A method and system for detecting a water flow structure of an indoor sink sand wave surface

Abstract

The application relates to the technical field of water flow structure detection, and discloses a detection method and system for the water flow structure of an indoor water tank sand wave surface, which comprises the following steps: arranging a sand wave surface and a test sand bed, arranging a camera system and a flow velocity collection probe, collecting image data for timestamp alignment, performing image recognition to extract sand wave contours and calculate sand wave state parameters, performing wavelet denoising and reconstruction on the flow velocity sequence of measuring points, performing frequency domain analysis to obtain energy spectrum distribution and calculate energy indexes, determining a backflow candidate area and a boundary position, performing energy criterion verification on the energy spectrum distribution and the energy indexes corresponding to the flow velocity sequence of the measuring points in the backflow candidate area, merging the measuring point grids that pass the verification according to connectivity to generate a water flow structure partition map and determine water flow structure characteristics, and establishing a response relationship model based on the water flow structure characteristics and the sand wave state parameters. The application improves the recognition robustness of turbulent and vortex structures and improves the accuracy of sand wave surface water flow structure detection.

Description

Technical Field

[0001] This invention relates to the field of water flow structure detection technology, and more specifically, to a method and system for detecting the water flow structure of an indoor water tank with a sand-like surface. Background Technology

[0002] In studies of riverbed evolution and sediment transport mechanisms, engineering hydraulic simulations, and indoor physical model experiments, the sand wave surface is a typical movable bed boundary. Changes in the morphology of its upstream and downstream slopes induce complex flow structures such as separation zones, backflow zones, and vortex structures, thereby affecting local shear, momentum exchange, and sediment transport flux. To obtain repeatable and quantifiable experimental conclusions, it is usually necessary to conduct detailed detection of sand wave motion morphology and near-bed flow structure in indoor flume observation sections. Furthermore, morphological evolution, tracer visualization, and velocity information at measurement points should be correlated on a spatiotemporal scale to support modeling and analysis of flow structure zoning, key feature localization, and the response relationship between morphology and structure.

[0003] In existing technologies, the focus is often on a single observation link or only one aspect of in-situ morphological observation or synchronous acquisition of flow field visualization: for example, patent document CN109115273B discloses an experimental system that uses visualization methods and synchronizes flow field information through a synchronizer; patent document CN115615659A discloses an image recognition and calibration scheme for dynamic in-situ observation of bank profiles or bedbed in flume tests; and patent document CN107290129A discloses a hydrological observation scheme that uses tracing and camera arrays to invert flow field and other parameters. However, in the above schemes, the determination of the backflow boundary by tracer visualization is easily affected by tracer diffusion, illumination and viewing angle, and there is no candidate region screening mechanism that verifies the frequency domain energy characteristics of the flow velocity at the measurement point, resulting in insufficient stability of key boundaries such as separation zone and reattachment point; although point flow velocity measurement can reflect the distribution of local disturbance and eddy energy, it lacks the rules for coupling with video measurement results into regions, making it difficult to generate reproducible water flow structure partition maps; there is no feasible response relationship modeling link between morphological parameters and water flow structure characteristics, making it difficult to support comparison and prediction under different working conditions.

[0004] Therefore, it is necessary to design a detection method and system for the water flow structure of an indoor water tank with a sand-like surface to solve the problems existing in the current technology. Summary of the Invention

[0005] In view of this, the present invention proposes a method and system for detecting the water flow structure on the sand-wave surface of an indoor water tank, aiming to solve the problems that the water flow structure boundary is easily affected by tracer diffusion and measurement noise under the working condition of an indoor water tank, making it difficult to stably partition and locate features, and that the morphological parameters and water flow structure features lack reproducible correlation modeling.

[0006] This invention proposes a method for detecting the water flow structure of an indoor water tank with a sand-like surface, comprising: A sand wave surface and a test sand bed were set up in the indoor water tank observation section. A camera system was set up and the camera system was spatially calibrated. A flow velocity acquisition probe was installed at the preset measuring point grid in the sand wave surface. Collect sand wave motion image sequences, tracer flow field videos, and flow velocity sequences corresponding to the measuring point grid, and align the timestamps of each collected data stream; Image recognition is performed on the sand wave motion image sequence to extract the sand wave contour and calculate the sand wave state parameters, which include morphological parameters and motion parameters. The flow velocity sequence at the measurement points is subjected to wavelet denoising and reconstruction, and frequency domain analysis is performed to obtain the energy spectrum distribution and calculate the energy index. Based on the tracer flow field video, the candidate regions and boundary locations of the return flow are determined. The energy criterion is verified by the energy spectrum distribution and energy index corresponding to the flow velocity sequence of the measurement points in the candidate regions of the return flow. The measurement point grids that pass the verification are merged according to connectivity to generate a water flow structure partition map and determine the water flow structure characteristics. A response relationship model is established based on the water flow structure characteristics and sand wave state parameters.

[0007] Furthermore, when determining the candidate reflow region and boundary location based on the tracer flow field video, the process includes: Based on the tracer flow field video, the displacement of the tracer trajectory is calculated frame by frame and the mainstream direction is determined. Spatial units that present tracer trajectories opposite to the mainstream direction for no less than a preset number of consecutive frames are determined as the reflow candidate region, and the outer edge of the reflow candidate region is determined as the boundary position.

[0008] Furthermore, the spatial unit is determined by the spatial calibration result of the camera system, so that the image area of ​​the tracer flow field video corresponds one-to-one with the measurement point grid; based on the spatial unit, the displacement of the tracer trajectory is statistically assigned, and the candidate backflow area and the boundary position are determined.

[0009] Furthermore, when performing energy criterion verification, the following are included: The recirculation candidate region is mapped to the measurement point grid. The low-frequency energy ratio, the main frequency peak value, and the energy decay index are extracted from the energy spectrum distribution corresponding to the flow velocity sequence of the measurement point in the recirculation candidate region as energy indicators. When the low-frequency energy ratio is not less than the preset ratio threshold and the energy decay index is not greater than the preset decay threshold, the corresponding measurement point grid is determined to pass the energy criterion verification.

[0010] Furthermore, the energy spectrum distribution is divided into low-frequency band and non-low-frequency band according to a preset frequency band, the low-frequency energy ratio is the proportion of low-frequency band energy to the total frequency band energy, the main frequency peak is the peak value at the point of maximum amplitude in the energy spectrum distribution, and the energy attenuation index is determined by the attenuation trend of the high-frequency band of the energy spectrum distribution.

[0011] Furthermore, when generating a flow structure zoning map by merging the measurement point grid verified by the energy criterion according to the four-neighbor connectivity, the following is included: The outer edge of the water-facing side of the connected region is determined as the boundary of the separation zone, and the outer edge of the backwater side of the connected region is determined as the reattachment point. A sample set is constructed using the sand wave state parameters and the water flow structure features. A response relationship model is obtained by fitting the model, and the prediction results of the water flow structure features are output.

[0012] Furthermore, the preset proportion threshold and the preset attenuation threshold are determined by the energy index of the flow velocity sequence of the measurement point corresponding to the reference area outside the backflow candidate area; the response relationship model uses the sand wave state parameters aligned with the timestamp and the water flow structure features to form a sample set, and uses regression fitting to obtain the model used to output the prediction results of the water flow structure features.

[0013] Furthermore, the morphological parameters include any one or more of wavelength, wave height, and wave angle, and the motion parameters include wave speed.

[0014] Furthermore, the sampling frequency of the flow velocity acquisition probe is not less than 20Hz, and a grid of measurement points is set up at the crests, troughs, upstream slopes, downstream slopes, and separation zones; the tracer is uniformly released through the outlet pre-embedded on the upstream side of the sand wave, and multi-channel independent control is adopted, with the tracer flow rate adjustable from 0.1 to 2 ml / s.

[0015] Compared with existing technologies, the advantages of this invention are as follows: Based on the spatial calibration and timestamp alignment of the camera system, the sand wave motion image sequence, tracer flow field video, and measuring point velocity sequence are unified into the same detection link. This not only allows the extraction of morphological and motion parameters from the sand wave contour to form boundary conditions, but also enables wavelet denoising and frequency domain energy spectrum analysis of the measuring point velocity sequence to obtain energy indices, thereby improving the robustness of identifying non-stationary features such as turbulence and vortex structures. Furthermore, the tracer flow field video provides candidate recirculation regions and boundary locations, and the energy spectrum distribution and energy indices of the measuring point velocities within the candidate recirculation regions are then used for further analysis. The system performs energy criterion verification and merges the data according to connectivity to generate a water flow structure partition map. This allows the location of key features such as separation zones, reattachment points, backflow ranges, and vortex structure evolution to no longer rely on single visual judgments or single-point measurement inferences. It can effectively suppress boundary drift and misjudgment caused by tracer diffusion, illumination changes, and local measurement noise. After obtaining stable partitions and structural features, a response relationship model between water flow structure features and sand wave state parameters is further established. This elevates the detection results from phenomenon descriptions to quantifiable correlations and predictable outputs, thereby providing a consistent and verifiable data foundation and model for comparative analysis under different working conditions.

[0016] On the other hand, this application also provides a detection system for the water flow structure of an indoor water tank with a sand-like surface, for applying the above-mentioned detection method for the water flow structure of an indoor water tank with a sand-like surface, including: An indoor water tank with an observation section, wherein a test sand bed is set up in the observation section and a sand wave surface is formed; The camera system is configured to acquire a sequence of sand wave motion images and a video of the tracer flow field, and output calibration data for spatial calibration of the sand wave motion image sequence and the video of the tracer flow field; The flow velocity acquisition unit includes flow velocity acquisition probes arranged at preset measurement point grids in the sand wave surface, used to acquire the flow velocity sequence at the measurement points; The synchronous acquisition control unit is used to synchronously trigger the acquisition of the camera system and the flow velocity acquisition unit, and to align the timestamps of the sand wave motion image sequence, the tracer flow field video and the measurement point flow velocity sequence. The image processing unit is used to perform image recognition on the sand wave motion image sequence to extract the sand wave contour and calculate the sand wave state parameters; The velocity analysis unit is used to perform wavelet denoising and reconstruction on the velocity sequence of the measurement point, and to perform frequency domain analysis to obtain the energy spectrum distribution and calculate the energy index. The structural partitioning and modeling unit is used to determine the candidate region and boundary position of the return flow based on the tracer flow field video, and to verify the energy criterion by using the energy spectrum distribution and energy index corresponding to the flow velocity sequence of the measurement points in the candidate region of the return flow. The measured point grids that pass the verification are merged according to connectivity to generate a water flow structure partitioning map and determine the water flow structure characteristics. A response relationship model is established based on the water flow structure characteristics and the sand wave state parameters.

[0017] It is understandable that the above-mentioned detection methods and systems for indoor water tanks with wavy surface water flow structures have the same beneficial effects, and will not be elaborated further here. Attached Figure Description

[0018] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention. Furthermore, the same reference numerals denote the same parts throughout the drawings. In the drawings: Figure 1 A flowchart of a method for detecting the water flow structure of an indoor water tank with a sand-like surface, provided in an embodiment of the present invention; Figure 2 This is a functional block diagram of a detection system for an indoor water tank with a sand-wave surface water flow structure, provided in an embodiment of the present invention. Figure 3 This is a schematic diagram of the arrangement of sand wave surface velocity measuring points and the water flow structure zoning provided in an embodiment of the present invention. Detailed Implementation

[0019] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the disclosure to those skilled in the art. It should be noted that, unless otherwise specified, embodiments and features in the embodiments of the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0020] In some embodiments of this application, see Figure 1 As shown, this application proposes a method for detecting the water flow structure of an indoor water tank with a sand-like surface, comprising: S100: Set up a sand wave surface and a test sand bed in the indoor water tank observation section, set up a camera system and perform spatial calibration of the camera system, and install flow velocity acquisition probes at the preset measurement point grid in the sand wave surface.

[0021] S200: Acquires sand wave motion image sequences, tracer flow field videos, and flow velocity sequences corresponding to the measurement point grid, and aligns the timestamps of each acquired data stream.

[0022] S300: Performs image recognition on the sequence of sand wave motion images, extracts the sand wave contours and calculates the sand wave state parameters, which include morphological parameters and motion parameters.

[0023] S400: Perform wavelet denoising and reconstruction on the flow velocity sequence at the measurement point, and perform frequency domain analysis to obtain the energy spectrum distribution and calculate the energy index.

[0024] S500: Based on the tracer flow field video, the candidate region and boundary position of the recirculation are determined. The energy criterion is verified by the energy spectrum distribution and energy index corresponding to the flow velocity sequence of the measurement points in the candidate region. The measurement point grids that pass the verification are merged according to connectivity to generate a water flow structure partition map and determine the water flow structure characteristics. A response relationship model is established based on the water flow structure characteristics and sand wave state parameters.

[0025] Specifically, the measuring point grid is a spatial arrangement of multiple measuring points pre-set along the sand wave surface, and each grid cell in the measuring point grid corresponds to one measuring point or a group of measuring points; the spatial cell is an image region in the tracer flow field video determined by the spatial calibration, and corresponds one-to-one with the measuring point grid; the backflow candidate region is a set of spatial cells that present tracer trajectories opposite to the mainstream direction for no less than a preset number of consecutive frames within the spatial cell; the reference region is a spatial region located in the mainstream stable region and does not contain the backflow candidate region, used to determine the mainstream direction or to determine a preset proportion threshold and a preset attenuation threshold; the water flow structure features include at least one or more of the following: separation zone boundary, reattachment point location, backflow range, or vortex street evolution.

[0026] Specifically, in this embodiment, the indoor water tank can be a rectangular tank with a total length of 25m, a width of 1m, and a depth of 0.8m, with an adjustable bottom slope range of 0–1%. The observation section is set as a high-transmittance glass structure, and the experimental sand bed can be laid with non-uniform plastic sand with a median particle size of 0.75mm. After forming a sand wave surface in the observation section, a sand wave sequence is created and fixed based on the research objective. For example, a sand wave sequence with a wave height of 3.6cm, an upstream wavelength of 20cm, and a downstream wavelength of 4cm can be used as the boundary for repeatable experiments. Subsequently, a grid of measuring points is set up on the sand wave surface, for example, 7 vertical lines with 5 measuring points on each vertical line. A flow velocity acquisition probe is installed at the measuring point grid. The flow velocity acquisition probe can be installed on a three-dimensional moving mechanism to achieve grid-based positioning, and the sampling frequency can be 20Hz. At the same time, a camera system is set up. The system includes a high-speed camera for acquiring tracer flow field video and a top-down camera for acquiring sand wave motion image sequences. A 10cm×10cm grid calibration plate is placed in the shooting area to complete spatial calibration, for example, to obtain a pixel calibration result of 1 pixel corresponding to 0.12mm. Under the target working conditions, the flow rate can be set to 15.6L / s and the water depth to 12.7cm. After the flow field stabilizes, the velocity acquisition probe continuously acquires a 180-second flow velocity sequence at the measuring point. Simultaneously, the tracer injection and the high-speed camera acquire and record a 60-second tracer flow field video are triggered. The top-down camera takes sand wave motion image sequences at set intervals (for example, one image every 10 seconds). The control console synchronizes and timestamps the above data to ensure the repeatability and consistency of subsequent fusion analysis.

[0027] Specifically, in the data processing stage, image recognition is first performed on the sand wave motion image sequence to extract the sand wave contour and form sand wave state parameters. Among them, the morphological parameters can be calculated from the spatial geometric relationship of the continuous contour, and the motion parameters can be calculated from the displacement and time interval of the sand wave contour at adjacent time points, such as obtaining the temporal changes of wave velocity and wave height. At the same time, wavelet denoising and reconstruction are first performed on the flow velocity sequence of the measuring points to reduce the impact of acquisition noise on the frequency domain criteria. After denoising, statistical quantities such as time-averaged flow velocity, pulsation intensity, and turbulence intensity are extracted to characterize the spatial distribution. Furthermore, multi-scale decomposition can be performed on the flow velocity sequence of the measuring points in the high-turbulence region near the wave trough to identify eddy structure components at different scales. Then, fast Fourier transform is performed to obtain the energy spectrum distribution. By fitting the attenuation trend of the high-frequency band, the energy attenuation index is calculated, and the cumulative energy distribution can be plotted simultaneously to characterize the energy accumulation characteristics in the frequency band. In the structural zoning and modeling stage, the spatial distribution of streamlines, separation vortices, and recirculation zones is first identified based on tracer flow field videos to determine candidate recirculation regions and boundary locations, and these candidate regions are mapped to a measuring point grid. Subsequently, energy criteria are validated using the energy spectrum distribution and energy indices corresponding to the velocity sequences at measuring points within the candidate recirculation regions. For example, the proportion of low-frequency energy can be used as one of the key energy indicators to eliminate false recirculation candidate regions caused by tracer diffusion, illumination changes, or instantaneous disturbances. The validated measuring point grids are then merged according to connectivity to generate a flow structure zoning map. Based on the zoning map, flow structure features such as separation zones, reattachment points, recirculation ranges, recirculation vortex centers, and vortex street evolution are further identified, and vortex street frequencies and other characterizing quantities can be extracted. Finally, a sample set is constructed by combining flow structure features with sand wave state parameters to establish a response relationship model, such as establishing the relationship between separation zone length and wave height, to achieve comparative analysis and predictive output of flow structure features under different operating conditions.

[0028] Understandably, by forming a dual-constraint mechanism by determining the candidate region of the tracer flow field video and verifying the frequency domain energy criterion of the flow velocity at the measurement point, the extraction of the backflow boundary and key structural features no longer relies on a single visual judgment or a single measurement point fluctuation. This improves the stability and reproducibility of the identification results of the separation zone, reattachment point, backflow vortex core, and vortex street frequency. At the same time, based on wavelet multi-scale decomposition and indicators such as energy spectrum attenuation index and cumulative energy distribution, the multi-scale structure and energy dissipation characteristics of turbulence can be quantitatively characterized, and a groundable response relationship can be established with the sand wave state parameters. This provides a repeatable experimental data link and model output for revealing the evolution of the water flow structure and energy transfer law on the sand wave surface.

[0029] In some embodiments of this application, when determining the reflow candidate region and boundary position based on the tracer flow field video, the method includes: calculating the tracer trajectory displacement frame by frame based on the tracer flow field video and determining the mainstream direction, determining the spatial unit that presents a tracer trajectory opposite to the mainstream direction for no less than a preset number of consecutive frames as the reflow candidate region, and determining the outer edge of the reflow candidate region as the boundary position.

[0030] In some embodiments of this application, the spatial units are determined by the spatial calibration results of the camera system, so that the image area of ​​the tracer flow field video corresponds one-to-one with the measurement point grid; the displacement of the tracer trajectory is statistically assigned based on the spatial units, and the candidate return flow area and boundary position are determined.

[0031] Specifically, the determination of the candidate backflow region and boundary location is based on the tracer flow field video. The tracer is released uniformly through a microtube injection system embedded in the upstream surface of the sand wave. The tracer flow field is recorded by a high-speed camera system in conjunction with underwater illumination. For example, the tracer can be a 0.05% potassium permanganate solution, propelled uniformly by a stepper motor-driven injection pump at a flow rate of 0.5 ml / s. The high-speed camera operates at a frame rate of 500 fps to obtain continuous and clear images of streamlines, separation eddies, and backflow areas. The tracer trajectory is extracted and its displacement is calculated on a per-frame basis, using adjacent frames of the tracer flow field video. Specifically, adjacent frames are segmented by difference and thresholding to obtain the connected regions of the tracer trajectory. The centerline or centroid of the connected region is extracted as the tracer trajectory position, and the displacement vector of the corresponding position in adjacent frames is taken as the tracer trajectory displacement. Multiple tracer trajectory displacement vectors are statistically analyzed within the upstream reference region of the sand wave, and the dominant direction is taken as the mainstream direction. Subsequently, the displacement vector of the tracer trajectory is assigned to each spatial unit. If the spatial unit shows a tracer trajectory displacement opposite to the mainstream direction within a continuous period of no less than a preset number of frames, the spatial unit is determined to be a backflow candidate region. Based on this, the spatial units in the backflow candidate region are aggregated according to connectivity, and the outer edge of the aggregated region is taken as the boundary position, which is used to connect with the energy criterion verification link of the flow velocity sequence of the measurement point.

[0032] Specifically, in some embodiments of this application, the spatial units are determined by the spatial calibration results of the camera system to achieve a one-to-one correspondence between the image area of ​​the tracer flow field video and the measurement point grid. Specifically, a 10cm × 10cm grid calibration plate can be placed in the shooting area to complete pixel calibration, for example, obtaining a conversion relationship of 1 pixel to 0.12mm, and establishing a correspondence between the image plane coordinates and the actual size coordinates accordingly. The measurement point grid can be arranged as follows, for example, setting 7 vertical lines with 5 measurement points on each vertical line, and the flow velocity acquisition probe is installed on a three-dimensional electric slide to achieve grid-based positioning; after calibration, the projection range of each measurement point grid under the actual size coordinates is mapped to the image plane coordinates, and the spatial unit is defined as "the image projection area corresponding to each measurement point grid," thereby ensuring that the statistical attribution of the tracer trajectory displacement and the measurement point flow velocity sequence have a consistent spatial correspondence. Therefore, when determining the candidate recirculation area, the set of measurement point grids corresponding to the candidate recirculation area can be directly obtained, providing spatial constraints for energy criterion verification and structural partitioning map generation.

[0033] Understandably, by transforming the reflow determination in the tracer video from a purely visual result into a gridded result that can be directly used for verification, the reflow candidate region and boundary position have a clear spatial scale and positioning basis. By using frame-by-frame displacement statistics to determine the mainstream direction and using the reverse displacement of consecutive frames as the reflow candidate criterion, the instantaneous pseudo-reflow caused by tracer diffusion and local illumination fluctuations can be effectively suppressed, thereby improving the stability and repeatability of the reflow boundary.

[0034] In some embodiments of this application, the energy criterion verification includes: mapping the recirculation candidate region to the measurement point grid, extracting the low-frequency energy ratio, the main frequency peak and the energy decay index as energy indicators from the energy spectrum distribution corresponding to the flow velocity sequence of the measurement points in the recirculation candidate region, and determining the corresponding measurement point grid as passing the energy criterion verification when the low-frequency energy ratio is not less than a preset ratio threshold and the energy decay index is not greater than a preset decay threshold.

[0035] In some embodiments of this application, the energy spectrum distribution is divided into low-frequency and non-low-frequency bands according to a preset frequency band. The low-frequency energy proportion is the ratio of low-frequency energy to the total energy of the entire frequency band, and the dominant frequency peak is the peak value at the point of maximum amplitude in the energy spectrum distribution. The energy decay index is determined by the high-frequency band decay trend of the energy spectrum distribution. The energy decay index is a decay rate characterization quantity obtained by fitting the amplitude variation trend of the high-frequency band energy spectrum curve with frequency.

[0036] The energy attenuation index is determined as follows: Using the non-low frequency band as the high frequency band, frequency points with zero amplitude and isolated peak frequencies formed by instantaneous acquisition spikes are removed from the high frequency band. The logarithms of the frequency values ​​and energy spectrum amplitudes of the remaining frequency points are then taken, and linear fitting is performed on a double logarithmic coordinate system. When the slope of the linearly fitted line is negative, the absolute value of the slope is determined as the energy attenuation index. When the slope of the linearly fitted line is zero or positive, the corresponding measurement point is marked as an abnormal spectral attenuation measurement point and is not considered a measurement point that passes the energy criterion verification. The number of effective frequency points participating in the linear fitting is no less than 5, and the goodness of fit is no less than 0.80. When the number of effective frequency points is less than 5 or the goodness of fit is less than 0.80, the frequency point with the largest deviation from the fitting trend is removed, and the fitting is refitted. If the above conditions are still not met after refitting, the corresponding measurement point is marked as a measurement point to be reviewed. Measurement points to be reviewed are not included in connectivity merging. Through the above method, the energy attenuation index has a fixed high-frequency band range, fitting model, frequency point number requirements, and error tolerance conditions.

[0037] In some embodiments of this application, when generating a flow structure partition map by merging the measurement point grid verified by the energy criterion according to the four-neighbor connectivity, the following steps are taken: determining the outer edge of the upstream side of the connected region as the boundary of the separation zone, determining the outer edge of the downstream side of the connected region as the reattachment point location, constructing a sample set with sand wave state parameters and flow structure features, obtaining a response relationship model by fitting, and outputting the prediction results of the flow structure features.

[0038] In some embodiments of this application, the preset proportion threshold and the preset attenuation threshold are determined by the energy index of the velocity sequence of the measuring point corresponding to the reference area outside the backflow candidate area; the response relationship model uses the sand wave state parameters aligned with the timestamp and the water flow structure features to form a sample set, and uses regression fitting to obtain the model for outputting the prediction results of the water flow structure features.

[0039] Specifically, after completing the spatial correspondence between the candidate backflow region and the measurement point grid, and aligning the timestamps of each data stream, energy criterion verification is performed on the velocity sequences of the measurement points within the candidate backflow region. First, wavelet denoising and reconstruction are performed on the velocity sequences. Then, a Fast Fourier Transform is performed on the velocity sequences after denoising to obtain the energy spectrum distribution. The energy attenuation index is calculated by fitting the high-frequency band attenuation trend of the energy spectrum distribution, and a cumulative energy distribution is simultaneously formed to characterize the concentration of energy in the frequency band. Based on this, the energy spectrum distribution is divided into low-frequency and non-low-frequency bands according to preset frequency bands. The low-frequency energy proportion is defined as the ratio of low-frequency energy to the total frequency band energy, and the dominant frequency peak is defined as the peak value at the point of maximum amplitude in the energy spectrum distribution. For example, the low-frequency band is selected as a band no higher than 10Hz to enhance the sensitivity to the dominant frequency components of the backflow vortex and vortex sheave.

[0040] When determining the low-frequency and non-low-frequency bands, the Nyquist frequency is first determined based on the actual sampling frequency of the flow velocity acquisition probe, which is half of the actual sampling frequency. The low-frequency band cutoff frequency is determined jointly based on the dominant frequency peak, a fixed upper limit, and the Nyquist frequency: first, the dominant frequency peak with the largest energy spectrum amplitude is selected within the candidate reflux region, and 1.5 times the dominant frequency peak is calculated; the minimum value among 1.5 times the dominant frequency peak, 10Hz, and half of the Nyquist frequency is taken as the candidate cutoff frequency; when the candidate cutoff frequency is lower than 5Hz, the low-frequency band cutoff frequency is taken as 5Hz. When the flow velocity acquisition probe's sampling frequency is 20Hz and the sand wave scale, water depth, and flow rate conditions described in this embodiment are used, the Nyquist frequency is 10Hz, and the low-frequency band cutoff frequency is taken as 5Hz. The low-frequency band is from 0Hz to the low-frequency band cutoff frequency, and the non-low-frequency band is the frequency band greater than the low-frequency band cutoff frequency but not higher than the Nyquist frequency. Therefore, the proportion of low-frequency energy is calculated according to the same frequency boundary determination rules at different sampling frequencies, while ensuring that non-low-frequency bands can be used for energy attenuation index calculation, thus avoiding arbitrary selection of the low-frequency band range by humans.

[0041] Specifically, the preset proportion threshold and preset attenuation threshold are determined by a reference region outside the candidate backflow region. The reference region is selected as a spatial region located in the stable mainstream upstream of the sand wave, separated from the candidate backflow region by at least one measurement point grid, and in which no continuous reverse tracer trajectories appear in the tracer flow field video. The reference region has no fewer than 5 valid measurement points; a valid measurement point is defined as a measurement point that has completed timestamp alignment, has a non-zero total energy spectrum, and is not marked as a measurement point with abnormal spectral attenuation or a measurement point to be verified. If the number of valid measurement points in the reference region is less than 5, the reference region is redefined after extending one measurement point grid column upstream along the mainstream direction; if there are still fewer than 5 valid measurement points after extension, the velocity sequence of the measurement points in the reference region is re-acquired.

[0042] For each effective measuring point within the reference area, the low-frequency energy proportion and energy attenuation index are calculated. The upper quartile of the low-frequency energy proportion within the reference area is determined as the preset proportion threshold, and the lower quartile of the energy attenuation index within the reference area is determined as the preset attenuation threshold. The upper quartile is used to characterize the higher level of the low-frequency energy proportion in the mainstream stable region, and the lower quartile is used to characterize the lower level of high-frequency energy attenuation in the mainstream stable region. Subsequently, the energy indicators of each measuring point grid within the return flow candidate area are verified using criteria. When the low-frequency energy proportion is not less than the preset proportion threshold and the energy attenuation index is not greater than the preset attenuation threshold, the corresponding measuring point grid is determined to have passed the energy criterion verification; when either condition is not met, the corresponding measuring point grid is removed. Thus, the spatial selection of the reference area, the requirements for effective measuring points, the handling of abnormal measuring points, and the threshold statistics method all have unified rules, which can reduce the inclusion of false return flow candidate areas in the flow structure zoning map caused by tracer diffusion, local illumination changes, or instantaneous disturbances.

[0043] The measuring point grids that passed the energy criterion verification were merged according to four-neighbor connectivity to form a flow structure zoning map. On the zoning map, the outer edge of the upstream side of the connected region was determined as the boundary of the separation zone, and the outer edge of the downstream side of the connected region was determined as the reattachment point location. Flow structure features such as the return vortex core and Karman vortex street frequency were further identified by combining flow field video. Finally, a sample set was constructed using the timestamp-aligned sand wave state parameters and flow structure features, and a response relationship model was established using regression fitting. For example, the response relationship between the separation zone length and wave height can be established to output the predicted results of the flow structure features and support comparative analysis under different operating conditions.

[0044] Understandably, this embodiment makes the water flow structure partitioning no longer rely on a single visual discrimination or single-point time-domain fluctuation, and enhances the stability of the identification of dominant structures such as backflow vortices and vortex streets from the frequency domain perspective. At the same time, a reproducible structure partitioning map is obtained by merging connectivity, and a response relationship model between the separation zone length and state parameters such as sand wave morphology is further established, realizing the expression from "structure detection" to "association modeling and prediction output", thereby improving the repeatability and portability of experimental conclusions.

[0045] In some embodiments of this application, morphological parameters include any one or more of wavelength, wave height, and wave angle, and motion parameters include wave velocity. In some embodiments of this application, the sampling frequency of the flow velocity acquisition probe is not less than 20Hz, and a grid of measurement points is arranged at wave crests, troughs, upstream slopes, downstream slopes, and separation zones; the tracer is uniformly released through an outlet pre-embedded on the upstream side of the sand wave, and multi-channel independent control is adopted, with the tracer flow rate adjustable from 0.1 to 2 ml / s.

[0046] Specifically, in this embodiment, the acquisition of sand wave state parameters is based on a sequence of sand wave motion images: time-series images are acquired by a high-definition camera system looking down and sideways, and an image recognition and morphology extraction unit automatically tracks the sand wave contour, thereby outputting morphological and motion parameters. The morphological parameters may include any one or more of wavelength, wave height, and wave angle. Wavelength is characterized by the distance between adjacent wave crests, wave height by the elevation difference between wave crests and troughs, and wave angle by the angle between the wave crest line and the mainstream direction; all of these can be calculated based on contour and terrain simulation results. Motion parameters may include wave velocity, which can be obtained by tracking the displacement of wave crest positions or characteristic contours at adjacent times and calculating the time interval, forming a temporal change sequence of wave velocity and wave height, used for correlation analysis with water flow structure characteristics. To ensure the reliability of velocity feature extraction, the velocity acquisition probe is mounted on a three-dimensional moving support or a three-dimensional electric sliding table to achieve gridded sampling points in key areas of the sand wave surface. The preferred grid covers wave crests, troughs, upstream slopes, downstream slopes, and separation zones, and acquires long-term instantaneous velocity data at a sampling frequency of at least 20Hz. Specifically, a grid structure with 7 vertical lines and 5 sampling points per line can be used, or a grid arrangement of 5cm longitudinally and 1cm vertically can be employed to improve the resolution of near-bed boundary layer and separation zone pulsations. Simultaneously, to obtain stable and controllable tracer visualization, the tracer is uniformly released from a pre-embedded outlet through a microtube injection system embedded in the upstream surface of the sand wave, with multi-channel independent control to achieve adjustable flow rate. The tracer can be a staining solution (e.g., potassium permanganate solution), and the tracer flow rate can be adjusted within the range of 0.1 to 2 ml / s to ensure clear streamline recording while minimizing disturbance to the mainstream structure.

[0047] Understandably, this embodiment enables the sand wave state parameters to not only reflect static geometry but also continuously characterize evolution velocity and amplitude changes, thus providing input for response relationship modeling. At the same time, the uniform release of the tracer on the upstream side and the multi-channel independent flow control ensure consistency of the tracer visualization results under different operating conditions and in different batches of experiments. Combined with the grid layout of measurement points at key locations such as the separation zone and high-frequency sampling of no less than 20Hz, the ability to capture low-frequency dominant components related to backflow and vortex structure and near-bed pulsation characteristics can be improved, thereby enhancing the stability of the water flow structure feature localization and the correlation conclusion of "sand wave state parameters - water flow structure characteristics".

[0048] In one specific embodiment, the indoor water tank is a rectangular tank with a total length of 25m, a width of 1m, and a depth of 0.8m, with a bottom slope set at 0.5%. The observation section uses a transparent sidewall structure for imaging. The experimental sand bed is laid with non-uniform plastic sand with a median particle size of 0.75mm. After the sand wave surface is formed in the observation section, a set of repeatable sand wave sequences is selected as the experimental boundary conditions: wave height 3.6cm, upstream wavelength 20cm, and downstream wavelength 4cm. The measuring point grid covers the wave crest, wave trough, upstream slope, downstream slope, and separation zone. Exemplarily, 7 vertical lines are arranged longitudinally, with 5 measuring points on each vertical line, and a flow velocity acquisition probe is installed at each measuring point. The sampling frequency is set to 20Hz. The camera system includes a top-down camera for acquiring image sequences of sand wave motion and a high-speed camera for acquiring video of the tracer flow field. A 10cm × 10cm grid calibration plate is placed in the shooting area to complete spatial calibration, obtaining the conversion relationship between pixels and actual dimensions (e.g., 1 pixel corresponds to 0.12mm). Based on this, the "coverage range of each measuring point grid in actual space" is mapped to the corresponding image area in the tracer flow field video, and this corresponding image area is used as the spatial unit. The operating conditions are set at a flow rate of 15.6 L / s and a water depth of 12.7 cm. After the flow rate and water depth stabilize, image sequences of sand wave motion, video of the tracer flow field, and flow velocity sequences at measuring points are acquired simultaneously. The flow velocity sequence at measuring points is acquired continuously for 180 seconds, the high-speed camera acquires 60 seconds of tracer flow field video at 500 fps, and the top-down camera acquires image sequences of sand wave motion at 10-second intervals. The timestamps of each acquired data stream are then aligned. The tracer is released uniformly through an outlet pre-embedded on the water-facing side of the sandbar. The tracer flow rate is adjustable within the range of 0.1–2 ml / s using multi-channel independent control; in this embodiment, 0.5 ml / s is used to balance tracer clarity and disturbance control.

[0049] In the data processing stage, image recognition is first performed on the sand wave motion image sequence to extract the sand wave contour, and sand wave state parameters are calculated based on the conversion relationship between pixels and actual size. Morphological parameters include, for example, any one or more of wavelength, wave height, and wave angle: for instance, on the contour at the same moment, if the horizontal pixel distance between two adjacent wave peaks is selected as 1667 pixels, then the wavelength is 1667 × 0.12 mm, approximately 200 mm; if the vertical pixel difference between the wave peak and trough on the same cross-section is selected as 300 pixels, then the wave height is 300 × 0.12 mm, approximately 36 mm; the wave angle can be determined by the angle between the direction of the wave peak line in the image coordinates and the mainstream direction. Motion parameters include, for instance, wave velocity: for instance, if the interval between two adjacent image acquisitions is 10 seconds, and a certain characteristic wave peak displaces 125 pixels along the mainstream direction, then the wave peak displacement is 125 × 0.12 mm, approximately 15 mm, corresponding to a wave velocity of approximately 1.5 mm / s. Subsequently, wavelet denoising and reconstruction are performed on the flow velocity sequence at the measurement points to reduce the impact of random noise and isolated outliers on the frequency domain analysis. Then, frequency domain analysis is performed on the reconstructed flow velocity sequence to obtain the energy spectrum distribution and calculate the energy index. Taking 180s sampling as an example, the frequency resolution is approximately 0.0056Hz, which can support stable identification of the low-frequency dominant component. The energy spectrum distribution is divided into low-frequency and non-low-frequency bands according to a preset frequency band. The low-frequency energy proportion is defined as the ratio of low-frequency energy to the total energy of the entire frequency band, and the dominant frequency peak is defined as the peak value at the point of maximum amplitude in the energy spectrum distribution. In this embodiment, the sampling frequency of the flow velocity acquisition probe is 20Hz, and the Nyquist frequency is 10Hz. According to the aforementioned low-frequency band cutoff frequency determination rule, the low-frequency band cutoff frequency is set to 5Hz, the low-frequency band is from 0Hz to 5Hz, and the non-low-frequency band is the band greater than 5Hz and not higher than 10Hz. The low-frequency energy proportion is the ratio of energy in the 0Hz to 5Hz band to the total energy of the 0Hz to 10Hz band. The energy attenuation index is calculated according to the aforementioned double logarithmic linear fitting rule: The non-low frequency band greater than 5Hz and not higher than 10Hz is considered the high frequency band. After removing frequency points with zero amplitude and isolated peak frequency points formed by instantaneous acquisition spikes, the logarithms of the frequency values ​​and energy spectrum amplitudes of the remaining frequency points are taken, and linear fitting is performed in double logarithmic coordinates. When the slope of the fitted line is negative, the absolute value of the slope is taken as the energy attenuation index. When the slope of the fitted line is zero or positive, the number of effective frequency points is less than 5, or the goodness of fit is less than 0.80, it is handled according to the aforementioned rules for abnormal measurement points or measurement points to be verified. Therefore, in this embodiment, the low-frequency energy ratio and energy attenuation index are obtained according to a fixed frequency band and a fixed fitting rule.

[0050] In the structural partitioning and modeling stage, the candidate regions and boundary positions for recirculation are first determined based on the tracer flow field video. Specifically, for each spatial unit, the tracer trajectory position is extracted frame by frame, and the displacement vector of adjacent frames is calculated. The dominant direction of the displacement vector in the upstream reference region of the sand wave is taken as the mainstream direction. When a spatial unit exhibits a tracer trajectory displacement opposite to the mainstream direction for no less than a preset number of consecutive frames, the spatial unit is determined as a candidate region for recirculation, and the boundary position is determined by the outer edge of the candidate region for recirculation. In this embodiment, the preset number of frames is 100, corresponding to approximately 0.2 seconds of continuous reverse motion under 500fps conditions, to suppress false recirculation caused by instantaneous eddy currents and tracer diffusion. Subsequently, energy criterion verification is performed: the candidate regions for recirculation are mapped to the measurement point grid, and the low-frequency energy proportion, peak frequency, and energy attenuation index corresponding to the velocity sequence of each measurement point in the candidate regions for recirculation are extracted and compared with the thresholds determined according to the reference region. The preset proportion threshold and the preset attenuation threshold are determined according to the aforementioned reference region selection rules and effective measurement point rules. Specifically, the reference region is selected as a spatial region located in the stable upstream mainstream of the sand wave, separated from the candidate backflow region by at least one measuring point grid, and in which no continuous reverse tracer trajectories appear in the tracer flow field video. The reference region must have at least 5 valid measuring points, which are those with completed timestamp alignment, non-zero total energy of the energy spectrum, and not marked as measuring points with abnormal spectral attenuation or those awaiting verification. If the number of valid measuring points in the reference region is less than 5, the reference region is redefined after extending one measuring point grid column upstream along the mainstream direction; if there are still fewer than 5 valid measuring points after the extension, the velocity sequence of the measuring points in the reference region is re-acquired.

[0051] After determining the reference region, the low-frequency energy proportion and energy attenuation index are calculated for each valid measuring point within the reference region. The upper quartile of the low-frequency energy proportion within the reference region is determined as the preset proportion threshold, and the lower quartile of the energy attenuation index within the reference region is determined as the preset attenuation threshold. For example, there are 7 valid measuring points in the reference region. The upper quartile of the low-frequency energy proportion of the 7 valid measuring points, after sorting, is 0.32, so the preset proportion threshold is 0.32; the lower quartile of the energy attenuation index of the 7 valid measuring points, after sorting, is 1.5, so the preset attenuation threshold is 1.5. For a certain return flow candidate measuring point, if its low-frequency energy proportion is 0.41 and its energy attenuation index is 1.2, then it simultaneously satisfies the conditions that the low-frequency energy proportion is not less than the preset proportion threshold and the energy attenuation index is not greater than the preset attenuation threshold, and the measuring point grid is determined to pass the energy criterion verification; if either condition is not met, the corresponding measuring point grid is removed and not included in the subsequent four-neighbor connectivity merging.

[0052] The validated measuring point grid is merged according to four-neighbor connectivity to generate a flow structure zoning map, and the flow structure characteristics are determined by the zoning map: the outer edge of the upstream side of the connected region is determined as the boundary of the separation zone, and the outer edge of the downstream side of the connected region is determined as the location of the reattachment point; the backflow range is determined by the coverage area of ​​the connected region. For example, under the condition of a wavelength of about 200 mm, the boundary of the separation zone is about 70 mm from the upstream side of the wave crest, and the reattachment point is about 160 mm from the downstream side of the wave crest, then the length of the separation zone is about 90 mm.

[0053] Finally, a response relationship model is established based on the characteristics of the water flow structure and the state parameters of the sand wave. For example, a sample set is formed from multiple operating conditions or repeated experiments, using wave height as the input variable and separation zone length as the output variable for regression fitting; for example, samples are (wave height 30mm, separation zone length 80mm), (wave height 36mm, separation zone length 95mm), and (wave height 42mm, separation zone length 110mm). The regression yields an approximately linear response relationship, which can be expressed as "for every 1mm increase in wave height, the separation zone length increases by approximately 2.5mm." Based on this, the predicted separation zone length can be output for a given wave height and used for comparative evaluation under different operating conditions. In this embodiment, all calculations are based on the calibrated actual size conversion, the frequency resolution corresponding to the sampling duration, and the statistical results of the threshold from the reference area.

[0054] In summary, this application, based on the spatial calibration and timestamp alignment of the camera system, unifies the sand wave motion image sequence, the tracer flow field video, and the measurement point velocity sequence into a single detection link. This not only extracts morphological and motion parameters from the sand wave contour to form boundary conditions, but also performs wavelet denoising and frequency domain energy spectrum analysis on the measurement point velocity sequence to obtain energy indices, thereby improving the robustness of identifying non-stationary features such as turbulence and vortex structures. Furthermore, the tracer flow field video is used to provide candidate recirculation regions and boundary locations, and the energy criterion is verified using the energy spectrum distribution and energy indices of the measurement point velocities within the candidate recirculation regions. The system merges the data according to connectivity to generate a water flow structure partition map, so that the location of key features such as separation zone, reattachment point, backflow range and vortex structure evolution no longer depends on a single visual judgment or single-point measurement inference. This can effectively suppress boundary drift and misjudgment caused by tracer diffusion, illumination changes and local measurement noise. After obtaining stable partitions and structural features, a response relationship model between water flow structure features and sand wave state parameters is further established, so that the detection results are improved from phenomenon description to quantifiable correlation and predictable output, thereby providing a consistent and verifiable data foundation and model for comparative analysis under different working conditions.

[0055] Based on another preferred embodiment described above, see [link to preferred embodiment]. Figure 2 As shown, this embodiment provides a detection system for the water flow structure of an indoor water tank with a sand-like surface, used to apply the above-described detection method for the water flow structure of an indoor water tank with a sand-like surface, including: An indoor water tank with an observation section, in which a test sand bed is set up and a sand wave surface is formed; The camera system is configured to acquire sand wave motion image sequences and tracer flow field videos, and output calibration data for spatial calibration of the sand wave motion image sequences and tracer flow field videos; The velocity acquisition unit includes velocity acquisition probes deployed at preset measurement point grids in the sand wave surface, used to acquire velocity sequences at the measurement points; The synchronous acquisition control unit is used to synchronously trigger the acquisition of the camera system and the flow velocity acquisition unit, and to align the timestamps of the sand wave motion image sequence, the tracer flow field video and the flow velocity sequence at the measurement point. The image processing unit is used to perform image recognition on the sequence of sand wave motion images to extract the sand wave contour and calculate the sand wave state parameters. The velocity analysis unit is used to perform wavelet denoising and reconstruction on the velocity sequence at the measurement point, and to perform frequency domain analysis to obtain the energy spectrum distribution and calculate the energy index. The structural partitioning and modeling unit is used to determine the candidate regions and boundary locations of the recirculation based on the tracer flow field video, and to verify the energy criteria by using the energy spectrum distribution and energy index corresponding to the flow velocity sequence of the measurement points in the candidate regions. The measured point grids that pass the verification are merged according to connectivity to generate a flow structure partitioning map and determine the flow structure characteristics. Based on the flow structure characteristics and sand wave state parameters, a response relationship model is established.

[0056] Specifically, see Figure 3 As shown in the figure, a grid of measuring points is laid out along key locations such as the upstream slope, crests, troughs, and downstream slope of the sand wave, using the sand wave cross-section as a reference (for example, the embodiment provides a grid layout of "7 vertical lines with 5 measuring points on each line"). This is used to obtain velocity sequences in the near-bed region of the sand wave and support energy spectrum and energy index analysis; Additionally, [the figure includes...] Figure 3 The separation and backflow structure between the main flow zone and the backwater side of the sand wave is schematically expressed in the form of "zone / boundary" and corroborated with the results of the tracer flow field visualization. This makes it easier to mark the separation point, reattachment zone (or area near the reattachment point), backflow vortex core and vortex street evolution and other flow structure characteristics on the same zone map, and serve as the basis for establishing the relationship between flow structure characteristics and sand wave state parameter response.

[0057] Understandably, by integrating the sand wave motion image sequence, tracer flow field video, and measuring point velocity sequence into the same detection link based on the spatial calibration and timestamp alignment of the camera system, it is possible to extract morphological and motion parameters from the sand wave contour to form boundary conditions, and to perform wavelet denoising and frequency domain energy spectrum analysis on the measuring point velocity sequence to obtain energy indices, thereby improving the robustness of identifying non-stationary features such as turbulence and vortex structures. The tracer flow field video is used to provide candidate recirculation regions and boundary locations, and then the energy criterion is corrected using the energy spectrum distribution and energy indices of the measuring point velocities within the candidate recirculation regions. The system verifies and merges the data according to connectivity to generate a water flow structure partition map. This allows the location of key features such as separation zones, reattachment points, backflow ranges, and vortex structure evolution to no longer rely on single visual judgments or single-point measurement inferences. It can effectively suppress boundary drift and misjudgment caused by tracer diffusion, illumination changes, and local measurement noise. After obtaining stable partitions and structural features, a response relationship model between water flow structure features and sand wave state parameters is further established. This elevates the detection results from phenomenon descriptions to quantifiable correlations and predictable outputs, thereby providing a consistent and verifiable data foundation and model for comparative analysis under different working conditions.

[0058] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A method for detecting the water flow structure of an indoor water tank with a sand-like surface, characterized in that, include: A sand wave surface and a test sand bed were set up in the indoor water tank observation section. A camera system was set up and the camera system was spatially calibrated. A flow velocity acquisition probe was installed at the preset measuring point grid in the sand wave surface. Collect sand wave motion image sequences, tracer flow field videos, and flow velocity sequences corresponding to the measuring point grid, and align the timestamps of each collected data stream; Image recognition is performed on the sand wave motion image sequence to extract the sand wave contour and calculate the sand wave state parameters, which include morphological parameters and motion parameters. The flow velocity sequence at the measurement points is subjected to wavelet denoising and reconstruction, and frequency domain analysis is performed to obtain the energy spectrum distribution and calculate the energy index. Based on the tracer flow field video, the candidate region and boundary position of the return flow are determined. The energy criterion is verified by the energy spectrum distribution and energy index corresponding to the flow velocity sequence of the measurement points in the candidate region of the return flow. The measurement point grids that pass the verification are merged according to connectivity to generate a water flow structure partition map and determine the water flow structure characteristics. A response relationship model is established based on the aforementioned water flow structure characteristics and sand wave state parameters.

2. The method for detecting the water flow structure of an indoor water tank with a sand-like surface according to claim 1, characterized in that, When determining the candidate reflow region and boundary location based on the tracer flow field video, the following are included: Based on the tracer flow field video, the displacement of the tracer trajectory is calculated frame by frame and the mainstream direction is determined. Spatial units that present tracer trajectories opposite to the mainstream direction for no less than a preset number of consecutive frames are determined as the reflow candidate region, and the outer edge of the reflow candidate region is determined as the boundary position.

3. The method for detecting the water flow structure of an indoor water tank with a sand-like surface according to claim 2, characterized in that, The spatial unit is determined by the spatial calibration result of the camera system, so that the image area of ​​the tracer flow field video corresponds one-to-one with the measurement point grid; Based on the spatial unit, the displacement of the tracer trajectory is statistically assigned, and the candidate return region and the boundary position are determined.

4. The method for detecting the water flow structure of an indoor water tank with a sand-like surface according to claim 2, characterized in that, When performing energy criterion verification, the following are included: The recirculation candidate region is mapped to the measurement point grid. The low-frequency energy ratio, the main frequency peak value, and the energy decay index are extracted from the energy spectrum distribution corresponding to the flow velocity sequence of the measurement point in the recirculation candidate region as energy indicators. When the low-frequency energy ratio is not less than the preset ratio threshold and the energy decay index is not greater than the preset decay threshold, the corresponding measurement point grid is determined to pass the energy criterion verification.

5. The method for detecting the water flow structure of an indoor water tank with a sand-like surface according to claim 4, characterized in that, The energy spectrum distribution is divided into low-frequency band and non-low-frequency band according to a preset frequency band. The low-frequency energy ratio is the proportion of low-frequency band energy to the total frequency band energy. The main frequency peak is the peak value at the point of maximum amplitude in the energy spectrum distribution. The energy attenuation index is determined by the attenuation trend of the high-frequency band of the energy spectrum distribution.

6. The method for detecting the water flow structure of an indoor water tank with a sand-like surface according to claim 4, characterized in that, When generating a flow structure partition map by merging the measurement point grids verified by the energy criterion according to the four-neighbor connectivity, the following is included: The outer edge of the water-facing side of the connected region is determined as the boundary of the separation zone, and the outer edge of the backwater side of the connected region is determined as the reattachment point. A sample set is constructed using the sand wave state parameters and the water flow structure features. A response relationship model is obtained by fitting the model, and the prediction results of the water flow structure features are output.

7. The method for detecting the water flow structure of an indoor water tank with a sand-like surface according to claim 4, characterized in that, The preset proportion threshold and the preset attenuation threshold are determined by the energy index of the flow velocity sequence of the measurement point corresponding to the reference area outside the backflow candidate area; the response relationship model uses the sand wave state parameters aligned with the timestamp and the water flow structure features to form a sample set, and uses regression fitting to obtain the model used to output the prediction results of the water flow structure features.

8. The method for detecting the water flow structure of an indoor water tank with a sand-like surface according to claim 1, characterized in that, The morphological parameters include any one or more of wavelength, wave height, and wave angle, and the motion parameters include wave speed.

9. The method for detecting the water flow structure of an indoor water tank with a sand-like surface according to claim 1, characterized in that, The sampling frequency of the flow velocity acquisition probe is not less than 20Hz, and a grid of measurement points is set up at the crests, troughs, upstream slopes, downstream slopes, and separation zones. The tracer is uniformly released through the outlet pre-embedded on the upstream side of the sand wave, and is controlled independently by multiple channels. The tracer flow rate is adjustable from 0.1 to 2 ml / s.

10. A detection system for the water flow structure of an indoor water tank with a sand-like surface, used in applying the detection method for the water flow structure of an indoor water tank with a sand-like surface as described in any one of claims 1-9, characterized in that, include: An indoor water tank with an observation section, wherein a test sand bed is set up in the observation section and a sand wave surface is formed; The camera system is configured to acquire a sequence of sand wave motion images and a video of the tracer flow field, and output calibration data for spatial calibration of the sand wave motion image sequence and the video of the tracer flow field; The flow velocity acquisition unit includes flow velocity acquisition probes arranged at preset measurement point grids in the sand wave surface, used to acquire the flow velocity sequence at the measurement points; The synchronous acquisition control unit is used to synchronously trigger the acquisition of the camera system and the flow velocity acquisition unit, and to align the timestamps of the sand wave motion image sequence, the tracer flow field video and the measurement point flow velocity sequence. The image processing unit is used to perform image recognition on the sand wave motion image sequence to extract the sand wave contour and calculate the sand wave state parameters; The velocity analysis unit is used to perform wavelet denoising and reconstruction on the velocity sequence of the measurement point, and to perform frequency domain analysis to obtain the energy spectrum distribution and calculate the energy index. The structural partitioning and modeling unit is used to determine the candidate region and boundary position of the return flow based on the tracer flow field video, and to verify the energy criterion by using the energy spectrum distribution and energy index corresponding to the flow velocity sequence of the measurement points in the candidate region of the return flow. The measured point grids that pass the verification are merged according to connectivity to generate a water flow structure partitioning map and determine the water flow structure characteristics. A response relationship model is established based on the water flow structure characteristics and the sand wave state parameters.

Images