CN-121982070-A - Synchronous detection system and method for three-dimensional flow field of sand wave surface

CN121982070ACN 121982070 ACN121982070 ACN 121982070ACN-121982070-A

Abstract

The invention discloses a sand wave surface three-dimensional flow field synchronous detection system and method, which comprise the steps of obtaining a time sequence synchronous particle image data set and a morphology image data set, obtaining an initial three-dimensional three-component speed vector field and a corresponding vector confidence level field, obtaining a three-dimensional three-component speed vector field after confidence level screening and continuity constraint correction, generating a multi-scale turbulence feature set, outputting a three-dimensional vortex structure evolution feature set based on a vortex structure recognition criterion, extracting sand wave contours and sand wave motion parameters based on the morphology image data set, constructing a sand wave surface macroscopic skeleton curved surface model and a sand wave surface time-varying bed surface coordinate system, obtaining a bed surface registration aligned three-dimensional flow field model, and generating a sand wave surface three-dimensional flow field synchronous detection report. The invention establishes the corresponding relation between the macroscopic bed surface undulating stripping and the bed surface phase-flow field phase, so that flow fields of different time frames can be aligned on the wave crest-wave trough phase axis in a comparable way, thereby realizing the real bed surface phase alignment.

Inventors

LIU XIAOMIN
YANG JIAYUN
MENG SHAOMIN
YANG YAOTIAN
Shi Cuixiang
JI HONGLAN

Assignees

内蒙古农业大学

Dates

Publication Date: 20260505
Application Date: 20260408

Claims (8)

1. The sand wave surface three-dimensional flow field synchronous detection method is characterized by comprising the following steps of: Acquiring uncalibrated multi-view particle image sequences and sand wave form image sequences in a sand wave surface adjacent area in an experimental tank, continuously acquiring a plurality of groups of particle image sequences and sand wave form time sequence images under a unified time stamp trigger mechanism, and writing a time stamp of a unified clock into each frame of data to obtain a time sequence synchronous particle image data set and a time sequence synchronous form image data set; Performing three-dimensional reconstruction of the chromatographic particles based on the particle image dataset to obtain initial three-dimensional particle intensity volume data and a corresponding reconstruction confidence volume, and performing three-dimensional cross-correlation and vector solving based on the initial three-dimensional particle intensity volume data to obtain an initial three-dimensional three-component velocity vector field and a corresponding vector confidence field; Removing a low-confidence vector from the initial three-dimensional three-component speed vector field according to the reconstruction confidence body and the vector confidence field, and performing continuity constraint complementation and smooth correction on the hollow region to obtain a three-dimensional three-component speed vector field after confidence screening and continuity constraint correction; performing time-averaged statistics and pulsation decomposition on the three-dimensional three-component speed vector field subjected to confidence level screening and continuity constraint correction to generate a multi-scale turbulence characteristic set; Performing energy spectrum analysis and time-frequency decomposition on the multi-scale turbulence feature set, extracting energy distribution, dominant frequency and energy attenuation indexes of different scale vortex structures, and outputting a three-dimensional vortex structure evolution feature set based on a vortex structure identification criterion; Extracting sand wave contours and sand wave motion parameters based on a morphological image data set, and constructing a sand wave surface macroscopic skeleton curved surface model and a bed surface coordinate system of the sand wave surface changing along with time; carrying out flow field-bed surface flexible registration on the three-dimensional three-component velocity vector field and the sand wave surface macroscopic skeleton curved surface model under a bed surface coordinate system to obtain a three-dimensional flow field model with aligned bed surface registration; Based on the spatial correlation between the bed surface registration aligned three-dimensional flow field model and the sand wave surface macroscopic skeleton curved surface model, multidimensional flow field-bed surface coupling characteristic data are obtained, and grading judgment is carried out on the multidimensional flow field-bed surface coupling characteristic data, so that a sand wave surface three-dimensional flow field synchronous detection report is generated.
2. The method for synchronous detection of three-dimensional flow field of sand wave surface according to claim 1, wherein the step of writing a time stamp of a unified clock for each frame of data to obtain a time-sequence synchronous particle image dataset and a morphology image dataset comprises the steps of: constructing a trigger sequence taking a synchronous controller as a main clock, sending homologous trigger signals to a double-camera synchronous acquisition module, a laser chip light illumination module, a trace particle release and regulation module and a multi-angle form acquisition module, and writing nanosecond time stamps when triggering each time; Performing gate control matching on the windowing time of the laser film light and the exposure time of the camera according to the trigger sequence, so that the exposure effective interval and the film light stable interval are overlapped on a time axis; Setting an acquisition frame rate and an acquisition duration according to the sand wave migration speed and the time resolution requirement of a target turbulence scale, so that the continuous frame number of the particle image sequence is not less than a preset threshold value and the morphological image sequence covers the migration process of at least one sand wave wavelength; And performing frame sequence number alignment and timestamp interpolation correction on the multi-source data stream to obtain a time-sequence synchronous particle image data set and a morphology image data set.
3. The method for synchronous detection of a three-dimensional flow field of a sand wave surface according to claim 1, wherein the performing three-dimensional cross-correlation and vector solving based on initial three-dimensional particle intensity volume data comprises: Performing stereo matching and line-of-sight back projection on the dual-camera particle image, constructing a three-dimensional voxel grid, and solving particle intensity distribution on the voxel grid to obtain initial three-dimensional particle intensity volume data; calculating a reconstruction confidence body of the initial three-dimensional particle intensity body data based on voxel re-projection consistency, iteration residual error convergence degree and particle density uniformity; performing three-dimensional cross-correlation between three-dimensional particle intensity volume data at adjacent moments to obtain a displacement peak value and sub-pixel displacement, and dividing the displacement by a time interval to obtain an initial three-dimensional three-component velocity vector field; and calculating a vector confidence coefficient field based on the cross-correlation peak value ratio, the peak sharpness and the local consistency of the matching residual error and the vector field, so that the vector confidence coefficient field and the initial three-dimensional three-component speed vector field are in one-to-one correspondence in the voxel index dimension.
4. The method for synchronous detection of a three-dimensional flow field of a sand wave surface according to claim 1, wherein the removing the low confidence vector from the initial three-dimensional three-component velocity vector field according to the reconstructed confidence volume and the vector confidence field comprises: reading the reconstruction confidence coefficient and the vector confidence coefficient of each voxel position, and eliminating the speed vector at the voxel when the reconstruction confidence coefficient or the vector confidence coefficient is lower than a corresponding threshold value to obtain a speed vector field after confidence coefficient screening; Converting the speed vector field subjected to confidence level screening into regular grid topological structure data, wherein the regular grid topological structure data comprises a voxel node set, an adjacent side set and a side weight set, and the side weight is determined by adjacent vector differential amplitude values and local shear strength; Performing constraint optimization correction on the velocity vector field, wherein a data fidelity term is used for limiting the correction vector not to deviate from an original high-confidence vector, a continuity constraint term is used for inhibiting non-physical mutation caused by cavity complementation, and a near-bed boundary condition term is used for constraining velocity gradient and shearing direction near a bed surface; and outputting the three-dimensional three-component speed vector field after confidence level screening and continuity constraint correction.
5. The method for synchronously detecting the three-dimensional flow field of the sand wave surface according to claim 1, wherein the time-averaged statistics and pulsation decomposition of the three-dimensional three-component velocity vector field after the opposite confidence screening and the continuity constraint correction comprises the following steps: Time-averaging the velocity vector field to obtain a time-averaged flow velocity field, and subtracting the time-averaged flow velocity field from the instantaneous flow velocity field to obtain a pulsating velocity field; Calculating a reynolds stress tensor component, turbulence intensity, turbulence energy and dissipation rate estimator based on the pulsation velocity field; Performing wavelet decomposition on the velocity time sequence of the selected near-bed high-turbulence region to obtain a multi-scale velocity component; Performing fast Fourier transform on the multi-scale velocity components to obtain an energy spectrum curve, fitting a high-frequency-band slope, calculating an energy attenuation index, and simultaneously performing POD modal decomposition to extract a dominant coherent structure, and outputting a multi-scale turbulence characteristic set and a dominant modal set.
6. The method for synchronously detecting the three-dimensional flow field of the sand wave surface according to claim 1, wherein the steps of extracting the sand wave contour and the sand wave motion parameters based on the morphological image data set, constructing a sand wave surface macroscopic skeleton curved surface model and a bed surface coordinate system of the sand wave surface changing along with time are as follows: performing distortion correction and edge enhancement on the overlook and side-view morphological images, and extracting sand wave crest lines, wave trough lines and bed surface contour curves based on threshold segmentation and contour tracking to obtain a morphological key point set; performing multi-view geometric fusion and time consistency constraint reconstruction on the form key point set to obtain a bed surface three-dimensional form point cloud and a bed surface form confidence map; Performing flow type surface fitting on the bed surface three-dimensional form point cloud to obtain a sand wave surface macroscopic skeleton curved surface model updated with time, and constructing a bed surface coordinate system according to the tangential direction, the normal direction and the main migration direction of the macroscopic skeleton curved surface model; Binding the macroscopic skeleton curved surface model with the bed surface coordinate system and the time stamp of the speed vector field.
7. The method for synchronous detection of a three-dimensional flow field of a sand wave surface according to claim 1, wherein the step of performing flow field-bed surface flexible registration of the three-dimensional three-component velocity vector field and a sand wave surface macroscopic skeleton curved surface model under a bed surface coordinate system comprises the following steps: Projecting the velocity vector field to a normal-tangential component space under a bed surface coordinate system, executing macroscopic bed surface fluctuation stripping, eliminating low-frequency background components caused by overall water surface fluctuation and tank body slight inclination, and only retaining a local high-frequency flow structure related to sand wave morphology; Performing flexible registration on the stripped speed vector field, and establishing a corresponding relation between a bed surface phase and a flow field phase, so that flow fields of different time frames can be aligned on a wave crest-wave trough phase axis to obtain a three-dimensional flow field model with the bed surface in registration alignment; Identifying a position set of a separation point and a reattachment point in a three-dimensional flow field model with the bed surface aligned, calculating the length of the separation region, the volume fraction of the reflux region, the reflux vortex core track and the vortex peak distribution, and calculating a near-bed shear stress index and a change curve of the near-bed shear stress index along with the sand wave phase; and establishing a response relation model between the three-dimensional vortex evolution characteristics and the sand wave migration speed and wave height changes, and outputting multidimensional flow field-bed surface coupling characteristic data for structured field filling and grading judgment of a detection report.
8. A sand wave surface three-dimensional flow field synchronous detection system for executing the sand wave surface three-dimensional flow field synchronous detection method according to any one of claims 1 to 7, characterized by comprising: The image acquisition module is used for acquiring uncalibrated multi-view particle image sequences in the observation section of the experimental tank, synchronously acquiring morphological image sequences, and performing light path-imaging robust preprocessing on the particle image sequences to generate an illumination homogenization particle characteristic set; The synchronous calibration and triggering module is used for executing the stereoscopic calibration of the two cameras to output a pixel-world coordinate mapping matrix set and a calibration confidence coefficient parameter, and is used for constructing a unified time stamp triggering mechanism to synchronously trigger the release of the cameras, the laser film light and the particles and simultaneously finish the alignment of frame sequence numbers and time stamps; the three-dimensional reconstruction module is used for executing three-dimensional reconstruction of the chromatographic particles to output initial three-dimensional particle intensity volume data and a reconstruction confidence volume, and executing three-dimensional cross-correlation and vector solving to output an initial three-dimensional three-component velocity vector field and a vector confidence volume; The confidence level screening and constraint correction module is used for eliminating low confidence level vectors according to the reconstruction confidence level body and the vector confidence level field, and performing complementation and smooth correction on the velocity vector field based on the continuity constraint and the near-bed boundary condition so as to output a corrected three-dimensional three-component velocity vector field; The turbulence multi-scale analysis module is used for outputting a vortex structure evolution feature set; The bed surface morphology modeling module is used for extracting sand wave contours and motion parameters from the morphology image sequence, constructing a sand wave surface macroscopic skeleton curved surface model and a bed surface coordinate system and outputting a bed surface morphology confidence coefficient map; The flow field-bed surface registration and coupling characteristic extraction module is used for flexibly registering the three-dimensional three-component speed vector field and the sand wave surface macroscopic skeleton curved surface model under a bed surface coordinate system to obtain multidimensional flow field-bed surface coupling characteristic data; The grading judgment and report generation module is used for carrying out grading judgment on the multidimensional flow field-bed surface coupling characteristic data and generating a sand wave surface three-dimensional flow field synchronous detection report.

Description

Synchronous detection system and method for three-dimensional flow field of sand wave surface Technical Field The invention relates to the technical field of sand wave surfaces, in particular to a sand wave surface three-dimensional flow field synchronous detection system and method. Background Along with the continuous improvement of river bed evolution research and hydraulic structure safety assessment requirements, the refined measurement of a three-dimensional flow field structure near a sand wave surface has become an important technical means for revealing sediment starting, transporting and bed surface morphology feedback mechanisms. Especially under the conditions of open channel water tanks or experimental tanks, the sand wave migration process is accompanied with complex three-dimensional separation flow, a reflux zone and a multi-scale vortex structure, and the spatial distribution and the time evolution of the sand wave migration process directly influence the shear stress distribution of a bed surface and the change rule of the sand wave morphology. In the prior art, an experimental measurement method for a sand wave surface flow structure generally adopts a planar particle image velocimetry or two-dimensional section measurement mode to collect and analyze a velocity field in a certain fixed plane. However, the flow near the sand wave surface has obvious three-dimensional property and space non-uniformity, and the single plane measurement method is difficult to completely capture the space topological relation of the three-dimensional separation area, the reflux area and the vortex structure, and cannot reflect the coupling change of the flow field in the normal direction and the tangential direction. Meanwhile, the traditional two-dimensional PIV measurement often relies on single visual angle imaging, and accurate reconstruction of a shielding area and a complex fluctuation area of a bed surface is difficult, so that the speed vector is in missing or misjudgment in a near-bed area, and accurate identification of near-bed shear stress and separation point positions is affected. In the prior art, three-dimensional flow field data are obtained by adopting a three-dimensional PIV or chromatography PIV technology, but in the practical application process, three-dimensional particle reconstruction is highly sensitive to synchronization precision, calibration precision and particle density distribution. In the prior art, an independent triggering or simple hardware synchronization mode is often adopted between multi-source acquisition equipment, the time alignment precision is limited, the time dislocation phenomenon is easy to generate under the conditions of a high-frequency turbulence structure or rapid sand wave migration, the strict correspondence between a particle image sequence and a bed surface morphological image is lacked, and the spatial consistency of a flow field and the bed surface at the same moment is difficult to ensure. In addition, the traditional tomographic reconstruction method is easily influenced by particle shielding, uneven illumination and reprojection errors in the voxel inversion process to generate low confidence vector or local abnormal speed peak value, but the traditional method usually only carries out post-processing through simple threshold value screening or neighborhood averaging, lacks structural continuity constraint, is easy to introduce non-physical mutation in the cavity complement process, and reduces the reliability of flow field data. In terms of the coupling analysis of the flow field and the bed surface morphology, the prior art mostly adopts the method of measuring the flow field and the bed surface morphology respectively, and then carries out superposition analysis through later stage manual alignment or simple coordinate conversion. However, the sand wave surface continuously migrates and changes in wave height in the experimental process, if a bed surface reference coordinate system updated along with time is not established, flow fields of different time frames are difficult to carry out comparative analysis on a unified phase axis, macroscopic bed surface fluctuation and a local flow structure are mutually overlapped, and low-frequency background components such as overall water surface fluctuation or tank body slight inclination and the like are easily mistaken as local flow characteristics related to the sand wave, so that the accurate judgment of the separation area length, the reflux area volume fraction and the vortex core track is influenced. Meanwhile, the existing analysis method stays at a statistical average level, lacks a systematic extraction means for the response relation between the multi-scale turbulence structure and the sand wave migration parameter, and is difficult to reveal the evolution rule of the three-dimensional vortex structure from the energy spectrum distribution, dominant frequency or coherent structure modal angl