled outflow at time t.

[0071] Change in water storage = Inflow × Time step - Outflow × Time step (ignoring minor losses such as evaporation and seepage, or uniformly deducting the loss coefficient through historical data fitting); Water storage at time t = Water storage at time t-1 + Change in water storage; If the water storage exceeds the designed maximum reservoir capacity, the outflow is forced to be corrected (increase the discharge); If the water storage is below the dead storage (non-adjustable lower limit), an alarm is triggered and the outflow is frozen (only maintaining the minimum discharge), generating a curve of water storage change for each reservoir within 24 hours (e.g., for Reservoir R03, the water storage drops to 15 million cubic meters at t = 6 hours and rises to 18 million cubic meters at t = 12 hours), visually showing the water volume fluctuations during the scheduling period.

[0072] Refers to the maximum range within which a reservoir can be flexibly scheduled (increase or decrease water storage) under the condition of meeting flood control / water supply constraints. The calculation formula is: Adjustable storage capacity = Maximum allowable water storage - Minimum allowable water storage (where "Maximum allowable water storage" is "Total storage - Pre-discharge storage" during the flood season and the designed maximum storage capacity during the dry season; "Minimum allowable water storage" is the dead storage or the water storage corresponding to meeting the minimum discharge).

[0073] Traverse the curve of water storage change obtained by iterative calculation to find whether the water storage at each moment touches the constraint boundary: If the water storage of a certain reservoir reaches "Total storage - Pre-discharge storage" (upper limit during the flood season) at t = 8 hours, the adjustable storage capacity at this moment is the difference between the current water storage and the dead storage; If during the dry season, the water storage of a certain reservoir is always higher than the dead storage and meets the minimum discharge, the adjustable storage capacity is the maximum value of "Current water storage - Dead storage" (usually appears at the start of the scheduling period). Combining the joint scheduling requirements of the reservoir group, the adjustable storage capacity of a single reservoir is corrected collaboratively. For example, if the adjustable storage capacity of the upstream reservoir R01 is large, more water can be allocated to the downstream water-deficient reservoir R07 to avoid overloading a single reservoir during scheduling.

[0074] Through the dynamic balance equation, multiple factors such as real-time water storage, water volume transfer between upstream and downstream, and demand priority are integrated to achieve accurate calculation of the adjustable storage capacity of each reservoir. The error rate is reduced from ±15% of traditional static estimation to ±5%, providing data support for refined water allocation. For different constraint conditions (minimum downstream discharge / pre-discharge storage ratio) in the dry season and flood season, the calculation logic is dynamically adjusted, enabling the model to increase the storage utilization rate by 30% in extreme hydrological scenarios while ensuring flood control safety and water supply stability. Through constraint checks in iterative calculations (such as dead storage protection and gate flow capacity limitation), it is avoided that reservoir operation exceeds the physical limit, reducing the risk of dam break and downstream flow interruption, and enhancing the security of water resources management. Based on the water volume transfer relationship among the reservoir group, coordinated allocation of the adjustable storage capacity across reservoirs (such as weight allocation of upstream / midstream / downstream reservoirs) is realized, reducing the load of a single reservoir and improving the scheduling efficiency and balance of the entire basin.

[0075] The predicted value of the inflow in the dynamic balance equation is dynamically corrected through the water volume transfer relationship matrix, and the controlled value of the outflow is calculated by segmented limiting according to the ecological flow threshold and the priority of irrigation area water demand. This enables the rational allocation of water resources among different demands, ensuring both the water demand for the ecological environment and the production and domestic water demand of the irrigation area, and improving the utilization efficiency of water resources.

[0076] In a preferred embodiment of the present invention, in step 23, a water volume transfer relationship matrix between the main stream and tributaries is constructed based on the river channel runoff continuity equation, and the water volume conduction time parameters of each river section are determined, including: Step 231, according to the river channel runoff continuity equation, map the topological structure of the main stream and tributaries of the basin into a node-river section connection matrix, and the matrix elements represent the water volume conduction coefficient between adjacent river sections; Step 232, based on the historical runoff data and cross-section topographic parameters of each river section in the standardized data pool, calculate the water flow propagation time at different flow levels to determine the water volume conduction time parameters; Step 233, for the river sections with water conservancy project regulation, dynamically compensate and correct the conduction time parameters according to the gate dam operation log data to form a spatio-temporal coupling parameter set including natural conduction and artificial intervention effects.

[0077] In an embodiment of the present invention, for the above step 231 of constructing a node-river reach connection matrix, the specific implementation process is as follows: Conduct a detailed analysis of the topological structures of the main stream and tributaries of the basin. This includes determining information such as the flow direction of the river channel, the confluence positions of tributaries, and the starting and ending points of each river reach. For example, through Geographic Information System (GIS) data or on-site surveys, clarify the specific path of the main stream of a certain basin from the source to the estuary, as well as the confluence points of each tributary on the main stream. Define the key positions in the river channel as nodes, such as the outlet of a reservoir, the confluence point of a tributary, the location of an important hydrological monitoring station, etc. The river channel section between each two nodes is defined as a river reach. For example, the river channel from the outlet of reservoir A to the first tributary confluence point downstream is a river reach. According to the river channel runoff continuity equation, for two adjacent river reaches, analyze the water volume conduction relationship between them. Considering the influence of factors such as the slope, roughness coefficient, and cross-sectional area of the river channel on water volume conduction, determine a coefficient representing the water volume conduction capacity between adjacent river reaches. This coefficient can be obtained through theoretical calculations, empirical formulas, or historical data fitting.

[0078] For example, if the water flow in river reach A is relatively smooth, with a large slope and a relatively stable cross-sectional area, while the water flow in river reach B is relatively gentle and has a large roughness coefficient, then the water volume conduction coefficient between them will be adjusted according to these factors. Assume that through calculation or experience, it is known that the water volume flowing from river reach A to river reach B will decrease by a certain proportion without the influence of other factors, and this proportion is the water volume conduction coefficient.

[0079] Construct a matrix with nodes as rows and columns. The elements in the matrix represent the water volume conduction coefficients between the river reaches corresponding to adjacent nodes. If there is a river reach connection between node i and node j, then the matrix element M[i][j] represents the water volume conduction coefficient from the river reach connected to node i to the river reach connected to node j; if there is no direct river reach connection between node i and node j, then M[i][j] = 0. By constructing the node-river reach connection matrix, the topological relationship and water volume conduction path between the main stream and tributaries in the basin can be clearly described, providing a basic framework for subsequent water volume calculation and scheduling analysis. This matrix-form data structure is convenient for storage and calculation in a computer model, can efficiently process complex river channel network systems, and improve the calculation efficiency and accuracy of the model.

[0080] Step 232: Calculate the water volume conduction time parameter. The specific implementation process is as follows: Obtain the historical runoff data of each river section from the standardized data pool. This data includes information such as the flow rate and water level height at different time points. At the same time, collect the cross-sectional terrain parameters of each river section, such as the width, depth, and shape of the river channel. These data can be obtained through on-site measurement, topographic mapping, or historical materials. According to the range of historical runoff data, divide the flow rate into different levels. For example, the flow rate can be divided into low flow rate level, medium flow rate level, and high flow rate level in ascending order. The specific division criteria can be determined according to the actual situation of the basin and the analysis requirements. For each flow rate level, use the hydraulics principle and relevant formulas, combined with the cross-sectional terrain parameters of the river section, to calculate the propagation speed of the water flow in this river section.

[0081] For example, according to Manning's equation, the average flow velocity of the water flow under different flow rate levels can be calculated. Then, according to the length of the river section, calculate the time required for the water flow to propagate from the starting point to the end point of the river section, that is, the water flow propagation time. Assume that the length of a certain river section is L, and the average flow velocity calculated at a certain flow rate level is V. Then the water flow propagation time T = L / V. Organize and analyze the calculated water flow propagation times under different flow rate levels to determine the water volume conduction time parameters of each river section under different flow conditions. These parameters can be stored in the form of a table or a function for use in subsequent model calculations. For example, a table can be established to record the water volume conduction times of each river section under different flow rate levels.

[0082] By calculating the water volume conduction time parameters under different flow rate levels, the influence of flow rate changes on the water flow propagation time is fully considered, enabling the model to more accurately simulate the actual water flow movement. Accurate water volume conduction time parameters are crucial for constructing an accurate spatio-temporal coupling model. It can more realistically reflect the propagation process of water flow in the river channel, thereby improving the prediction and decision-making capabilities of the model for water resource scheduling and management.

[0083] The above-mentioned step 233: Dynamically compensate and correct the conduction time parameter. The specific implementation process is as follows: For the river sections with water conservancy project regulation, collect the scheduling log data of relevant sluice dams. These data record information such as the opening time, opening degree, and closing time of the sluice dams. For example, a certain sluice dam opened a certain opening degree during a specific period on a specific date to regulate the water volume downstream; According to the sluice dam scheduling log data, analyze the impact of the sluice dam regulation behavior on the water flow propagation time. When the sluice dam is opened, the flow velocity and flow rate of the water will change, thus affecting the water flow propagation time in the river section. Through the analysis of historical data and the application of hydraulics principles, determine the specific impact mode and degree of the sluice dam regulation on the conduction time. For example, by comparing the water flow velocity and propagation time data before and after the sluice dam is opened, it is found that after the sluice dam is opened, the water flow velocity increases and the propagation time is shortened by a certain proportion.

[0084] According to the impact of the sluice dam regulation on the conduction time obtained from the analysis, dynamically compensate and correct the water volume conduction time parameter determined in step 232. For each sluice dam regulation event, calculate the corresponding conduction time correction value according to its specific regulation parameters (such as opening degree, duration, etc.), and apply it to the corresponding water volume conduction time parameter. For example, if it is calculated based on experience or model that the water flow propagation time is shortened by 30% after the sluice dam is opened, then during the sluice dam regulation period, the water volume conduction time parameter of the corresponding river section needs to be adjusted to reflect the effect of this artificial intervention. Integrate the water volume conduction time parameter after dynamic compensation and correction with the parameter under natural conduction conditions to form a spatio-temporal coupling parameter set including natural conduction and artificial intervention effects. This parameter set can comprehensively reflect the spatio-temporal characteristics of water flow propagation in the river channel, including the comprehensive impact of natural factors and artificial water conservancy project regulation on the water flow.

[0085] Through steps 231 to 233, various factors affecting the water flow propagation in the river channel are comprehensively considered, from the description of the river channel topology structure, the calculation of water flow propagation time under different flow levels to the dynamic compensation and correction of water conservancy project regulation, which can more accurately simulate the movement process of the water flow. The accurate water volume conduction time parameter and the spatio-temporal coupling parameter set including artificial intervention effects enable the model to more precisely predict and analyze the distribution and change of water resources in the basin, contribute to the realization of refined water resources management, improve the utilization efficiency of water resources, and ensure flood control safety and ecological environment requirements.

[0086] In a preferred embodiment of the present invention, in step 24, convert the water demand of the ecological protection area and the canal head of the irrigation area into the ecological flow threshold constraint conditions of the corresponding river section, and establish a spatio-temporal coupling model with the joint operation of the reservoir group as the core, the water volume transfer in the river channel as the link, and the ecological flow compliance as the boundary, including: Step 241: Convert the monthly average water demand of the ecological reserve area and the daily water use plan of the canal head of the irrigation area into the lower limit threshold of the ecological flow and the water use flow demand curve for the corresponding river section respectively. Combine the water volume conduction time parameters generated in Step 23 to construct a multi-period coupling model with the joint scheduling instructions of the reservoir group as the decision variable, the water transfer process of the river channel as the state variable, and the compliance with the ecological flow threshold as the boundary condition. Among them, the ecological flow threshold sets a dynamic threshold range according to the flood season and non-flood season scheduling rules extracted in Step 21 at the position of the river channel control section, and establishes a linkage mechanism for triggering the water replenishment strategy when the threshold is exceeded through the emergency scheduling response time parameters in the historical optimal water replenishment strategy case library.

[0087] In the embodiment of the present invention, for the above Step 241 of constructing the multi-period coupling model, the specific implementation process is as follows: Obtain the monthly average water demand data of the ecological reserve area, which can be obtained through long-term ecological monitoring and research. For example, the monthly average water demand of a certain ecological reserve area is 500,000 cubic meters. According to the length of the river section where the ecological reserve area is located, the cross-sectional area of the river channel, and other information, convert the monthly average water demand into the lower limit threshold of the ecological flow for the corresponding river section. For example, after calculation, the lower limit threshold of the ecological flow for the corresponding river section of this ecological reserve area is 8 cubic meters per second (m 3 / s), which means that in order to maintain the ecological function of this ecological reserve area, the flow of this river section should not be lower than this value at any time.

[0088] Collect the daily water use plan data of the canal head of the irrigation area. These data are usually formulated by the irrigation area management department according to factors such as the crop planting area, irrigation method, and meteorological conditions. For example, the daily water use plan of the canal head of a certain irrigation area is 1,000,000 cubic meters of water per day during the irrigation peak period. Convert the daily water use plan into a water use flow demand curve. According to the change of water use demand at different times within a day, allocate the daily water volume to each time period to form a water use flow demand curve. For example, during the irrigation peak period from 8 am to 6 pm, the water use flow demand is relatively high, possibly reaching 15m 3 / s; while at night, the water use flow demand is relatively low, possibly 5m 3 / s. Invoke the water volume conduction time parameters generated in Step 23. These parameters record the time required for water flow to propagate between different river sections. For example, it takes 3 hours for the water flow from Reservoir A to the river section where the ecological reserve area is located, and 2 hours for the water flow from Reservoir B to the river section where the canal head of the irrigation area is located, etc. Considering the time delay of water volume conduction, associate the scheduling instructions of the reservoir group with the water transfer process of the river channel. For example, if Reservoir A plans to increase the outflow flow at 10 am, then according to the water volume conduction time parameters, the river section where the ecological reserve area is located may not feel the change in flow until around 1 pm.

[0089] Decision variables: The joint operation instructions of the reservoir group are used as decision variables. This includes decision factors such as the outflow discharge of each reservoir at different time periods and the water release time. For example, the outflow discharge of Reservoir A is 20 m 3 / s from 8 am to 10 am, and the outflow discharge is adjusted to 25 m 3 / s from 10 am to 12 pm, etc. Taking the water transfer process in the river channel as the state variable, this involves the propagation of water flow in the river channel, the distribution and change of water volume, etc. By considering the topological structure of the river channel, the water volume conduction coefficient, and the water volume conduction time parameter, the dynamic change process of water flow in different river reaches is simulated. For example, according to the node-river reach connection matrix and the water volume conduction time parameter, the water volume of a certain river reach and the situation of water flow propagating downstream at a certain moment are calculated, with the ecological flow threshold compliance as the boundary condition. Ensure that the flow in the river reach where the ecological protection area is located is not lower than the lower limit threshold of the ecological flow at any time period, and at the same time meet the water demand curve of the canal head of the irrigation area. If the flow in the ecological protection area is lower than the threshold during a certain time period, or the water demand of the canal head of the irrigation area cannot be met, the model will trigger the corresponding adjustment mechanism; according to the flood season and non-flood season operation rules extracted in Step 21, set the dynamic threshold range of the ecological flow threshold according to the position of the river control section.

[0090] During the flood season, due to the large rainfall and relatively sufficient river water volume, in order to avoid adverse impacts on the ecological environment caused by excessive restriction of reservoir water release and ensure flood control safety, the ecological flow threshold may be appropriately increased. For example, during the flood season, the ecological flow threshold of a certain river control section increases from 8 m 3 / s usually to 10 m 3 / s. During the non-flood season, the river water volume is relatively small. In order to ensure the ecological water demand, the ecological flow threshold may be appropriately decreased. For example, during the non-flood season, the ecological flow threshold of this river control section decreases to 6 m³ / s. Through the emergency dispatch response time parameter in the historical optimal water replenishment strategy case base, establish a linkage mechanism for triggering the water replenishment strategy when the threshold is exceeded. When the ecological flow monitoring data shows that the flow of a certain river reach is lower than the set ecological flow threshold, the system will determine the emergency dispatch response time according to the data in the historical optimal water replenishment strategy case base. For example, if historical data shows that when the ecological flow of a certain river reach is lower than the threshold, starting the water replenishment strategy within 2 hours can effectively restore the ecological flow, then the system will trigger the water replenishment strategy within 2 hours after detecting that the flow is lower than the threshold.

[0091] After the triggering of the water replenishment strategy, the model will calculate the increased outflow of each reservoir and the water replenishment schedule based on factors such as the adjustable storage capacity of the reservoir group and the water volume conduction time in the river channel, so as to quickly restore the ecological flow above the threshold. The water demand of the ecological protection area and the canal head of the irrigation area is converted into specific flow thresholds and demand curves, which are used as the boundary conditions of the model, ensuring that the water use requirements of the ecological environment and agricultural irrigation are fully considered during the water resources scheduling process, protecting the stability of the ecosystem and the normal progress of agricultural production. A multi-period coupling model is constructed by combining the water volume conduction time parameter, fully considering the time delay of water flow propagation in the river channel and the changes in water use demand in different periods, enabling the model to more accurately simulate the actual water resources scheduling situation, improving the accuracy and reliability of the model. Dynamic threshold intervals are set for the flood season and non-flood season, enabling the model to flexibly adjust the ecological flow threshold according to different hydrological conditions, avoiding the problems of over-protection or under-protection of the ecosystem that may be caused by fixed thresholds in different seasons. By establishing a linkage mechanism for triggering the water replenishment strategy when the threshold is exceeded and using the data in the historical optimal water replenishment strategy case library, a quick response can be made when there are problems with the ecological flow, a reasonable water replenishment strategy can be formulated, the joint scheduling of the reservoir group can be optimized, and the utilization efficiency of water resources and the compliance rate of ecological flow can be improved.

[0092] In a preferred embodiment of the present invention, the scheduling scheme includes: Step 311: Classify the rainfall intensity in the real-time weather forecast data into three scenarios: light rain, moderate rain, and heavy rain. After inputting it into the spatio-temporal coupling model, based on the joint scheduling rules of the reservoir group and the river channel water volume transfer relationship matrix, simulate the changes in reservoir storage capacity, the flow of main and tributary sections, and the ecological protection area flow propagation process hour by hour for each scenario; Step 312: According to the dynamic ecological flow threshold interval set in Step 24, extract the river sections where the flow is continuously lower than the lower limit of the threshold for 3 hours in the simulation results as water shortage river sections, calculate the cumulative flow gap of each water shortage river section in combination with the water volume conduction time parameter, and divide the water replenishment urgency level according to the river basin location where the river section is located; Step 313: Based on the historical optimal water replenishment strategy case library, use the case feature vector matching algorithm to perform similarity retrieval on the water volume gap, conduction time parameter, and urgency level of the current water shortage river section with the feature vectors in the cases, and extract the water replenishment path topological relationship, reservoir water replenishment response time, and water replenishment efficiency parameters; Step 314: Construct a multi-objective function with the minimum total time consumption of the water replenishment path, the least cross-reservoir water transfer volume, and the highest ecological flow compliance rate as the optimization objectives. Based on the adjustable storage capacity data of the reservoir calculated in Step 22, use the non-dominated sorting genetic algorithm to search for the Pareto front solution set of the water replenishment time sequence, path, and water volume, and finally select the scheme that meets the water replenishment urgency level and has a scheduling cost lower than the average value of historical cases as the scheduling scheme.

[0093] In the embodiment of the present invention, the above-mentioned step 311: simulation under different rainfall scenarios is specifically implemented as follows: Obtain rainfall intensity information from real-time weather forecast data, and classify it into three scenarios of light rain, moderate rain, and heavy rain according to established standards. For example, the rainfall intensity range of light rain may be 0.1 - 2.5 millimeters per hour, moderate rain is 2.6 - 8 millimeters, and heavy rain is above 8 millimeters. Input the three classified rainfall scenarios into the spatio-temporal coupling model established in step 24, and based on the joint operation rules of the reservoir group and the river water volume transfer relationship matrix, conduct simulations on an hourly basis. At each hour, the model determines the outflow of the reservoir according to the rainfall amount determined by the rainfall scenario, combined with the joint operation rules of the reservoir group, and then calculates the propagation process of the main and tributary section flows and the ecological protection area flow according to the river water volume transfer relationship matrix. At the same time, continuously track the changes in the water storage of each reservoir. For example, in the heavy rain scenario, the reservoir may increase its water storage due to a large amount of rainfall, and the main and tributary section flows will also increase accordingly. The model will simulate the specific values of these changes per hour. By simulating different rainfall scenarios, it is possible to comprehensively understand the changes in the water storage of the reservoir, the main and tributary section flows, and the ecological protection area flow under various meteorological conditions, providing a more abundant reference basis for subsequent water resource scheduling and enhancing the adaptability of the scheduling plan to different weather conditions.

[0094] The above-mentioned step 312: identification and classification of water shortage reaches is specifically implemented as follows: Analyze the simulation results based on the dynamic ecological flow threshold interval set in step 24. If the flow of a certain reach is continuously lower than the lower limit of the threshold for 3 hours, mark this reach as a water shortage reach, and calculate the cumulative flow gap of each water shortage reach within 3 consecutive hours in combination with the water volume conduction time parameter generated in step 23. The cumulative flow gap reflects the severity of water shortage in this reach. For example, if the hourly flow gaps of a certain reach within 3 hours are 5 cubic meters per second, 6 cubic meters per second, and 7 cubic meters per second respectively, then the cumulative flow gap of this reach is the sum of these three values. Classification of water replenishment urgency level: According to factors such as the basin location where the reach is located, such as whether it is an ecologically sensitive area and whether it is the water source reach of an important irrigation area, classify the water shortage reaches into different water replenishment urgency levels. For example, the water shortage reach in the core area of the ecological protection area may have a high water replenishment urgency level; while the water shortage reach in a general tributary with low surrounding water demand may have a low water replenishment urgency level. Accurately identifying water shortage reaches and classifying the water replenishment urgency levels helps to clarify the key points and priorities of water resource scheduling, giving priority to ensuring the water replenishment needs of reaches that have a greater impact on the ecological environment and water demand, and improving the utilization efficiency of water resources and the pertinence of scheduling.

[0095] Step 313, "Extracting Similar Case Features," is specifically implemented using a case feature vector matching algorithm based on a historical optimal water replenishment strategy case library. This algorithm combines information such as the water deficit, transmission time parameters, and urgency level of the current water-deficient river section into a feature vector. This algorithm then performs a similarity search with the feature vectors of each case in the case library. From the cases with high similarity, the topological relationship of the water replenishment path, the reservoir water replenishment response time, and the water replenishment efficiency parameters are extracted. The topological relationship of the water replenishment path describes the water flow path from the reservoir to the water-deficient river section; the reservoir water replenishment response time represents the time it takes for the reservoir to begin replenishing water to the water-deficient river section after receiving a water replenishment command; and the water replenishment efficiency parameter reflects the efficiency of water resource utilization during the replenishment process. For example, if a similar case shows that the topological relationship of the water replenishment path from reservoir A to the water-deficient river section B is that water is transported through tributary C, the reservoir water replenishment response time is 2 hours, and the water replenishment efficiency is that each cubic meter of water can increase the river section flow by 0.1 cubic meters per second, then these parameters are extracted as a reference, and the characteristic parameters of similar cases are extracted through the case feature vector matching algorithm. This can draw on previous successful water replenishment experiences, provide feasible plans and parameter references for current water replenishment scheduling, and reduce the blindness of scheduling decisions.

[0096] In step 314 of the present invention, the construction of the multi-objective function requires the quantitative calculation of the water replenishment path time, water transfer volume, and ecological compliance rate. The specific calculation method is described as follows: The total duration of the water replenishment path is the total time from when the reservoir initiates the water replenishment command until the water actually reaches the water-deficient river section. It consists of two parts: the cumulative water flow propagation time of all adjacent river sections along the replenishment path, based on the river water flow transmission time parameter (for example, if the reservoir needs to pass through three river sections to reach the water-deficient river section, and the transmission time of each section is 2 hours, 1.5 hours, and 1 hour, the path transmission time is 4.5 hours); and the time from the reservoir receiving the dispatching command to the actual opening of the gate to replenish water (for example, the response time of a certain reservoir is 0.5 hours). The goal: to minimize the sum of these two time components by optimizing the water replenishment path and dispatching sequence to quickly alleviate the water shortage problem in the river section.

[0097] For each reservoir participating in water transfer, multiply the water transfer flow rate (unit: cubic meters per second) at each moment during the scheduling period by the time (seconds) to convert it into the single water transfer volume, and then accumulate the water transfer volumes throughout the period (for example, if a reservoir transfers 10 cubic meters per second on average per hour within 24 hours, the total water transfer volume is 10×3600×24 = 864000 cubic meters). Add up the water transfer volumes of all reservoirs participating in water transfer, while ensuring that the water transfer volume of each reservoir does not exceed its adjustable storage capacity (the maximum adjustable water volume of the reservoir calculated in step 22 under the premise of meeting flood control / water supply constraints). On the premise of meeting the water replenishment demand, minimize the total water transfer volume of all reservoirs, reduce the scheduling cost and reservoir load. During the scheduling period (for example, 24 hours), count the proportion of time when the flow rate of the water shortage section reaches or exceeds the ecological flow threshold, and average the compliance time proportions of all water shortage sections. By optimizing the water replenishment volume and timing, maximize the average flow rate compliance rate of all water shortage sections to ensure the water demand for the ecological environment.

[0098] Each candidate scheduling plan contains three core parameters: the water replenishment timing (the time point when each reservoir starts to replenish water), the water replenishment path (the specific water flow path from the reservoir to the water shortage section, such as "Reservoir A → Tributary X → Main Stream Y → Water Shortage Section Z"), and the water replenishment volume (the water transfer flow rate and total volume of each reservoir during the scheduling period). Based on the historical optimal water replenishment strategy case library, extract the topological relationship of the water replenishment path of similar cases (such as the direct connection path across 2 reservoirs, the detour path across 3 reservoirs), the reservoir response time (such as 0.5 hours, 1 hour), and the water replenishment efficiency parameter (such as 0.1 m³ / s of river section flow can be increased per cubic meter of water) to generate the initial candidate plan. At the same time, randomly generate some new plans to expand the search space to ensure that more possible scheduling combinations are covered.

[0099] Calculate the total time consumption of the water replenishment path in the plan (path conduction time + reservoir response time), and compare it with the average time consumption of similar water shortage scenarios in the historical case library. If the total time consumption exceeds 120% of the historical average (for example, the historical average time consumption is 5 hours, and the current plan time consumption exceeds 6 hours), it is marked as "time consumption exceeding the limit" and enters the constraint filtering link. According to the adjustable storage capacity of each reservoir calculated in step 22 (such as the adjustable storage capacity of Reservoir A is 1 million cubic meters, and that of Reservoir B is 800,000 cubic meters), accumulate the total water transfer volume of each reservoir in the plan. If the water transfer volume of a certain reservoir exceeds its adjustable storage capacity, or the total water transfer volume of the entire basin exceeds the sum of the adjustable storage capacities of the reservoir group, it is determined as "water transfer volume exceeding the limit". Based on the dynamic threshold interval set in step 24 (such as the lower limit of the ecological flow compliance rate is 80% in the non-flood season and 85% in the flood season), simulate the proportion of time when the flow rates of each water shortage section reach the standard after the implementation of the plan, and calculate the average compliance rate of the entire basin. If the compliance rate is lower than the lower limit requirement of the corresponding scenario (such as the compliance rate of the non-flood season plan is only 75%), it is marked as "ecological non-compliance".

[0100] For each effective solution, calculate the total time consumption as "path conduction time + reservoir response time". The smaller the value, the better. Accumulate the water transfer volume (cubic meters) of each reservoir. The smaller the value, the lower the scheduling cost. Calculate the average percentage of the up-to-standard time of the water shortage sections in the whole basin (percentage). The higher the value, the better the ecological protection effect. Compare all effective solutions. If solution A is better than solution B in at least one objective (such as A has shorter time consumption and higher up-to-standard rate), and is not inferior to B in other objectives, then solution A "dominates" solution B. Through multi-level sorting, divide the solutions into different Pareto front levels. The first level (non-dominated solutions) constitutes the initial Pareto solution set. Perform "selection, crossover, mutation" operations on the solutions in the Pareto solution set: preferentially retain solutions with high front levels (such as non-dominated solutions in the first level), ensure diversity by combining the roulette wheel method, integrate the water supply paths of different solutions (such as exchanging the reservoir scheduling order of two solutions), water supply time sequences (such as adjusting the start water supply time), randomly fine-tune the water supply volume (such as ±10%) or switch a certain section of the river in the water supply path (such as switching from the main stream to the tributary), generate new solutions and re-check the constraint conditions, iterate the above process (such as setting 50 iterations), gradually approach the optimal solution, and finally form a Pareto front solution set containing 10 - 20 solutions. Each solution reaches a balance among "time consumption, water transfer volume, up-to-standard rate" (such as solution 1 has a time consumption of 6 hours, a water transfer volume of 2 million m 3 ³, and an up-to-standard rate of 90%; solution 2 has a time consumption of 7 hours, a water transfer volume of 1.5 million m 3 ³, and an up-to-standard rate of 88%).

[0101] According to factors such as the importance of the water shortage sections, the impact on the surrounding ecology and social economy, etc., divide the water shortage sections into different urgency levels. For example, the water shortage sections in the core area of the ecological protection area, the sections undertaking important water supply tasks, etc. can be classified as high urgency levels; while some secondary sections with less impact on ecology and social economy can be classified as low urgency levels. For different urgency levels, formulate corresponding target weight distribution strategies. Taking the high-urgency sections as an example, since it is necessary to give priority to ensuring the up-to-standard of ecological flow, a higher weight of 0.4 is assigned to the "up-to-standard rate of ecological flow"; while the weights of "water supply time consumption" and "water transfer volume" are relatively reduced, each being 0.3. For low-urgency sections, the weights can be adjusted according to the actual situation. For example, the weight of "water supply time consumption" is 0.4, the weight of "water transfer volume" is 0.3, and the weight of "up-to-standard rate of ecological flow" is 0.3.

[0102] For each solution in the Pareto solution set, obtain the specific values of its "water replenishment time", "water transfer volume", and "ecological flow compliance rate". According to the weight distribution strategy determined above, perform weighted calculations on the three objective values of each solution. Suppose the "water replenishment time" of a certain solution is t, the "water transfer volume" is v, and the "ecological flow compliance rate" is r. When evaluating for high-urgency river sections (with weights of 0.3, 0.3, and 0.4 respectively), the weighted score S of this solution is S = 0.3×f(t) + 0.3×g(v) + 0.4×h(r), where f(t), g(v), and h(r) are the function values after standardizing the "water replenishment time", "water transfer volume", and "ecological flow compliance rate" respectively, to ensure the comparability of different objective values in the weighted calculation. Sort the solutions in the Pareto solution set from high to low according to the weighted scores.

[0103] Collect historical scheduling cases similar to the current water shortage scenario. These cases should have certain similarities in aspects such as the location, scale, and time of the water shortage river section. For each historical case, calculate its scheduling cost. The scheduling cost comprehensively considers two factors: "water replenishment time" and "water transfer volume", and the two can be combined in a reasonable way, for example, setting a comprehensive calculation rule, such as the scheduling cost C = a×t + b×v (this is only for illustration), where a and b are coefficients determined according to the actual situation. Calculate the average value Cavg of the scheduling costs of all historical cases, and screen out the solutions with costs lower than the average value of historical cases; for the solutions in the Pareto solution set sorted by priority matching, calculate their scheduling costs in turn.

[0104] Compare the scheduling cost of each solution with the average value Cavg of historical cases, and screen out the solutions with scheduling costs lower than Cavg. For example, if the average water transfer volume of historical cases is 2.5 million m 3 , and the upper limit of the water transfer volume corresponding to the average scheduling cost of historical cases calculated in combination with the "water replenishment time" is 2.25 million m 3 , then screen out the solutions with a water transfer volume less than or equal to 2.25 million m 3 .

[0105] After priority matching and screening by comparison with historical cases, if there is only one solution left, then this solution is the final optimal solution. If there are still multiple solutions, factors such as expert experience and practical operation feasibility can be further combined for comprehensive judgment to finally determine an optimal solution that takes into account water replenishment efficiency, scheduling cost, and ecological protection.

[0106] For example, in the integration of expert experience, if there are multiple solutions with similar performance, the solution that conforms to the historical successful strategy (such as the optimal water replenishment path for similar water shortage scenarios in the case library) is preferentially selected to avoid repeated trial and error. For example, if the water replenishment path of a certain solution is the same as the path with the fastest response speed in historical cases, even if the water transfer volume is slightly higher, it is still selected as the preferred solution; for the final decision, select the solution that simultaneously meets the conditions of "water replenishment time ≤ 120% of the historical average", "water transfer volume ≤ the adjustable storage capacity of each reservoir", "compliance rate ≥ the lower limit of the dynamic threshold", and has the highest comprehensive score under the target weight as the multi-objective optimal scheduling solution for final implementation.

[0107] Through the above calculation process, the final solution needs to achieve the following balance among the three goals: The total time-consuming of the water replenishment path is shortened by 40% - 80% compared with manual decision-making (such as from 10 hours to 4 - 6 hours) to ensure timely response to water shortage problems; the cross-reservoir water transfer volume is reduced by 20% - 30% compared with the traditional solution (such as from 3 million m 3 down to 2 - 2.4 million m 3 ) to avoid overloading the scheduling of a single reservoir; the ecological flow compliance rate is increased by 10% - 20% compared with the fixed threshold solution (such as from 70% to 85% - 90%), while meeting the dynamic protection requirements during the flood season / non-flood season. Through quantitative constraints and multi-objective optimization, the intelligent scheduling goal of "rapid water replenishment, economical water transfer, and efficient water protection" is achieved.

[0108] In a preferred embodiment of the present invention, the corrected scheduling solution includes: Step 411, obtain the actual value of the ecological flow in real time through the monitoring point at the downstream section of the ecological protection area, compare it hour by hour with the dynamic threshold interval set in Step 24, and trigger the water replenishment time sequence correction mechanism. Based on the water volume conduction time parameter from the downstream section to the upstream water replenishment node of the ecological protection area, recalculate the water replenishment start time window; Step 412, use the water level monitoring data of the flood control control section, combined with the real-time rainfall forecast information, to dynamically evaluate the flood control risk level: when the water level rising rate exceeds the 90th percentile of the historical same period, preferentially adjust the discharge flow of the reservoir upstream of this section in the water replenishment path, and re-plan the topology of the cross-reservoir water replenishment path through the branch flood diversion threshold constraint; Step 413, according to the channel water level data of the irrigation area head monitoring point, judge the actual water use efficiency of the irrigation area: if the fluctuation range of the channel water level exceeds the set tolerance range, then dynamically adjust the distribution weight of the water replenishment volume according to the water use priority coefficient, calculate the deviation value between the water demand and the water replenishment volume in the irrigation area in the past 6 hours, and inversely correct the water transfer volume objective function in Step 314 according to the deviation ratio; In step 414, the modified parameters of steps 411 to 413 are input into the spatiotemporal coupling model, and the water allocation process for the next 12 hours is re-simulated using the current monitoring data as the initial condition, and a final scheduling plan including the delayed water replenishment period, detour path identification, and water redistribution ratio is generated.

[0109] In the embodiment of the present invention, the above step 411, the timing correction of water replenishment driven by ecological flow, may include: real-time data acquisition and threshold comparison, through the monitoring point of the downstream section of the ecological protection zone, real-time collection of the actual value of ecological flow in hours (for example, the monitoring value at a certain moment is 9m 3 / s, and the dynamic threshold interval set in step 24 (e.g. the non-flood season threshold is 10-15m 3 / s, 15-20m in flood season 3 If the monitoring value is lower than the lower threshold for 2 consecutive hours (e.g., lower than 10m / s in non-flood season), 3 / s), triggering the water replenishment timing correction mechanism; if it is higher than the upper threshold (such as more than 20m in flood season), 3 / s), then temporarily maintain the current water replenishment sequence or reduce the water replenishment amount. Call the water transmission time parameter of the "downstream section of the ecological protection zone to the upstream water replenishment node" generated in step 23 (for example, the transmission time from reservoir A to the section is 3 hours), based on the current monitored flow gap (for example, the target threshold is 10m 3 / s, actual 9m 3 / s, gap 1m 3 / s), combined with the adjustable storage capacity of the reservoir (calculated in step 22), calculate the continuous water replenishment time corresponding to the required water replenishment amount (for example, continuous water replenishment is required for 6 hours to fill the gap). If the current time is T and the conduction time is 3 hours, then the water replenishment must be started before T + (target time to reach the standard - T - conduction time) (for example, if the standard is required to be reached in T + 4 hours, the start time is T + 1 hour, and 3 hours of conduction time is reserved).

[0110] Step 412: Adjust the water replenishment path in response to flood risk: Collect real-time water level data of the flood control section (e.g., the current water level is 50.2m, and the 90% percentile in the same period in history is 50.0m) and the rainfall forecast for the next 6 hours (e.g., expected rainfall of 20mm), and calculate the water level rise rate (e.g., 0.15m / h). If it exceeds the rate corresponding to the 90% percentile in the same period in history (e.g., 0.1m / h), it is determined to be a high flood risk level; if it is between the 50%-90% percentile, it is determined to be a medium risk; if it is below the 50% percentile, it is determined to be a low risk.

[0111] Prioritize adjusting the discharge of the reservoirs upstream of the section in the water replenishment path (e.g. reducing the outflow of Reservoir B to avoid exacerbating the rise of the downstream water level), and adjust the discharge of the reservoirs upstream of the section according to the maximum flow capacity of the tributary flood diversion gate (e.g. 200m 3( / s), re-plan the water replenishment path. For example, if the original path is "Reservoir A → Main Stream → Flood Control Section → Water-scarce River Section", it can be adjusted to "Reservoir A → Tributary X → Flood Diversion Channel → Water-scarce River Section", bypassing the high-risk main stream area, and only slightly adjusting the conduction time parameters of the water replenishment path (such as correcting the main stream propagation time from 2 hours to 2.5 hours to reserve a safety buffer), without large-scale path switching.

[0112] The above step 413: Correction of the water replenishment volume driven by the water use efficiency in the irrigation area can include: collecting the channel water level data at the monitoring point at the head of the irrigation area canal (such as the daily fluctuation range is ±0.8m, and the set tolerance range is ±0.5m). If the fluctuation exceeds the tolerance, it indicates that the actual water use efficiency in the irrigation area has decreased (such as pipeline leakage or abnormal irrigation equipment). According to the water use priority coefficient (extracted in step 212, such as the ecological water demand priority 0.6 and the irrigation area priority 0.3), dynamically adjust the water replenishment volume distribution weight. For example, if the water use efficiency in the irrigation area decreases, lower the irrigation area priority coefficient by 0.1 (to 0.2), and correspondingly increase the ecological water demand weight; calculate the actual water demand in the irrigation area within the past 6 hours (inferred from the water level change) and the deviation between the water replenishment volume in the plan in step 314 (such as the actual water demand is 1 million m 3 ³, and the planned water replenishment in the plan is 1.2 million m 3 ³, with a deviation of +200,000 m 3 ³).

[0113] Reverse-correct the water transfer volume objective function according to the deviation ratio (such as 20%): If the actual water demand is less than the planned water replenishment volume, reduce the water replenishment volume of the corresponding irrigation area (such as reducing by 20%), and re-allocate the saved water volume to the ecological protection area or other high-priority areas.

[0114] Step 414: Re-simulation of the spatio-temporal coupling model and generation of the plan, including: correcting parameter input and model initialization, inputting the water replenishment start time in step 411, the new water replenishment path topology in step 412 (such as "Reservoir A → Tributary X → Flood Diversion Channel"), and the water volume distribution weight in step 413 (such as the irrigation area priority 0.2 and the ecology 0.7) into the spatio-temporal coupling model. Using the current monitoring data (ecological flow rate 9 m³ / s, flood control water level 50.2 m, irrigation area water level fluctuation ±0.8 m) as the initial conditions of the model to overwrite the original plan parameters generated in step 3, simulate the water volume distribution process in the next 12 hours, focusing on verifying the corrected water replenishment time sequence (such as the start time is advanced by 3 hours), the conduction efficiency of the bypass path (such as the conduction time of Tributary X is 4 hours), and the water volume re-allocation ratio (such as the irrigation area water replenishment volume is reduced by 20% and the ecological water replenishment volume is increased by 15%), determine the start / end time of water replenishment for each reservoir (such as Reservoir A replenishes water from T + 1 hour to T + 7 hours), mark the key nodes passed by the water replenishment (such as "Flood Diversion Gate Z of Tributary X is opened"), and the water volume re-allocation ratio: the adjustment range of the water replenishment volume in each target area (ecological, irrigation area, flood control) (such as the ecological water replenishment increases by 100,000 m³ and the irrigation area decreases by 80,000 m³).

[0115] In a preferred embodiment of the present invention, for step 5 above, encoding the scheduling scheme after correction in step 4 to generate an executable scheduling instruction may include: extracting core data such as the delayed water replenishment period, the bypass path identifier, and the water volume redistribution ratio from the scheduling scheme after correction in step 4. For example, clarifying that reservoir A replenishes water from T+1 hour to T+7 hours, and the water replenishment path is "reservoir A → tributary X → flood diversion channel → water shortage section", with an increase of 100,000 m³ in ecological water replenishment and a reduction of 80,000 m³ in the irrigation area, etc.; classifying the extracted data according to the scheduling objects (such as each reservoir, sluice, channel, etc.) and scheduling actions (such as opening, closing, adjusting the flow rate, etc.). For example, classifying the water replenishment time and flow rate adjustment data related to the reservoir into one category, and classifying the opening or closing information of the sluice into another category; assigning a unique code to each scheduling object. For example, reservoir A is coded as "R01", and the flood diversion sluice Z on tributary X is coded as "G02", etc. The coding rule should be concise and clear for easy identification and management; setting corresponding codes for different scheduling actions. For example, "opening" is coded as "O", "closing" is coded as "C", "increasing the flow rate" is coded as "I", "decreasing the flow rate" is coded as "D", etc.; for the specific parameters involved in the scheduling action, such as the water replenishment time, flow rate value, etc., encoding is performed in a unified format. For example, the time can be in the format of "YYYY-MM-DD HH:MM", and the flow rate can be represented by a specific value.

[0116] According to the encoding rules determined above, combine the scheduling actions and related parameters of each scheduling object into a single instruction code. For example, for the instruction that reservoir A starts to replenish water at T+1 hour (assuming T is the current time 2024-05-03 10:00, and T+1 hour is 2024-05-03 11:00) with a flow rate increase of 10 m³ / s, the code can be "R01-I-10-2024-05-03 11:00".

[0117] Processing complex instructions: For complex scheduling involving multiple actions or multiple parameters, encoding is performed in a certain logical order. For example, for the instruction that the flood diversion sluice Z is opened and adjusted to a specified opening at a specific time, the code can be "G02-O-50%-2024-05-03 12:30", indicating that the flood diversion sluice Z is opened to 50% of the opening at 12:30 on May 3, 2024.

[0118] According to the logic and actual operation requirements of the scheduling plan, determine the execution order of each instruction. For example, first open the relevant sluice gates on the water replenishment path, and then start the water replenishment operation of the reservoir. Arrange the generated individual instructions into a complete instruction sequence according to the determined execution order. For example, the instruction sequence "G02 - O - 50% - 2024 - 05 - 03 12:30; R01 - I - 10 - 2024 - 05 - 03 13:00" means to open the flood - discharge sluice Z to 50% opening at 12:30 first, and then increase the flow of reservoir A by 10 m³ / s at 13:00. Check whether the generated instruction sequence conforms to the logic and actual operation requirements of the scheduling plan. For example, check whether there are instruction conflicts, such as the same sluice being required to be opened and closed at the same time; check whether the execution order of the instructions is reasonable and whether it will lead to scheduling chaos. Ensure that the instruction sequence contains all the necessary scheduling information in the scheduling plan without missing key scheduling actions or parameters. For the problems found during the verification process, make corrections and improvements in a timely manner. For example, adjust the execution order of the instructions, supplement the missing parameters, etc. Convert the verified and improved instruction sequence into a format that can be recognized and executed by the relevant scheduling system. The finally generated executable scheduling instructions will be used for actual water resource scheduling operations.

Claims

1. An intelligent management method for natural resource engineering data, characterized in that, The method includes: Step 1: Construct a semantic parsing model based on a knowledge graph to perform structured conversion on heterogeneous data. Parse the semantic ambiguity of keyword fields through a preset watershed ontology library, establish a standardized data pool under a unified spatio-temporal coordinate system, and process missing values using a spatio-temporal correlation interpolation algorithm for upstream and downstream stations to obtain the standardized data pool. Step 2: Based on the standardized data pool output in Step 1, combined with the scheduling rule constraint conditions in the historical scheduling case library, establish a spatio-temporal coupling model including a dynamic balance equation for reservoir storage, a continuity equation for river runoff, and an ecological flow threshold constraint. Step 3: Input the spatio-temporal coupling model constructed in Step 2 into real-time meteorological forecast data, simulate the water volume allocation plan under different rainfall scenarios, automatically identify the river section positions where the ecological flow is lower than the design threshold, and generate a multi-objective optimal scheduling plan including the water replenishment timing, water replenishment path, and water replenishment volume in combination with the historical water replenishment strategy case library. Step 4: Set up three monitoring points at the downstream section of the ecological protection area, the flood control control section, and the canal head of the irrigation area in the watershed to respectively monitor the ecological flow, flood control section water level, and canal water level data in real time. According to the spatial position characteristics of the monitoring points and the real-time monitoring data, dynamically correct the water replenishment timing, water replenishment path, and water replenishment volume in the multi-objective optimal scheduling plan generated in Step 3 to obtain the corrected scheduling plan. Step 5: Encode the scheduling plan corrected in Step 4 to generate an executable scheduling instruction.

2. The intelligent management method for natural resource engineering data according to claim 1, wherein The heterogeneous data includes: Through satellite remote sensing monitoring terminals, hydrological station data acquisition modules, and irrigation area Internet of Things sensors, synchronously obtain 9 types of dynamic data including the water level elevation data, water quality parameters, real-time runoff data of 15 reservoirs in the watershed, as well as the water consumption of the irrigation area, the water demand of the ecological protection area, and the flood control scheduling instructions. Among them, the water level elevation data includes two data formats: the absolute elevation value based on the altitude reference plane and the relative depth value based on the dam body reference point.

3. The intelligent management method for natural resource engineering data according to claim 2, characterized in that Step 2: Based on the standardized data pool output in Step 1, combined with the scheduling rule constraint conditions in the historical scheduling case library, establish a spatio-temporal coupling model including a dynamic balance equation for reservoir storage, a continuity equation for river runoff, and an ecological flow threshold constraint, including: Step 21: Based on the spatio-temporally calibrated reservoir water level, runoff, and water demand data in the standardized data pool, extract the scheduling rule constraint conditions for two typical scenarios of preferential supply guarantee during the dry season and flood control peak shifting during the flood season in the historical scheduling case library. Step 22: Use the real-time storage volume of 15 reservoirs in the standardized data pool as the input variable, and calculate the adjustable storage capacity of each reservoir within the scheduling period through the dynamic balance equation. Step 23: Construct a water volume transfer relationship matrix between the main stream and tributaries based on the river runoff continuity equation, and determine the water volume conduction time parameters for each river section. Step 24: Convert the water demands of the ecological protection area and the canal head of the irrigation area into ecological flow threshold constraint conditions for the corresponding river sections, and establish a spatio-temporal coupling model with the joint scheduling of the reservoir group as the core, river water volume transfer as the link, and ecological flow compliance as the boundary, where the ecological flow threshold sets a dynamic threshold range according to the flood season and non-flood season.

4. The intelligent management method for natural resource engineering data according to claim 3, characterized in that, The scheduling rule constraint conditions include: Step 211: Based on the reservoir water level, runoff, and water demand data after spatio-temporal calibration in the standardized data pool, perform morphological matching on the time series of the current hydrological characteristics and the meteorological and hydrological characteristic parameters in the dry season and flood season in the historical scheduling case base, calculate the similarity index between the sequences, and screen out the corresponding case set from the similarity index; Step 212: For the screened case set in the scenario of ensuring water supply with priority in the dry season, analyze and extract the 5% quantile of the minimum discharge constraint of the reservoir as the safety boundary value, calculate the weight distribution matrix of the water use priority coefficient of the irrigation area, and determine the water demand weight adjustment coefficient based on the frequency distribution of historical water shortage events in the ecological protection area; for the screened case set in the scenario of flood control and peak shifting in the flood season, extract the median of the pre-discharge ratio as the benchmark parameter according to the ratio distribution of the pre-discharge storage capacity and the total storage capacity of the reservoir, statistically calculate the probability density of the duration of the peak-shifting scheduling time window through a time-sliding window, and set the upper limit of the flood diversion flow threshold according to the maximum flow capacity of the flood diversion gate of the tributary; Step 213: Perform feature clustering on the water supply parameters in the dry season and the flood control parameters in the flood season respectively to obtain the clustering results; generate a set of constraint conditions for the linear programming model according to the clustering results, where the constraints in the dry season scenario include the weighted sum of the water use priority coefficients of the irrigation area ≥ 0.8, and the product of the ecological water demand weight and the minimum discharge ≥ the design threshold, and the constraints in the flood season scenario include the pre-discharge storage capacity ratio × the total storage capacity ≤ the flood control safety storage capacity, and the flood diversion flow ≤ the tributary flood diversion threshold × the time window adjustment coefficient.

5. The intelligent management method for natural resource engineering data according to claim 4, characterized in that Step 22: Using the real-time water storage of 15 reservoirs in the standardized data pool as the input variable, calculate the adjustable storage capacity of each reservoir within the scheduling period through the dynamic balance equation, including: Step 221: Taking the real-time water storage of 15 reservoirs in the standardized data pool as the initial input, and combining the minimum discharge or pre-discharge storage capacity ratio parameter of the reservoir in the scheduling rule constraints extracted in Step 21, construct a dynamic balance equation; The dynamic balance equation takes the predicted inflow and the controlled outflow of each reservoir within the scheduling period as the core variables, and determines the maximum adjustable storage capacity that meets the flood control or water supply constraint conditions by iteratively calculating the reservoir water storage change curve; among them, the predicted inflow is dynamically corrected through the water volume transfer relationship matrix of the upstream and downstream reservoir groups, and the controlled outflow is calculated by segmenting the limit according to the priority of the ecological flow threshold and the water demand of the irrigation area.

6. The intelligent management method for natural resource engineering data according to claim 5, wherein Step 23: Based on the river channel runoff continuity equation, construct a water volume transfer relationship matrix between the main stream and the tributaries, and determine the water volume conduction time parameters of each river section, including: Step 231: According to the river channel runoff continuity equation, map the topological structure of the main stream and tributaries of the basin to a node-river section connection matrix, and the matrix elements represent the water volume conduction coefficient between adjacent river sections; Step 232: Based on the historical runoff data and cross-section terrain parameters of each river section in the standardized data pool, calculate the water flow propagation time under different flow levels, and determine the water volume conduction time parameters; Step 233: For the river sections with water conservancy project regulation, dynamically compensate and correct the conduction time parameters according to the gate and dam scheduling log data to form a spatio-temporal coupling parameter set including natural conduction and artificial intervention effects.

7. The intelligent management method for natural resource engineering data according to claim 6, characterized in that, Step 24: Convert the water demand of the ecological reserve area and the canal head of the irrigation area into the ecological flow threshold constraints for the corresponding river sections, and establish a spatio-temporal coupling model with the joint operation of the reservoir group as the core, the water transfer in the river channel as the link, and the compliance of the ecological flow as the boundary, including: Step 241: Convert the monthly average water demand of the ecological reserve area and the daily water use plan of the canal head of the irrigation area into the lower limit threshold of the ecological flow and the water use flow demand curve for the corresponding river sections respectively. Combining with the water transfer time parameters generated in Step 23, construct a multi-period coupling model with the joint operation instructions of the reservoir group as the decision variable, the water transfer process in the river channel as the state variable, and the compliance of the ecological flow threshold as the boundary condition; Among them, the ecological flow threshold sets a dynamic threshold interval according to the flood season and non-flood season operation rules extracted in Step 21 based on the positions of the river channel control sections, and establishes a linkage mechanism for triggering the water replenishment strategy when the threshold is exceeded through the emergency scheduling response time parameters in the historical optimal water replenishment strategy case library.

8. The intelligent management method for natural resource engineering data according to claim 7, wherein The said scheduling plan includes: Step 311: Classify the rainfall intensity in the real-time weather forecast data into three scenarios: light rain, moderate rain, and heavy rain. After inputting it into the spatio-temporal coupling model, based on the joint operation rules of the reservoir group and the river channel water transfer relationship matrix, simulate the changes in reservoir storage, the flow rates of the main and tributary sections, and the ecological reserve area flow propagation process hour by hour for each scenario; Step 312: According to the dynamic ecological flow threshold interval set in Step 24, mark the river sections where the flow rate is continuously lower than the threshold lower limit for 3 hours in the simulation results as water shortage river sections. Calculate the cumulative flow gap of each water shortage river section in combination with the water transfer time parameters, and divide the water replenishment urgency levels according to the river basin positions where the river sections are located; Step 313: Based on the historical optimal water replenishment strategy case library, use the case feature vector matching algorithm to perform similarity retrieval on the water gap volume, transfer time parameters, and urgency level of the current water shortage river sections with the feature vectors in the cases, and extract the water replenishment path topological relationship, reservoir water replenishment response time, and water replenishment efficiency parameters; Step 314: Construct a multi-objective function with the optimization objectives of minimizing the total time consumption of the water replenishment path, minimizing the cross-reservoir water transfer volume, and maximizing the ecological flow compliance rate. Based on the reservoir adjustable storage capacity data calculated in Step 22, use the non-dominated sorting genetic algorithm to search for the Pareto front solution set of the water replenishment time sequence, path, and volume, and finally select the plan that meets the water replenishment urgency level and the scheduling cost is lower than the average value of historical cases as the scheduling plan.

9. The intelligent management method for natural resource engineering data according to claim 8, characterized in that, The said revised scheduling plan includes: Step 411: Obtain the actual value of the ecological flow in real time through the monitoring points of the downstream section of the ecological reserve area, compare it with the dynamic threshold interval set in Step 24 hour by hour, and trigger the water replenishment time sequence correction mechanism. Based on the water transfer time parameters from the downstream section to the upstream water replenishment node of the ecological reserve area, recalculate the water replenishment start time window; Step 412: Use the water level monitoring data of the flood control control section, combined with the real-time rainfall forecast information, to dynamically evaluate the flood control risk level: when the water level rising rate exceeds the 90th percentile of the same period in history, preferentially adjust the discharge of the reservoir upstream of this section in the water replenishment path, and re-plan the cross-reservoir water replenishment path topology through the tributary flood diversion threshold constraint; Step 413: Based on the channel water level data of the monitoring points at the head of the irrigation area, judge the actual water use efficiency of the irrigation area. If the fluctuation range of the channel water level exceeds the set tolerance range, dynamically adjust the distribution weight of the make-up water volume according to the water use priority coefficient, calculate the deviation value between the water demand and the make-up water volume in the irrigation area within the past 6 hours, and reverse-correct the water transfer objective function in Step 314 according to the deviation ratio. Step 414: Input the correction parameters in Steps 411 to 413 into the spatio-temporal coupling model, re-simulate the water volume distribution process for the next 12 hours with the current monitoring data as the initial condition, and generate the final scheduling plan including the delayed water supply period, the bypass path identifier, and the water volume re-distribution ratio.

Images