A method and system for multi-stage coordinated protection of a hydropower station flood discharge atomization

Abstract

The application provides a kind of water power station flood discharge atomization multistage collaborative protection method and system, it is related to water conservancy and hydropower engineering management technical field, the method includes: real-time acquisition flood discharge atomization area sparse rain intensity observation data and flood discharge gate operating condition data;According to sparse rain intensity observation data and operating condition data, the model parameters of the preset flood discharge atomization physical mathematics model are dynamically corrected, and the full-field atomization rain intensity spatial distribution is obtained;According to the full-field atomization rain intensity spatial distribution, the risk assessment is carried out in combination with the digital model of the plant area in the flood discharge atomization area, and the real-time risk index of each key partition in the plant area is obtained;Through multi-objective optimization algorithm, according to the real-time risk index and operating condition data, the gate operation strategy of flood discharge gate and the plant area protection instruction are determined;According to the gate operation strategy and plant area protection instruction, the flood discharge gate is controlled and the plant area is protected respectively.The application improves the protection effect for flood discharge atomization.

Description

Technical Field

[0001] This invention relates to the field of water conservancy and hydropower engineering management technology, and more specifically, to a method and system for multi-level coordinated protection of flood discharge misting in hydropower stations. Background Technology

[0002] When a high-head hydropower station releases floodwater, the jet of water splashes into the water, generating a large amount of water mist, forming atomized rainfall far exceeding the intensity of natural rainstorms. This atomized rainfall can lead to water accumulation in the power plant area, equipment flooding, and structural leakage, seriously threatening power generation safety and project stability. Currently, flood discharge atomization is typically predicted using offline models (e.g., analyzing the impact of wind fields on atomization distribution), and the flood discharge gates are then controlled and the plant area protected based on the prediction results.

[0003] In related technologies, offline models are only used for offline analysis during the design phase and post-construction review and verification. The prediction parameters of the model are relatively fixed, so they cannot be adaptively updated according to real-time operating conditions such as reservoir water level, gate opening, wind field, and inflow tongue shape. As a result, the prediction error accumulates over time, making the prediction accuracy of the offline model worse and worse. Consequently, the floodgate control and plant protection based on the prediction results cannot guarantee effective protection, thus affecting the safety of the plant. Summary of the Invention

[0004] The problem addressed by this invention is how to improve the protective effect against flood discharge atomization.

[0005] To address the aforementioned problems, this invention provides a multi-level collaborative protection method and system for flood discharge atomization in hydropower stations.

[0006] In a first aspect, the present invention provides a multi-level coordinated protection method for flood discharge atomization in hydropower stations, comprising: Real-time acquisition of multi-source observation data of the flood discharge atomization area, including sparse rainfall intensity observation data and flood discharge gate operation status data; Based on the sparse rainfall intensity observation data and the operational condition data, the model parameters of the preset flood discharge atomization physical-mathematical model are dynamically corrected to obtain the spatial distribution of the full-field atomized rainfall intensity in the flood discharge atomization area. The dynamic correction includes: using the sparse rainfall intensity observation data as an observation constraint, employing an ensemble Kalman filter algorithm to dynamically update the water droplet parameters of the flood discharge atomization physical-mathematical model based on the observation constraint, to obtain the corrected water droplet parameters; and driving the flood discharge atomization physical-mathematical model to perform analytical calculations based on the corrected water droplet parameters and the operational condition data to obtain the spatial distribution of the full-field atomized rainfall intensity in the flood discharge atomization area. Based on the spatial distribution of the atomized rainfall intensity across the entire field, and combined with the digital model of the plant area within the flood discharge atomization zone, a risk assessment is conducted to obtain the real-time risk index for each key zone within the plant area. Using a multi-objective optimization algorithm, based on the real-time risk index and the operating condition data, with the objective of minimizing the overall weighted risk value of the plant area, and with constraints such as the total discharge flow, the upper and lower limits of the gate opening, and the gate adjustment rate limit, the gate operation strategy and plant area protection instructions of the flood discharge gate are solved simultaneously. The floodgates are controlled and the plant area is protected according to the gate operation strategy and the plant area protection instructions, respectively.

[0007] Optionally, the real-time acquisition of multi-source observation data of the flood discharge atomization area includes: By setting up a rainfall monitoring array based on the vertical attenuation law of atomized rainfall intensity and the elevation gradient of the vulnerability of the protection target of the plant area, the rainfall intensity data corresponding to multiple preset areas in the flood discharge atomization area are acquired in real time according to a preset sampling frequency; wherein, the rainfall monitoring array includes at least three rain gauges arranged along different elevations, and the elevation difference between adjacent rain gauges decreases as the elevation increases; The rainfall intensity data corresponding to all the preset areas are used as the sparse rainfall intensity observation data. The control unit of the flood discharge gate acquires the gate opening, reservoir water level, and total discharge flow in real time according to the preset sampling frequency. The gate opening, the reservoir water level, and the total discharge flow are used as the operating condition data of the flood discharge gate.

[0008] Optionally, the sparse rainfall intensity observation data is used as an observation constraint, and an ensemble Kalman filter algorithm is employed to dynamically update the droplet parameters of the flood discharge atomization physical-mathematical model based on the observation constraints, to obtain the corrected droplet parameters, including: Based on the aforementioned flood discharge atomization physical and mathematical model, the water droplet parameters are initialized, including the average droplet size, droplet size variance, average initial velocity, velocity variance, and mean exit angle. Based on the physical and mathematical model of flood discharge atomization and the operating condition data, the water droplet parameters are calculated in the forward direction to obtain the model-predicted rainfall intensity set corresponding to the sparse rainfall intensity observation data. The observation error covariance is obtained by comparing the predicted rainfall intensity set based on the model with the sparse rainfall intensity observation data. By using the Kalman filter analysis equations, the water droplet parameters are iteratively updated according to the observation error covariance until the water droplet parameters meet the convergence condition, thus obtaining the corrected water droplet parameters; wherein, the convergence condition is that the state increment of the particle size parameter is less than 0.1 mm and the state increment of the angle parameter is less than 1°.

[0009] Optionally, the step of driving the flood discharge atomization physical-mathematical model to perform analytical calculations based on the corrected water droplet parameters and the operating condition data to obtain the spatial distribution of the full-field atomization rainfall intensity in the flood discharge atomization area includes: The corrected water droplet parameters are used as input parameters for the joint distribution function in the flood discharge atomization physical and mathematical model. The gate opening, reservoir water level and total discharge flow in the operating condition data are used as model boundary conditions. The motion trajectory and spatial distribution of water droplets in the flood discharge atomization area are determined by the water droplet splashing motion differential equation. Based on the trajectory and spatial distribution of the water droplets, the rainfall intensity contribution value of each spatial grid cell in the flood discharge atomization area is determined by calculation according to the spatial grid cells in the flood discharge atomization area. The rainfall intensity contribution values ​​of all spatial grid cells within the flood discharge atomization area are spatially combined and mapped to generate the spatial distribution of the full-field atomized rainfall intensity.

[0010] Optionally, the risk assessment is performed based on the spatial distribution of the overall atomized rainfall intensity, combined with a digital model of the plant area within the flood discharge atomization zone, to obtain a real-time risk index for each key zone within the plant area, including: Based on the digital model of the factory area, the location range, equipment waterproofing level, and asset importance information of key zones within the factory area are extracted; Based on the location range of the key partitions within the factory area, the waterproofing level of the equipment, and the asset importance information, determine the vulnerability coefficient corresponding to each key partition; Based on the spatial distribution of the overall fogged rainfall intensity, the maximum predicted rainfall intensity corresponding to the key zone is determined; The real-time risk index of the key partition is obtained by using a preset risk assessment model:

[0011] in, Let be the real-time risk index for the i-th critical partition. The maximum 10-minute rainfall intensity is retrieved from the key partition. The preset baseline rainfall intensity (range 80-120 mm / h) is used. This represents the vulnerability coefficient of the partition.

[0012] Optionally, the gate operation strategy includes the opening combination mode of the flood discharge gate, the gate opening degree, and the adjustment sequence; the gate operation strategy and plant protection instructions for the flood discharge gate are simultaneously solved using a multi-objective optimization algorithm, based on the real-time risk index and the operating condition data, with the objective of minimizing the overall weighted risk value of the plant area, and with constraints such as the total discharge flow, the upper and lower limits of the gate opening degree, and the gate adjustment rate limit, including: Using the operating parameters of the flood discharge gate as decision variables, minimizing the weighted risk value of the plant area as the optimization objective, and the total discharge flow, gate opening limits, and gate adjustment rate limits as constraints, an optimization model is constructed through an objective function.

[0013] The optimization model is iteratively optimized and solved using the multi-objective optimization algorithm based on the real-time risk index and the operating condition data to obtain the opening combination mode, gate opening degree and adjustment sequence of the flood discharge gate; The plant area protection command is generated based on the opening combination method, the gate opening degree, and the adjustment sequence.

[0014] Optionally, the objective function is: ; in, Let ∑ωi=1 be the importance weight of the i-th key partition, X be the activation combination method, and R be the value of the key partition. i (X) is the real-time risk index of the i-th key partition under the opening combination mode X, and m is the total number of key partitions; The constraints are as follows: , , ; in, Let Q(X) be the total discharge capacity of the gates, representing the opening combinations of n gates. Let G be the target discharge flow rate, ε be the allowable error, and G be the allowable discharge flow rate. j G represents the opening degree of the j-th gate. min G max These are the lower and upper limits of the gate's opening, ΔG and ΔG, respectively. j / Δt is the gate adjustment rate of the gate, V max ΔG is the maximum permissible adjustment rate of the gate. j For the gate at time intervals The change in opening degree within.

[0015] Optionally, generating the plant protection command according to the gate operation strategy includes: Based on the gate operation strategy, the risk index of the key partition is predicted to obtain the predicted risk index value of the key partition under the gate operation strategy. Based on the predicted risk index value, the control commands for the active protection facilities of the plant area are generated by matching according to the preset risk-command matching rules. The control commands of the active protection facilities shall be used as the plant area protection commands.

[0016] Optionally, the step of controlling the floodgate and protecting the plant area according to the gate operation strategy and the plant area protection command respectively includes: The gate operation strategy is sent to the control unit of the flood discharge gate, and the flood discharge gate is regulated according to the opening combination method, the gate opening degree and the adjustment sequence. The plant protection command is sent to the control unit of the active protection facility in the plant area for regulation via an industrial communication protocol.

[0017] Secondly, the present invention provides a multi-level coordinated protection system for flood discharge atomization in hydropower stations, comprising: The data acquisition module is used to acquire multi-source observation data of the flood discharge atomization area in real time. The multi-source observation data includes sparse rainfall intensity observation data and the operating status data of the flood discharge gate. The model calibration module is used to dynamically calibrate the model parameters of a preset flood discharge atomization physical-mathematical model based on the sparse rainfall intensity observation data and the operating condition data, thereby obtaining the spatial distribution of the full-field atomized rainfall intensity in the flood discharge atomization area. The dynamic calibration includes: using the sparse rainfall intensity observation data as an observation constraint, employing an ensemble Kalman filter algorithm to dynamically update the water droplet parameters of the flood discharge atomization physical-mathematical model based on the observation constraint, thereby obtaining the calibrated water droplet parameters; and driving the flood discharge atomization physical-mathematical model to perform analytical calculations based on the calibrated water droplet parameters and the operating condition data, thereby obtaining the spatial distribution of the full-field atomized rainfall intensity in the flood discharge atomization area. The risk assessment module is used to conduct risk assessment based on the spatial distribution of the atomized rainfall intensity across the entire site, combined with the digital model of the plant area within the flood discharge atomization area, to obtain the real-time risk index of each key zone within the plant area. The strategy generation module is used to simultaneously solve for the gate operation strategy of the flood discharge gate and the protection instructions of the plant area by using a multi-objective optimization algorithm, based on the real-time risk index and the operating condition data, with the goal of minimizing the overall weighted risk value of the plant area, and with the constraints of the total discharge flow, the upper and lower limits of the gate opening and the gate adjustment rate limit. The control module is used to control the flood discharge gate and protect the plant area according to the gate operation strategy and the plant area protection command, respectively.

[0018] The present invention relates to a multi-level collaborative protection method and system for flood discharge atomization in hydropower stations. First, it acquires multi-source observation data in real time, including sparse rainfall intensity observation data and floodgate operation data, of the flood discharge atomization area. This rapidly captures the actual state of the atomization site and key parameters of flood discharge operation, avoiding biased protection decisions due to missing or delayed data. Second, it dynamically corrects the model parameters of a pre-set physical-mathematical model of flood discharge atomization based on sparse rainfall intensity observation data and operation data, obtaining the spatial distribution of atomization rainfall intensity across the entire site. This avoids the shortcomings of traditional offline models with fixed parameters that cannot adapt to real-time conditions, allowing model parameters to adaptively update with operating conditions, significantly reducing prediction errors and restoring the entire atomization distribution, including areas that cannot be monitored. Finally, it conducts a risk assessment based on the spatial distribution of atomization rainfall intensity across the entire site combined with the plant's digital model, obtaining the results of various... The real-time risk index for key zones transforms abstract atomized rainfall intensity into quantifiable and intuitive plant safety risk indicators, clearly identifying high-risk areas and risk levels for precise risk identification. Then, a multi-objective optimization algorithm determines the floodgate operation strategy and plant protection instructions based on the real-time risk index and operational data, thus balancing flood discharge safety requirements with plant protection needs. This simultaneously optimizes source-level flood discharge operation and end-point protection measures, achieving coordinated source-sink control and avoiding protection failures caused by single-control methods. Finally, by controlling the floodgates and plant area to perform coordinated protection according to the gate operation strategy and plant protection instructions, optimized decisions can be quickly translated into actual protection actions, achieving proactive protection and comprehensively improving the timeliness, accuracy, and effectiveness of flood discharge atomization protection, thereby enhancing the protective effect against flood discharge atomization at hydropower stations. Attached Figure Description

[0019] Figure 1 This is a schematic flowchart of the multi-level coordinated protection method for flood discharge atomization in hydropower stations according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the three-dimensional gradient layout of the flood discharge atomization area according to another embodiment of the present invention; Figure 3 This is a flowchart of the atomization field inversion process according to another embodiment of the present invention; Figure 4 This is an optimized flowchart of another embodiment of the present invention; Figure 5 This is a schematic diagram of the structure of a multi-level coordinated protection system for flood discharge atomization in a hydropower station, according to another embodiment of the present invention. Detailed Implementation

[0020] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Although some embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of the present invention. It should be understood that the accompanying drawings and embodiments of the present invention are for illustrative purposes only and are not intended to limit the scope of protection of the present invention.

[0021] It should be understood that the various steps described in the method embodiments of the present invention may be performed in different orders and / or in parallel. Furthermore, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present invention is not limited in this respect.

[0022] The term "comprising" and its variations as used herein are open-ended, meaning "including but not limited to"; the term "based on" means "at least partially based on"; the term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments"; and the term "optionally" means "optional embodiments". Definitions of other terms will be given in the following description. It should be noted that the concepts of "first," "second," etc., mentioned in this invention are used only to distinguish different devices, modules, or units, and are not intended to limit the order of functions performed by these devices, modules, or units or their interdependencies.

[0023] It should be noted that the terms "a" and "a plurality of" used in this invention are illustrative rather than restrictive. Those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0024] Combination Figure 1 As shown in the figure, an embodiment of the present invention provides a multi-level coordinated protection method for flood discharge atomization in hydropower stations, comprising: Real-time acquisition of multi-source observation data of the flood discharge atomization area, including sparse rainfall intensity observation data and flood discharge gate operation status data.

[0025] Specifically, for sparse rainfall intensity observation data from multi-source observation data, rainfall monitoring equipment is typically deployed using a three-dimensional characteristic elevation gradient based on the vertical attenuation abrupt change points of the atomized rainfall intensity and the vulnerability elevation of the protected target, to collect discrete and sparse measured rainfall intensity information. For the operational status data of the flood discharge gates from multi-source observation data, the data is integrated into the power station monitoring system to obtain operational status data such as gate opening, reservoir water level, and total discharge flow. In a preferred embodiment of the invention, the multi-source observation data also includes image information related to the water tongue morphology simultaneously acquired by a high-definition fog-penetrating pan-tilt camera on the dam crest, forming multi-source real-time sensing data covering the atomization site and flood discharge operation.

[0026] Based on the sparse rainfall intensity observation data and the operational condition data, the model parameters of the preset flood discharge atomization physical-mathematical model are dynamically corrected to obtain the spatial distribution of the full-field atomized rainfall intensity in the flood discharge atomization area. The dynamic correction includes: using the sparse rainfall intensity observation data as an observation constraint, employing an ensemble Kalman filter algorithm to dynamically update the water droplet parameters of the flood discharge atomization physical-mathematical model based on the observation constraint, to obtain the corrected water droplet parameters; and driving the flood discharge atomization physical-mathematical model to perform analytical calculations based on the corrected water droplet parameters and the operational condition data to obtain the spatial distribution of the full-field atomized rainfall intensity in the flood discharge atomization area.

[0027] Specifically, the pre-defined physical and mathematical model for flood discharge atomization in engineering applications is a fixed-parameter model calibrated offline, including the differential equations of water droplet splashing motion and the joint distribution function of initial water droplet parameters. Initial parameter calibration based solely on hydraulic model tests and historical operating data during the design phase cannot adapt to the dynamic changes in real-time operating conditions such as reservoir water level, gate opening, wind field, and water jet morphology during flood discharge. Directly using fixed parameters for calculations would lead to a continuously accumulating deviation between the model's predicted values ​​and the actual atomized rainfall intensity on site, failing to accurately reflect the real-time spatial distribution of flood discharge atomization. Therefore, dynamic correction of the model parameters is necessary to ensure that the model output closely matches the real-time observations from sparse measuring points, thereby inverting the actual atomized rainfall intensity distribution across the entire field. Specifically, using sparse measured rainfall intensity as the observation constraint, an ensemble Kalman filter data assimilation algorithm is employed to dynamically iteratively correct the internal state parameters (such as average droplet size, droplet size variance, average initial velocity, velocity variance, and mean exit angle) in the physical mathematical model of flood discharge atomization. Then, based on the corrected model parameters and operational data, the spatial distribution of atomized rainfall intensity across the entire field, including areas where measurement points cannot be deployed, is obtained through analytical calculation using the differential equation of droplet splashing motion. For dynamic correction, the sparsely distributed measured rainfall intensity is used as the observation constraint, rather than the traditional static interpolation basis. An ensemble Kalman filter algorithm is used, with a state vector θ (containing key statistics from the joint distribution function of droplet initial parameters, such as average droplet size, droplet size variance, average initial velocity, velocity variance, and mean exit angle) representing the internal state of the physical model (e.g., a model constructed based on the joint distribution function of droplet motion differential equations and maximum entropy). The ensemble Kalman filter algorithm uses sparse measured values ​​as true observations and dynamically corrects these state parameters through iterative updates (e.g., 20 iterations), making the model output approximate the measured values, thus solving the problem of error accumulation caused by the model not recognizing data and fixed parameters. Secondly, after obtaining the corrected droplet parameters, instead of mathematical interpolation, it drives the corrected physical model to perform analytical calculations. Specifically, the corrected parameters are used as input to the joint distribution function, and the operating condition data (gate opening, reservoir water level) are used as boundary conditions. The motion trajectory of each droplet (or droplet set) is calculated through the differential equation of droplet splashing motion, and then the rainfall intensity contribution value of each spatial grid unit (e.g., 10m×10m) is accumulated and calculated. Finally, a high-precision spatial distribution map of rainfall intensity across the entire field, including the atomization center area where measurement points cannot be deployed, is synthesized. This achieves a leap from sparse point observations to a full-field distribution supported by physical mechanisms.

[0028] Based on the spatial distribution of the atomized rainfall intensity across the entire field, and combined with the digital model of the plant area within the flood discharge atomization zone, a risk assessment is conducted to obtain the real-time risk index for each key zone within the plant area.

[0029] Specifically, by overlaying and matching the spatial distribution of fogging rainfall intensity across the entire site with the digital model, risk calculations are performed on each key zone. The digital model is a three-dimensional digital model of the plant area constructed using oblique photogrammetry and BIM technology, containing complete information such as topography, building structure, drainage network, electrical equipment location, waterproofing level, and asset importance. Key zones refer to areas within the plant area that are significantly affected by flood discharge fogging and are crucial to the safe operation of the power station, including the main transformer room, power distribution room, GIS room, entrance gate, main plant roof, access road, and drainage confluence area. The real-time risk index is a dimensionless quantitative indicator obtained based on inverted rainfall intensity and zone vulnerability calculations. It completes the quantitative assessment of fogging risk in the plant area, intuitively reflecting the degree of safety threat posed by current fogging rainfall to the corresponding zone. It is used to determine whether to initiate collaborative optimization and serves as the core input of the multi-objective optimization algorithm, providing a quantitative decision-making basis for the linkage between flood discharge gate control and plant protection facilities.

[0030] Using a multi-objective optimization algorithm, based on the real-time risk index and the operating condition data, with the objective of minimizing the overall weighted risk value of the plant area, and with constraints such as the total discharge flow, the upper and lower limits of the gate opening, and the gate adjustment rate limit, the gate operation strategy and plant area protection instructions of the flood discharge gate are solved simultaneously.

[0031] Specifically, the multi-objective optimization algorithm includes, but is not limited to, the non-dominated sorting genetic algorithm with elitist strategy (NSGA-II), the strength Pareto evolutionary algorithm (SPEA2), or the decomposition-based multi-objective evolutionary algorithm (MOEA / D), etc. The present invention preferably adopts the non-dominated sorting genetic algorithm with elitist strategy (NSGA-II). Specifically, the decision variables are the combinations of opening degrees of each flood discharge gate, the objective is to minimize the overall weighted risk of the plant area, and the constraints are the achievement of the total discharge target, the upper and lower limits of gate opening degree, and the gate opening and closing rate. A multi-objective optimization algorithm is used for iterative solution to obtain the optimized gate operation strategy. The gate operation strategy refers to the scheme for coordinated control of each flood discharge gate under the premise of meeting the total discharge requirement. It includes the gate opening combination method, the opening value of each gate, the gate adjustment sequence, and the opening and closing rate. For example, adjusting a single gate to be fully open to a combination of multiple gates, reducing the opening degree of a certain surface gate from 100% to 70% and simultaneously opening another surface gate to 30%, or adjusting the gate opening degree in steps at a rate not exceeding 5% per minute. Plant area protection instructions refer to the plant area active protection facility control instructions automatically generated according to the risk level. They include drainage facility control instructions and building airtight protection instructions, such as starting a designated standby drainage pump, opening the electric valve of the drainage ditch, closing the intercepting ditch valve, lowering the waterproof roller shutter at the entrance of the main transformer room or GIS room, and closing the louvers of the ventilation opening.

[0032] The floodgates are controlled and the plant area is protected according to the gate operation strategy and the plant area protection instructions, respectively.

[0033] Specifically, the gate operation strategy is pushed to the flood discharge operation control system for execution. At the same time, the plant protection instructions are sent to the corresponding execution units through industrial communication protocols such as Modbus TCP and OPC UA, which automatically complete protective actions such as starting and stopping drainage pumps, adjusting electric valves, and raising and lowering waterproof roller shutters.

[0034] The multi-level collaborative protection method for flood discharge atomization in hydropower stations of the present invention firstly acquires multi-source observation data in real time, including sparse rainfall intensity observation data and flood discharge gate operation data, in the flood discharge atomization area. This rapidly captures the actual state of the atomization site and key parameters of flood discharge operation, avoiding biased protection decisions due to missing or delayed data. Secondly, the model parameters of the preset flood discharge atomization physical-mathematical model are dynamically corrected based on the sparse rainfall intensity observation data and operation data to obtain the spatial distribution of atomization rainfall intensity across the entire site. This avoids the defects of traditional offline models with fixed parameters that cannot adapt to real-time conditions, allowing the model parameters to adaptively update with the operating conditions, significantly reducing prediction errors and restoring the entire atomization distribution, including areas that cannot be monitored. Finally, a risk assessment is conducted based on the spatial distribution of atomization rainfall intensity across the entire site combined with the plant's digital model to obtain the key parameters. The zoned real-time risk index transforms abstract atomized rainfall intensity into quantifiable and intuitive plant area safety risk indicators, clearly identifying high-risk areas and risk levels for precise risk identification. Then, a multi-objective optimization algorithm determines the floodgate operation strategy and plant area protection instructions based on the real-time risk index and operational data. This balances flood discharge safety requirements with plant area protection needs, simultaneously optimizing source flood discharge operation and end-point protection measures to achieve coordinated source-sink control and avoid protection failures caused by single-control measures. Finally, coordinated protection is implemented on the floodgates and plant area according to the gate operation strategy and plant area protection instructions. This allows optimized decisions to be quickly translated into actual protection actions, achieving proactive protection and comprehensively improving the timeliness, accuracy, and effectiveness of flood discharge atomization protection, thereby enhancing the protective effect against flood discharge atomization at hydropower stations.

[0035] Optionally, the real-time acquisition of multi-source observation data of the flood discharge atomization area includes: By setting up a rainfall monitoring array based on the vertical attenuation law of atomized rainfall intensity and the elevation gradient of the vulnerability of the protection target of the plant area, the rainfall intensity data corresponding to multiple preset areas in the flood discharge atomization area are acquired in real time according to a preset sampling frequency; wherein, the rainfall monitoring array includes at least three rain gauges arranged along different elevations, and the elevation difference between adjacent rain gauges decreases as the elevation increases; The rainfall intensity data corresponding to all the preset areas are used as the sparse rainfall intensity observation data. The control unit of the flood discharge gate acquires the gate opening, reservoir water level, and total discharge flow in real time according to the preset sampling frequency. The gate opening, the reservoir water level, and the total discharge flow are used as the operating condition data of the flood discharge gate.

[0036] Specifically, firstly, based on the physical laws of the vertical attenuation and abrupt changes in atomized rainfall intensity and the vulnerability of protection targets at different elevations within the power plant area, a three-dimensional gradient-deployed rainfall monitoring array was planned, rather than a uniform or random distribution. Secondly, differentiated equipment selection principles were defined for zones with different atomization intensities (e.g., dense fog and rainstorm zones, light fog and precipitation zones). For example, in dense fog and rainstorm zones (Level I), electromagnetic valve continuous measurement rain gauges were selected to handle ultra-high rainfall intensities up to 1500 mm / h, while in light fog and precipitation zones (Level II), expanded-capacity tipping bucket rain gauges were selected. Finally, gate operation data from the power plant monitoring system, including gate opening, reservoir water level, and total discharge flow, were integrated. By synchronously acquiring sparse but representative gradient rainfall intensity observations with continuous gate operation data, a high spatiotemporal resolution, reliable, and comprehensive multi-source data foundation was provided for subsequent dynamic inversion of physical models and collaborative decision-making. In a preferred embodiment of the invention, combined with... Figure 2As shown, taking the Sanhekou Water Conservancy Project of the Hanjiang-to-Weihe River Water Diversion Project as an example, based on the vertical attenuation law of flood discharge atomization and the elevation gradient of the vulnerability of the plant protection targets, the monitoring area is divided into multiple elevation and functional zones, including elevations of 580m, 566m, 546.5m, the No. 4 road on the left bank, the second dam platform, the roof of the main plant, the roof of the GIS room, the roof of the main transformer room, and the road leading to the plant. Among them, the Level II zones at elevations of 580m and 566m are equipped with expanded-capacity tipping bucket rain gauges, the Level I zones at elevations of 546.5m, the No. 4 road on the left bank, and the second dam platform are equipped with electromagnetic valve continuous measurement rain gauges, the Level II zones at elevations of the main plant roof, the roof of the GIS room, the roof of the main transformer room, and the road leading to the plant are also equipped with expanded-capacity tipping bucket rain gauges, and the Level III zone at the end of the drainage ditch, which is a key area for the convergence of atomized rainfall, is equipped with weighing rain gauges. Explanation of the sparse rainfall intensity observation data deployment density: In this embodiment, the deployment density of the rainfall monitoring array does not exceed one monitoring point per hectare. Within the approximately 8.5-hectare flood discharge atomization impact area (0+000 to 0+500m downstream of the dam, with a width of approximately 170m on both banks), a total of 15 rain gauges were deployed, with an average density of approximately 0.57 gauges / hectare. This density falls into the category of "sparse," based on the principle that it does not aim for full spatial coverage, but rather uses a three-dimensional characteristic gradient deployment according to the vertical attenuation abrupt change points of atomized rainfall intensity and the vulnerability elevation of the plant area protection targets, deploying points only at key control sections. Actual measurement and inversion results show that this sparse deployment scheme is sufficient to obtain high-precision atomization distribution across the entire field through data assimilation and inversion, reducing the number of monitoring points by more than 60% compared to traditional gridded dense deployments (such as one monitoring point per 10m × 10m). Each measuring point synchronously collects rainfall intensity data at a preset sampling frequency of 1 minute to form sparse rainfall intensity observation data; at the same time, the power station's local control unit (LCU) acquires real-time operating condition data such as the opening degree of the three-gate flood discharge gate, reservoir water level, and total discharge flow rate at the same sampling frequency. All data adopt a unified time reference (NTP synchronization), providing a complete and synchronous multi-source observation basis for the dynamic correction of the parameters of the subsequent flood discharge atomization physical and mathematical model.

[0037] In this embodiment of the invention, by employing a gradient deployment and differentiated selection based on physical laws, the most critical control sections are covered with a minimum number of highly reliable sensors. While ensuring the accuracy of the overall atomization field inversion, this significantly reduces system construction and maintenance costs compared to traditional grid-based dense deployment schemes. Simultaneously, the deep integration of real-time, sparse rainfall intensity observations with precise gate operation data enables subsequent model inversions to accurately correlate specific flood discharge patterns with the generated atomization fields.

[0038] Optionally, the sparse rainfall intensity observation data is used as an observation constraint, and an ensemble Kalman filter algorithm is employed to dynamically update the droplet parameters of the flood discharge atomization physical-mathematical model based on the observation constraints, to obtain the corrected droplet parameters, including: The sparse rainfall intensity observation data is used as an observation constraint. An ensemble Kalman filter algorithm is employed to dynamically update the water droplet parameters of the flood discharge atomization physical-mathematical model based on the observation constraints, resulting in corrected water droplet parameters, including: Based on the aforementioned flood discharge atomization physical and mathematical model, the water droplet parameters are initialized, including the average droplet size, droplet size variance, average initial velocity, velocity variance, and mean exit angle. Based on the physical and mathematical model of flood discharge atomization and the operating condition data, the water droplet parameters are calculated in the forward direction to obtain the model-predicted rainfall intensity set corresponding to the sparse rainfall intensity observation data. The observation error covariance is obtained by comparing the predicted rainfall intensity set based on the model with the sparse rainfall intensity observation data. By using the Kalman filter analysis equations, the water droplet parameters are iteratively updated according to the observation error covariance until the water droplet parameters meet the convergence condition, thus obtaining the corrected water droplet parameters; wherein, the convergence condition is that the state increment of the particle size parameter is less than 0.1 mm and the state increment of the angle parameter is less than 1°.

[0039] Specifically, firstly, based on the specific engineering background, the key statistics of the joint distribution function of the initial parameters of the water droplets in the physical and mathematical model of flood discharge atomization are initialized and assigned values ​​to form a state vector θ, which specifically includes the average particle size, particle size variance, average initial velocity, velocity variance, and exit angle. By adding random perturbations that conform to a preset covariance structure (usually a diagonal matrix, whose diagonal elements are set according to the possible variation range of each parameter, such as the standard deviation of the average particle size perturbation being 0.5 mm) near the initial values, an initial state set containing N (e.g., N=40) set members is generated. Secondly, using the current gate operation data (e.g., gate opening, reservoir water level) as the boundary conditions of the model, and taking the water droplet parameters corresponding to each set member of the aforementioned state vector as input, the physical model (i.e., performing the water droplet trajectory calculation and grid rainfall intensity contribution value accumulation process) is driven to perform calculations, thereby generating a set of model-predicted rainfall intensities corresponding to the locations of each sparse observation point. Then, the model-predicted rainfall intensities set is compared with the actual observed rainfall intensities at the sparse observation points to calculate the observation error covariance. By applying the analytical equations of ensemble Kalman filtering and utilizing the calculated observation error covariance, the optimal estimation and adjustment of all ensemble members of the state vector (i.e., the droplet parameters) are performed, completing one iterative update. This process is repeated until the incremental change of the state vector is less than a preset threshold (e.g., Δθ < 0.01), at which point the convergence condition is considered met, or the number of iterations reaches a preset number, thus obtaining a set of droplet parameters calibrated from the observation data that best represents the current actual flood discharge conditions. Therefore, in summary, combining... Figure 3As shown in the flowchart of the atomization field inversion, the dynamic correction of the physical and mathematical model of flood discharge atomization is achieved through the ensemble Kalman filter algorithm. Taking the sparse measured rainfall intensity collected from characteristic elevation measuring points and the preset initial parameters of the physical and mathematical model (including water droplet size, initial velocity, and exit angle) as input, an initial set consisting of a state vector θ (average particle size, variance, initial velocity, and exit angle) is first generated. Then, the prediction step is entered, and the physical model is run to obtain the predicted rainfall intensity value of the corresponding measuring point. Then, the state vector is corrected by the measured rainfall intensity as a constraint through the update step. After convergence judgment (the increment Δθ of the state vector is less than the threshold or the maximum number of iterations is reached), if convergence is not achieved, the prediction step is returned for iterative optimization. If convergence is achieved, the corrected model parameters are output. Finally, based on the corrected parameters, the model is driven to calculate the spatial distribution of the atomization rainfall intensity of the entire field, including the atomization center area where measuring points cannot be deployed, with a grid accuracy of 10m×10m and an update frequency of 1~10min. This realizes the adaptive update of model parameters with real-time operating conditions and the accurate inversion of the entire field rainfall intensity.

[0040] In a preferred embodiment of the invention, the state vector θ is first initialized to [2.5, 0.6, 3.2, 5.6, 35°], these initial values ​​being derived from historical model test data of the project. Under the flood discharge condition (right orifice fully open, total discharge 1870 m³ / s),... 3 At a speed of 1000 m / s, two key observations were collected: a rainfall intensity of 280 mm / h at the 546.5 m embankment and 85 mm / h on the factory roof. The ensemble Kalman filter algorithm then uses the current gate opening and reservoir water level as input to drive the physical model (droplet motion differential equation and maximum entropy joint distribution function) to perform forward calculations on 40 ensemble members, obtaining the predicted rainfall intensity sets at the two measuring points. Next, the algorithm compares these predicted sets with the two measured values ​​to calculate the observation error covariance matrix R. Then, using the analytical equations of the ensemble Kalman filter algorithm, this covariance matrix is ​​used to optimally correct and update the state vector θ, which includes parameters such as average particle size and average initial velocity. After 20 iterations, the convergence condition is met: for particle size parameters (average particle size, particle size variance), the state increment Δθ < 0.1 mm; for angle parameters (mean emission angle), Δθ < 1°, resulting in the final output corrected droplet parameters.

[0041] In this embodiment of the invention, by using the online iterative correction of the Kalman filter algorithm, the key microphysical parameters of the model (droplet size, velocity distribution, etc.) can be adaptively adjusted according to the specific working conditions of each flood discharge event (such as different gate opening and closing combinations and water levels), thereby significantly improving the model's accuracy in depicting the current real atomization field.

[0042] Optionally, the step of driving the flood discharge atomization physical-mathematical model to perform analytical calculations based on the corrected water droplet parameters and the operating condition data to obtain the spatial distribution of the full-field atomization rainfall intensity in the flood discharge atomization area includes: The corrected water droplet parameters are used as input parameters for the joint distribution function in the flood discharge atomization physical and mathematical model. The gate opening, reservoir water level and total discharge flow in the operating condition data are used as model boundary conditions. The motion trajectory and spatial distribution of water droplets in the flood discharge atomization area are determined by the water droplet splashing motion differential equation. Based on the trajectory and spatial distribution of the water droplets, the rainfall intensity contribution value of each spatial grid cell in the flood discharge atomization area is determined by calculation according to the spatial grid cells in the flood discharge atomization area. The rainfall intensity contribution values ​​of all spatial grid cells within the flood discharge atomization area are spatially combined and mapped to generate the spatial distribution of the full-field atomized rainfall intensity.

[0043] Specifically, the corrected droplet parameters (average particle size, average initial velocity, mean ejection angle, etc.) were used as inputs to the joint distribution function of the initial parameters of the droplets (a joint probability density function established based on the maximum entropy principle), thereby determining the statistical characteristics of the droplet swarm in the source region. Simultaneously, real-time operational data (gate opening determines the shape and position of the water jet, reservoir water level affects upstream energy, and total discharge flow is the overall constraint) were used as the boundary conditions and driving inputs of the model. Based on this, kinematic calculations were performed on a large number of droplets with different initial parameters by solving the differential equations of droplet splashing motion, thereby determining the complete trajectory of each droplet (or a representative droplet) from its splash point, through its movement in the air, to its final landing point and its spatial distribution. Secondly, after obtaining the motion trajectories and landing points of a massive number of droplets, the entire flood discharge atomization influence area was divided into regular spatial grid cells on a horizontal plane. For each spatial grid cell, the total volume or mass of all water droplets falling into that grid cell within a given time period (e.g., 10 minutes for calculating rainfall intensity) is statistically analyzed and converted into the rainfall intensity contribution value received by that grid cell. This value is essentially the theoretical rainfall calculated by the physical model. Finally, the rainfall intensity contribution values ​​of all grid cells within the calculation area are combined and mapped according to their spatial coordinates to generate a full-field fogged rainfall intensity spatial distribution map covering the entire region of interest with high spatial resolution (determined by the grid size).

[0044] In a preferred embodiment of the present invention, the differential equation of motion for water droplet splashing is: ; in, Let be the velocity vector of the water droplet. For ambient wind speed vector, The diameter of the water droplet. , These are the densities of water and air, respectively. The drag coefficient, This is the acceleration due to gravity.

[0045] drag coefficient The formula for fitting the drag coefficient curve of standard spherical particles is as follows: ; Among them, the Reynolds number Re , Aerodynamic viscosity. Ambient wind speed vector. The wind field of the water tongue was obtained by interpolation based on the measured data from the anemometer at the dam crest, using an empirical model superposition. The method for determining the above parameters is common knowledge in this field; for details, please refer to "Research on the Improvement of Mathematical Model for Atomization of Diversion Flood Discharge and Analysis of Influencing Factors" (2004).

[0046] Using operational data (gate opening, reservoir water level, total discharge flow) as boundary conditions, for each water droplet sample, the dynamic process of the water droplet under the influence of drag, buoyancy, and gravity in the air medium was described using the differential equation of water droplet splashing motion. The trajectory (position change over time) and spatial distribution of each water droplet were determined by numerical integration (such as the fourth-order Runge-Kutta method). Then, the flood discharge atomization influence area (such as the range from 0+000 to 0+500m downstream of the dam and the elevation range from 510m to 580m) was divided into 10m×10m spatial grid cells. For each grid cell, the total volume or total mass of all water droplets falling into the cell within a 10-minute calculation period (the time base for rainfall intensity calculation) was statistically analyzed.

[0047] The joint distribution function of the initial parameters of the water droplets adopts the joint probability density function established based on the maximum entropy principle, which is: ; in, The diameter of the water droplet. The initial velocity of the water droplet. The angle of departure (the angle between the angle and the horizontal plane). Let λ0, λ1, ..., λ be the azimuth angle (deflection angle in the horizontal plane). m The Lagrange multipliers are determined by constraints such as conservation of mass, momentum, and energy. Based on the joint probability density function form derived from the maximum entropy principle in the image, the corrected droplet parameters are substituted to generate a set of droplet samples with the statistical characteristics of the current operating conditions.

[0048] In another preferred embodiment of the present invention, taking the Sanhekou Water Conservancy Project as the application background, the joint distribution function of the initial parameters of the water droplets is simplified to the following engineering-expressible form: The particle size distribution adopts a log-normal distribution: ; in, The logarithmic mean is... This is the logarithmic standard deviation. The initial values ​​were determined based on historical data from hydraulic model tests: =2.5, =0.6 (corresponding to an average particle size of approximately 12.2 mm).

[0049] The velocity distribution adopts a Gamma distribution: ; in, For shape parameters, Let Γ(k) be the scaling parameter and Γ(k) be the gamma function. The initial values ​​are calibrated as follows: =3.2, =5.6 (corresponding to an average initial velocity of approximately 17.9 m / s).

[0050] The emission angle ψ follows a normal distribution initial value =35°, =8°; Azimuth It follows a uniform distribution in the horizontal plane. .

[0051] In this embodiment of the invention, through analytical calculation using a physical model, the peak rainfall intensity and its precise location in areas that cannot be directly monitored can be predicted from a mechanistic perspective, providing a complete risk baseline map for risk assessment. Since the rainfall intensity of each grid originates from the same calibrated physical model and unified microphysical parameters, the entire distribution field conforms to the physical laws of water droplet movement, diffusion, and sedimentation, and the spatial gradient changes continuously and reasonably. This avoids physical contradictions or abrupt changes that may occur from simple interpolation of data from different measuring points, making the generated risk cloud map more realistic and reliable.

[0052] Optionally, the risk assessment is performed based on the spatial distribution of the overall atomized rainfall intensity, combined with a digital model of the plant area within the flood discharge atomization zone, to obtain a real-time risk index for each key zone within the plant area, including: Based on the digital model of the factory area, the location range, equipment waterproofing level, and asset importance information of key zones within the factory area are extracted; Based on the location range of the key partitions within the factory area, the waterproofing level of the equipment, and the asset importance information, determine the vulnerability coefficient corresponding to each key partition; Based on the spatial distribution of the overall fogged rainfall intensity, the maximum predicted rainfall intensity corresponding to the key zone is determined; The real-time risk index of the key partition is obtained by using a preset risk assessment model:

[0053] in, Let be the real-time risk index for the i-th critical partition. The maximum 10-minute rainfall intensity is retrieved from the key partition. The preset baseline rainfall intensity (range 80-120 mm / h) is used. This represents the vulnerability coefficient of the partition.

[0054] Specifically, firstly, based on a pre-built, high-precision 3D digital base map of the plant area, geographic information system (GIS) technology is used to extract the 3D spatial location range and boundaries of each key area (such as the main transformer room, power distribution room, GIS room, and entrance gate), and associate the static attributes of all equipment / assets within the key areas, namely, the equipment's waterproof rating (e.g., Class I, Class II) and the asset's importance (determined based on its criticality in the power generation, power supply, and safety systems). Secondly, the extracted static attributes are comprehensively calculated to assign a quantified vulnerability coefficient (λ) to each key area. i The vulnerability coefficient reflects the strength of the area's own resistance to impact; the lower the waterproof rating and the more important the asset, the higher the vulnerability coefficient. Next, the spatial distribution map of the overall fogged rainfall intensity obtained in the previous step is spatially overlaid with the spatial range of the key areas. Specifically, by statistically analyzing the rainfall intensity values ​​of all grid cells falling within the area, the maximum value is found, which is the maximum predicted rainfall intensity P corresponding to that area. i This represents the most severe direct physical threat posed by the current atomization field to this area. Finally, through a pre-set risk assessment model, the quantified external threat (maximum predicted rainfall intensity P) is... i ) and internal vulnerability (vulnerability coefficient λ) i Multiplying these together, a comprehensive, dimensionless real-time risk index R is ultimately generated. i The real-time risk index intuitively indicates the severity level of the risk currently faced by each partition, providing a precise and differentiated decision-making basis for whether to trigger optimization and how to allocate protection resources. The preset risk assessment model expression is as follows: ; in, Let be the real-time risk index for the i-th critical partition. The maximum 10-minute rainfall intensity is retrieved from the key partition. The preset baseline rainfall intensity (range 80-120 mm / h) is used. This represents the vulnerability coefficient of the partition.

[0055] In a preferred embodiment of the invention, the main transformer room is designated as Waterproof Level I, with extremely high asset importance, and is assigned a vulnerability coefficient λ=1.5; the power distribution room is designated as Waterproof Level II, with high importance, and λ=1.3; the GIS room has λ=1.4; and the internal roads have λ=1.0. During flood discharge, spatial analysis is performed based on the inverted full-field atomized rainfall intensity distribution map. By overlaying the three-dimensional model range of the main transformer room, the maximum predicted 10-minute rainfall intensity P within this spatial range is extracted. i The value was 86 mm / h (calculated from the inversion model). Subsequently, substituting this value into the risk assessment model, the real-time risk index R for the main transformer room was calculated to be 0.86. Similarly, the risk indices for other zones such as the factory area and roads were calculated. Since the R value for the main transformer room exceeded the preset alarm threshold of 0.8, the risk was deemed unacceptable, automatically triggering a subsequent multi-objective collaborative optimization process to find a gate operation scheme that could reduce the risk index.

[0056] In this embodiment of the invention, by integrating a three-dimensional digital model of the plant area with a misting field distribution map, precise spatial positioning of risks is achieved. Secondly, by introducing a vulnerability coefficient that combines equipment waterproofing level and asset importance, static engineering attributes are multiplied by dynamic physical threats to generate a real-time risk index that scientifically reflects the combined impact of threat magnitude and the vulnerability of the affected entity. This provides operators with an objective risk quantification standard that transcends personal experience.

[0057] Optionally, the gate operation strategy includes the opening combination mode of the flood discharge gate, the gate opening degree, and the adjustment sequence; the gate operation strategy and plant protection instructions for the flood discharge gate are simultaneously solved using a multi-objective optimization algorithm, based on the real-time risk index and the operating condition data, with the objective of minimizing the overall weighted risk value of the plant area, and with constraints such as the total discharge flow, the upper and lower limits of the gate opening degree, and the gate adjustment rate limit, including: Using the operating parameters of the flood discharge gate as decision variables, minimizing the weighted risk value of the plant area as the optimization objective, and the total discharge flow, the upper and lower limits of the gate opening, and the gate adjustment rate limit as constraints, an optimization model is constructed through the objective function. The optimization model is iteratively optimized and solved using the multi-objective optimization algorithm based on the real-time risk index and the operating condition data to obtain the opening combination mode, gate opening degree and adjustment sequence of the flood discharge gate; The plant area protection command is generated based on the opening combination method, the gate opening degree, and the adjustment sequence.

[0058] Specifically, in combination Figure 4As shown in the optimization flowchart, firstly, using the operating parameters of the floodgates as adjustable control buttons, and based on the real-time risk indices of each key zone calculated in previous steps, different importance weights are assigned to different key zones. By calculating the weighted sum of the risk indices and weights of each zone, the multi-dimensional safety requirements are transformed into a single-objective optimization problem. The optimization algorithm aims to minimize F, meaning it prioritizes gate operation schemes that can significantly reduce the risk in important areas while having a smaller impact on secondary areas. Secondly, a multi-objective optimization algorithm is used as the solver, including but not limited to the Non-Dominated Ranking Genetic Algorithm with Elite Strategy (NSGA-II). With the decision variable X as the individual and the objective function F as the fitness, within the solution space that satisfies the above constraints, iterative operations such as population initialization, selection, crossover, mutation, non-dominated ranking, and elite retention are performed to simulate the natural evolution process, ultimately converging to a set of optimal or near-optimal solutions (Pareto front). From this, specific gate opening combinations, gate opening values, and recommended adjustment sequences can be recommended. Finally, the optimal solution selected from the Pareto front has its decision variable values ​​directly parsed into an executable gate operation strategy (such as reducing the opening of the currently fully open right orifice (G1) to 70%, while opening the closed left orifice (G3) to 30%, and keeping the middle orifice (G2) closed). At the same time, a new and expected overall fogging distribution is predicted based on this strategy, and then a new risk index for each zone is calculated. Based on the preset risk-instruction matching rules, corresponding plant protection instructions that are accurate to specific equipment actions are automatically generated (such as starting the No. 2 backup drainage pump in the right plant area and automatically lowering the electric waterproof roller shutter at the entrance of the GIS room), thus completing the transformation from optimization calculation to executable work order.

[0059] In this embodiment of the invention, by setting the gate operation as an adjustable variable and performing reverse optimization with the actual risks of the plant area as the target, the spatial distribution of atomization generation can be proactively adjusted from the source to avoid highly vulnerable plant areas as much as possible, thus achieving proactive prevention and control. Secondly, it provides operators with scientific, intuitive, and quantifiable decision support, transforming complex multi-objective trade-offs (ensuring discharge volume, protecting equipment, and protecting the plant area) into clear comparisons of optimization schemes, thereby improving the resilience and intelligent operation level of hydropower stations in the face of extreme atomization disasters.

[0060] Optionally, the objective function is: ; in, This represents the minimum overall weighted risk value for the aforementioned factory area. Let ∑ωi=1 be the importance weight of the i-th key partition, X be the activation combination method, and R be the value of the key partition. i (X) is the real-time risk index of the i-th key partition under the opening combination mode X, and m is the total number of key partitions; The constraints are as follows: , , ; in, Let Q(X) be the total discharge flow rate of n gate opening combinations. Let G be the target discharge flow rate, ε be the allowable error, and G be the allowable discharge flow rate. j G represents the opening degree of the j-th gate. min G max These are the lower and upper limits of the gate's opening, ΔG and ΔG, respectively. j / Δt is the gate adjustment rate of the gate, V max ΔG is the maximum permissible adjustment rate of the gate. j For the gate at time intervals The change in opening degree within.

[0061] Specifically, the weighted summation is achieved through the objective function, and its expression is as follows: ; Wherein, minF is the minimum overall weighted risk value of the plant area. Let ∑ωi=1 be the importance weight of the i-th key partition, X be the activation combination method, and R be the value of the key partition. i (X) is the real-time risk index of the i-th critical partition under the opening combination mode X, and m is the total number of critical partitions, such as main transformer room, power distribution room, entrance gate, GIS room, etc.

[0062] In a preferred embodiment of the present invention, the objective function is minF = 0.6R. 厂房 +0.3R 主变 +0.1R 道路 The factory building is the most important (weight 0.6), followed by the main transformer room (0.3), and the road is relatively less important (0.1). The optimization result recommends reducing the opening of the right meter hole to 70% and increasing the opening of the left meter hole to 30%. The minF value decreased from 0.6×0.86+0.3×0.86+0.1×0.7≈0.82 before optimization to 0.6×0.52+0.3×0.52+0.1×0.42≈0.51, reducing the overall risk by about 38%.

[0063] At the same time, the basic safety and functional constraints of flood discharge operations were strictly incorporated, including: The total discharge capacity is constrained by an equation to ensure that optimization does not affect the reservoir's predetermined discharge schedule. Its expression is: ; The upper and lower limits of the gate opening are constrained to ensure that the equipment operates within the mechanical safety range. Their expression is: ; The single-step adjustment rate constraint of the gate, considering the dynamic response capability and safety of the gate opening and closing mechanism, and avoiding operational shock, is expressed as follows: ; in, Let Q(X) be the total discharge flow rate of n gate opening combinations. Let G be the target discharge flow rate, ε be the allowable error, and G be the allowable discharge flow rate. j Let G be the opening degree of the j-th gate. min G max These represent the lower and upper limits of the gate opening, ΔG. j / Δt is the gate adjustment rate, V max ΔG is the maximum permissible regulating rate of the gate. j For the gate at time intervals The change in opening degree within.

[0064] In this embodiment of the invention, by constructing an objective function with the core objective of minimizing the overall weighted risk value of the plant area, and combining it with safety constraints such as total discharge flow, gate opening and regulation rate, multi-objective collaborative optimization is carried out. This not only allocates protection weights according to the importance of key zones and prioritizes the safety of core equipment and plant buildings, but also significantly reduces the overall atomization risk of the plant area without changing the reservoir discharge task or exceeding the gate operation safety boundary. This achieves deep synergy between flood discharge operation scheduling and active protection of the plant area, while taking into account engineering safety, flood control function and equipment protection, and greatly improving the accuracy, reliability and economy of flood discharge atomization protection.

[0065] Optionally, generating the plant protection command according to the gate operation strategy includes: Based on the gate operation strategy, the risk index of the key partition is predicted to obtain the predicted risk index value of the key partition under the gate operation strategy. Based on the predicted risk index value, the control commands for the active protection facilities of the plant area are generated by matching according to the preset risk-command matching rules. The control commands of the active protection facilities shall be used as the plant area protection commands.

[0066] Specifically, after determining a set of gate opening combinations, instead of directly issuing commands, this optimized set of gate opening combinations is used as new operating condition data for a new round of forward simulation prediction. Specifically, using this gate opening combination as the new operating condition and boundary conditions, a predictive spatial distribution of the entire field's atomized rainfall intensity is recalculated using the current physical model and water droplet parameters. Based on this predicted distribution, the predicted risk index value for each key zone under the new gate strategy is recalculated. The calculated predicted risk index value is compared with a preset, configurable risk-command matching rule. This preset risk-command matching rule defines the mapping relationship between the predicted risk index value ranges for different key zones and specific protective action commands. For example, if the predicted risk index value for the "main transformer room" is greater than 0.8, a command to start the backup drainage pump and lower the waterproof roller shutter is generated; if the predicted risk index value for the "access road" is greater than 0.5, a command to activate the drainage facilities along the route is generated. The risk-command matching rule is a quantitative representation of engineering logic, typically using an if-then rule table. For example, if R... 主变室 If R > 0.8, then start the No. 2 standby drainage pump and close the waterproof roller shutter in the main transformer room; if R 厂房 If R > 0.7, then start drainage pump #1; if R 道路预测 If the value is greater than 0.6, the electric valve of the intercepting ditch on the access road to the plant will be opened. The rules integrate remote control logic for the plant's protective facilities. Finally, control commands for the active protective facilities are generated based on the matching results.

[0067] In this embodiment of the invention, by predicting the risk index first and then matching the instructions, the plant protection instructions are forward-looking and targeted. The protection intensity can be configured in advance based on the expected risk level after the gate is controlled, avoiding over-protection or under-protection. At the same time, the instructions are automatically generated by relying on standardized matching rules, ensuring that the protection decision is unified, standardized and reproducible, effectively improving the response efficiency and reliability of the protection system, and realizing seamless connection and precise coordination between flood discharge control and plant protection.

[0068] Optionally, the step of controlling the floodgate and protecting the plant area according to the gate operation strategy and the plant area protection command respectively includes: The gate operation strategy is sent to the control unit of the flood discharge gate, and the flood discharge gate is regulated according to the opening combination method, the gate opening degree and the adjustment sequence. The plant protection command is sent to the control unit of the active protection facility in the plant area for regulation via an industrial communication protocol.

[0069] Specifically, firstly, for the flood discharge gates, the gate operation strategy, including the opening combination method, the precise opening value of each gate, and the recommended adjustment sequence, is pushed to the flood discharge gate control unit in a structured data format (such as JSON or specific protocol messages) for visualization and confirmation. The flood discharge gate control unit typically refers to the local control unit or monitoring system of the hydropower station. The control unit generates a gate control command sequence according to the opening combination method, target gate opening value, and adjustment sequence (considering the single-step adjustment rate constraint of the gate) specified in the strategy, driving the hydraulic or electric hoist to execute, thereby physically changing the water jet morphology and optimizing the atomization distribution at its source. Secondly, for the control of plant protection facilities, the generated, specific plant protection instructions are sent to the control units of each active protection facility via industrial communication protocols (such as Modbus TCP or OPC UA). These control units can be PLC cabinets for drainage pumps, controllers for electric valves, or intelligent control boxes for roller shutters. The issuance and execution of commands are automated, requiring no human intervention, thus achieving a rapid response at the millisecond to second level, ensuring that the corresponding waterproofing and drainage measures are in place before or at the same time as the atomized rainfall intensifies.

[0070] In this embodiment of the invention, by automatically converting high-level instructions generated through optimized calculations into standardized control commands executable by lower-level devices, the final mile of the "decision-control" process is bridged, achieving closed-loop automatic operation of the protection system. This not only greatly improves response speed but also reduces delays and misjudgments caused by manual operation, ensuring the accurate implementation of protective measures at critical moments.

[0071] Combination Figure 5 As shown, another embodiment of the present invention provides a multi-level coordinated protection system for hydropower station flood discharge atomization, comprising: The data acquisition module is used to acquire multi-source observation data of the flood discharge atomization area in real time. The multi-source observation data includes sparse rainfall intensity observation data and the operating status data of the flood discharge gate. The model calibration module is used to dynamically calibrate the model parameters of a preset flood discharge atomization physical-mathematical model based on the sparse rainfall intensity observation data and the operating condition data, thereby obtaining the spatial distribution of the full-field atomized rainfall intensity in the flood discharge atomization area. The dynamic calibration includes: using the sparse rainfall intensity observation data as an observation constraint, employing an ensemble Kalman filter algorithm to dynamically update the water droplet parameters of the flood discharge atomization physical-mathematical model based on the observation constraint, thereby obtaining the calibrated water droplet parameters; and driving the flood discharge atomization physical-mathematical model to perform analytical calculations based on the calibrated water droplet parameters and the operating condition data, thereby obtaining the spatial distribution of the full-field atomized rainfall intensity in the flood discharge atomization area. The risk assessment module is used to conduct risk assessment based on the spatial distribution of the atomized rainfall intensity across the entire site, combined with the digital model of the plant area within the flood discharge atomization area, to obtain the real-time risk index of each key zone within the plant area. The strategy generation module is used to simultaneously solve for the gate operation strategy of the flood discharge gate and the protection instructions of the plant area by using a multi-objective optimization algorithm, based on the real-time risk index and the operating condition data, with the goal of minimizing the overall weighted risk value of the plant area, and with the constraints of the total discharge flow, the upper and lower limits of the gate opening and the gate adjustment rate limit. The control module is used to control the flood discharge gate and protect the plant area according to the gate operation strategy and the plant area protection command, respectively.

[0072] The advantages of the hydropower station flood discharge atomization multi-level coordinated protection system of the present invention compared with the prior art are the same as those of the above-mentioned hydropower station flood discharge atomization multi-level coordinated protection method compared with the prior art, and will not be repeated here.

[0073] While the present invention has been disclosed above, its scope of protection is not limited thereto. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of the present invention, and all such changes and modifications will fall within the scope of protection of the present invention.

Claims

1. A multi-stage coordinated protection method for flood discharge atomization in hydropower stations, characterized in that, include: Real-time acquisition of multi-source observation data of the flood discharge atomization area, including sparse rainfall intensity observation data and flood discharge gate operation status data; Based on the sparse rainfall intensity observation data and the operational condition data, the model parameters of the preset flood discharge atomization physical-mathematical model are dynamically corrected to obtain the spatial distribution of the full-field atomized rainfall intensity in the flood discharge atomization area. The dynamic correction includes: using the sparse rainfall intensity observation data as an observation constraint, employing an ensemble Kalman filter algorithm to dynamically update the water droplet parameters of the flood discharge atomization physical-mathematical model based on the observation constraint, to obtain the corrected water droplet parameters; and driving the flood discharge atomization physical-mathematical model to perform analytical calculations based on the corrected water droplet parameters and the operational condition data to obtain the spatial distribution of the full-field atomized rainfall intensity in the flood discharge atomization area. Based on the spatial distribution of the atomized rainfall intensity across the entire field, and combined with the digital model of the plant area within the flood discharge atomization zone, a risk assessment is conducted to obtain the real-time risk index for each key zone within the plant area. Using a multi-objective optimization algorithm, based on the real-time risk index and the operating condition data, with the objective of minimizing the overall weighted risk value of the plant area, and with constraints such as the total discharge flow, the upper and lower limits of the gate opening, and the gate adjustment rate limit, the gate operation strategy and plant area protection instructions of the flood discharge gate are solved simultaneously. The floodgates are controlled and the plant area is protected according to the gate operation strategy and the plant area protection instructions, respectively.

2. The multi-level coordinated protection method for flood discharge atomization in hydropower stations according to claim 1, characterized in that, The real-time acquisition of multi-source observation data in the flood discharge atomization area includes: By setting up a rainfall monitoring array based on the vertical attenuation law of atomized rainfall intensity and the elevation gradient of the vulnerability of the protection target of the plant area, the rainfall intensity data corresponding to multiple preset areas in the flood discharge atomization area are acquired in real time according to a preset sampling frequency; wherein, the rainfall monitoring array includes at least three rain gauges arranged along different elevations, and the elevation difference between adjacent rain gauges decreases as the elevation increases; The rainfall intensity data corresponding to all the preset areas are used as the sparse rainfall intensity observation data. The control unit of the flood discharge gate acquires the gate opening, reservoir water level, and total discharge flow in real time according to the preset sampling frequency. The gate opening, the reservoir water level, and the total discharge flow are used as the operating condition data of the flood discharge gate.

3. The multi-level coordinated protection method for flood discharge atomization in hydropower stations according to claim 1, characterized in that, The sparse rainfall intensity observation data is used as an observation constraint. An ensemble Kalman filter algorithm is employed to dynamically update the water droplet parameters of the flood discharge atomization physical-mathematical model based on the observation constraints, resulting in corrected water droplet parameters, including: Based on the aforementioned flood discharge atomization physical and mathematical model, the water droplet parameters are initialized, including the average droplet size, droplet size variance, average initial velocity, velocity variance, and mean exit angle. Based on the physical and mathematical model of flood discharge atomization and the operating condition data, the water droplet parameters are calculated in the forward direction to obtain the model-predicted rainfall intensity set corresponding to the sparse rainfall intensity observation data. The observation error covariance is obtained by comparing the predicted rainfall intensity set based on the model with the sparse rainfall intensity observation data. By using the Kalman filter analysis equations, the water droplet parameters are iteratively updated according to the observation error covariance until the water droplet parameters meet the convergence condition, thus obtaining the corrected water droplet parameters; wherein, the convergence condition is that the state increment of the particle size parameter is less than 0.1 mm and the state increment of the angle parameter is less than 1°.

4. The multi-level coordinated protection method for flood discharge atomization in hydropower stations according to claim 1, characterized in that, The step of driving the flood discharge atomization physical-mathematical model to perform analytical calculations based on the corrected water droplet parameters and the operating condition data to obtain the spatial distribution of the full-field atomization rainfall intensity in the flood discharge atomization area includes: The corrected water droplet parameters are used as input parameters for the joint distribution function in the flood discharge atomization physical and mathematical model. The gate opening, reservoir water level and total discharge flow in the operating condition data are used as model boundary conditions. The motion trajectory and spatial distribution of water droplets in the flood discharge atomization area are determined by the water droplet splash motion differential equation. Based on the trajectory and spatial distribution of the water droplets, the rainfall intensity contribution value of each spatial grid cell in the flood discharge atomization area is determined by calculation according to the spatial grid cells in the flood discharge atomization area. The rainfall intensity contribution values ​​of all spatial grid cells within the flood discharge atomization area are spatially combined and mapped to generate the spatial distribution of the full-field atomized rainfall intensity.

5. The multi-level coordinated protection method for flood discharge atomization in hydropower stations according to claim 1, characterized in that, The risk assessment is conducted based on the spatial distribution of the overall atomized rainfall intensity and a digital model of the plant area within the flood discharge atomization zone, resulting in a real-time risk index for each key area within the plant area, including: Based on the digital model of the factory area, the location range, equipment waterproofing level, and asset importance information of key zones within the factory area are extracted; Based on the location range of the key partitions within the factory area, the waterproofing level of the equipment, and the asset importance information, determine the vulnerability coefficient corresponding to each key partition; Based on the spatial distribution of the overall fogged rainfall intensity, the maximum predicted rainfall intensity corresponding to the key zone is determined; The real-time risk index of the key partition is obtained by using a preset risk assessment model: in, Let be the real-time risk index for the i-th critical partition. The maximum 10-minute rainfall intensity is retrieved from the key partition. The preset baseline rainfall intensity (range 80-120 mm / h) is used. This represents the vulnerability coefficient of the partition.

6. The multi-level coordinated protection method for flood discharge atomization in hydropower stations according to claim 1, characterized in that, The gate operation strategy includes the opening combination mode, gate opening degree, and adjustment sequence of the flood discharge gate; the multi-objective optimization algorithm, based on the real-time risk index and the operating condition data, aims to minimize the overall weighted risk value of the plant area, and uses the total discharge flow, upper and lower limits of gate opening degree, and gate adjustment rate limits as constraints to simultaneously solve for the gate operation strategy and plant area protection instructions for the flood discharge gate, including: Using the operating parameters of the flood discharge gate as decision variables, minimizing the weighted risk value of the plant area as the optimization objective, and the total discharge flow, the upper and lower limits of the gate opening, and the gate adjustment rate limit as constraints, an optimization model is constructed through the objective function. The optimization model is iteratively optimized and solved using the multi-objective optimization algorithm based on the real-time risk index and the operating condition data to obtain the opening combination mode, gate opening degree and adjustment sequence of the flood discharge gate; The plant area protection command is generated based on the opening combination method, the gate opening degree, and the adjustment sequence.

7. The multi-level coordinated protection method for flood discharge atomization in hydropower stations according to claim 6, characterized in that, The objective function is: ； in, Let ∑ωi=1 be the importance weight of the i-th key partition, X be the activation combination method, and R be the value of the key partition. i (X) is the real-time risk index of the i-th key partition under the opening combination mode X, and m is the total number of key partitions; The constraints are as follows: ， ， ； in, Let Q(X) be the total discharge capacity of the gates, representing the opening combinations of n gates. Let G be the target discharge flow rate, ε be the allowable error, and G be the allowable discharge flow rate. j G represents the opening degree of the j-th gate. min G max These are the lower and upper limits of the gate's opening, ΔG and ΔG, respectively. j / Δt is the gate adjustment rate of the gate, V max ΔG is the maximum permissible adjustment rate of the gate. j For the gate at time intervals The change in opening degree within.

8. The multi-level coordinated protection method for flood discharge atomization in hydropower stations according to claim 6, characterized in that, The step of generating the plant area protection command according to the gate operation strategy includes: Based on the gate operation strategy, the risk index of the key partition is predicted to obtain the predicted risk index value of the key partition under the gate operation strategy. Based on the predicted risk index value, the control commands for the active protection facilities of the plant area are generated by matching according to the preset risk-command matching rules. The control commands of the active protection facilities shall be used as the plant area protection commands.

9. The multi-level coordinated protection method for flood discharge atomization in hydropower stations according to claim 8, characterized in that, The process of controlling the floodgate and protecting the plant area according to the gate operation strategy and the plant area protection command respectively includes: The gate operation strategy is sent to the control unit of the flood discharge gate, and the flood discharge gate is regulated according to the opening combination method, the gate opening degree and the adjustment sequence. The plant protection command is sent to the control unit of the active protection facility in the plant area for regulation via an industrial communication protocol.

10. A multi-level coordinated protection system for flood discharge atomization in hydropower stations, characterized in that, include: The data acquisition module is used to acquire multi-source observation data of the flood discharge atomization area in real time. The multi-source observation data includes sparse rainfall intensity observation data and the operating status data of the flood discharge gate. The model calibration module is used to dynamically calibrate the model parameters of a preset flood discharge atomization physical-mathematical model based on the sparse rainfall intensity observation data and the operating condition data, thereby obtaining the spatial distribution of the full-field atomized rainfall intensity in the flood discharge atomization area. The dynamic calibration includes: using the sparse rainfall intensity observation data as an observation constraint, employing an ensemble Kalman filter algorithm to dynamically update the water droplet parameters of the flood discharge atomization physical-mathematical model based on the observation constraint, thereby obtaining the calibrated water droplet parameters; and driving the flood discharge atomization physical-mathematical model to perform analytical calculations based on the calibrated water droplet parameters and the operating condition data, thereby obtaining the spatial distribution of the full-field atomized rainfall intensity in the flood discharge atomization area. The risk assessment module is used to conduct risk assessment based on the spatial distribution of the atomized rainfall intensity across the entire site, combined with the digital model of the plant area within the flood discharge atomization area, to obtain the real-time risk index of each key zone within the plant area. The strategy generation module is used to simultaneously solve for the gate operation strategy of the flood discharge gate and the protection instructions of the plant area by using a multi-objective optimization algorithm, based on the real-time risk index and the operating condition data, with the goal of minimizing the overall weighted risk value of the plant area, and with the constraints of the total discharge flow, the upper and lower limits of the gate opening and the gate adjustment rate limit. The control module is used to control the flood discharge gate and protect the plant area according to the gate operation strategy and the plant area protection command, respectively.

Images