Construction method of water flow control model of torrent training water area

Abstract

The invention relates to the technical field of model construction, in particular to a construction method of a water flow control model of a torrent training water area. The method comprises the following steps: acquiring torrent training water area environment sensing data; performing spatial three-dimensional modeling on the torrent training water area environment sensing data to generate a torrent training water area spatial three-dimensional model; performing water flow dynamic simulation on the torrent training water area space three-dimensional model to generate torrent training water area water flow simulation data; acquiring actual hydrological data of a torrent training water area; and performing simulation water flow data correction on the torrent training water area water flow simulation data through the torrent training water area actual hydrological data to generate simulation water flow correction data. The precision and adaptability of the torrent training water area water flow control model are improved by combining three-dimensional modeling, water flow simulation, actual hydrological data correction, ship driving mode analysis and a convolutional neural network algorithm.

Description

Technical Field

[0001] The present invention relates to the technical field of model construction, and particularly to a method for constructing a water flow control model for a rapids training water area. Background Art

[0002] In the early stage, water flow control mainly relied on natural terrain and artificial dams, and it was difficult to accurately regulate the water flow speed and direction, resulting in limited training effects. With the development of water conservancy engineering technology, technologies based on water flow simulation emerged. Through water flow simulators and numerical calculation methods, water flow control could gradually be realized to a certain extent in experimental environments. However, these methods were mostly used for static water flows and were difficult to adapt to complex and changing training requirements. After entering the 21st century, water flow control technology began to introduce the combination of automated control systems and computational fluid dynamics (CFD) simulations. By arranging multiple devices for adjusting water flow in the water area, such as water pumps, flow deflectors, fountain systems, etc., more accurate water flow control could be achieved. The development of CFD technology enabled better prediction and adjustment of water flow behavior, providing a more flexible control scheme for rapids training. However, currently, the construction of water flow control models mostly relies on static water areas or simplified water flow simulations, making it difficult to accurately reflect the dynamic characteristics of rapids training water areas. At the same time, water flow control often lacks real-time feedback mechanisms and dynamic adjustment capabilities, unable to respond to the demand changes under different training scenarios, thereby resulting in low accuracy and adaptability of the water flow control model for rapids training water areas. Summary of the Invention

[0003] Based on this, it is necessary to provide a method for constructing a water flow control model for a rapids training water area to solve at least one of the above technical problems.

[0004] To achieve the above object, a method for constructing a water flow control model for a rapids training water area, the method includes the following steps:

[0005] Step S1: Obtain environmental perception data of the rapids training water area; perform three-dimensional spatial modeling on the environmental perception data of the rapids training water area to generate a three-dimensional spatial model of the rapids training water area; perform water flow dynamics simulation on the three-dimensional spatial model of the rapids training water area to generate water flow simulation data of the rapids training water area;

[0006] Step S2: Obtain actual hydrological data of the rapids training water area; correct the water flow simulation data of the rapids training water area through the actual hydrological data of the rapids training water area to generate corrected simulation water flow data; perform regional water flow regime adaptation analysis on the three-dimensional spatial model of the rapids training water area based on the corrected simulation water flow data to generate a first type of flow regime area and a second type of flow regime area; set water flow control valve openings for the first type of flow regime area and the second type of flow regime area to obtain water flow control valve opening data;

[0007] Step S3: Obtain the position information data of the water area training vessels; import the position information data of the water area training vessels into the analysis of the vessel driving mode in the three-dimensional model of the rapids training water area to generate the driving mode of the training vessels, where the driving model of the training vessels includes a normal driving mode and an abnormal driving mode; use the water flow control valve data to perform vessel guidance control on the normal driving mode and the abnormal driving mode to generate water flow control guidance data;

[0008] Step S4: Extract the water flow control features from the water flow control guidance data to obtain the water flow control feature data; use the convolutional neural network algorithm to construct a model for the water flow control feature data, thereby generating a water flow control model.

[0009] The present invention generates an accurate three-dimensional model of the rapids training water area by obtaining environmental perception data and performing three-dimensional spatial modeling, and combines hydrodynamic simulation to generate water flow simulation data, which can accurately reproduce the water flow characteristics of the training water area, providing real and reliable basic data for subsequent water flow control. This process ensures the accuracy and comprehensiveness of the model, avoiding the limitations of traditional models relying on simplified assumptions. By correcting the actual hydrological data with the simulated water flow data, the difference between the simulation and the actual situation can be eliminated, improving the accuracy of the model. The flow state adaptation analysis further refines the water flow characteristics of different regions, ensuring that reasonable water flow control valves can be set according to the regional characteristics, making the water flow control more targeted and precise. By importing the vessel position information into the model, the driving mode of the vessel can be tracked in real time, and the vessel guidance control can be realized based on the water flow control valve data. By making corresponding water flow adjustments according to the normal and abnormal driving modes of the vessel, the water flow environment can be dynamically optimized, improving the safety and effectiveness of the training process and ensuring the best driving route of the training vessels. By extracting the features from the water flow control guidance data and using the convolutional neural network algorithm to construct an intelligent water flow control model, the water flow control strategy can be automatically learned and optimized from the training data, making the water flow control more intelligent and adaptive. This intelligent model can make real-time adjustments according to different training requirements and environmental changes, greatly improving the flexibility and accuracy of the training water area. Therefore, the present invention improves the accuracy and adaptability of the water flow control model in the rapids training water area by combining three-dimensional modeling, water flow simulation, actual hydrological data correction, vessel driving mode analysis, and convolutional neural network algorithm.

[0010] Preferably, step S1 includes the following steps:

[0011] Step S11: Perform water area environment perception on the rapids training water area to obtain the environmental perception data of the rapids training water area;

[0012] Step S12: Perform data preprocessing on the rapid training water area environmental perception data to generate standard water area environmental perception data, where the data preprocessing includes data cleaning, data denoising, missing value filling, and data standardization;

[0013] Step S13: Extract the terrain height, waterway shape, and obstacle distribution data from the standard water area environmental perception data, and perform spatial three-dimensional modeling based on the terrain height, waterway shape, and obstacle distribution data to generate a spatial three-dimensional model of the rapid training water area;

[0014] Step S14: Extract the deployment position data of the water discharge outlet of the rapid training water area from the spatial three-dimensional model of the rapid training water area to obtain the deployment position data of the water discharge outlet; perform hydrodynamic simulation on the spatial three-dimensional model of the rapid training water area based on the deployment position data of the water discharge outlet to generate the hydrodynamic simulation data of the rapid training water area.

[0015] Through environmental perception of the rapid training water area, the present invention can comprehensively understand key information such as the geographical environment, obstacles, and river channel morphology of the water area, ensuring the accuracy of subsequent analysis and modeling. Through data preprocessing, such as data cleaning, denoising, missing value filling, and data standardization, the quality of the environmental perception data can be effectively improved, the influence of noise can be reduced, and the usability and reliability of the data can be enhanced. Extracting the terrain height, waterway shape, and obstacle distribution data in the water area and performing three-dimensional modeling can accurately reproduce the physical characteristics of the water area, providing support for hydrodynamic simulation and related decisions. Performing hydrodynamic simulation based on the deployment position data of the water discharge outlet can simulate the dynamic behavior of the water flow in the water area, predict the changes in the water flow and its impact on the training process, and contribute to the optimization and design of the training scenario. Through the implementation of this series of steps, accurate environmental data support can be provided for rapid training, effectively optimizing the training plan, and improving the training effect and safety.

[0016] Preferably, performing spatial three-dimensional modeling based on the terrain height, waterway shape, and obstacle distribution data includes:

[0017] Construct terrain grid data according to the terrain height, waterway shape, and obstacle distribution data to obtain terrain grid data; extract water area streamline analysis data from the terrain grid data to obtain water area streamline analysis data;

[0018] Construct a water area profile based on the water area streamline analysis data to obtain water area profile data; construct a refined obstacle model of the terrain grid data through the water area profile data to obtain refined obstacle model data;

[0019] Generate a three-dimensional water surface of the water area for the refined model data of the obstacle using the water flow line analysis data of the water area, so as to obtain the three-dimensional water surface data of the water area; perform a three-dimensional model mapping on the three-dimensional water surface data of the water area, so as to obtain a three-dimensional spatial model of the rapids training water area.

[0020] Through the construction of terrain grid data, the present invention can accurately describe the terrain features of the rapids training water area. The terrain grid data provides a basis for subsequent streamline analysis and three-dimensional modeling, ensuring the accuracy and authenticity of the spatial model. The extraction of water flow line analysis data of the water area helps to understand the distribution and trend of water flow in the training water area, which provides detailed guidance for the simulation of water flow, and can better predict and optimize the dynamic changes of water flow, especially in rapids training. The water area profile data can show the lateral changes of the water area at different positions, providing a clear perspective for subsequent streamline and obstacle analysis. It enhances the accurate grasp of features such as water depth and width of the water area. By constructing a refined model of the obstacle, the shape and position of the obstacle in the water area can be described in depth, making the simulation of the influence of the obstacle on water flow more accurate and providing strong support for the training plan. The generation of the three-dimensional water surface of the water area can more intuitively display the shape of the training water area, enhancing the actual perception and control accuracy during the training process, which provides valuable data for various water area simulations, risk predictions, and optimization of training strategies. Through three-dimensional model mapping in space, all modeling data can be integrated into a complete three-dimensional water area model, facilitating subsequent water flow simulation, obstacle handling, and dynamic decision-making support during the training process, and improving the overall training effect and efficiency. Through highly refined three-dimensional modeling, various situations in the water area can be analyzed more accurately, potential training risks can be identified in advance, so as to provide real-time decision-making support for coaches and trainers, ensuring the safety and maximization of the training effect during the training process.

[0021] Preferably, step S2 includes the following steps:

[0022] Step S21: Use sensors to obtain the actual hydrological data of the rapids training water area;

[0023] Step S22: Calibrate the simulated water flow data of the rapids training water area through the actual hydrological data of the rapids training water area to generate calibrated simulated water flow data; based on the calibrated simulated water flow data, perform segmentation of the water flow pattern adaptation area of the three-dimensional spatial model of the rapids training water area to generate the data of the segmented area of the water flow pattern of the training water area;

[0024] Step S23: Detect the number of water flow pattern adaptations in the segmented area data of the water flow pattern of the training water area. When the number of water flow pattern adaptations in the area is greater than or equal to the preset standard number of water flow pattern adaptations in the standard area, the corresponding data of the segmented area of the water flow pattern of the training water area is marked as the first type of flow pattern area;

[0025] Step S24: When the number of regional water flow pattern adaptations is less than the preset standard number of regional water flow pattern adaptations, the corresponding training water area water flow pattern segmentation area data is marked as the second type of flow pattern area; the water flow control valve openings are set for the first type of flow pattern area and the second type of flow pattern area, so as to obtain the water flow control valve opening data.

[0026] By using the actual hydrological data to correct the water flow simulation data, the present invention can ensure that the simulation results are closer to the actual situation and improve the simulation accuracy. This makes the simulation of water flow during rapids training more realistic and credible, enhancing the reliability of training. Through the segmentation of the water flow pattern adaptation area, the flow pattern characteristics of different areas within the water area can be refined. The data after area division can help identify the dynamic changes of the water flow, better adapt to different flow pattern requirements, and ensure the optimization of the water flow conditions in the training water area. Through the detection of the number of regional water flow pattern adaptations, the system can automatically determine the flow pattern adaptation situation of each area and mark the area as the first type or the second type of flow pattern area according to the adaptation number. This classification mechanism makes the water flow control in the training water area more targeted and improves the flow pattern adjustment efficiency during training. Based on the classification of the flow pattern areas, appropriate water flow control valve openings can be set for different areas to achieve refined water flow regulation. By setting the valve openings for the first type of flow pattern area and the second type of flow pattern area, the water flow can be controlled more effectively to meet different training requirements, improving the training effect and safety. This step can flexibly adjust the water flow conditions in the water area, making the rapids training water area more adaptable to the needs of different training stages. Trainers can adjust their strategies in a timely manner according to the actual performance of the water flow, improving the adaptability and efficiency of training. Through careful flow pattern control and valve opening setting, the occurrence of extreme water flow situations can be effectively avoided, ensuring the safety of trainers during rapids training, especially in training scenarios with large dynamic changes in water flow, providing safety guarantees.

[0027] Preferably, the segmentation of the water flow pattern adaptation area for the three-dimensional space model of the rapids training water area based on the simulated water flow correction data includes:

[0028] Confirm the spatial boundary of the three-dimensional space model of the rapids training water area to obtain the training water area spatial boundary data; based on the training water area spatial boundary data, perform water area terrain elevation analysis on the three-dimensional space model of the rapids training water area to generate water area terrain elevation data;

[0029] Perform spatial area division on the three-dimensional space model of the rapids training water area through the training water area spatial boundary data and the water area terrain elevation data, and grid the divided spatial areas to obtain the gridded area data of the training water area; extract the hydrological characteristics of the simulated water flow correction data to obtain the simulated water flow characteristics data, where the hydrological characteristics extraction includes flow velocity extraction, flow direction extraction, and vortex extraction;

[0030] Analyze the water flow pattern distribution of the gridded area data of the training water area by using the flow velocity characteristics and flow direction characteristics in the simulated water flow characteristics data to generate the regional water flow pattern distribution data; perform flow pattern transition analysis on the regional water flow pattern distribution data based on the eddy current characteristics in the simulated water flow characteristics data to generate the regional water flow pattern transition data;

[0031] Perform grid cell clustering adaptation on the gridded area data of the training water area according to the regional water flow pattern distribution data and the regional water flow pattern transition data, so as to generate the water flow pattern segmentation area data of the training water area.

[0032] Through spatial boundary confirmation, the actual scope of the training water area can be accurately identified, providing reliable basic data for subsequent analysis and modeling. Ensure the consistency between the model analysis and the real environment, laying a solid foundation for water flow adaptation and flow pattern analysis. Conduct elevation analysis on the water area terrain to accurately grasp the terrain undulation and water depth distribution of the water area, which plays a crucial role in water flow pattern analysis and the spatial division of the training water area, making the subsequent extraction of water flow characteristics and flow pattern distribution more refined. Through the division and grid processing of the spatial area, the training water area can be subdivided into multiple small areas, which helps to conduct more detailed water flow pattern analysis. The gridded area data can effectively support the adaptation and analysis of water flow patterns, improving the flexibility and accuracy of the model. By extracting hydrological characteristics such as flow velocity, flow direction, and eddy current, the basic dynamic characteristics of the water flow can be detailedly understood, which provides rich information for the subsequent water flow pattern distribution analysis, ensuring the true reflection and simulation of water flow behavior. Based on the flow velocity, flow direction, and eddy current data of the water flow characteristics, conduct flow pattern distribution analysis and transition analysis, which can reveal the change rules of water flow in different regions. The flow pattern transition analysis can help identify the complexity and potential problems of the flow patterns in the water area, providing a scientific basis for flow pattern adaptation. Through grid cell clustering adaptation, the water flow patterns in different regions can be reasonably classified and adjusted to ensure that the water flow state in each region meets the training requirements. This precise adaptation makes the flow pattern of the training water area more stable and controllable, optimizing the training conditions. This process provides a powerful dynamic adjustment ability for rapids training. It can adjust the water flow pattern of the training water area in real time according to the changes in the actual water flow to adapt to different training needs. In this way, the trainer can obtain a more challenging and diverse training experience in the changing water flow environment.

[0033] Preferably, the setting of the water flow control valve openings for the first type of flow pattern area and the second type of flow pattern area includes:

[0034] Perform initial water flow control valve orifice settings for the first type of flow regime area to obtain the first type of water flow control valve orifice setting data; conduct regional water flow regime diffusion analysis on the first type of flow regime area to generate regional water flow regime diffusion data; based on the regional water flow regime diffusion data, perform dynamic flow restriction on the first type of water flow control valve orifice setting data and set a diversion guide plate, thereby obtaining the first type of water flow control valve orifice data;

[0035] Perform initial water flow control valve orifice settings for the second type of flow regime area to obtain the second type of water flow control valve orifice setting data; conduct regional water flow regime aggregation analysis on the second type of flow regime area to generate regional water flow regime aggregation data; based on the regional water flow regime aggregation data, perform dynamic flow compensation on the second type of water flow control valve orifice setting data and set a flow regime guiding structure, thereby obtaining the second type of water flow control valve orifice data;

[0036] Integrate the first type of water flow control valve orifice data and the second type of water flow control valve orifice data to obtain the water flow control valve orifice data.

[0037] Through the valve orifice settings and dynamic adjustment for different flow regime regions, the present invention can precisely control the change of water flow, ensuring that the water flow state in each region meets the training requirements. This makes the water flow management in the entire training water area more efficient and avoids the impact of water flow instability on the training effect. Conducting diffusion analysis on the first type of flow regime region and aggregation analysis on the second type of flow regime region helps to scientifically understand the diffusion and aggregation characteristics of water flow in different regions. Such analysis can assist in optimizing the water flow control strategy, ensuring uniform distribution of water flow in different regions, and avoiding excessive concentration or dispersion of water flow, thereby improving the stability of the training environment. Through dynamic flow rate limitation and flow compensation of the valve orifice, dynamic adjustment can be performed according to the real-time change of water flow, ensuring that the water flow in each region always remains within the ideal range. This dynamic regulation helps to adapt to the change of water flow conditions, enabling the training water area to continuously and stably provide suitable water flow conditions. By setting diversion guide plates and flow regime guiding structures, the movement of water flow can be guided within different flow regime regions, optimizing the water flow path and avoiding local water flow disorder. This makes the water flow in the training water area more orderly, improving the training effect and controllability. Through targeted valve orifice settings and optimization processing for different types of flow regime regions, the water flow state of the training water area can be flexibly adjusted to adapt to different training requirements and scenarios. Whether in regions with a large flow rate or a small flow rate, a suitable water flow environment can be provided, enhancing the diversity and challenge of training. By precisely controlling the valve orifice settings of the water flow control in the flow regime region, potential safety hazards caused by excessive or too small flow rates can be effectively avoided. Especially in regions with complex flow regimes, the setting of dynamic flow rate limitation and guiding structures can reduce potential water flow impacts and irregular changes, thus ensuring safety during the training process. Through dynamic flow compensation and valve orifice control, the flow regime of the training water area can be adjusted more flexibly to quickly respond to different training requirements or changes. This reduces the time for manual intervention and adjustment and improves the overall training efficiency.

[0038] Preferably, step S3 includes the following steps:

[0039] Step S31: Obtain water area training vessel position information data;

[0040] Step S32: Import the water area training vessel position information data into the three-dimensional model of the white-water training water area for mapping the vessel driving position, and generate training vessel driving position mapping data;

[0041] Step S33: Analyze the vessel driving mode of the training vessel driving position mapping data to generate a training vessel driving mode, where the training vessel driving model includes a normal driving mode and an abnormal driving mode;

[0042] Step S34: Use the water flow control valve orifice data to perform vessel guidance control on the normal driving mode and the abnormal driving mode, and generate water flow control guidance data.

[0043] By obtaining the position information of the water area training vessel and importing it into the three-dimensional space model for mapping, the present invention can track the position change of the training vessel in real time. It ensures the accurate recording and visualization of the vessel movement during the training, providing real-time data support for subsequent dynamic control and analysis. By analyzing the mapping data of the traveling position of the training vessel, the normal traveling mode and abnormal traveling mode of the vessel can be identified. This analysis helps to understand the performance of the vessel under different conditions, providing a basis for further optimizing the vessel training plan and adjusting the water flow control strategy. Using the water flow control valve data to guide and control different traveling modes can make dynamic adjustments according to the normal and abnormal traveling conditions of the vessel. It ensures the stable operation of the vessel in the water flow environment and effectively avoids dangerous or abnormal behaviors. By identifying and controlling the abnormal traveling mode, irregular behaviors or potential risks in the training can be discovered and handled in a timely manner, ensuring the safety of the training process. Especially for uncontrolled abnormal behaviors, rapid guidance and control can effectively avoid accidents. By guiding and controlling the water flow to provide personalized support in different traveling modes, the training efficiency can be improved. When the vessel is traveling normally, the water flow control can ensure its smooth progress; while in the abnormal traveling mode, the guidance of the water flow control helps to correct the traveling path of the vessel, improving the accuracy and effectiveness of the training. This step can not only optimize the movement of the vessel, but also dynamically adjust the water flow control strategy according to different water flow conditions and vessel states, enabling the training scenario to more flexibly adapt to various environmental changes, thereby enhancing the diversity and challenge of the training.

[0044] Preferably, step S34 includes the following steps:

[0045] Step S341: Based on the normal traveling mode, use the mapping data of the traveling position of the training vessel to confirm the area in the three-dimensional space model of the rapids training water area, and obtain the vessel location area data; match the vessel location area data with the water flow regime segmentation area data of the training water area. If the area match is the first type of flow regime area, then control the water flow control valve data in the flow regime area in the direction of the ship's water flow to generate the water flow guidance data for the normal traveling flow regime area;

[0046] Step S342: If the area match is the second type of flow regime area, then do not process the corresponding water flow control valve data in the flow regime area;

[0047] Step S343: Based on the abnormal traveling mode, mark the abnormal area of the vessel location area data to obtain the vessel abnormal traveling area marking data; perform reverse water flow control on the water flow control valve data according to the vessel abnormal traveling area marking data, so as to generate the water flow guidance data for the abnormal traveling flow regime area;

[0048] Step S344: Integrate the water flow guiding data of the normal driving flow state area and the water flow guiding data of the abnormal driving flow state area to generate water flow control guiding data.

[0049] Through the confirmation of the area where the ship is located based on the normal driving mode and the matching with the water flow state segmentation area data in the present invention, it is ensured that the water flow control strategy in a specific area can be accurately applied to the ship. The control of the water flow direction along the ship in the first type of flow state area can better assist the ship to sail smoothly, improving the sailing efficiency and stability. By setting the control of the water flow direction along the ship for the first type of flow state area, it helps the ship to adapt to the flow state in the water area, thereby reducing the flow state interference, improving the training accuracy and the smoothness of the ship's sailing. Such a control mechanism effectively improves the controllability and accuracy of the ship's sailing during the training process. For the handling of the abnormal driving mode, by marking the abnormal driving area of the ship and performing reverse water flow control, the ship's driving trajectory can be corrected in time to avoid further deviation of the ship in abnormal situations. Through this mechanism, the training interruption or non-standard behaviors caused by abnormal driving are effectively reduced, improving the training safety. The strategy of not dealing with the second type of flow state area ensures the flexibility and efficiency during the training process. This processing method reduces excessive operations on areas that do not require intervention, optimizes the accuracy of water flow control and avoids unnecessary resource waste. By integrating the water flow guiding data of the normal driving flow state area and the water flow guiding data of the abnormal driving flow state area, a comprehensive water flow control scheme can be provided for the training process. This data integration ensures that the ship can obtain appropriate water flow guidance in different states, optimizing the decision support system of the entire training process. Through different control methods for normal and abnormal driving modes, the training system can dynamically adapt to the water area environment according to the different states of the ship. This adaptability makes the training environment more flexible, capable of not only supporting conventional training requirements but also coping with sudden abnormal situations.

[0050] Preferably, the reverse water flow control of the water flow control valve port data according to the marked data of the abnormal driving area of the ship includes:

[0051] Conduct spatial distribution analysis of the abnormal area according to the marked data of the abnormal driving area of the ship to obtain the boundary data of the abnormal area; confirm the water flow control valve port area affected according to the boundary data of the abnormal area to obtain the affected valve port area data;

[0052] Conduct reverse analysis of the direction and intensity of the water flow based on the affected valve port area data to obtain reverse water flow control data; adjust the water flow direction of the water flow control valve port data according to the reverse water flow control data to obtain water flow reverse adjustment data; optimize the flow state guidance of the marked data of the abnormal driving area of the ship through the water flow reverse adjustment data, and finally obtain the water flow guiding data of the abnormal driving flow state area.

[0053] Through the analysis of the spatial distribution of abnormal navigation areas, the present invention can accurately identify the areas where vessels navigate abnormally, and based on this data, identify the affected water flow control valve areas. This precise identification mechanism can ensure that the water flow control strategy is only applied to the abnormal navigation areas, thereby improving the accuracy of system response. Through the reverse analysis based on the data of the affected valve areas, the direction and intensity of the water flow can be effectively adjusted. This strategy helps to change the adverse water flow direction or velocity, correct the deviation of the vessel from the normal navigation track, and ensure that the vessel re-enters the correct navigation path. By making reverse adjustments through the water flow control valve data, the water flow direction can be flexibly changed to guide the vessel back to the correct waterway in the opposite direction. This dynamic water flow adjustment mechanism can respond in real time to the abnormal conditions of vessel navigation, improving the adaptability and flexibility of the training environment. Through the optimization of the flow pattern guidance in the abnormal navigation areas based on the reverse water flow adjustment data, it can be ensured that the vessel is timely adjusted in case of abnormal navigation and guided back to the normal navigation state. This not only improves the safety of vessel training but also effectively avoids the training deviation caused by abnormal navigation. The reverse water flow control strategy can correct the navigation track of the vessel in abnormal situations in real time, preventing the vessel from further deviating from the predetermined waterway. This improves the safety during the training process, reduces the negative impact of abnormal navigation on the training effect, and ensures the controllability and stability of the training. Through precise reverse water flow control and flow pattern guidance optimization, unnecessary intervention in the entire training water area is avoided, thus reducing resource waste. The precise allocation of resources ensures the efficiency and effect during the training process and improves the utilization rate of the water flow control valve data.

[0054] Preferably, step S4 includes the following steps:

[0055] Step S41: Conduct a water flow control feedback assessment on the water flow control guidance data to generate water flow control feedback assessment data;

[0056] Step S42: Extract water flow control features from the water flow control feedback assessment data to obtain water flow control feature data; divide the water flow control feature data into data sets to generate a model training set and a model test set; use the convolutional neural network algorithm to train the model training set to generate a water flow control pre-model;

[0057] Step S43: Optimize and iterate the water flow control pre-model through the model test set to generate a water flow control model.

[0058] Through the water flow control feedback evaluation of the water flow control guidance data, the operating state of the water flow control system can be monitored in real time, problems can be discovered and corrected in a timely manner, and this feedback mechanism improves the response speed and control accuracy of the system. Through feature extraction from the water flow control feedback evaluation data, the key factors affecting the water flow control effect can be identified, thus providing high-quality input data for subsequent model training. This process helps optimize the control strategy and improve the intelligent level of the water flow control system. Dividing the water flow control feature data into a model training set and a test set helps avoid overfitting problems and improve the generalization ability of the model. By training the model training set using the Convolutional Neural Network (CNN) algorithm, efficient feature learning and model optimization can be achieved, enabling the water flow control system to automatically adapt to complex water environments. The convolutional neural network can achieve deeper water flow pattern recognition through hierarchical feature extraction, enhancing the intelligent level of water flow control. Through the optimization and iteration of the model, the water flow control pre-model can continuously improve its prediction and control accuracy, adapt to more complex water environments, and the generated water flow control model can accurately predict and adjust the water flow pattern, apply appropriate control strategies in different flow pattern regions, thereby achieving precise water flow control. This model not only helps improve the stability of ship navigation but also provides scientific decision-making support for future training. Through repeated optimization and iteration, the water flow control model can achieve automatic adjustment, enhancing the system's adaptability in a dynamically changing environment. The continuously improved model during the training process will be able to quickly adapt to different training scenarios and provide more accurate control strategies. Through the efficient learning ability of the convolutional neural network, the water flow control pre-model can reduce human intervention and quickly adapt to changes in the water environment. During the training process, through the feedback evaluation of data, the system can continuously improve the control accuracy and reliability, thereby enhancing the overall training efficiency and quality. After model optimization and iteration, the water flow control model can more accurately predict and guide the water flow, thus improving the stability and reliability of system control. The continuous optimization of the model ensures that the water flow control system can operate efficiently in complex water environments and meet the training requirements to the greatest extent. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 It is a schematic flowchart of the steps of a method for constructing a water flow control model for a rapids training water area;

[0060] Figure 2 is Figure 1 a detailed implementation step flowchart of step S2 in

[0061] Figure 3 is Figure 1 a detailed implementation step flowchart of step S3 in

[0062] The realization, functional features, and advantages of the present invention will be further described in conjunction with embodiments and with reference to the accompanying drawings. Detailed Embodiment

[0063] The technical method of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those skilled in the art within the scope of the present invention without creative efforts belong to the scope of protection of the present invention.

[0064] In addition, the accompanying drawings are only schematic diagrams of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and thus repeated descriptions of them will be omitted. Some of the block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor methods and / or microcontroller methods.

[0065] It should be understood that although terms such as "first" and "second" may be used here to describe various units, these units should not be limited by these terms. These terms are only used to distinguish one unit from another. For example, without departing from the scope of the exemplary embodiments, the first unit can be called the second unit, and similarly the second unit can be called the first unit. The term "and / or" used here includes any and all combinations of one or more of the listed related items.

[0066] To achieve the above object, please refer to Figures 1 to 3 , a method for constructing a water flow control model of a rapids training water area, the method comprising the following steps:

[0067] Step S1: Obtain the environmental perception data of the rapids training water area; perform a three-dimensional spatial modeling on the environmental perception data of the rapids training water area to generate a three-dimensional spatial model of the rapids training water area; perform a water flow dynamics simulation on the three-dimensional spatial model of the rapids training water area to generate the water flow simulation data of the rapids training water area;

[0068] Step S2: Obtain the actual hydrological data of the rapids training water area; correct the simulated water flow data with the actual hydrological data of the rapids training water area to generate the corrected simulated water flow data; perform a regional water flow pattern adaptation analysis on the three-dimensional spatial model of the rapids training water area based on the corrected simulated water flow data to generate a first type of flow pattern area and a second type of flow pattern area; set the water flow control valve openings for the first type of flow pattern area and the second type of flow pattern area to obtain the water flow control valve opening data;

[0069] Step S3: Obtain the position information data of the water area training vessels; import the position information data of the water area training vessels into the analysis of the vessel driving mode in the three-dimensional model of the rapids training water area to generate the vessel driving mode, where the vessel driving model includes a normal driving mode and an abnormal driving mode; use the water flow control valve data to conduct vessel guidance control on the normal driving mode and the abnormal driving mode to generate water flow control guidance data;

[0070] Step S4: Extract the water flow control features from the water flow control guidance data to obtain the water flow control feature data; use the convolutional neural network algorithm to construct a model for the water flow control feature data, thereby generating a water flow control model.

[0071] The present invention generates an accurate three-dimensional model of the rapids training water area by obtaining environmental perception data and performing three-dimensional spatial modeling, and combines hydrodynamic simulation to generate water flow simulation data, which can accurately reproduce the water flow characteristics of the training water area, providing real and reliable basic data for subsequent water flow control. This process ensures the accuracy and comprehensiveness of the model, avoiding the limitations of traditional models that rely on simplified assumptions. By calibrating the actual hydrological data with the simulated water flow data, the differences between the simulation and the actual situation can be eliminated, improving the accuracy of the model. The flow state adaptation analysis further refines the water flow characteristics of different regions, ensuring that reasonable water flow control valves can be set according to the regional characteristics, making the water flow control more targeted and precise. By importing the vessel position information into the model, the vessel driving mode can be tracked in real time, and vessel guidance control can be achieved based on the water flow control valve data. By making corresponding water flow adjustments according to the normal and abnormal driving modes of the vessels, the water flow environment can be dynamically optimized, improving the safety and effectiveness of the training process and ensuring the best driving route for the training vessels. By extracting features from the water flow control guidance data and using the convolutional neural network algorithm to construct an intelligent water flow control model, the water flow control strategy can be automatically learned and optimized from the training data, making the water flow control more intelligent and adaptive. This intelligent model can make real-time adjustments according to different training requirements and environmental changes, greatly improving the flexibility and accuracy of the training water area. Therefore, the present invention improves the accuracy and adaptability of the water flow control model in the rapids training water area by combining three-dimensional modeling, water flow simulation, actual hydrological data calibration, vessel driving mode analysis, and convolutional neural network algorithm.

[0072] In the embodiment of the present invention, refer to Figure 1 As shown, it is a schematic diagram of the step flow of a method for constructing a water flow control model in a rapids training water area of the present invention. In this example, the method for constructing a water flow control model in a rapids training water area includes the following steps:

[0073] Step S1: Obtain the environmental perception data of the white-water training water area; perform 3D spatial modeling on the environmental perception data of the white-water training water area to generate a 3D spatial model of the white-water training water area; perform hydrodynamic simulation on the 3D spatial model of the white-water training water area to generate white-water training water area flow simulation data;

[0074] In the embodiments of the present invention, environmental perception data of the white-water training waters is obtained through a variety of sensors and devices. Common devices include water flow sensors, depth detectors, temperature sensors, humidity sensors, meteorological monitoring devices, etc. Select a suitable sensor array, including flow velocity sensors, water level sensors, weather stations, etc., to ensure the comprehensiveness and accuracy of the data. Regularly and real-time collect various data of the water area environment, including water flow velocity, flow direction, water depth, temperature, humidity, meteorological data, etc. Data collection can be carried out based on fixed positions and dynamic measurement methods. Perform preprocessing such as denoising, filtering, and interpolation on the collected raw data to ensure the cleanliness and consistency of the data. Utilize the obtained environmental perception data, especially information such as water depth, geographical location, and obstacles, to construct a three-dimensional spatial model of the water area. Combine the two-dimensional data (such as water flow, temperature, etc.) obtained by the sensors with a three-dimensional coordinate system to form environmental data with spatial distribution. High-precision modeling can be achieved through lidar (LiDAR), topographic map data, etc. Use computer graphics or CAD software (such as AutoCAD, Revit, etc.) to convert the collected environmental perception data into a digital three-dimensional model. This model needs to accurately represent the topography and geomorphology of the water area, the fluid motion area, and obstacles, etc. Mesh the three-dimensional model to ensure that the model has sufficient accuracy and operability during computational simulation. The meshing process can be refined according to the different regional characteristics of the water area (such as deep water areas, shoal areas). Based on the constructed three-dimensional spatial model of the white-water training waters, conduct water flow simulation using the principles of fluid mechanics. This simulation model should consider various factors of the water flow, such as flow velocity, flow direction, eddy current, turbulence, etc. Select a suitable computational fluid dynamics simulation software (such as ANSYS Fluent, OpenFOAM, etc.) for simulation. According to the specific situation of the training waters, select a suitable solver and boundary conditions. Set the necessary physical parameters for the water flow simulation, including water density, viscosity, flow velocity, flow direction, etc., and at the same time set the fluid boundary conditions for different regions, such as water depth, friction coefficient of the shoreline, external force acting on the fluid, etc. Set the simulation boundary conditions according to the actual situation of the water area, such as water surface flow velocity, inlet flow rate, resistance, etc. Adjust the boundaries in the simulation model according to the flow velocity, flow direction, etc. information obtained by the sensors. Through simulation calculations, obtain information such as the flow state, velocity distribution, and turbulence area of the water flow in different regions. The simulation results will include water flow field data (such as velocity distribution, pressure distribution, etc.) of each region. Use the visualization tools provided by the simulation software to generate dynamic water flow diagrams and three-dimensional flow field animations to help understand the change rules of the water flow and its impact on training. Extract relevant data from the simulation process, covering water flow simulation data in multiple dimensions such as water flow velocity, flow direction, pressure, and turbulence intensity.

[0075] Step S2: Obtain the actual hydrological data of the rapids training water area; correct the simulated water flow data of the rapids training water area through the actual hydrological data of the rapids training water area to generate simulated water flow correction data; perform regional water flow pattern adaptation analysis on the three-dimensional spatial model of the rapids training water area based on the simulated water flow correction data to generate a first type of flow pattern area and a second type of flow pattern area; set the water flow control valve openings for the first type of flow pattern area and the second type of flow pattern area to obtain water flow control valve opening data;

[0076] In the embodiments of the present invention, actual hydrological data of the rapids training water area is collected through deployed hydrological monitoring devices (such as water level gauges, current meters, temperature sensors, weather stations, etc.). The data items usually include: regularly measuring the speed and direction of the water flow, especially at key positions in the training area. Recording the water level changes at different positions to understand the water volume and depth of the water area, including meteorological factors such as wind speed, air temperature, precipitation, etc., which will affect the water flow and other water quality indicators that affect hydrodynamics. Using technical means such as hydrological monitoring networks and unmanned surface vehicles (USVs) to collect water area data regularly or in real-time to ensure the integrity and timeliness of the data. Comparing the actually collected hydrological data with the water flow simulation data to determine the errors and deviations between the two. The comparison content includes: comparing the actually measured water flow speed with the water flow speed calculated by the simulation model to find the differences between the two. Comparing the actually measured water flow direction with the water flow direction calculated in the simulation flow field to evaluate the accuracy of the water flow direction of the simulation model. Comparing the actual water level with the simulated water level and analyzing the accuracy of the water level change. Based on the error analysis, data correction methods such as the least squares method, Kalman filtering, regression analysis, etc. are used to adjust the simulated water flow data. The correction goal is to reduce the deviation between the simulation and the actual data, making the simulation results closer to the actual water area situation. For the water flow speed, direction and other data at different times and regions, the simulation results are dynamically adjusted to reflect the flow characteristics in the actual water area. Through the corrected water flow data, the simulated water flow corrected data is obtained, providing more accurate water flow field information and being able to better interface with the actual hydrological data. Using the simulated water flow corrected data, a detailed water flow pattern adaptation analysis is carried out on the three-dimensional spatial model of the training water area. The specific analysis content includes: analyzing the water flow states (such as laminar flow, turbulent flow, etc.) in different regions and classifying the flow patterns. According to the simulated water flow corrected data, spatial distribution analysis of parameters such as water flow speed and direction in different regions is carried out to identify the flow pattern characteristics of different regions. Analyzing the factors affecting the water flow pattern in the water area (such as obstacles, shoreline morphology, water depth changes, etc.) and integrating these factors into the water flow adaptation model. Based on the results of the adaptation analysis, the three-dimensional spatial model of the training water area is divided into different types of flow pattern regions. The flow pattern regions are usually divided into the following two categories: The first type of flow pattern region refers to the region with relatively low water flow speed or complex flow patterns such as vortex flow and turbulent flow, usually a special region where the flow pattern needs to be controlled. The second type of flow pattern region refers to the region with relatively large water flow speed, relatively stable water flow direction and no obvious turbulent flow characteristics such as vortex flow, usually a normal water flow region. According to the analysis results, the first type of flow pattern region and the second type of flow pattern region are generated, and the specific positions and areas of each region are marked to form the flow pattern region data. For the first type of flow pattern region, due to the complex flow pattern or the existence of turbulent flow phenomenon, different control strategies are adopted: when setting the valve opening, the opening and flow rate of the valve are dynamically adjusted according to the flow pattern change to avoid excessive vortex flow or unstable water flow.In the first type of flow regime area, a special flow regime guiding structure (such as a flow direction guiding plate or a flow velocity regulating device) is set to guide the water flow direction to ensure the stability and control effect of the water flow. For the second type of flow regime area, conventional water flow control strategies are adopted for appropriate water flow regulation. When setting the valve port data, the following factors are considered: a water flow control valve port is set in the second type of area to control the flow velocity and direction to optimize the flow regime. The opening and closing degree of the valve port can be adjusted according to the flow velocity and water flow demand to ensure that the flow rate is within an appropriate range. When necessary, in order to improve the flow regime, a diversion guide plate can be set near the valve port to guide the water flow in a specific direction, thus avoiding water flow disorder. According to the different requirements of the first type and the second type of flow regime areas, the water flow control valve ports are set respectively, and specific water flow control valve port data are generated, including the type, position, flow rate, control method, etc. of the valve port. The water flow control valve port data of all the first type and the second type of flow regime areas are integrated to generate a complete set of water flow control valve port data, providing a basis for subsequent water area control and vessel navigation.

[0077] Step S3: Obtain the water area training vessel position information data; import the water area training vessel position information data into the analysis of the vessel driving mode in the three-dimensional model of the rapids training water area to generate a training vessel driving mode, where the training vessel driving model includes a normal driving mode and an abnormal driving mode; use the water flow control valve port data to conduct vessel guidance control on the normal driving mode and the abnormal driving mode to generate water flow control guidance data;

[0078] In the embodiments of the present invention, the real-time position information of water area training vessels is collected through a positioning system (such as GPS or Real-Time Location System (RTLS)). Specifically, it includes: the latitude and longitude data of each training vessel, which is used to mark the exact position of the vessel in the water area. The heading angle (direction) and speed of the vessel are obtained to assist in analyzing the driving state of the vessel. Through continuous positioning information, the movement trajectory of the vessel in the water area is recorded, providing a basis for subsequent driving mode analysis. The collected position information contains noise or errors, and data cleaning and filtering are required to ensure the accuracy of the data. Kalman filtering or other filtering techniques are used to remove unnecessary noise and error points to ensure the high precision of the position information. The position information data of the water area training vessels is imported into the three-dimensional spatial model of the white-water training water area, and this three-dimensional model already includes the terrain, environmental characteristics, and water flow data of the water area. According to the real-time position information of the vessel, the current position of the vessel is matched with the water area terrain in the three-dimensional model, so that the driving state of the vessel in different water area regions can be accurately judged. Based on the real-time position, heading, speed, and other information of the vessel, its driving mode is analyzed. It is mainly divided into the following two modes: When the vessel sails steadily along the predetermined route and is not affected by unexpected factors, it is defined as the normal driving mode. When the vessel deviates from the predetermined route and there are abnormal behaviors such as sharp turning, drifting, and reverse driving, it is defined as the abnormal driving mode. Through the trajectory and speed changes of the vessel, the stability of its driving is analyzed to determine whether it conforms to the normal driving mode. According to the preset driving path and behavior rules, a deviation threshold is set. When the driving trajectory of the vessel exceeds this threshold, it is determined as the abnormal driving mode. The analysis result is converted into driving mode data, indicating the current driving mode (normal or abnormal) of the vessel, as well as the deviation path or risk factors. According to the result of the vessel driving mode analysis, specific training vessel driving mode data is generated, and this data includes: information such as the sailing route, speed, and flow direction of the vessel in the normal driving mode. The behavioral characteristics of the vessel in the abnormal driving mode are recorded, such as sharp turning, out-of-control drifting, and collision with obstacles. In the normal driving mode, the vessel can rely on the preset route and flow regime to maintain its stability. Using the water flow control valve data (such as water flow direction and velocity), the vessel is guided along the ideal path. The water flow in the normal driving area is appropriately controlled to ensure that the water flow direction and velocity meet the requirements of the vessel's driving, so that the vessel can sail smoothly. According to the analysis result of the normal driving mode, the water flow intensity and direction are adjusted by controlling the valve to guide the vessel to sail along the predetermined route. For vessels entering the abnormal driving mode, special guiding control is required. If the vessel deviates from the route or shows unstable driving, the water flow direction can be adjusted through the water flow control valve data, and reverse water flow control is applied to help the vessel return to the normal sailing state. According to the analysis result of the abnormal driving mode, it is necessary to slow down the vessel speed and adjust the water flow direction through the water flow control valve to prevent the vessel from further deviating from the channel or encountering danger.According to the above control strategies for normal and abnormal driving modes, corresponding water flow control guidance data are generated, including: the opening and closing states of the control valve ports, which determine the intensity and direction of the water flow. Based on different driving modes, the flow direction and velocity of the water flow are adjusted to meet the driving requirements of the vessel, including water flow guidance strategies, command data for vessel navigation, etc., to ensure that the vessel sails smoothly according to the expected mode.

[0079] Step S4: Extract water flow control features from the water flow control guidance data to obtain water flow control feature data; use the convolutional neural network algorithm to construct a model for the water flow control feature data, thereby generating a water flow control model.

[0080] In the embodiments of the present invention, the collected water flow control and guidance data is preprocessed, including removing outliers, filling in missing values, and normalizing, to make the data more suitable for feature extraction. Water flow control features such as flow velocity and flow direction are extracted from the water flow control and guidance data. These features directly affect the navigation state of the ship and play a key role especially when controlling the ship's travel path. Based on the water flow simulation data, environmental features related to the water flow are extracted, such as the stability of the water flow, the rate of change of the flow direction, and the fluctuation of the flow velocity. These extracted features help to further understand the impact of the water flow on the ship's travel mode. Valve opening adjustment features are extracted from the control and guidance data, and the opening and closing states and adjustment amplitudes of different valves are analyzed to judge their impact on water flow control. According to the correlation between the ship's travel mode and water flow control, features with a greater impact on water flow control are selected, and redundant or irrelevant features are removed. Methods such as correlation analysis and information gain are used to ensure that the most effective control features are selected. Dimensionality reduction algorithms such as PCA (Principal Component Analysis) or t-SNE are used to reduce the dimensionality of the features to reduce the data dimension and extract key features, facilitating subsequent neural network modeling. According to the extracted water flow control feature data, a suitable convolutional neural network architecture is designed. Common CNN structures include convolutional layers, pooling layers, fully connected layers, and output layers. The input layer receives the processed water flow control feature data. Multiple convolutional kernels are used to perform convolutional operations on the input features to extract local patterns and correlations in the water flow feature data. Through pooling operations (such as max pooling or average pooling), the size of the feature map is reduced while important feature information is retained. After the data processed by convolution and pooling is flattened, it is further processed through the fully connected layer and gradually transformed into output predictions. The output layer generates the final control strategy according to the goals of water flow control (such as the opening and closing states of the valves, and the adjustment of the flow direction and flow velocity). According to the extracted water flow control feature data and the corresponding training goals (such as the accurate path of the ship's travel, the water flow control and guidance effect, etc.), a training set is generated. The training set should include different scenarios of water flow control so that the model can learn different water flow control strategies. A suitable loss function (such as Mean Squared Error MSE or cross-entropy) is selected to measure the difference between the prediction effect of the model and the actual control goal. An optimization algorithm (such as Adam or SGD) is used to optimize the model parameters, gradually adjusting the convolutional kernels and weight parameters to minimize the model prediction error. An independent validation set is used to validate the trained model to check the performance of the model on unseen data. According to the validation results, the hyperparameters of the model, such as the convolutional kernel size, the number of layers, and the learning rate, are adjusted to further optimize the model performance. Through the cross-validation method, the generalization ability of the model is further improved to avoid overfitting. The performance of the convolutional neural network model is evaluated to ensure that it has sufficient accuracy and stability and can effectively perform water flow control and guidance prediction. The trained water flow control model is saved as a file (such as a TensorFlow model file) and deployed to a real-time system to support actual water flow control tasks.Use the trained convolutional neural network model for real-time water flow control to guide the vessel to sail along a predetermined path or perform countercurrent guidance control in case of anomalies.

[0081] Preferably, step S1 includes the following steps:

[0082] Step S11: Perform water area environment perception on the rapids training water area to obtain rapids training water area environment perception data;

[0083] Step S12: Perform data preprocessing on the rapids training water area environment perception data to generate standard water area environment perception data, where data preprocessing includes data cleaning, data denoising, missing value filling, and data standardization;

[0084] Step S13: Extract the terrain height, waterway shape, and obstacle distribution data from the standard water area environment perception data, and perform three-dimensional spatial modeling based on the terrain height, waterway shape, and obstacle distribution data to generate a three-dimensional spatial model of the rapids training water area;

[0085] Step S14: Extract the deployment position of the water discharge outlet of the rapids training water area from the three-dimensional spatial model of the rapids training water area to obtain the deployment position data of the water discharge outlet of the water area; perform hydrodynamic simulation on the three-dimensional spatial model of the rapids training water area based on the deployment position data of the water discharge outlet of the water area to generate rapids training water area water flow simulation data.

[0086] In the embodiments of the present invention, the environment of the rapids training water area is perceived by drones, satellite images or ground sensors, and these sensors may include water quality sensors, meteorological monitoring devices, water flow sensors, etc. Environmental perception data including water flow velocity, temperature, air pressure, humidity, and pollutant data present in the water body is obtained through these devices. Invalid data, abnormal data, and error data are removed. The influence of environmental noise is removed by using a filter or an algorithm (such as wavelet transform). An interpolation algorithm (such as linear interpolation or Kriging interpolation) is used to fill in the missing values in the data. The data from different sources is uniformly standardized to ensure that the data is analyzed under the same dimension. Key information is extracted from the preprocessed standard water area environmental perception data: the water area terrain height data is extracted by using lidar or a digital elevation model (DEM). The flow direction and width of the waterway are extracted through remote sensing images and hydrological data. The spatial distribution data of water area obstacles is generated by using obstacle scanning within the water area (such as an obstacle detection system). Based on these data, spatial modeling is carried out by using 3D modeling technology (such as 3D modeling software or GIS-based modeling tools) to generate a three-dimensional spatial model of the water area. According to the three-dimensional model of the water area and the water flow analysis, the position suitable for setting the water flow outlet is determined. Position selection is carried out through an optimization algorithm (such as a genetic algorithm or a simulated annealing algorithm) to ensure the optimization of the water flow regulation function. Using a CFD (Computational Fluid Dynamics) simulation tool, the water flow is simulated based on the water area outlet deployment position data. The simulation process simulates the characteristics of water flow velocity, direction, turbulence, etc., as well as the influence of the water flow at the outlet, thereby generating detailed water flow simulation data.

[0087] Preferably, the spatial three-dimensional modeling based on the terrain height, waterway shape, and obstacle distribution data includes:

[0088] Construct terrain grid data according to the terrain height, waterway shape, and obstacle distribution data to obtain terrain grid data; extract water area streamline analysis data from the terrain grid data to obtain water area streamline analysis data;

[0089] Construct a water area profile based on the water area streamline analysis data to obtain water area profile data; construct a refined obstacle model for the terrain grid data through the water area profile data to obtain refined obstacle model data;

[0090] Generate a water area three-dimensional surface from the refined obstacle model data by using the water area streamline analysis data to obtain water area three-dimensional surface data; map the water area three-dimensional surface data to obtain a three-dimensional spatial model of the rapids training water area.

[0091] In the embodiments of the present invention, three-dimensional reconstruction of water areas is performed by using Light Detection and Ranging (LiDAR) data, Digital Elevation Model (DEM), and remote sensing images. The terrain data is transformed into a grid model through a hierarchical grid method (e.g., using uniform grids or adaptive grid division) for subsequent analysis. By fusing watercourse shape and obstacle distribution data, the characteristics of grid regions are defined. For example, information such as watercourse width, height variation, and obstacle position is introduced into the grid. Based on the terrain grid data and a hydrological model (such as a hydrodynamic model or a hydraulic model), water flow streamlines are calculated through numerical simulation or analytical methods. These streamlines represent the main flow directions and velocities of water flow in the water area. The streamline tracking algorithm (such as the streamline tracking method or the discrete streamline method) is used to extract streamline data in the water area and convert it into analysis data. These streamline data provide important information on the water flow pattern in the water area. According to the streamline analysis data, profiles at different positions in the water area (such as the middle of the watercourse, the edge, near obstacles, etc.) are selected and vertically dissected. The profiles contain topographic features such as water depth, flow velocity, and obstacles, providing detailed longitudinal information on the water area. The profile construction algorithm (such as contour generation, outline generation) is used to generate spatial profiles of the water area to ensure that the three-dimensional structure of the water area can be comprehensively captured. Based on the water area profile data and using the obstacle information in the terrain grid data, the obstacles are refined. The refinement model considers the influence of water flow on the obstacles and adjusts the spatial position, shape, and interaction relationship with the water flow of the obstacles. A refinement algorithm (such as the finite element method or the grid refinement algorithm) is used to further accurately detail the obstacles and generate more accurate obstacle model data. Combining the water area streamline analysis data, a three-dimensional surface of the water area is generated using the obstacle refinement model data. The surface represents the result of the interaction between water flow and obstacles and can be smoothed and optimized through interpolation methods (such as B-spline interpolation or Kriging interpolation). The generated three-dimensional surface is optimized and adjusted to more accurately reflect the changes in water flow and the influence of obstacles in the water area. Based on the water area three-dimensional surface data, spatial three-dimensional mapping is performed. Through three-dimensional modeling software (such as AutoCAD, Blender, or GIS tools), the terrain and water flow information of the water area are integrated into a unified three-dimensional model. Finally, the obstacle, streamline, and water area profile data are mapped into three-dimensional space to generate a complete three-dimensional model of the white-water training water area. This model can reflect the water flow changes, obstacle positions, and water area morphology and serve as the basis for subsequent simulation and training.

[0092] As an example of the present invention, refer to Figure 2 shown. In this example, step S2 includes:

[0093] Step S21: Use sensors to obtain actual hydrological data of the white-water training water area;

[0094] Step S22: Correct the simulated water flow data of the rapids training water area through the actual hydrological data of the rapids training water area to generate simulated water flow correction data; based on the simulated water flow correction data, perform segmentation of the water flow pattern adaptation area on the three-dimensional spatial model of the rapids training water area to generate the training water area water flow pattern segmentation area data;

[0095] Step S23: Detect the number of regional water flow pattern adaptations for the training water area water flow pattern segmentation area data. When the number of regional water flow pattern adaptations is greater than or equal to the preset standard number of regional water flow pattern adaptations, mark the corresponding training water area water flow pattern segmentation area data as the first type of flow pattern area;

[0096] Step S24: When the number of regional water flow pattern adaptations is less than the preset standard number of regional water flow pattern adaptations, mark the corresponding training water area water flow pattern segmentation area data as the second type of flow pattern area; set the water flow control valve openings for the first type of flow pattern area and the second type of flow pattern area to obtain the water flow control valve opening data.

[0097] In the embodiments of the present invention, by deploying a variety of sensors in the white-water training waters, including water flow sensors, temperature sensors, pressure sensors, water quality monitoring sensors, etc., it is used to monitor physical and chemical parameters such as water flow velocity, flow direction, water level, water temperature, salinity, etc. of the waters in real time. These hydrological data are collected in real time through sensors to ensure high-frequency and high-precision acquisition of the data, so as to reflect the dynamic changes of the waters. Based on the three-dimensional spatial model of the white-water training waters, preliminary water flow simulation data are generated through hydrodynamic simulation (such as CFD simulation). These simulation data represent the water flow behavior of the waters under different conditions, including flow velocity, flow direction, turbulence, etc. By comparing the actual hydrological data collected by the sensors with the simulated water flow data, the simulation data are corrected using error correction methods (such as the least squares method, Kalman filter). This process aims to reduce the difference between the simulation data and the actual data, making the simulation results more accurate. According to the corrected simulated water flow data, the three-dimensional spatial model of the training waters is divided into multiple regions through flow pattern analysis methods. These regions are classified according to characteristics such as the flow velocity, flow direction, and turbulence of the water flow. For example, it can be segmented according to the change of the flow velocity gradient and flow pattern to ensure that the water flow characteristics of each region are similar. The water flow pattern adaptation region data of the training waters are generated through spatial division algorithms (such as Voronoi division or grid method). In each segmented region, the detection of the number of water flow pattern adaptations is carried out. Specifically, it is detected whether there are a sufficient number of flow pattern adaptation samples in this region (for example, detecting whether the water flow velocity and turbulence degree in this region meet the flow pattern standard). The preset standard number of water flow pattern adaptations in the standard region is usually determined by the characteristics of the water flow model and the training requirements. When the number of water flow pattern adaptations in a certain region is greater than or equal to the preset standard, this region is marked as the first type of flow pattern region, indicating that the water flow condition in this region meets the training requirements. If the adaptation number is lower than the preset standard, it is marked as the second type of flow pattern region. According to the different flow pattern regions, water flow control valves (or regulating valves) are configured at key positions in the training waters to control the flow direction and flow velocity of the water. For the first type of flow pattern region, valves suitable for maintaining a stable training environment are set, for example, in a region where the water flow velocity is relatively stable; for the second type of flow pattern region, valves are set to optimize regions with uneven water flow or low water flow velocity. The water flow control valve data are generated according to the position, adjustment range, and flow control parameters of the valve, and these data will be used for subsequent water flow regulation and the realization of training goals.

[0098] Preferably, the segmentation of the water flow pattern adaptation region of the three-dimensional spatial model of the white-water training waters based on the corrected simulated water flow data includes:

[0099] Confirm the spatial boundary of the three-dimensional spatial model of the white-water training waters to obtain the spatial boundary data of the training waters; based on the spatial boundary data of the training waters, conduct an analysis of the water area terrain elevation of the three-dimensional spatial model of the white-water training waters to generate the water area terrain elevation data;

[0100] The spatial region of the rapids training water area three-dimensional model is divided by training water area space boundary data and water area terrain elevation data, and the divided spatial regions are meshed to obtain the meshed region data of the training water area; the hydrological characteristics of the simulated water flow correction data are extracted to obtain the simulated water flow characteristics data, where the hydrological characteristics extraction includes flow velocity extraction, flow direction extraction, and vortex extraction;

[0101] The water flow pattern distribution analysis of the meshed region data of the training water area is carried out by using the flow velocity characteristics and flow direction characteristics in the simulated water flow characteristics data to generate the regional water flow pattern distribution data; the flow pattern transition analysis of the regional water flow pattern distribution data is carried out based on the vortex characteristics in the simulated water flow characteristics data to generate the regional water flow pattern transition data;

[0102] The grid cell clustering adaptation of the meshed region data of the training water area is carried out according to the regional water flow pattern distribution data and the regional water flow pattern transition data, so as to generate the water flow pattern segmentation region data of the training water area.

[0103] In the embodiments of the present invention, spatial boundary data of the white-water training waters is obtained through high-precision Geographic Information System (GIS) data. This step involves using remote sensing images, satellite data, or existing geographical data models to ensure accurate identification of the geographical boundaries of the waters and the edge features of the waters. The confirmed spatial boundary data of the waters serves as the basis for subsequent analysis and can effectively guide subsequent analysis of the terrain elevation of the waters and spatial area division. After confirming the spatial boundary data, a Digital Elevation Model (DEM) is used to analyze the terrain elevation of the training waters. By analyzing features such as elevation, slope, and water flow convergence areas in different regions of the waters, terrain elevation data of the waters is generated. The terrain elevation data of the waters provides an important basis for subsequent spatial area division and grid generation, especially in judging water flow changes and their flow patterns. Combining the spatial boundary data of the waters with the terrain elevation data of the waters, spatial analysis methods (such as multi-dimensional spatial division or K-means clustering algorithm) are used to divide the white-water training waters into regions. Different regions can be partitioned based on factors such as elevation differences, flow obstacles, or flow velocity distributions. The divided spatial regions are further gridded. By using regular grids (such as square or rectangular grids) within each region, gridded area data is generated. Each grid cell will contain corresponding water flow feature data according to its specific position in the waters, generating gridded area data of the training waters for subsequent analysis of water flow patterns. By simulating water flow correction data, key hydrological features are extracted. This process involves extracting features such as flow velocity, flow direction, and eddies from the simulation data: water flow velocity data of each region is extracted through velocity field analysis; the flow direction features of the water flow are extracted according to the water flow direction field data; a hydrodynamic model is used to identify the eddies existing in the waters and their impacts, generating simulation water flow feature data, which provides a basis for subsequent analysis of water flow pattern distribution and flow pattern transition analysis. Using the flow velocity and flow direction features, the water flow patterns of different grid cells are analyzed. The specific methods include: analyzing the flow state by comparing the water flow velocities and flow directions in different regions; using the flow velocity and flow direction features to generate water flow pattern distribution data for each grid cell to ensure reflection of the differences in flow patterns in the waters (such as steady flow, accelerating flow, turbulent flow, etc.), generating regional water flow pattern distribution data, which will contribute to subsequent flow pattern transition analysis and area division. Based on the eddy features, flow pattern transition analysis is performed on the regional water flow pattern distribution data. This process aims to identify the transition regions of the water flow in the waters, such as the transition region from a stable flow pattern to a turbulent flow pattern. The eddy features can help identify regions where the water flow is unstable or undergoes drastic changes, and then analyze the transition of the water flow patterns, generating regional water flow pattern transition data for adjusting the flow pattern adaptation of the regions.According to the regional water flow regime distribution data and flow regime transition data, the grid cells are adapted using a clustering algorithm (such as the K-means clustering or DBSCAN algorithm). In this step, based on the distribution characteristics of the flow regime, the grid cells with similar flow regimes are grouped into the same class, realizing the adaptation of the water flow regime between regions and generating the water flow regime segmentation region data of the training water area, which reflects the specific distribution of different flow regime regions within the water area. Through clustering, the training water area is finally divided into multiple regions with similar flow regime characteristics, providing an accurate regional division basis for water flow control and training.

[0104] Preferably, the setting of the water flow control valve openings for the first type of flow regime region and the second type of flow regime region includes:

[0105] Perform an initial water flow control valve opening setting for the first type of flow regime region to obtain the first type of water flow control valve opening setting data; conduct a regional water flow regime diffusion analysis for the first type of flow regime region to generate regional water flow regime diffusion data; based on the regional water flow regime diffusion data, perform a dynamic flow rate limit on the first type of water flow control valve opening setting data and set a diversion guide plate, thereby obtaining the first type of water flow control valve opening data;

[0106] Perform an initial water flow control valve opening setting for the second type of flow regime region to obtain the second type of water flow control valve opening setting data; conduct a regional water flow regime aggregation analysis for the second type of flow regime region to generate regional water flow regime aggregation data; based on the regional water flow regime aggregation data, perform a dynamic flow rate compensation on the second type of water flow control valve opening setting data and set a flow regime guiding structure, thereby obtaining the second type of water flow control valve opening data;

[0107] Integrate the first type of water flow control valve opening data and the second type of water flow control valve opening data to obtain the water flow control valve opening data.

[0108] In the embodiments of the present invention, for the first type of flow regime area, the position and size of the initial water flow control valve orifice are determined based on the water flow regime characteristics (such as flow velocity, flow direction, etc.) of the water area. This setting needs to ensure that the valve orifice can effectively control and regulate the water flow according to the flow regime characteristics. The initial setting is based on numerical simulations and existing water area control experience to obtain the first type of water flow control valve orifice setting data, including initial control parameters such as valve orifice position, size, and opening degree. Through simulation or actual hydrological data analysis, the diffusion behavior of water flow in the first type of flow regime area is studied. The diffusion analysis involves the propagation mode of water flow in this area, such as vortex expansion, flow velocity change, etc. Numerical simulations (such as fluid mechanics models) or experimental data are used to capture the diffusion process of water flow, generating regional water flow regime diffusion data to describe information such as the flow pattern, diffusion direction, and velocity of water flow in the first type of flow regime area. Based on the regional water flow regime diffusion data, dynamic flow rate limitation of the valve orifice is performed. By dynamically adjusting the opening degree of the valve orifice, precise control of the water flow is achieved. For example, a flow rate threshold is set, and the flow rate is automatically adjusted according to the actual situation of the water flow to avoid excessive flow rate or uneven water flow. According to the water flow diffusion characteristics, flow splitting guide plates are installed at the valve orifice. These guide plates can guide the water flow to flow along a predetermined direction, optimize the water flow distribution, avoid too high or too low flow velocity, and obtain the first type of water flow control valve orifice data, including dynamic flow control parameters and the specific design of the flow splitting guide plates. For the second type of flow regime area, the initial setting of the valve orifice is also determined by analyzing the flow regime characteristics. Considering that the water flow characteristics of the second type of area are different from those of the first type (such as lower flow velocity, stronger turbulence, etc.), the design of the valve orifice should adapt to these different flow characteristics to obtain the second type of water flow control valve orifice setting data, including initial control parameters such as valve orifice position, size, and opening degree. Regional water flow regime aggregation analysis is carried out on the second type of flow regime area. This analysis focuses on studying the aggregation behavior of water flow in this area, such as the convergence area and turbulence area of the water flow. Water flow simulation tools are used to analyze the aggregation phenomenon. For example, by calculating the contraction area of the flow field, the aggregation area of the water flow is found, generating regional water flow regime aggregation data to depict the spatial distribution and intensity of the aggregation phenomenon. Based on the regional water flow regime aggregation data, dynamic compensation of the flow rate of the second type of water flow control valve orifice is performed. In areas with weak flow regime or low flow velocity, the flow rate is increased to enhance the water flow intensity and ensure the uniformity and stability of the water flow in the water area. In order to guide the water flow to flow along a predetermined path, flow regime guiding structures such as flow direction conduits and vortex inductors can be designed around the valve orifice. These structures help to optimize the flow mode of the water flow and avoid flow regime chaos or uneven flow velocity, obtaining the second type of water flow control valve orifice data, including flow rate compensation parameters and the design of the flow regime guiding structures. The first type of water flow control valve orifice data and the second type of water flow control valve orifice data are integrated to ensure coordinated water flow control throughout the white-water training water area. This integration process not only needs to consider the water flow rate but also the water flow relationship between different areas.The integrated data includes the final positions, opening degrees, flow control parameters, etc. of all water flow control valve ports, obtaining complete water flow control valve port data, which serves as the basis for subsequent water flow regulation and training processes.

[0109] As an example of the present invention, referring to Figure 3 as shown, in this example, step S3 includes:

[0110] Step S31: Obtain water area training vessel position information data;

[0111] Step S32: Import the water area training vessel position information data into the three-dimensional model of the rapids training water area for vessel travel position mapping, generating training vessel travel position mapping data;

[0112] Step S33: Analyze the training vessel travel position mapping data for the vessel travel mode, generating a training vessel travel mode, where the training vessel travel model includes a normal travel mode and an abnormal travel mode;

[0113] Step S34: Use the water flow control valve port data to perform vessel guidance control on the normal travel mode and the abnormal travel mode, generating water flow control guidance data.

[0114] In the embodiments of the present invention, the position information of the vessel is obtained in real time through sensors (such as GPS, inertial navigation system, radar, sonar, etc.) installed on the training vessel. These sensors can provide data such as the precise position, heading, speed, and acceleration of the vessel, ensuring comprehensive and accurate acquisition of the vessel's position information, and obtaining the position information data of the water area training vessel, including key data such as timestamp, longitude and latitude, vessel heading, and speed. The vessel position information data obtained in step S31 is imported into the three-dimensional spatial model of the rapids training water area. The three-dimensional spatial model provides a complete environmental background for the vessel's travel by including information such as the terrain, elevation, and obstacles of the water area. Based on the water area training vessel position information data, the actual travel position of the vessel is mapped in the three-dimensional model. This process combines the real-time position of the vessel and the spatial model data, displays the movement trajectory and position change of the vessel in the training water area, and generates the training vessel travel position mapping data, including information such as the path, heading, and moment position of the vessel in the three-dimensional spatial model. Based on the travel position mapping data of the training vessel, the travel behavior of the vessel in the training water area is analyzed. This analysis can identify the normal travel mode and abnormal travel mode of the vessel through algorithm models (such as machine learning, pattern recognition algorithms, etc.). Analyze the travel trajectory and behavior of the vessel under the condition of no obstacles and smooth water flow to identify the normal navigation mode of the vessel. Identify the abnormal travel behavior of the vessel, such as sharp turns, stagnation, deviation from the route, etc., which is usually related to obstacles, flow rate changes, or other external factors, and generate the training vessel travel mode, including the specific characteristics and data of the normal travel mode and abnormal travel mode. Using the water flow control valve data generated in step S24, water flow guidance is performed for the normal travel mode and abnormal travel mode. For the normal travel mode, use the water flow control valve to ensure that the vessel sails smoothly along the predetermined waterway; for the abnormal travel mode, by dynamically adjusting the valve opening flow rate and guiding structure, help the vessel adjust its heading or speed to avoid deviation caused by obstacles or water flow disturbances. Maintain stable water flow conditions to ensure smooth navigation of the vessel. In the abnormal travel mode, real-time water flow control is performed, including dynamically adjusting the flow rate, direction, etc. of the water flow, assisting the vessel to resume normal travel, and generating water flow control guidance data, including water flow control parameters, flow rate adjustment, guiding structure setting, etc. information for the normal and abnormal travel modes.

[0115] Preferably, step S34 includes the following steps:

[0116] Step S341: Based on the normal driving mode, use the training vessel's form position mapping data to confirm the area where the rapids training water area three-dimensional model is located, and obtain the vessel's location area data; match the vessel's location area data with the flow pattern segmentation area data of the training water area. If the area match is the first type of flow pattern area, then control the corresponding water flow control valve port data in the flow pattern area in the direction of the vessel's water flow to generate the water flow guidance data for the normal driving flow pattern area;

[0117] Step S342: If the area match is the second type of flow pattern area, then do not process the corresponding water flow control valve port data in the flow pattern area;

[0118] Step S343: Based on the abnormal driving mode, mark the abnormal area of the vessel's location area data to obtain the vessel's abnormal driving area marking data; according to the vessel's abnormal driving area marking data, perform reverse water flow control on the water flow control valve port data, so as to generate the water flow guidance data for the abnormal driving flow pattern area;

[0119] Step S344: Integrate the water flow guidance data for the normal driving flow pattern area and the water flow guidance data for the abnormal driving flow pattern area to generate the water flow control guidance data.

[0120] In the embodiments of the present invention, according to the normal driving mode, by training the position mapping data of the vessel, the position of the vessel is docked with the three-dimensional spatial model of the rapids training water area to determine the specific area where the vessel is currently located. Through the docking of the GPS coordinates of the vessel and the data of the three-dimensional spatial model, the precise position of the vessel in the water area is obtained. Using the three-dimensional space matching technology, the position of the vessel is mapped to the grid area of the training water area to confirm the area where the vessel is located, and the data of the area where the vessel is located is generated, including information such as the area number where the vessel is located and the water flow state in the area. The data of the area where the vessel is located is matched with the data of the water flow pattern segmentation area of the rapids training water area. According to the data of the flow pattern segmentation area, the current position of the vessel is compared with the flow pattern area of the water area to confirm the type of the flow pattern area where the vessel is located. Whether the area where the vessel is located belongs to the first type of flow pattern area is determined through a comparison algorithm. If the area where the vessel is located matches the first type of flow pattern area, it is a downstream area; if it matches the second type of flow pattern area, it is an upstream or adverse flow pattern area, and the matching result is obtained. If the area matches the first type of flow pattern area, the next step of processing is entered; otherwise, it jumps to step S342. When the area matches the first type of flow pattern area, it indicates that the vessel is sailing smoothly and is not affected by adverse flow patterns. Therefore, by adjusting the data of the water flow control valve port, the water flow direction can be ensured to be consistent with the vessel's heading. According to the driving direction of the vessel, the water flow direction of the corresponding flow pattern area is adjusted to ensure a smooth docking of the water flow and the vessel's driving direction, and to avoid situations where the flow rate is too fast or too slow. This process calculates the most suitable water flow direction control scheme and generates the water flow guiding data for the normal driving flow pattern area, including specific parameters such as flow rate and flow direction adjustment. If the area matching result is the second type of flow pattern area, it indicates that the water flow conditions in this area are not suitable for the normal driving of the vessel. At this time, no adjustment is made to the data of the water flow control valve port in the second type of flow pattern area. In such an area, the direction of the water flow is opposite to the vessel's heading, and the vessel will encounter vortexes or strong water flow interference. Therefore, no processing is performed on the data of the water flow control valve port to ensure that the system does not affect the vessel's driving under complex water flow conditions. Through an abnormal driving mode recognition algorithm (such as the judgment based on the vessel's sharp turns, speed changes, stagnation, etc.), the abnormal driving area of the vessel is marked. This mark helps to identify whether the vessel is in an upstream or adverse water area, and the marked data of the vessel's abnormal driving area is obtained, including information such as the type and position of the abnormal area. When the vessel is in the abnormal driving mode and enters the abnormal area, the water flow in this area needs to be controlled reversely. The reverse water flow control is to help the vessel correct its heading and avoid deviating from the track. According to the marked data of the vessel's abnormal area, the data of the water flow control valve port is automatically adjusted to reversely adjust the flow rate, flow direction, etc., so as to change the water flow direction and intensity and assist the vessel to pass through the upstream area smoothly. By adjusting the water flow parameters, the water flow guiding data for the abnormal area is generated to ensure that the vessel can pass through the upstream area and resume normal navigation.Integrate the water flow guidance data for the normal driving flow state area and the abnormal driving flow state area to form a complete set of water flow control and guidance strategies. Dynamically adjust the water flow control strategy according to the real-time driving mode, location area of the vessel, and the flow state matching result. The integrated data includes water flow guidance schemes for different flow state areas to ensure that the vessel can sail smoothly under various water flow conditions, generate complete water flow control and guidance data, including parameters such as water flow velocity, direction, and flow rate adjustment, provide real-time guidance and control for the vessel, and ensure the smooth completion of the training task.

[0121] Preferably, the reverse water flow control of the water flow control valve port data according to the marked data of the abnormal driving area of the vessel includes:

[0122] Conduct a spatial distribution analysis of the abnormal area based on the marked data of the abnormal driving area of the vessel to obtain the boundary data of the abnormal area; confirm the affected water flow control valve port area based on the boundary data of the abnormal area to obtain the affected valve port area data;

[0123] Conduct a reverse analysis of the direction and intensity of the water flow based on the affected valve port area data to obtain the reverse water flow control data; adjust the water flow direction of the water flow control valve port data according to the reverse water flow control data to obtain the water flow reverse adjustment data; optimize the flow state guidance of the marked data of the abnormal driving area of the vessel through the water flow reverse adjustment data, and finally obtain the water flow guidance data for the abnormal driving flow state area.

[0124] In the embodiments of the present invention, the abnormal area where the vessel is located is determined by the marked data of the abnormal driving mode of the vessel. Information such as the movement trajectory and speed change of the vessel in these areas is analyzed to conduct a spatial distribution analysis of the abnormal areas. Combining the vessel position and the abnormal driving mode, the abnormal areas are defined through a spatial analysis algorithm to identify the areas where the vessel deviates from the normal shipping lane, and the boundary data of the abnormal areas is generated. The boundary data includes the spatial position, size, and the influence range on the water flow control orifices of the abnormal area. This data is used to further determine which water flow control orifices are affected by the abnormal area. According to the boundary data of the abnormal area, it is analyzed which water flow control orifices are within the influence range of the abnormal area. The water area is divided through a spatial overlap algorithm to find the water flow control orifices that coincide with the boundary of the abnormal area, and these orifice areas are the key areas affected by the water flow interference in the abnormal area. Through a spatial allocation algorithm, the data of each affected water flow control orifice area is obtained to determine which orifices need to conduct reverse water flow control. For the affected water flow control orifice areas, a reverse analysis of the water flow is carried out. Through a hydrodynamic model and numerical simulation tools, the water flow direction and intensity that need to be controlled in this area are deduced in reverse. Appropriate flow equations, such as the Navier-Stokes equation, are used to simulate the change in the water flow direction to determine the adjustment force required for the reverse water flow. This step involves simulating the reverse control effects under different flow velocity and flow direction conditions. The water flow adjustment data obtained through the reverse analysis, including parameters such as the flow velocity that needs to be reduced and the adjusted flow direction, are used for the specific adjustment of the orifices. Based on the reverse water flow control data, the water flow direction of the corresponding water flow control orifices is adjusted. The opening and closing degree of the orifices is changed through an orifice adjustment device (such as an automatic control valve, a flow regulator, etc.) to ensure that the water flow direction is opposite to the navigation direction of the vessel, thereby forming a reverse flow effect. Precise control of the affected orifices is carried out to reverse the water flow adjustment, which involves the opening control, flow rate adjustment, and flow direction optimization of the orifices, generating water flow reverse adjustment data, including information such as the opening degree, flow rate change, and flow direction control of each orifice. These data provide a basis for subsequent guidance optimization. According to the water flow reverse adjustment data, the abnormal driving area of the vessel is optimized and guided. The effect of the water flow reverse adjustment is analyzed, and the driving path and water flow guidance direction of the vessel are optimized according to the marked data of the abnormal driving area of the vessel. Combining with the fluid state guidance optimization data, the driving mode of the vessel is corrected so that it can pass through the reverse flow area more smoothly and resume the normal driving mode. The finally generated water flow guidance data, including parameters such as the water flow direction and flow velocity, optimizes the driving effect of the vessel in the abnormal area to ensure the safety and smoothness of the vessel.

[0125] Preferably, step S4 includes the following steps:

[0126] Step S41: Conduct a water flow control feedback evaluation on the water flow control guidance data to generate water flow control feedback evaluation data;

[0127] Step S42: Extract the water flow control feature from the water flow control feedback evaluation data to obtain the water flow control feature data; divide the water flow control feature data into data sets to generate a model training set and a model test set; use the convolutional neural network algorithm to train the model training set to generate a water flow control pre-model;

[0128] Step S43: Optimize and iterate the water flow control pre-model through the model test set to generate a water flow control model.

[0129] In an embodiment of the present invention, the generated water flow control guidance data is evaluated by using a feedback evaluation method. The goal of the evaluation is to confirm the effect of the water flow control measures, including whether the water flow velocity, flow direction, and ship path guidance are reasonable. The feedback evaluation includes aspects such as water flow stability, ship driving efficiency, and flow velocity change. The guidance effect can be quantitatively analyzed using a fluid mechanics model and ship motion trajectory data. The evaluation results generate water flow control feedback evaluation data through an algorithm, including information such as flow state evaluation results and control errors. Key water flow control feature data are extracted from the water flow control feedback evaluation data, which include key factors affecting the water flow control effect such as water flow velocity, flow direction change, vortex behavior, and fluctuation. The feedback evaluation data is preprocessed to remove noise, standardize the data, and extract feature information for model training. The key features in the data are extracted by signal processing, time series analysis, and other methods to ensure that the extracted data can reflect the main influencing factors of water flow control. The extracted water flow control feature data is divided into a model training set and a test set. The training set is used to train the model, and the test set is used to verify the model performance. The data set is usually divided into 7:3 or 8:2 ratios to ensure that the training set contains enough samples for learning, while the test set is representative to test the model effect. The data in the training set and the test set are normalized to ensure that the data distribution of the two is consistent, so as to ensure the effectiveness of model training and evaluation. The training set is trained using the convolutional neural network (CNN) algorithm. CNN is a deep learning algorithm that is particularly suitable for processing image data and spatial data, and can efficiently extract spatial features from the data. The network structure of CNN is set, including convolutional layers, pooling layers, and fully connected layers. During the training process, the model optimizes parameters through back propagation and gradient descent, and adjusts the network weights so that it can better fit the training data. After training, a water flow control pre-model is generated, which can predict or infer the optimal control scheme under different water flow conditions. The pre-model is tested and evaluated using the model test set. The prediction effect of the model is evaluated by calculating indicators such as the accuracy of the model and the loss function. If the model effect is not good, it can be further optimized. According to the evaluation results on the test set, the model structure is optimized or the hyperparameters, such as the learning rate and the number of convolutional layers, are adjusted, and the model is optimized through multiple iterations. After multiple rounds of optimization and iteration, the final water flow control model is generated. This model can accurately predict the water flow control strategy based on the water flow control characteristic data, and improve the accuracy and stability of water flow guidance.

[0130] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0131] The above are only specific embodiments of the present invention, enabling those skilled in the art to understand or implement the present invention. Various modifications to these embodiments will be obvious to those skilled in the art, and the general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to these embodiments shown herein, but rather will be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for constructing a water flow control model for rapids training waters, characterized in that: The following steps are involved: Step S1: Acquire rapid current training water area environment perception data; perform spatial three-dimensional modeling on the rapid current training water area environment perception data to generate a rapid current training water area spatial three-dimensional model; Conduct water flow dynamics simulation on the three-dimensional model of the rapids training water area to generate water flow simulation data for the rapids training water area; Step S2: obtaining actual hydrological data of the rapids training water area; performing simulated water flow data correction on the water flow simulation data of the rapids training water area by using the actual hydrological data of the rapids training water area to generate simulated water flow correction data; performing regional water flow pattern adaptation analysis on the spatial three-dimensional model of the rapids training water area based on the simulated water flow correction data to generate a first type of flow pattern area and a second type of flow pattern area; performing water flow control valve port setting on the first type of flow pattern area and the second type of flow pattern area to obtain water flow control valve port data; Step S3: obtaining the location information data of the training vessel in the water area; importing the location information data of the training vessel in the water area into the three-dimensional model of the rapids training water area to analyze the vessel driving mode, and generating a training vessel driving mode, wherein the training vessel driving model includes a normal driving mode and an abnormal driving mode; Using the water flow control valve port data to guide the ship in the normal driving mode and the abnormal driving mode, and generating water flow control guidance data; Step S4: extracting water flow control features from the water flow control guide data to obtain water flow control feature data; constructing a model for the water flow control feature data using a convolutional neural network algorithm to generate a water flow control model.

2. The method for constructing a water flow control model for rapids training water area according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: performing water environment perception on the rapids training water area to obtain rapids training water area environment perception data; Step S12: preprocessing the rapids training water environment perception data to generate standard water environment perception data, wherein the data preprocessing includes data cleaning, data denoising, missing value filling and data standardization; Step S13: extracting terrain height, waterway shape and obstacle distribution data from the standard water environment perception data, and performing spatial three-dimensional modeling based on the terrain height, waterway shape and obstacle distribution data to generate a spatial three-dimensional model of the rapids training water area; Step S14: extracting the deployment positions of water outlets in the three-dimensional model of the rapids training water area space to obtain the water outlet deployment position data; performing water flow dynamics simulation on the three-dimensional model of the rapids training water area space based on the water outlet deployment position data to generate water flow simulation data for the rapids training water area.

3. The method for constructing a water flow control model for rapids training water area according to claim 2, characterized in that: Spatial 3D modeling based on terrain height, waterway shape, and obstacle distribution data includes: Construct terrain grid data according to terrain height, waterway shape and obstacle distribution data to obtain terrain grid data; extract water area streamline analysis data from terrain grid data to obtain water area streamline analysis data; Based on the water area streamline analysis data, a water area profile is constructed to obtain water area profile data; an obstacle refinement model is constructed for the terrain grid data through the water area profile data to obtain obstacle refinement model data; The water area streamline analysis data is used to generate the water area three-dimensional surface for the obstacle refinement model data, thereby obtaining the water area three-dimensional surface data; the water area three-dimensional surface data is mapped into a spatial three-dimensional model, thereby obtaining the rapids training water area spatial three-dimensional model.

4. The method for constructing a water flow control model for rapids training water area according to claim 1, characterized in that: Step S2 includes the following steps: Step S21: using sensors to obtain actual hydrological data of rapids training waters; Step S22: performing simulated water flow data correction on the water flow simulation data of the rapids training water area by using the actual hydrological data of the rapids training water area to generate simulated water flow correction data; performing water flow pattern adaptation area segmentation on the three-dimensional model of the rapids training water area space based on the simulated water flow correction data to generate water flow pattern segmentation area data of the training water area; Step S23: performing regional water flow pattern adaptation quantity detection on the water flow pattern segmentation area data of the training water area, when the regional water flow pattern adaptation quantity is greater than or equal to the preset standard regional water flow pattern adaptation quantity, the corresponding training water area water flow pattern segmentation area data is marked as the first type of flow pattern area; Step S24: When the number of regional water flow pattern adaptations is less than the preset standard regional water flow pattern adaptation number, the corresponding training water area water flow pattern segmentation area data is marked as a second-class flow pattern area; the water flow control valve ports are set for the first-class flow pattern areas and the second-class flow pattern areas, thereby obtaining the water flow control valve port data.

5. The method for constructing a water flow control model for rapids training water area according to claim 4, characterized in that: Based on the simulated water flow correction data, the water flow pattern adaptation area segmentation of the three-dimensional model of the rapids training water area space includes: Confirm the spatial boundary of the three-dimensional model of the rapids training water area to obtain the spatial boundary data of the training water area; perform water area terrain elevation analysis on the three-dimensional model of the rapids training water area based on the spatial boundary data of the training water area to generate water area terrain elevation data; The spatial three-dimensional model of the rapids training water area is divided into spatial regions by using the training water area spatial boundary data and the water area terrain elevation data, and the divided spatial regions are gridded to obtain the gridded regional data of the training water area; the hydrological characteristics are extracted from the simulated water flow correction data to obtain the simulated water flow characteristic data, wherein the hydrological characteristics extraction includes flow velocity extraction, flow direction extraction and eddy current extraction; The flow velocity characteristics and flow direction characteristics in the simulated water flow characteristic data are used to analyze the water flow pattern distribution of the gridded regional data of the training water area, and generate regional water flow pattern distribution data; based on the eddy current characteristics in the simulated water flow characteristic data, the regional water flow pattern distribution data is analyzed for flow pattern transition, and regional water flow pattern transition data is generated; According to the regional water flow pattern distribution data and the regional water flow pattern transition data, the grid unit clustering and adaptation of the gridded regional data of the training water area are performed to generate the water flow pattern segmentation regional data of the training water area.

6. The method for constructing a water flow control model for rapids training water area according to claim 4, characterized in that: The water flow control valve port setting for the first type of flow pattern area and the second type of flow pattern area includes: Performing initial water flow control valve port settings on the first type of flow pattern area to obtain first type of water flow control valve port setting data; performing regional water flow pattern diffusion analysis on the first type of flow pattern area to generate regional water flow pattern diffusion data; performing valve port dynamic flow restriction on the first type of water flow control valve port setting data based on the regional water flow pattern diffusion data and setting a diversion guide plate, thereby obtaining first type of water flow control valve port data; Performing initial water flow control valve port settings on the second type of flow pattern area to obtain second type of water flow control valve port setting data; performing regional water flow pattern aggregation analysis on the second type of flow pattern area to generate regional water flow pattern aggregation data; performing valve port dynamic flow compensation on the second type of water flow control valve port setting data based on the regional water flow pattern aggregation data and setting a flow pattern guiding structure to obtain second type of water flow control valve port data; The first type of water flow control valve port data and the second type of water flow control valve port data are integrated to obtain the water flow control valve port data.

7. The method for constructing a water flow control model for rapids training water area according to claim 1, characterized in that: Step S3 includes the following steps: Step S31: Acquire the location information data of the training vessel in the water area; Step S32: importing the position information data of the training vessel in the water area into the three-dimensional model of the rapids training water area to map the vessel's driving position, thereby generating training vessel driving position mapping data; Step S33: performing a ship driving mode analysis on the training ship driving position mapping data to generate a training ship driving mode, wherein the training ship driving model includes a normal driving mode and an abnormal driving mode; Step S34: Use the water flow control valve port data to perform ship guidance control on the normal driving mode and the abnormal driving mode to generate water flow control guidance data.

8. The method for constructing a water flow control model for rapids training water area according to claim 7, characterized in that: Step S34 includes the following steps: Step S341: Based on the normal driving mode, the training ship form position mapping data is used to confirm the area where the rapids training water area space three-dimensional model is located, and the ship area data is obtained; the ship area data and the training water area water flow pattern segmentation area data are regionally matched, and if the area is matched to the first type of flow pattern area, the corresponding water flow control valve port data in the flow pattern area is controlled along the ship flow direction to generate the normal driving flow pattern area water flow guidance data; Step S342: if the area is matched to be the second type of flow pattern area, the corresponding water flow control valve port data in the flow pattern area is not processed; Step S343: marking the area where the vessel is located based on the abnormal driving mode to obtain abnormal driving area marking data of the vessel; performing reverse water flow control on the water flow control valve port data according to the abnormal driving area marking data of the vessel, thereby generating water flow guidance data of the abnormal driving flow pattern area; Step S344: Integrate the water flow guidance data of the normal driving flow pattern area and the water flow guidance data of the abnormal driving flow pattern area to generate water flow control guidance data.

9. The method for constructing a water flow control model for rapids training water area according to claim 8, characterized in that: The reverse flow control of the water flow control valve port data based on the abnormal ship driving area mark data includes: According to the marking data of the abnormal driving area of ​​the ship, the spatial distribution analysis of the abnormal area is performed to obtain the boundary data of the abnormal area; according to the boundary data of the abnormal area, the affected water flow control valve port area is confirmed for the water flow control valve port data, so as to obtain the affected valve port area data; Based on the data of the affected valve area, the direction and intensity of the water flow are reversely analyzed to obtain the countercurrent control data; the water flow direction of the water flow control valve data is adjusted according to the countercurrent control data to obtain the reverse flow adjustment data; the flow pattern guidance of the abnormal driving area marking data of the ship is optimized through the reverse flow adjustment data, and finally the water flow guidance data of the abnormal driving flow pattern area is obtained.

10. The method for constructing a water flow control model for rapids training water area according to claim 1, characterized in that: Step S4 includes the following steps: Step S41: performing water flow control feedback evaluation on the water flow control guidance data to generate water flow control feedback evaluation data; Step S42: extracting water flow control features from the water flow control feedback evaluation data to obtain water flow control feature data; dividing the water flow control feature data into data sets to generate a model training set and a model test set; and performing model training on the model training set using a convolutional neural network algorithm to generate a water flow control pre-model; Step S43: Performing model optimization iteration on the water flow control pre-model through the model test set, thereby generating a water flow control model.

Images