An intelligent water level prediction and regulation method and system for hydraulic engineering

Abstract

The application provides a water conservancy project intelligent water level prediction regulation method and system, and belongs to the technical field of water level prediction. The method comprises the following steps: constructing a three-dimensional model of a reservoir; loading a water level prediction model; acquiring water level prediction data; predicting a reservoir storage capacity change rate; and regulating the reservoir in stages. The application trains water level prediction models under different structural components based on the differences between the structural components of the reservoir, so that the water level prediction model of the reservoir is more accurate, and a solid data foundation is laid for subsequent water resource allocation balance; the application predicts the risk water level of the reservoir according to the quality and historical situation of the reservoir, guarantees the accuracy of subsequent early warning and staged calling, and avoids resource waste.

Description

Technical Field

[0001] This invention relates to the field of water level prediction technology, specifically to an intelligent water level prediction and control method and system for water conservancy projects. Background Technology

[0002] Water is inextricably linked to every aspect of life and production. From daily water supply to agricultural irrigation and industrial production, from inland waterway transportation to the maintenance of ecosystems, the importance of water is self-evident. Water level, as a key indicator reflecting the state of water bodies, is influenced by a complex interplay of factors including weather, hydrology, and geology. Therefore, accurate water level prediction is of immeasurable importance and urgency for the scientific management of water resources, effective response to natural disasters, ensuring stable socio-economic development, and maintaining ecological balance.

[0003] Existing technologies, such as the water level prediction method and device disclosed in invention application patent with announcement number CN118760956A, include: water level prediction by coupling multiple models, and the ability to extract temporal and spatial features affecting the water level, thereby improving the accuracy and efficiency of water level prediction. Another example is the reservoir water level prediction method disclosed in invention application patent with announcement number CN118674127A, which addresses the problem of low accuracy in reservoir water level prediction in existing technologies.

[0004] Based on the above solutions, it can be found that existing technologies rarely train water level prediction models for different structural components of reservoirs to make the water level prediction models more accurate. The neglect of this aspect in existing technologies leads to a higher deviation between the water level prediction models and the actual situation. At the same time, there are few predictions of reservoir risk water levels based on the quality and historical conditions of the reservoir. However, each reservoir has different risk water levels that it can withstand due to differences in construction quality, usage, water body shape, and wet / dry ratio. The insufficient understanding of this aspect in existing technologies leads to low accuracy in early warning and tiered dispatch, which easily results in resource waste. Summary of the Invention

[0005] The main objective of this invention is to provide an intelligent water level prediction and control method and system for water conservancy projects. Based on different structural components of the reservoir, a water level prediction model is trained under different structural components, making the water level prediction model of the reservoir more accurate and laying a solid data foundation for the subsequent water resource allocation balance. This invention predicts the risk water level of the reservoir based on the quality and historical conditions of the reservoir, ensuring the accuracy of subsequent early warning and graded dispatch, and avoiding resource waste.

[0006] To achieve the above-mentioned technical features, the objective of this invention is as follows: A method for intelligent water level prediction and control in water conservancy projects, comprising: 3D model construction of reservoir: Collect topographic data of the reservoir bank and surrounding area to generate a 3D real scene model of the reservoir, and collect underwater topographic data to construct a 3D model of the water area. Integrate the 3D real scene model of the reservoir and the 3D model of the water area to form an integrated reservoir topographic model, and extract the water level-reservoir capacity relationship curve. The water level prediction model is equipped with: based on the reservoir topography model, the reservoir is simulated under different environments to obtain reservoir simulation data, and a reservoir capacity change rate prediction model is trained based on the reservoir simulation data. Water level prediction data acquisition: Acquire soil infiltration data of the reservoir, as well as meteorological data and dispatch behavior data during the prediction period; Reservoir capacity change rate prediction: The reservoir's soil infiltration data, meteorological data, and dispatch behavior data during the prediction period are integrated and imported into the reservoir capacity change rate prediction model after data processing. The model outputs the reservoir capacity change rate during the prediction period, which is then converted into water level using the water level-capacity relationship curve. Water level regulation: Collect basic construction information of water reservoirs and set risk water levels for water reservoirs. Then, regulate water levels in stages based on the water level and the scheduling rule engine.

[0007] Preferably, the reservoir simulation data includes reservoir capacity, meteorological simulation data, soil infiltration simulation data, and scheduling behavior simulation data for each simulation. The meteorological simulation data includes the characteristic values ​​of each meteorological characteristic parameter; The soil infiltration simulation data includes the water content of each soil layer; The simulation data of the scheduling behavior includes water demand and water release flow. Specifically, the water demand refers to the total amount of water drawn from surrounding farmland, residential areas, and other sources of water from the reservoir.

[0008] Preferably, the specific training method for the reservoir capacity change rate prediction model trained based on reservoir simulation data is as follows: A1. Extract meteorological simulation data, soil infiltration data and scheduling behavior data from each reservoir simulation data, and obtain the meteorological characteristic parameters, soil infiltration characteristic parameters and scheduling characteristic parameters of each reservoir simulation through data processing. A2. Divide the reservoir simulation data into training dataset and validation dataset according to a set ratio. Divide the training set into training sets for each structural component according to the reservoir capacity, and divide the validation set into validation sets for each structural component. A3. Obtain several training base models from the data warehouse, and import the reservoir capacity, meteorological characteristic parameters, soil infiltration characteristic parameters and scheduling characteristic parameters of each structural component in the training dataset of the reservoir into several training base models, thereby determining the coefficients of several training base models of each structural component and obtaining the exact training base models of each structural component. The aforementioned basic training models are specifically combinations of the characteristic relationships between meteorological characteristic parameters, soil infiltration characteristic parameters, scheduling characteristic parameters, and reservoir capacity. The training base model coefficients are specifically the correction coefficients for the characteristic relationships between meteorological characteristic parameters, soil infiltration characteristic parameters, scheduling characteristic parameters, and reservoir capacity. A4. Import the meteorological characteristic parameters, soil infiltration characteristic parameters, and scheduling characteristic parameters of each simulation in the validation dataset of each structural component of the reservoir into several training basic models of the corresponding structural components, so as to obtain the reservoir capacity of each simulation of several training basic models of each structural component on the validation set, and thus evaluate the convergence constraint value of several training basic models of each structural component. A5. For each structural component, the training base model with the largest convergence constraint value is selected as the reservoir capacity change rate prediction model for the corresponding structural component, and the models are summarized to obtain the reservoir capacity change rate prediction model.

[0009] Preferably, the specific evaluation method for the soil infiltration characteristic parameters of each simulation of the reservoir is as follows: The water content of each soil layer was extracted from the soil infiltration data of each reservoir simulation. ,hardness Thickness, which is then imported into the empirical formula for permeability: In the process, the permeability coefficient of each soil layer is output. m, where a, b, and c are the fitting parameters. Let be the water saturation threshold of each soil layer for the reservoir soil type stored in the data warehouse, i be the number of each simulation, i=1,2,3...,n, and m be the number of each soil layer, m=1,2,3,...l; The thickness of each soil layer was extracted from the soil infiltration data of each reservoir simulation. And combining the permeability coefficients of each soil layer in each reservoir simulation, after thickness weighting: Subsequently, the overall permeability coefficient of the output reservoir in each simulation was calculated. This is used as a characteristic parameter of soil infiltration, among which Specifically, it is the percentage factor of the permeability coefficient of the m-th layer of soil stored in the data warehouse, where D is the total soil thickness.

[0010] Preferably, the specific evaluation method for the scheduling characteristic parameters of the reservoir in each simulation is as follows: The water demand and water release flow are extracted from the simulation data of each reservoir scheduling simulation. The net flow difference is obtained by subtracting the water release flow from the water demand. The net flow difference is divided by the historical average water demand to obtain the dimensionless scheduling characteristic parameter.

[0011] Preferably, the specific evaluation content of the convergence constraint values ​​of the several training base models for evaluating each structural component includes: The reservoir capacity simulated in each simulation of the verification dataset of each structural component of the reservoir and the reservoir capacity simulated in each simulation of several training basic models are denoted as actual reservoir capacity and predicted reservoir capacity, respectively. They are combined to obtain vector feature data of each simulation of each training basic model of each structural component of the reservoir. The vector feature data points from actual reservoir capacity to predicted reservoir capacity. The vector magnitudes of each simulation of each training model for each structural component of the reservoir are calculated using the vector magnitude calculation method. This constructs a data sequence of vector magnitudes for each training model of each structural component of the reservoir. Convergence constraint features are then identified, specifically including trend eigenvalues. Average vector magnitude The trend feature value specifically refers to the value obtained by trend fitting of the vector model data sequence, which represents the rate of change of the trend fitting line, j is the number of each structural component, j=1,2,3...,k, and h is the number of each training base model, h=1,2,3,...g; After normalizing the average vector magnitude of each training basic model for each structural component of the reservoir, it is imported into the convergence constraint value evaluation expression along with the trend feature values: In the formula, the convergence constraint values ​​of several training base models for each structural component are output, where... This indicates the appropriate range of trend characteristic values ​​stored in the data warehouse. It is the midpoint of the appropriate trend characteristic value range.

[0012] Preferably, the specific method for collecting the basic construction information of the reservoir and setting the risk water level of the reservoir is as follows: The feature values ​​of several construction characteristic parameters are extracted from the basic construction information of the reservoir, and the first risk characteristic parameter of the reservoir is obtained through data processing. ; Based on the reservoir's water level during the forecast period, combined with an integrated reservoir topography model, the water slope of the reservoir during the forecast period is obtained. Combined with the river's high-water-low-water ratio and downstream population density stored in the database, the second risk characteristic parameter of the reservoir is obtained through data processing. ; The first and second risk characteristic parameters of the reservoir are imported into the risk water level setting model, and the risk water level of the reservoir is output.

[0013] Preferably, the first risk characteristic parameter of the reservoir is specifically determined based on the characteristic values ​​of several construction characteristic parameters, and is used to quantify the risk construction quality of the reservoir; The second risk characteristic parameter of the reservoir is specifically determined based on the water body slope, river flood-dry ratio, and downstream population density within the prediction period. It is used to quantify the reservoir's carrying capacity against the impact of water bodies. The greater the water body slope, the greater the river flood-dry ratio, and the greater the downstream population density, the smaller the reservoir's carrying capacity against the impact of water bodies, and the greater the second risk characteristic parameter.

[0014] Preferably, the risk level setting model is specifically as follows: ; In the formula The first risk characteristic parameter as the benchmark Benchmark Second Risk Characteristic Parameter The corresponding risk level, , These are the corrected water level parameters corresponding to the first risk characteristic parameter of unit offset and the corrected water level parameters corresponding to the second risk characteristic parameter of unit offset, respectively, which are preset in the data warehouse.

[0015] Another aspect of the present invention provides a system for executing the intelligent water level prediction and control method for water conservancy projects, characterized in that it comprises: The reservoir 3D model construction module collects topographic data of the reservoir bank and surrounding areas, generates a 3D real-scene model of the reservoir, collects underwater topographic data, constructs a 3D model of the water area, integrates the 3D real-scene model of the reservoir and the 3D model of the water area to form an integrated reservoir topographic model, and extracts the water level-reservoir capacity relationship curve. The water level prediction model module is equipped with a reservoir topography model to simulate the reservoir under different environments, obtain reservoir simulation data, and train a reservoir capacity change rate prediction model based on the reservoir simulation data. Water level prediction data acquisition module: Acquires soil infiltration data of the reservoir, as well as meteorological data and dispatch behavior data during the prediction period; Reservoir capacity change rate prediction module: It integrates soil infiltration data of the reservoir with meteorological data and scheduling behavior data during the prediction period. After data processing, it is imported into the reservoir capacity change rate prediction model and outputs the reservoir capacity change rate during the prediction period. The data is converted into water level through the water level-capacity relationship curve. The reservoir hierarchical control module collects basic construction information of the reservoir and sets the risk water level of the reservoir. Based on the water level of the reservoir and the scheduling rule engine, it performs hierarchical control.

[0016] The present invention has the following beneficial effects: 1. Based on the different structural components of the reservoir, this invention trains water level prediction models under different structural components, making the water level prediction models of the reservoir more accurate and laying a solid data foundation for the subsequent water resource allocation balance.

[0017] 2. This invention predicts the risk level of a reservoir based on its quality and historical data, ensuring the accuracy of subsequent early warnings and tiered dispatching, and avoiding resource waste. Attached Figure Description

[0018] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0019] Figure 1 This is a schematic diagram of the implementation steps of the method of the present invention.

[0020] Figure 2 This is a schematic diagram of the system structure connection of the present invention. Detailed Implementation

[0021] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Reference Figure 1 As shown, the first aspect of the present invention provides a method for intelligent water level prediction and control in water conservancy projects, comprising: S1. Reservoir 3D Model Construction: Topographic data of the reservoir bank and surrounding area are collected using UAV oblique photography technology to generate a 3D real-scene model of the reservoir. Underwater topographic data is collected using a multibeam echo sounder system to construct a 3D model of the water area. The 3D real-scene model of the reservoir and the 3D model of the water area are integrated to form an integrated reservoir topographic model, and the water level-reservoir capacity relationship curve is extracted.

[0023] Among them, the existing technology for extracting the water level-reservoir capacity relationship curve through an integrated reservoir topography model is relatively mature and will not be elaborated here.

[0024] S2. Water level prediction model: Based on the reservoir topography model, the reservoir is simulated under different environments to obtain reservoir simulation data, and a reservoir capacity change rate prediction model is trained based on the reservoir simulation data.

[0025] In a specific embodiment of the present invention, the reservoir simulation data includes reservoir capacity, meteorological simulation data, soil infiltration simulation data, and scheduling behavior simulation data for each simulation.

[0026] The meteorological simulation data includes the characteristic values ​​of each meteorological characteristic parameter.

[0027] The soil infiltration simulation data includes the water content of each soil layer.

[0028] The simulation data of the scheduling behavior includes water demand and water release flow. Specifically, the water demand refers to the total amount of water drawn from surrounding farmland, residential areas, and other sources of water from the reservoir.

[0029] The characteristic values ​​of each meteorological parameter include rainfall, wind speed, temperature, humidity, etc.

[0030] In a specific embodiment of the present invention, the specific training method for training the reservoir capacity change rate prediction model based on reservoir simulation data is as follows: A1. Extract meteorological simulation data, soil infiltration data and scheduling behavior data from each reservoir simulation data, and obtain the meteorological characteristic parameters, soil infiltration characteristic parameters and scheduling characteristic parameters of each reservoir simulation through data processing.

[0031] Among them, the meteorological characteristic parameters, soil infiltration characteristic parameters, and scheduling characteristic parameters of each reservoir simulation specifically represent the impact of meteorology, soil, and scheduling on reservoir capacity.

[0032] The specific details of the meteorological characteristic parameters obtained from each simulation of the reservoir are as follows: After normalizing the characteristic values ​​of each meteorological feature parameter in the meteorological simulation data of each reservoir simulation, a weighted sum is performed to obtain the meteorological feature parameters of each reservoir simulation. Specifically, the weighted sum is obtained by accumulating the characteristic values ​​of each meteorological feature parameter with their corresponding weight coefficients and then adding them together. The corresponding weight coefficients of each meteorological feature parameter are obtained from the data warehouse. The weight coefficients of each meteorological feature parameter reflect the degree and direction of the influence of the meteorological feature parameter on the reservoir capacity. The direction of influence includes positive and negative influences.

[0033] In this embodiment, for example, the rainfall is calculated using meteorological simulation data from a specific reservoir simulation. ,temperature ,humidity For example, rainfall has the greatest impact on reservoir capacity, and its influence is positive. Temperature and humidity have a moderate impact on reservoir capacity, with temperature accelerating evaporation and having a negative influence, while humidity inhibits evaporation and has a positive influence. Therefore, the weighting coefficients for rainfall, temperature, and humidity can be set to 0.7, -0.15, and 0.15, respectively. The meteorological characteristic parameters for this reservoir meteorological simulation are then: .

[0034] A2. Divide the reservoir simulation data into training dataset and validation dataset according to a set ratio. Divide the training set into training sets for each structural component according to the reservoir capacity, and divide the validation set into validation sets for each structural component.

[0035] Specifically, the method of dividing the training set into training sets for each structural component based on the reservoir capacity is as follows: the discharge pattern of the reservoir is obtained from the data warehouse, which is specifically the discharge mode corresponding to each water level interval. Then, combined with the reservoir capacity of each simulation in the reservoir simulation data, the water level of each simulation is obtained by mapping through the water level-reservoir capacity relationship curve. Thus, each simulation of the reservoir is divided into training sets for each structural component, and the water level interval and the structural component have a one-to-one correspondence.

[0036] Specifically, each structural component of the reservoir is trained individually to improve the accuracy of water level prediction for each structural component.

[0037] A3. Obtain several training base models from the data warehouse, and import the reservoir capacity, meteorological characteristic parameters, soil infiltration characteristic parameters, and scheduling characteristic parameters of each structural component in the reservoir training dataset into several training base models. In this way, determine the coefficients of several training base models for each structural component, and obtain the exact training base models for each structural component.

[0038] Furthermore, the determination of several training base model coefficients for each structural component can specifically be achieved by minimizing the sum of squared residuals.

[0039] The aforementioned training models are specifically combinations of the characteristic relationships between meteorological characteristic parameters, soil infiltration characteristic parameters, scheduling characteristic parameters, and reservoir capacity. These characteristic relationships include, but are not limited to, radical, square, and linear relationships.

[0040] The training model coefficients are specifically the correction coefficients for the characteristic relationships between meteorological feature parameters, soil infiltration feature parameters, scheduling feature parameters, and reservoir capacity.

[0041] For example, if the reservoir structure consists of a bottom culvert and a spillway, then in the low water level range, only the bottom culvert discharges water; in the middle water level range, the culvert and part of the spillway discharge water; and in the high water level range, the water level discharge is fully open. Due to the different discharge methods, the water level prediction model for each structural component also differs. Therefore, it is necessary to train the water level prediction model for each structural component of the reservoir.

[0042] A4. Import the meteorological characteristic parameters, soil infiltration characteristic parameters, and scheduling characteristic parameters from each simulation in the validation dataset of each structural component of the reservoir into several training basic models of the corresponding structural component. This will give the reservoir capacity of each simulation of several training basic models of each structural component on the validation set, and thus evaluate the convergence constraint value of several training basic models of each structural component.

[0043] A5. For each structural component, the training base model with the largest convergence constraint value is selected as the reservoir capacity change rate prediction model for the corresponding structural component, and the models are summarized to obtain the reservoir capacity change rate prediction model.

[0044] In a specific embodiment of the present invention, the soil infiltration characteristic parameters of each reservoir simulation are specifically evaluated by extracting the water content of each soil layer from the soil simulation infiltration data of each reservoir simulation. ,hardness Thickness, which is then imported into the empirical formula for permeability: In the process, the permeability coefficient of each soil layer is output. m, where a, b, and c are fitting parameters (specifically determined through experiments or historical data). Let be the water saturation threshold of each soil layer for the reservoir soil type stored in the data warehouse, i be the number of each simulation, i=1,2,3...,n, and m be the number of each soil layer, m=1,2,3,...l.

[0045] The thickness of each soil layer was extracted from the soil infiltration data of each reservoir simulation. And combining the permeability coefficients of each soil layer in each reservoir simulation, after thickness weighting: Subsequently, the overall permeability coefficient of the output reservoir in each simulation was calculated. This is used as a characteristic parameter of soil infiltration, among which Specifically, it is the percentage factor of the permeability coefficient of the m-th layer of soil stored in the data warehouse, where D is the total soil thickness.

[0046] It should be noted that the proportion of the permeability coefficient of each soil layer specifically reflects the different soil height layers, and the contribution of their corresponding permeability coefficients to the overall permeability coefficient is also different. For example, the permeability coefficient of the surface layer contributes more to the permeability coefficient of the bottom layer than that of the bottom layer. The specific determination can be made by simulating the vertical permeation process using Darcy's law or by relevant professionals.

[0047] In a specific embodiment of the present invention, the specific evaluation method for the scheduling characteristic parameters of the reservoir in each simulation is as follows: extract the demand water volume and the discharge flow from the scheduling simulation data of the reservoir in each simulation, subtract the discharge flow from the demand water volume to obtain the net flow difference, divide the net flow difference by the historical average demand water volume to obtain the dimensionless scheduling characteristic parameters.

[0048] In a specific embodiment of the present invention, the evaluation of the convergence constraint values ​​of several training base models for each structural component includes the following evaluation content: the reservoir capacity of each simulation in the validation dataset of each structural component of the reservoir and the reservoir capacity of each simulation in several training base models are respectively denoted as the actual reservoir capacity and the predicted reservoir capacity, and they are combined to obtain the vector feature data of each simulation of each training base model for each structural component of the reservoir, wherein the vector feature data points from the actual reservoir capacity to the predicted reservoir capacity.

[0049] The vector magnitudes of each simulation of each training model for each structural component of the reservoir are calculated using the vector magnitude calculation method. This constructs a data sequence of vector magnitudes for each training model of each structural component of the reservoir. Convergence constraint features are then identified, specifically including trend eigenvalues. Average vector magnitude The trend feature value specifically refers to the value obtained by trend fitting of the vector model data sequence, which represents the rate of change of the trend fitting line, j is the number of each structural component, j=1,2,3,...,k, and h is the number of each training base model, h=1,2,3,...g.

[0050] After normalizing the average vector magnitude of each training basic model for each structural component of the reservoir, it is imported into the convergence constraint value evaluation expression along with the trend feature values: In the formula, the convergence constraint values ​​of several training base models for each structural component are output, where... This indicates the appropriate range of trend characteristic values ​​stored in the data warehouse. It is the midpoint of the appropriate trend characteristic value range.

[0051] This invention trains water level prediction models under different structural components of reservoirs, making the water level prediction models of reservoirs more accurate and laying a solid data foundation for the subsequent balance of water resource allocation.

[0052] S3. Water level prediction data acquisition: Acquire soil infiltration data of the reservoir, as well as meteorological data and dispatch behavior data during the prediction period.

[0053] It should be noted that meteorological data of the reservoir during the forecast period is obtained by a micro meteorological management center built in the reservoir, scheduling behavior data of the reservoir during the forecast period is obtained by a task management center of the reservoir, and soil infiltration data is obtained based on an integrated reservoir terrain model.

[0054] S4. Reservoir Capacity Change Rate Prediction: This model integrates soil infiltration data from the reservoir, as well as meteorological and scheduling data during the prediction period. After data processing, the data is imported into the reservoir capacity change rate prediction model, which outputs the reservoir capacity change rate during the prediction period. The change rate is then converted into water level using the water level-capacity relationship curve.

[0055] S5. Reservoir Classified Regulation: Collect basic construction information of reservoirs and set risk water levels for reservoirs. Based on the water level of the reservoirs and the scheduling rule engine, classified regulation is carried out.

[0056] In a specific embodiment of the present invention, the soil infiltration characteristic parameters of each reservoir simulation are specifically evaluated by extracting the water content of each soil layer from the soil simulation infiltration data of each reservoir simulation. ,hardness Thickness, which is then imported into the empirical formula for permeability: In the process, the permeability coefficient of each soil layer is output. m, where a, b, and c are fitting parameters (specifically determined through experiments or historical data). Let be the water saturation threshold of each soil layer for the reservoir soil type stored in the data warehouse, i be the number of each simulation, i=1,2,3...,n, and m be the number of each soil layer, m=1,2,3,...l.

[0057] The thickness of each soil layer was extracted from the soil infiltration data of each reservoir simulation, and combined with the permeability coefficient of each soil layer in each reservoir simulation, and then weighted by thickness. Subsequently, the overall permeability coefficient of the output reservoir in each simulation was calculated. This is used as a characteristic parameter of soil infiltration, among which Specifically, it is the percentage factor of the permeability coefficient of the m-th layer of soil stored in the data warehouse.

[0058] It should be noted that the proportion of the permeability coefficient of each soil layer specifically reflects the different soil height layers, and the contribution of their corresponding permeability coefficients to the overall permeability coefficient is also different. For example, the permeability coefficient of the surface layer contributes more to the permeability coefficient of the bottom layer than that of the bottom layer. The specific determination can be made by simulating the vertical permeation process using Darcy's law or by relevant professionals.

[0059] In a specific embodiment of the present invention, the specific evaluation method for the scheduling characteristic parameters of the reservoir in each simulation is as follows: extract the demand water volume and the discharge flow from the scheduling simulation data of the reservoir in each simulation, subtract the discharge flow from the demand water volume to obtain the net flow difference, divide the net flow difference by the historical average demand water volume to obtain the dimensionless scheduling characteristic parameters.

[0060] In a specific embodiment of the present invention, the evaluation of the convergence constraint values ​​of several training base models for each structural component includes the following evaluation content: the reservoir capacity of each simulation in the validation dataset of each structural component of the reservoir and the reservoir capacity of each simulation in several training base models are respectively denoted as the actual reservoir capacity and the predicted reservoir capacity, and they are combined to obtain the vector feature data of each simulation of each training base model for each structural component of the reservoir, wherein the vector feature data points from the actual reservoir capacity to the predicted reservoir capacity.

[0061] The vector magnitudes of each simulation of each training model for each structural component of the reservoir are calculated using the vector magnitude calculation method. This constructs a data sequence of vector magnitudes for each training model of each structural component of the reservoir. Convergence constraint features are then identified, specifically including trend eigenvalues. Average vector magnitude The trend feature value specifically refers to the value obtained by trend fitting of the vector model data sequence, which represents the rate of change of the trend fitting line, j is the number of each structural component, j=1,2,3,...,k, and h is the number of each training base model, h=1,2,3,...g.

[0062] After normalizing the average vector magnitude of each training basic model for each structural component of the reservoir, it is imported into the convergence constraint value evaluation expression along with the trend feature values: In the formula, the convergence constraint values ​​of several training base models for each structural component are output, where... This indicates the appropriate range of trend characteristic values ​​stored in the data warehouse. It is the midpoint of the appropriate trend characteristic value range.

[0063] It should be noted that the specific process of hierarchical regulation based on reservoir water level and scheduling rule engine is as follows: subtract the risk water level from the reservoir water level to obtain the excess water level of the reservoir, and compare it with the scheduling rules corresponding to each excess water level interval stored in the data warehouse to obtain the scheduling rules of the reservoir. The more excess water level, the stronger the scheduling rule.

[0064] For example, consider three water level exceedance intervals: a first exceedance interval, a second exceedance interval, and a third exceedance interval. If the dataset for the first exceedance interval is smaller than the dataset for the second exceedance interval, and vice versa, then the strength of the scheduling rule for the first exceedance interval is less than that for the second exceedance interval, and vice versa. For instance, the scheduling rule for the first exceedance interval might be to maintain regular scheduling and periodically monitor water level changes. The scheduling rule for the second exceedance interval might be to strengthen real-time monitoring, activate the early warning system to notify relevant upstream and downstream personnel, adjust the water release volume in advance, and reserve flood storage space. The scheduling rule for the third exceedance interval might be to activate emergency flood discharge facilities, increase the outflow, coordinate joint scheduling of upstream and downstream reservoirs, and share the pressure. This approach ensures water level prediction scheduling while reducing the resource investment in maintenance and management of water level prediction scheduling.

[0065] This invention predicts the risk level of a reservoir based on its quality and historical data, ensuring the accuracy of subsequent early warnings and tiered emergency response, and avoiding resource waste.

[0066] In a specific embodiment of the present invention, the method for collecting the basic construction information of the reservoir and setting the risk water level of the reservoir is as follows: extracting the feature values ​​of several construction feature parameters from the basic construction information of the reservoir, and obtaining the first risk feature parameter of the reservoir through data processing. .

[0067] Based on the reservoir's water level during the forecast period, combined with an integrated reservoir topography model, the water slope of the reservoir during the forecast period is obtained. Combined with the river's high-water-low-water ratio and downstream population density stored in the database, the second risk characteristic parameter of the reservoir is obtained through data processing. .

[0068] The first and second risk characteristic parameters of the reservoir are imported into the risk water level setting model, and the risk water level of the reservoir is output.

[0069] In a specific embodiment of the present invention, the first risk characteristic parameter of the reservoir is determined based on the characteristic values ​​of several construction characteristic parameters, and is used to quantify the risk construction quality of the reservoir.

[0070] It should be noted that the first risk characteristic parameter of the reservoir is obtained by: uniformizing the characteristic values ​​of several construction characteristic parameters of the reservoir and then performing a weighted summation to obtain the first risk characteristic parameter of the reservoir. The weighted summation specifically involves adding the characteristic values ​​of each construction characteristic parameter to their corresponding proportion coefficients. The corresponding proportion coefficients of each construction characteristic parameter are specifically obtained from the data warehouse. The proportion coefficients of each construction characteristic parameter reflect the degree and direction of the influence of the construction characteristic parameter on the risk water level. The direction of influence includes positive and negative influences.

[0071] The characteristic values ​​of each construction feature parameter include, but are not limited to, usage time, number of cracks, crack width, number of compaction passes, and compaction speed.

[0072] The second risk characteristic parameter of the reservoir is specifically determined based on the water body slope, river flood-dry ratio, and downstream population density within the prediction period. It is used to quantify the reservoir's carrying capacity against the impact of water bodies. The greater the water body slope, the greater the river flood-dry ratio, and the greater the downstream population density, the smaller the reservoir's carrying capacity against the impact of water bodies, and the greater the second risk characteristic parameter.

[0073] It should be noted that the specific analysis process of the second risk characteristic parameter of the reservoir is the same as that of the first risk characteristic parameter, and will not be repeated here.

[0074] In a specific embodiment of the present invention, the risk level setting model is specifically as follows: .

[0075] In the formula The first risk characteristic parameter as the benchmark Benchmark Second Risk Characteristic Parameter The corresponding risk level, , These are the corrected water level parameters corresponding to the first risk characteristic parameter of unit offset and the corrected water level parameters corresponding to the second risk characteristic parameter of unit offset, respectively, which are preset in the data warehouse.

[0076] Reference Figure 2 As shown, the second aspect of the present invention provides a system for executing the intelligent water level prediction and control method for water conservancy projects described in the present invention, comprising: a reservoir three-dimensional model construction module: collecting topographic data of the reservoir bank and surrounding areas, generating a three-dimensional real-scene model of the reservoir, collecting underwater topographic data, constructing a three-dimensional model of the water area, merging the three-dimensional real-scene model of the reservoir and the three-dimensional model of the water area to form an integrated reservoir topographic model, and extracting the water level-reservoir capacity relationship curve.

[0077] The water level prediction model module is equipped with a reservoir topography model to simulate the reservoir under different environments, obtain reservoir simulation data, and train a reservoir capacity change rate prediction model based on the reservoir simulation data.

[0078] Water level prediction data acquisition module: Acquires soil infiltration data of the reservoir, as well as meteorological data and scheduling behavior data during the prediction period.

[0079] The reservoir capacity change rate prediction module integrates soil infiltration data of the reservoir with meteorological data and scheduling behavior data during the prediction period. After data processing, the data is imported into the reservoir capacity change rate prediction model, and the reservoir capacity change rate during the prediction period is output. The data is then converted into water level using the water level-capacity relationship curve.

[0080] The reservoir hierarchical control module collects basic construction information of the reservoir and sets the risk water level of the reservoir. Based on the water level of the reservoir and the scheduling rule engine, it performs hierarchical control.

[0081] It should be noted that the present invention also includes a data warehouse for storing relevant data required for reservoir forecasting.

[0082] The present invention and its embodiments have been described above. This description is not restrictive. The accompanying drawings are only one embodiment of the present invention. The actual structure is not limited to this. In short, if a person skilled in the art is inspired by this description and designs a similar structure and embodiment without departing from the spirit of the present invention, such design should fall within the protection scope of the present invention.

Claims

1. A method for intelligent water level prediction and control in water conservancy projects, characterized in that, include: S1. Reservoir 3D Model Construction: Collect topographic data of the reservoir bank and surrounding area, generate a 3D real-scene model of the reservoir, collect underwater topographic data, construct a 3D model of the water area, integrate the 3D real-scene model of the reservoir and the 3D model of the water area to form an integrated reservoir topographic model, and extract the water level-reservoir capacity relationship curve. S2. Water level prediction model: Based on the reservoir topography model, the reservoir is simulated under different environments to obtain reservoir simulation data, and a reservoir capacity change rate prediction model is trained based on the reservoir simulation data. The reservoir simulation data includes reservoir capacity, meteorological simulation data, soil infiltration simulation data, and dispatch behavior simulation data for each simulation. The meteorological simulation data includes the characteristic values ​​of each meteorological characteristic parameter; The soil infiltration simulation data includes the water content of each soil layer; The simulation data of the scheduling behavior includes water demand and water release flow. Specifically, the water demand refers to the total amount of water drawn from the reservoir by surrounding farmland, residents, and other sources. The specific training method for the reservoir capacity change rate prediction model trained based on reservoir simulation data is as follows: A1. Extract meteorological simulation data, soil infiltration data and scheduling behavior data from each reservoir simulation data, and obtain the meteorological characteristic parameters, soil infiltration characteristic parameters and scheduling characteristic parameters of each reservoir simulation through data processing. A2. Divide the reservoir simulation data into training dataset and validation dataset according to a set ratio. Divide the training set into training sets for each structural component according to the reservoir capacity, and divide the validation set into validation sets for each structural component. A3. Obtain several training base models from the data warehouse, and import the reservoir capacity, meteorological characteristic parameters, soil infiltration characteristic parameters and scheduling characteristic parameters of each structural component in the training dataset of the reservoir into several training base models, thereby determining the coefficients of several training base models of each structural component and obtaining the exact training base models of each structural component. The aforementioned basic training models are specifically combinations of the characteristic relationships between meteorological characteristic parameters, soil infiltration characteristic parameters, scheduling characteristic parameters, and reservoir capacity. The training base model coefficients are specifically the correction coefficients for the characteristic relationships between meteorological characteristic parameters, soil infiltration characteristic parameters, scheduling characteristic parameters, and reservoir capacity. A4. Import the meteorological characteristic parameters, soil infiltration characteristic parameters, and scheduling characteristic parameters of each simulation in the validation dataset of each structural component of the reservoir into several training basic models of the corresponding structural components, so as to obtain the reservoir capacity of each simulation of several training basic models of each structural component on the validation set, and thus evaluate the convergence constraint value of several training basic models of each structural component. A5. For each structural component, select the training base model with the largest convergence constraint value as the reservoir capacity change rate prediction model for the corresponding structural component, and summarize them to obtain the reservoir capacity change rate prediction model. The specific evaluation method for the soil infiltration characteristic parameters of the reservoir in each simulation is as follows: The water content of each soil layer was extracted from the soil infiltration data of each reservoir simulation. ,hardness Thickness, which is then imported into the empirical formula for permeability: In the process, the permeability coefficient of each soil layer is output. m, where a, b, and c are the fitting parameters. Let be the water saturation threshold of each soil layer for the reservoir soil type stored in the data warehouse, i be the number of each simulation, i=1,2,3...,n, and m be the number of each soil layer, m=1,2,3,...l; The thickness of each soil layer was extracted from the soil infiltration data of each reservoir simulation. And combining the permeability coefficients of each soil layer in each reservoir simulation, after thickness weighting: Subsequently, the overall permeability coefficient of the output reservoir in each simulation was calculated. This is used as a characteristic parameter of soil infiltration, among which Specifically, it is the proportion factor of the permeability coefficient of the m-th soil layer stored in the data warehouse, where D is the total soil thickness; The specific evaluation method for the scheduling characteristic parameters of the reservoir in each simulation is as follows: The water demand and water release flow are extracted from the scheduling simulation data of each reservoir simulation. The net flow difference is obtained by subtracting the water release flow from the water demand. The net flow difference is divided by the historical average water demand to obtain the dimensionless scheduling characteristic parameter. The evaluation of the convergence constraint values ​​of several training base models for each structural component specifically includes: The reservoir capacity simulated in each simulation of the verification dataset of each structural component of the reservoir and the reservoir capacity simulated in each simulation of several training basic models are denoted as actual reservoir capacity and predicted reservoir capacity, respectively. They are combined to obtain vector feature data of each simulation of each training basic model of each structural component of the reservoir. The vector feature data points from actual reservoir capacity to predicted reservoir capacity. The vector magnitudes of each simulation of each training model for each structural component of the reservoir are calculated using the vector magnitude calculation method. This constructs a data sequence of vector magnitudes for each training model of each structural component of the reservoir. Convergence constraint features are then identified, specifically including trend eigenvalues. Average vector magnitude The trend feature value specifically refers to the value obtained by trend fitting of the vector model data sequence, which represents the rate of change of the trend fitting line, j is the number of each structural component, j=1,2,3...,k, and h is the number of each training base model, h=1,2,3,...g; After normalizing the average vector magnitude of each training basic model for each structural component of the reservoir, it is imported into the convergence constraint value evaluation expression along with the trend feature values: In the formula, the convergence constraint values ​​of several training base models for each structural component are output, where... This indicates the appropriate range of trend characteristic values ​​stored in the data warehouse. The midpoint of the appropriate trend characteristic value range; S3. Water level prediction data acquisition: Acquire soil infiltration data of the reservoir, as well as meteorological data and dispatch behavior data during the prediction period; S4. Reservoir capacity change rate prediction: Integrating soil infiltration data of the reservoir with meteorological data and scheduling behavior data during the prediction period, the data is processed and imported into the reservoir capacity change rate prediction model. The output is the reservoir capacity change rate during the prediction period, which is converted into water level through the water level-capacity relationship curve. S5. Reservoir Classified Regulation: Collect basic construction information of reservoirs and set risk water levels for reservoirs. Based on the water level of the reservoirs and the scheduling rule engine, classified regulation is carried out.

2. The intelligent water level prediction and control method for water conservancy projects according to claim 1, characterized in that, The specific method for collecting basic construction information of the reservoir and setting the risk water level of the reservoir is as follows: The feature values ​​of several construction characteristic parameters are extracted from the basic construction information of the reservoir, and the first risk characteristic parameter of the reservoir is obtained through data processing. ; Based on the reservoir's water level during the forecast period, combined with an integrated reservoir topography model, the water slope of the reservoir during the forecast period is obtained. Combined with the river's high-water-low-water ratio and downstream population density stored in the database, the second risk characteristic parameter of the reservoir is obtained through data processing. ; The first and second risk characteristic parameters of the reservoir are imported into the risk water level setting model, and the risk water level of the reservoir is output.

3. The intelligent water level prediction and control method for water conservancy projects according to claim 2, characterized in that, The first risk characteristic parameter of the reservoir is specifically determined based on the characteristic values ​​of several construction characteristic parameters, and is used to quantify the risk construction quality of the reservoir. The second risk characteristic parameter of the reservoir is specifically determined based on the water body slope, river flood-dry ratio, and downstream population density within the prediction period. It is used to quantify the reservoir's carrying capacity against the impact of water bodies. The greater the water body slope, the greater the river flood-dry ratio, and the greater the downstream population density, the smaller the reservoir's carrying capacity against the impact of water bodies, and the greater the second risk characteristic parameter.

4. The intelligent water level prediction and control method for water conservancy projects according to claim 2, characterized in that, The risk level setting model is as follows: ; In the formula The first risk characteristic parameter as the benchmark Benchmark Second Risk Characteristic Parameter The corresponding risk level, , These are the corrected water level parameters corresponding to the first risk characteristic parameter of unit offset and the corrected water level parameters corresponding to the second risk characteristic parameter of unit offset, respectively, which are preset in the data warehouse.

5. A system for executing the intelligent water level prediction and control method for water conservancy projects according to any one of claims 1-4, characterized in that, include: The reservoir 3D model construction module collects topographic data of the reservoir bank and surrounding area, generates a 3D real-scene model of the reservoir, collects underwater topographic data, constructs a 3D model of the water area, integrates the 3D real-scene model of the reservoir and the 3D model of the water area to form an integrated reservoir topographic model, and extracts the water level-reservoir capacity relationship curve. The water level prediction model module is equipped with a reservoir topography model to simulate the reservoir under different environments, obtain reservoir simulation data, and train a reservoir capacity change rate prediction model based on the reservoir simulation data. Water level prediction data acquisition module: Acquires soil infiltration data of the reservoir, as well as meteorological data and dispatch behavior data during the prediction period; Reservoir capacity change rate prediction module: It integrates soil infiltration data of the reservoir with meteorological data and scheduling behavior data during the prediction period. After data processing, it is imported into the reservoir capacity change rate prediction model and outputs the reservoir capacity change rate during the prediction period. The data is converted into water level through the water level-capacity relationship curve. The reservoir hierarchical control module collects basic construction information of the reservoir and sets the risk water level of the reservoir. Based on the water level of the reservoir and the scheduling rule engine, it performs hierarchical control.

Images