A test method and device for starting bed load particles

Abstract

The present invention discloses an experimental method and device for the initiation of bedload particles, which relates to the field of fluid mechanics technology. The method includes: laying sediment particles in a wave flume, and adjusting the water flow rate through flow control until the bed shear stress is less than the critical shear stress; exciting the fluorescence of tracer particles with a laser beam, and synchronously collecting the instantaneous flow velocity and sediment particle displacement data of the flow field using a camera array; analyzing the spatiotemporal distribution characteristics of turbulent kinetic energy dissipation and pressure energy dissipation based on the turbulent burst process, and establishing a mathematical model for the initiation of sediment particles; combining experimental data with computational fluid dynamics numerical simulation to verify and optimize the mathematical model. According to the technical solution of the present application, the analytical accuracy and experimental convergence speed of the sediment initiation criterion under complex hydrodynamic conditions can be significantly improved, and efficient coordination of the entire chain from micromechanical response to macro-engineering application can be achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of fluid mechanics, and in particular to a test method and device for starting bed load particles. Background Art

[0002] The mechanism of bedload sediment mobilization is a core research topic in river dynamics, coastal engineering, and environmental fluid dynamics. Its accurate prediction has important practical implications for riverbed evolution, waterway maintenance, and flood control. Traditional theories are often based on the Shields curve framework, describing the mobilization threshold of sediment particles using a critical shear stress model. This model assumes that sediment mobilization is dominated by the average bed shear stress and simplifies the critical mobilization condition to a function of the dimensionless shear stress and the particle Reynolds number.

[0003] However, the Shields curve shows significant limitations in complex hydrodynamic environments, and its prediction error can reach more than 30%. Studies have shown that such errors are due to the traditional model's simplified treatment of turbulent dynamic characteristics and energy dissipation mechanisms, which are specifically reflected in the following aspects: the method does not fully consider the instantaneous force of turbulent bursts in the near-wall area on sediment particles. Turbulent bursts cause pulsating velocity and pressure fluctuations near the bed surface, generating instantaneous drag and lift that far exceed the time-averaged shear stress. Traditional models only rely on time-averaged shear stress and cannot capture such dynamic effects; the Shields curve only considers the work done by fluid kinetic energy on sediment particles, ignoring the synergistic effect of turbulent kinetic energy dissipation and pressure energy dissipation on particle initiation; existing research Most studies have focused on single flow conditions, without systematically quantifying the coupled effects of multiple factors, such as wave period, bed roughness, and inter-particle collision friction. For example, in strong tidal areas of estuaries, the interaction between periodic flow reversals and dynamic adjustments in bed morphology significantly alters the sediment initiation threshold, but existing models lack adaptability to such complex scenarios. While traditional experimental devices can control a single variable, their data is difficult to directly embed into computational fluid dynamics models. Furthermore, the arrangement of pressure sensors easily interferes with the flow field and the sampling frequency is insufficient, leading to bottlenecks in the coupled verification of experimental data and numerical simulations. Therefore, in response to the aforementioned challenges, the present invention proposes an experimental method and device for bedload particle initiation. Summary of the Invention

[0004] Technical Purpose

[0005] To address the above-mentioned issues, the present invention aims to provide an experimental method and apparatus for bedload particle initiation, aiming to address the theoretical deficiencies of traditional bedload sediment initiation prediction models in complex hydrodynamic environments, including the neglect of the dynamic effects of turbulent bursts, the nonlinear mechanism of energy dissipation, and the coupling of multiple factors. By integrating the spatiotemporal analysis of turbulent kinetic energy and pressure energy dissipation, and the multi-scale collaboration of experimental data and numerical simulation, a high-precision mathematical model of sediment initiation is constructed, breaking through the Shields curve's sole reliance on time-averaged shear stress. This improves the reliability of predictions in complex scenarios such as strong turbulence and wave-current coupling, providing theoretical support for riverbed evolution analysis and water conservancy project optimization.

[0006] Technical Solution

[0007] To achieve the above objectives, the present invention provides an experimental method and apparatus for the initiation of bed load particles. By using sediment particles with embedded pressure sensors in a wave flume, a laser-PIV velocity measurement system, and a high-speed camera array, the instantaneous flow velocity and particle displacement in the flow field are synchronously acquired. Quadrant analysis and statistical methods are combined to quantify the contribution of turbulent coherent structures to drag and lift, and a nonlinear correlation model between the pressure energy dissipation rate and the turbulent kinetic energy spectrum is established. The method is expanded to low Reynolds number conditions through OpenFOAM fluid-structure coupling simulation, and a dynamic monitoring module is embedded to optimize parameter control, ultimately forming a universal prediction framework applicable to multi-scale hydrodynamic conditions.

[0008] In a first aspect, the present invention provides a test method for bed load particle initiation, comprising:

[0009] Sediment particles are laid in a wave tank, and the water flow rate is adjusted by flow control until the bed shear stress is less than the critical shear stress;

[0010] The laser beam is used to excite the fluorescence of the tracer particles, and a high-speed camera array is used to synchronously collect the instantaneous flow velocity and sediment particle displacement data of the flow field;

[0011] Based on the turbulent burst process, the temporal and spatial distribution characteristics of turbulent kinetic energy dissipation and pressure energy dissipation are analyzed, and a mathematical model of sediment particle initiation is established;

[0012] Combine experimental data with computational fluid dynamics numerical simulations to verify and optimize the mathematical model.

[0013] Furthermore, according to turbulence theory, the velocity and pressure at any point in the flow field show irregular pulsations over time. The formula for the instantaneous velocity of the fluid is:

[0014]

[0015] Where, is the instantaneous flow velocity; is the average flow velocity; Pulsating flow rate.

[0016] Furthermore, by calculating the Reynolds stress and combining it with the quadrant analysis method to divide the contribution of the turbulent burst process to the drag and lift of sediment particles, a statistical method is used to quantify the influence of each quadrant on the average Reynolds stress.

[0017] Furthermore, the calculation formula of the Reynolds stress is:

[0018]

[0019] Where, is the Reynolds stress; is the fluid density; It is the time-averaged product of turbulent pulsating flow velocity, which represents the intensity of turbulent bursts.

[0020] Furthermore, the contribution of Reynolds stress to the drag and lift of sediment particles during turbulent bursts is divided by the following formula:

[0021]

[0022] Where, is the contribution of the turbulence burst process to the average Reynolds stress; For time; is the pulsating flow velocity in the horizontal direction of the water flow at time t; is the pulsating flow velocity in the vertical direction of the water flow at time t; is the pulsating flow velocity in the horizontal direction of the water flow; is the pulsating flow velocity in the vertical direction of the water flow; is the behavior index of the turbulence burst process at the i-th sampling point.

[0023] Furthermore, the mathematical model of sediment particle initiation is based on the correlation between the pressure energy dissipation rate and the turbulent kinetic energy spectrum, and by synchronously collecting pressure sensor data and particle displacement data, a nonlinear relationship between the pressure energy fluctuation variance, skewness and kurtosis and the particle motion trajectory is established.

[0024] Furthermore, the turbulence intensity is quantified by the root mean square of velocity fluctuations in all directions, providing basic data for analyzing turbulent kinetic energy dissipation and pressure fluctuations. The calculation formula for turbulence intensity in different directions is:

[0025]

[0026] Where, is the turbulence intensity in the direction of water flow; is the horizontal turbulence intensity; is the turbulence intensity in the vertical direction; is the total number of samples, that is, the number of samples in the time series; is the instantaneous water flow velocity component at the i-th sampling point; is the instantaneous horizontal lateral velocity component of the i-th sampling point; is the instantaneous vertical velocity component of the i-th sampling point; Indicates the average flow velocity in the direction of water flow; It represents the average flow velocity in the horizontal direction; It represents the average flow velocity in the vertical direction.

[0027] Furthermore, the temporal characteristics of the drag force acting on the surface of the glass ball are represented by the drag force spectrum, which is calculated as follows:

[0028]

[0029] Where, is the drag force spectrum output; is the drag coefficient caused by flow velocity; is the area of the glass ball exposed to the pulsating flow area; is the sampling frequency; is the absolute value or modulus of a complex number; is the dimensionless frequency function of the drag force caused by the flow velocity; is the flow velocity spectrum; is the pressure coefficient of the drag force; is the surface area of the glass ball's pressure fluctuation zone; is the dimensionless frequency function of the drag force caused by pressure; is the fluid pressure spectrum.

[0030] Furthermore, the instantaneous pressure at any point is decomposed into the sum of the time-averaged pressure and the fluctuating pressure by statistical analysis method. The calculation formula is:

[0031]

[0032] Where, is the sum of the time-averaged pressure and the fluctuating pressure; is the average pressure value at any point in time; is the fluctuating pressure at the i-th sampling point.

[0033] Furthermore, the calculation formula for the pressure fluctuation variance is:

[0034]

[0035] Where, is the probability density function of pressure difference fluctuation; is the fluctuating pressure; is the total number of samples, that is, the number of samples in the time series.

[0036] Furthermore, the pressure deflection calculation formula is:

[0037]

[0038] Where, is the skewness of pressure.

[0039] Furthermore, the formula for calculating the kurtosis of pressure is:

[0040]

[0041] Where, is the kurtosis of pressure.

[0042] Furthermore, the OpenFOAM open source software was used to perform fluid-solid coupling numerical simulation and adjust the bed roughness and particle shape parameters, extending the model to low Reynolds number conditions to generate a general prediction model suitable for complex hydrodynamic conditions.

[0043] Furthermore, a real-time monitoring method for the microscopic strain of sediment particles based on fiber Bragg gratings is also included. The microscopic strain is measured by the reflection wavelength shift caused by the change of the grating period. The dynamic stress distribution on the particle surface is inverted by the wavelength shift. The transient response mechanism of the particle stress is quantified by combining the spatiotemporal characteristics of the turbulent burst process. The starting criterion formula is:

[0044]

[0045] Where, is the effective shear stress; is the strain coupling coefficient; is the real-time strain value; is the critical strain threshold.

[0046] Furthermore, it also includes a multi-parameter adaptive optimization control method based on reinforcement learning, which uses a deep reinforcement learning algorithm to build an interaction framework between the intelligent agent and the environment, dynamically optimizes parameters through a reward function, and the agent explores the parameter space through a Q-learning strategy. The reward function is defined as:

[0047]

[0048] Where, is the output of the reward function; || || is the Euclidean norm, which is used to quantify the absolute value of the deviation; is the particle displacement value; is the CFD predicted value of particle displacement; is the weight coefficient; is the effective shear stress value; is the shear stress of the theoretical model.

[0049] This method drives parameter space exploration through a reward function, achieving efficient collaboration between experimental data and numerical models, significantly shortening the experimental convergence time and improving prediction consistency. It deeply integrates data-driven adaptive control with physical models, providing a universal framework for parameter optimization under complex hydrodynamic conditions.

[0050] In a second aspect, the present invention further provides a test device for starting a load particle movement, wherein the device, when in operation, performs the method described in the first aspect, including:

[0051] A wave flume (1), a laser beam (2), sediment particles (3), a high-speed camera array (4), a water-filled tetrahedron (5), a USB data cable (6), an electronic computer (7), and a wave flow (8); the sediment particles (3) are placed in the middle section of the wave flume (1); the laser beam (2) is vertically directed toward the sediment particles (3) and is perpendicular to the direction of the wave flow (8); the high-speed camera array (4) is connected to the electronic computer (7) via the USB data cable (6) for synchronously collecting displacement and flow field data of the sediment particles (3); the water-filled tetrahedron (5) is arranged on the side wall of the wave flume (1) to reduce laser reflection interference.

[0052] Furthermore, the mud and sand particles (3) are a mixture of hollow glass balls and quartz sand, wherein a pressure sensor is embedded in the surface of the hollow glass ball, and the sensor circuit is hidden inside the glass ball and is connected to the data acquisition instrument through the side wall of the water tank.

[0053] Furthermore, the high-speed camera array (4) includes three high-speed cameras, which are respectively arranged on the side and above the water tank, two of which are at a non-perpendicular angle to the water tank and are used for three-dimensional PIV flow velocity measurement, and the other camera is used for real-time monitoring of the spatial displacement of sediment particles (3).

[0054] By embedding an optimized mathematical model of sediment initiation into the control system of a river dredging project, a cross-scale translation from theoretical model to engineering practice is achieved. This technology breaks through the barriers between laboratory conditions and engineering scenarios, enabling sediment initiation predictions to drive the adaptive control of dredging equipment in real time, significantly improving project efficiency and reducing the risk of environmental disturbances. It embodies a closed-loop innovation path from basic research to industrial implementation.

[0055] In a third aspect, the present invention further provides a computer-readable storage medium storing a computer program, wherein the computer program, when executed by a management platform, implements the aforementioned test method for starting load particles.

[0056] This method integrates a laser PIV velocity measurement system, a high-speed camera array, and an embedded pressure sensor to achieve simultaneous, high-resolution acquisition of instantaneous flow velocity and particle displacement. Combining a fiber Bragg grating microstrain monitoring system with reinforcement learning multi-parameter adaptive optimization control, it overcomes the bottlenecks of traditional methods in terms of spatiotemporal resolution and parameter adjustment efficiency. By integrating multi-physics field dynamic analysis, data-driven intelligent decision-making, and cross-scale model coupling, it significantly improves the analytical accuracy and experimental convergence speed of sediment initiation criteria under complex hydrodynamic conditions. This method achieves efficient collaboration across the entire chain, from micromechanical response to macro-engineering application, providing an innovative, integrated theoretical and practical solution for sediment dynamics research and water conservancy project optimization.

[0057] Beneficial effects

[0058] The following technical effects are achieved by implementing the test method and device for starting bed load particles provided by the present invention:

[0059] (1) This application achieves nanoscale resolution monitoring of the dynamic strain on the particle surface by embedding FBG sensors inside sediment particles, breaking through the limitations of traditional external sensors on flow field interference. By capturing the transient stress distribution induced by turbulent bursts in real time, the device can accurately identify the critical strain threshold for particle initiation, significantly improving the sensitivity and spatiotemporal resolution of the initiation criterion. Its technical contribution lies in coupling microscopic mechanical responses with macroscopic flow field characteristics, providing a new experimental observation dimension for the nonlinear dynamic analysis of sediment initiation mechanisms.

[0060] (2) An intelligent agent based on deep reinforcement learning dynamically optimizes parameters such as water velocity and wave frequency, solving the inefficiency of traditional experiments that rely on manual experience to adjust parameters. By driving parameter space exploration through a reward function, efficient collaboration between experimental data and numerical models is achieved, significantly shortening experimental convergence time and improving prediction consistency. This deeply integrates data-driven adaptive control with physical models, providing a universal framework for parameter optimization under complex hydrodynamic conditions.

[0061] (3) By integrating wireless pressure sensors with convolutional neural network algorithms, real-time automatic recognition of sediment particle motion patterns is achieved. This multimodal fusion analysis, combined with high-frequency image data and mechanical signals, enhances the ability to model the correlation between particle motion trajectories and turbulent energy dissipation, laying a data foundation for the dynamic prediction of refined initiation thresholds.

[0062] (4) By embedding the optimized mathematical model of sediment initiation into the control system of river dredging projects, a cross-scale transformation from theoretical models to engineering practice was achieved. This technology breaks through the barriers between laboratory conditions and engineering scenarios, enabling sediment initiation prediction to drive the adaptive control of dredging equipment in real time, significantly improving project efficiency and reducing the risk of environmental disturbances, embodying a closed-loop innovation path from basic research to industrial implementation. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] In order to make the above-mentioned test method and device for starting the bed load particles of the present invention more obvious and easy to understand, the following will briefly introduce the drawings required for use in the specific embodiments of the present invention. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative work.

[0064] Figure 1 A schematic diagram showing the technical route of this application;

[0065] Figure 2 Diagram showing the components of the experimental device for bedload particle initiation (1-wave flume; 2-laser beam; 3-sediment particle; 4-high-speed camera array, 5-water-filled tetrahedron, 6-USB data cable, 7-computer; 8-wave current);

[0066] Figure 3 Schematic diagram showing the installation position of the glass ball and pressure sensor;

[0067] Figure 4 It shows the front view of the sink arrangement;

[0068] Figure 5 Schematic diagram showing the arrangement of high-speed camera array;

[0069] Figure 6 Schematic diagram showing different buried depths of particles;

[0070] Figure 7 Schematic diagram of the X–Z plane quadrant division. DETAILED DESCRIPTION

[0071] Example 1:

[0072] Provides a test method and device for starting bed load particles, the technical route of the method is as follows Figure 1 The device is composed of Figure 2As shown, it specifically includes: a wave tank (1), a laser beam (2), sediment particles (3), a high-speed camera array (4), a water-filled tetrahedron (5), a USB data line (6), an electronic computer (7) and a wave water flow (8); the sediment particles (3) are placed in the middle section of the wave tank (1), the laser beam (2) is vertically shot at the sediment particles (3) and is perpendicular to the direction of the wave water flow (8), the high-speed camera array (4) is connected to the electronic computer (7) via the USB data line (6), and is used to synchronously collect the displacement and flow field data of the sediment particles (3); the water-filled tetrahedron (5) is arranged on the side wall of the wave tank (1) to reduce the interference of laser reflection. The details are as follows.

[0073] The installation position of glass ball and pressure sensor is as follows Figure 3 As shown, the mud and sand particles (3) are a mixture of hollow glass balls and quartz sand, wherein a pressure sensor is embedded in the surface of the hollow glass ball, and the sensor circuit is hidden inside the glass ball and is connected to the data acquisition instrument through the side wall of the water tank.

[0074] Quartz sand with a particle size of 8-25mm and hollow glass balls were selected as experimental materials. Quartz sand was laid in the front and back halves of the experimental area and at the same height as the glass balls. Glass balls were laid in 3 layers in the experimental area and used as model materials for the sediment initiation experiment. The glass balls in the experimental observation area were specially prepared with a density of about .like Figure 3 As shown, the surface of the glass ball at the pressure measurement positions P1, P2, P3 and P4 is marked, and the pressure sensor is installed after special production; there are three glass balls of the same size below the glass ball for starting experiment observation. Figure 3 Arrange the three glass balls where the pressure sensors are to be installed so that they are completely hidden inside the glass balls; the four pressure sensors are as follows: Figure 3 As shown, the dotted line is the sensor line path.

[0075] The front view of the sink layout is as follows Figure 4 As shown, through flow control and cross-sectional velocity measurement, the average flow velocity and its distribution along the water depth were obtained. Bed shear stress was calculated based on the logarithmic velocity curve along the water depth and compared with the critical shear stress of sediment particles based on the Shields number curve. By repeatedly adjusting the water flow rate and combining experimental observations, a flow rate at which the bed shear stress was less than the critical shear stress but the glass balls partially started to move was obtained. A flow rate of this velocity or less was considered the turbulent flow and bed condition before this experiment.

[0076] The high-speed camera array is arranged as follows Figure 5As shown, the high-speed camera array (4) includes three high-speed cameras, which are arranged on the side and above the water tank, respectively. Two of the cameras are at a non-perpendicular angle to the water tank and are used for three-dimensional PIV flow velocity measurement, and the other camera is used for real-time monitoring of the spatial displacement of sediment particles (3).

[0077] Two high-speed cameras were placed on the side of the flume to measure the instantaneous flow velocity in the x, y, and z directions within the plane. The x direction corresponds to the direction of water flow, the y direction is coplanar and perpendicular to the x direction, and the z direction is perpendicular to the direction of gravity. A high-speed camera was also placed on the side of the flume to monitor the spatial position of the glass ball over time. A computer controlled the high-speed cameras to perform synchronous high-frequency monitoring.

[0078] Tracer particles in the flow field emit fluorescence from a laser beam emitted from above the flume. Images are captured by two high-speed cameras on the sides of the flume. The PIV system's built-in software then calculates and displays the instantaneous flow velocity near the sediment particles. The two cameras are positioned at an angle to the flume, and a water-filled tetrahedron glass is installed perpendicular to the flume's side wall to minimize the effects of refraction.

[0079] The displacement image of the glass ball is collected by a high-speed camera at a perpendicular angle to the water tank. During the collection process, the collection start time and collection frequency are synchronously controlled by the software. Figure 6 The three particle arrangements shown were tested separately.

[0080] According to turbulence theory, the velocity and pressure at any point in the flow field show irregular pulsations over time, which is the sum of the time-averaged velocity and the pulsating velocity. The calculation formula is:

[0081]

[0082] Where, is the instantaneous flow velocity; is the average flow velocity; Pulsating flow rate.

[0083] The characteristics of turbulent burst processes are studied through the time series of turbulent instantaneous velocity. The instantaneous velocity data at different depths of the boundary layer are obtained by three-dimensional PIV velocity measurement and plotted on a graph with time as the horizontal axis and instantaneous velocity as the vertical axis. According to the properties of the turbulent coherent structure in time and space, the intensity and time of outward convection, injection, inward convection and sweeping effects, the characteristics of pulsating velocity and instantaneous Reynolds stress are obtained. The velocity variance is expressed as the ratio of the time-averaged value of the product of the pulsating velocity to the square of the friction velocity. The calculation formulas for each combination mode are respectively 、 、 、 、 and ,in It represents the spatial average, and the horizontal axis is expressed as the ratio of the height above the bed surface to the sediment particle size.

[0084] By calculating the Reynolds stress and combining it with the quadrant analysis method to divide the contribution of the turbulent burst process to the drag and lift of sediment particles, a statistical method is used to quantify the influence of each quadrant on the average Reynolds stress.

[0085] According to the momentum theorem and Newton's second law, the time-averaged Reynolds stress is calculated by the pulsating flow velocity of the flow field. The calculation formula of the Reynolds stress is:

[0086]

[0087] Where, is the Reynolds stress; is the fluid density; It is the time-averaged product of turbulent pulsating flow velocity, which represents the intensity of turbulent bursts.

[0088] According to the quadrant analysis, the contribution of Reynolds stress to the drag and lift of sediment particles during turbulent bursts is divided by the following formula:

[0089]

[0090] Where, is the contribution of the turbulence burst process to the average Reynolds stress; For time; is the pulsating flow velocity in the horizontal direction of the water flow at time t; is the pulsating flow velocity in the vertical direction of the water flow at time t; is the pulsating flow velocity in the horizontal direction of the water flow; is the pulsating flow velocity in the vertical direction of the water flow; is the behavior index of the turbulence burst process at the i-th sampling point.

[0091] Figure 7 In quadrant I, outward convection occurs. ; Quadrant II is injection, , Quadrant III is inward convection, ; Quadrant IV is sweeping,

[0092] .

[0093] The mathematical model of sediment particle initiation is based on the correlation between the pressure energy dissipation rate and the turbulent kinetic energy spectrum, and by synchronously collecting pressure sensor data and particle displacement data, a nonlinear relationship between the pressure energy fluctuation variance, skewness and kurtosis and the particle motion trajectory is established.

[0094] The turbulence intensity is quantified by the root mean square of velocity fluctuations in all directions, providing basic data for analyzing turbulent kinetic energy dissipation and pressure fluctuations. The calculation formula for turbulence intensity in different directions is:

[0095]

[0096] Where, is the turbulence intensity in the direction of water flow; is the horizontal turbulence intensity; is the turbulence intensity in the vertical direction; is the total number of samples, that is, the number of samples in the time series; is the instantaneous water flow velocity component at the i-th sampling point; is the instantaneous horizontal lateral velocity component of the i-th sampling point; is the instantaneous vertical velocity component of the i-th sampling point; Indicates the average flow velocity in the direction of water flow; It represents the average flow velocity in the horizontal direction; It represents the average flow velocity in the vertical direction.

[0097] During the experiment, pressure measurements were obtained using a pressure sensor on the surface of the glass ball, with the sampling frequency consistent with the PIV velocity measurement sampling frequency. The temporal characteristics of the drag force acting on the surface of the glass ball are represented by the drag force spectrum, which is calculated as follows:

[0098]

[0099] Where, is the drag force spectrum output; is the drag coefficient caused by flow velocity; is the area of the glass ball exposed to the pulsating flow area; is the sampling frequency; is the absolute value or modulus of a complex number; is the dimensionless frequency function of the drag force caused by the flow velocity; is the flow velocity spectrum; is the pressure coefficient of the drag force; is the surface area of the glass ball's pressure fluctuation zone; is the dimensionless frequency function of the drag force caused by pressure; is the fluid pressure spectrum.

[0100] A mathematical model of pressure energy dissipation rate and sediment displacement is constructed by monitoring the pressure at four points on the surface of sediment particles.

[0101] Using synchronously collected photos, the spatial displacement data of the glass ball over time is obtained, providing a basis for establishing a mathematical model related to the movement of the glass ball.

[0102] The pressure energy spectrum and turbulent kinetic energy spectrum are plotted on the same graph with frequency as the horizontal axis. Based on the changing trends of the two spectra, the response of pressure energy to the dissipation of turbulent kinetic energy is analyzed. This is then compared with the time series of the spatial position of the glass ball to obtain the regular characteristics of the relationship between the ball's motion position and the change in pressure energy.

[0103] The instantaneous pressure at any point is decomposed into the sum of the time-averaged pressure and the fluctuating pressure by statistical analysis method. The calculation formula is:

[0104]

[0105] Where, is the sum of the time-averaged pressure and the fluctuating pressure; is the average pressure value at any point in time; is the fluctuating pressure at the i-th sampling point.

[0106] The calculation formula for pressure fluctuation variance is:

[0107]

[0108] Where, is the probability density function of pressure difference fluctuation; is the fluctuating pressure; is the total number of samples, that is, the number of samples in the time series.

[0109] The formula for calculating the deflection of pressure is:

[0110]

[0111] Where, is the skewness of pressure.

[0112] The formula for calculating the kurtosis of pressure is:

[0113]

[0114] Where, is the kurtosis of pressure.

[0115] The obtained pressure fluctuation parameters were plotted on the x, y, and z coordinate systems, and the effects of drag and lift were divided according to the horizontal and vertical directions. Furthermore, the temporal and spatial positional distribution curves of the instantaneous pressure were compared. Combined with the temporal and spatial displacement functions of the glass ball, the characteristics of the pressure change with the distance and direction of the glass ball's movement were analyzed. Finally, the dissipation of turbulent kinetic energy and pressure energy was incorporated into the mathematical model of the glass ball's displacement.

[0116] Focusing on the aforementioned turbulent burst and boundary layer breakdown processes, the time series of outward convection, jet, inward convection, and sweeping of the turbulent burst were characterized. A statistical analysis of the turbulent kinetic energy based on the pulsating velocity and the pressure energy dissipation based on pressure fluctuations was conducted by considering the coherent structure scale and its motion, momentum transfer, and turbulent kinetic and pressure energy dissipation. Correlation studies were conducted using established mathematical models, linking the temporal and spatial displacements of the glass sphere.

[0117] Using models of drag and lift caused by turbulent pulsating flow velocity and pressure fluctuations, and based on the energy required for glass sphere displacement changes and the dissipation of turbulent kinetic and pressure energies, mathematical models for the initiation of bedload sediment particles, namely sliding, rolling, and saltation, were constructed. Using the open-source OpenFOAM software, the experimentally derived bedload sediment initiation model was incorporated into the numerical model, and a fluid-structure coupling model code was developed for direct numerical simulation. The accuracy of the numerical model was verified by comparing the calculated results with experimental data. Hydrodynamic parameters such as water depth and flow velocity, as well as parameters such as sediment particle size, particle shape, and bed roughness, were adjusted to simulate a wider range of experimental conditions at low Reynolds numbers, expand experimental data, and refine the mathematical model.

[0118] Example 2:

[0119] Building on the previous examples, a real-time monitoring method for sediment particle microscopic strain based on fiber Bragg gratings (FBGs) has been added. Fiber Bragg grating sensors precisely measure microscopic strain by shifting the reflected wavelength caused by changes in the grating period. When sediment particles are displaced or deformed by fluid, the FBG sensor embedded within the particle uses the wavelength shift to infer the dynamic stress distribution on the particle surface. This wavelength shift is then combined with the spatiotemporal characteristics of the turbulent burst process to quantify the transient response mechanism of the particle to stress.

[0120] An FBG sensor array is embedded inside the sediment particles, with a grating spacing of 1 mm and a wavelength resolution of 0.1 nm. The sensor circuit is led out through the microchannel inside the sediment particles to avoid interfering with the flow field.

[0121] The FBG demodulator is connected to an electronic computer to analyze the wavelength shift data in real time and convert it into particle surface strain values. The sampling frequency of the FBG demodulator is 10 kHz.

[0122] A strain-displacement correlation model is established by combining high-speed camera displacement data to define the critical strain threshold. , when the local strain exceeds , it is determined that the particle starts. The starting criterion formula is:

[0123]

[0124] Where, For effective shear stress, the effects of turbulent pulsation, strain energy and turbulent kinetic energy dissipation are integrated; is the fluid density, reflecting the contribution of fluid inertia to shear stress; is the Reynolds stress, which is the time-averaged product of turbulent pulsating velocity and represents the intensity of turbulent bursts; is the strain coupling coefficient, which is determined to be 0.3 through FBG calibration experiments; It is the real-time strain value, reflecting the local stress deformation; is the critical strain threshold, which is determined by the elastic modulus of the material and the particle size.

[0125] Verification shows that while achieving an average error similar to that of the above-mentioned embodiment, the time resolution of particle initiation detection is improved to 10μs, and the error rate is reduced to 2%. The results show that this method can achieve non-invasive, high-precision monitoring of the micromechanical behavior of sediment particles, effectively capturing transient stress fluctuations on the particle surface during turbulent bursts. Compared with traditional external sensors, the experimental results show that the spatiotemporal resolution of particle forces is significantly improved, and flow field interference is greatly suppressed. Through the correlation analysis of dynamic strain and displacement, the device reveals the coupling mechanism of strain accumulation and energy dissipation during particle initiation, providing a more sensitive physical basis for determining critical initiation conditions. It further verifies the technology's ability to analyze the nonlinear mechanical response of particles in complex turbulent environments, enabling the model to more accurately characterize the impact of actual hydrodynamic conditions on sediment movement.

[0126] Example 3:

[0127] Building on the previous examples, a multi-parameter adaptive optimization control method based on reinforcement learning was added. A deep reinforcement learning algorithm was used to construct an interactive framework between the intelligent agent and the environment. Parameters such as water flow rate and wave frequency were dynamically optimized through a reward function, enabling the experiment to quickly converge to the target starting conditions. The agent then explored the parameter space using a Q-learning strategy to maximize the agreement between experimental data and CFD simulations.

[0128] Deploy the DRL module in an electronic computer, and the input state space includes real-time flow rate , turbulence intensity , particle displacement , the action space is the flow valve opening and wave generator frequency .

[0129] The reward function is defined as:

[0130]

[0131] Where, is the output of the reward function; || || is the Euclidean norm, which is used to quantify the absolute value of the deviation; is the particle displacement value; is the CFD predicted value of particle displacement; is the weight coefficient, which is an adjustment factor used to balance the displacement error and stress error; is the effective shear stress value; is the shear stress of the theoretical model.

[0132] The optimal control strategy is generated through offline training, and the experimental parameters are adjusted online in real time.

[0133] Assume that the multi-parameter adaptive optimization control method based on reinforcement learning is used to dynamically control the flow velocity and wave parameters of the bed load sediment start-up experiment. Optimize water flow rate under strong disturbance conditions and wave frequency , causing the sediment particles to move Compared with CFD simulation values Minimize the mean square error.

[0134] Assume initial flow rate , initial wave frequency , target critical flow rate ;

[0135] The DRL agent explores the parameter space through the Q-learning strategy and adjusts the flow valve opening and wave frequency , and updated the strategy according to the reward function. After 72 minutes of online optimization, it converged to the optimal parameter combination.

[0136] The effects of the multi-parameter adaptive optimization control method based on reinforcement learning are shown in Table 1.

[0137] Table 1. Summary of the effects of multi-parameter adaptive optimization control methods based on reinforcement learning

[0138]

[0139] According to the experimental table, the average optimization time of the DRL system is 71 minutes, which is 40.8% shorter than that of traditional manual adjustment. The flow valve opening control error is ±0.5%, and the wave frequency adjustment error is ±0.05 Hz. The adaptive formula calculates the critical flow velocity with an average prediction error of 2.5%. The particle displacement error is reduced from 0.45 mm² to 0.12 mm², a reduction of 73.3%. Under these conditions, the optimized mean velocity was 1.06 m / s, with a deviation of only 0.95% from the theoretical value. The results show that this method achieves efficient and autonomous optimization of experimental parameters through the dynamic interaction of the intelligent agent and the flow field environment. It demonstrates the ability to rapidly converge to the target experimental conditions under strong turbulence and unsteady conditions, significantly improving the robustness of parameter adjustment. During the experiment, the optimized flow velocity and wave frequency combination was highly consistent with the theoretical model predictions, verifying the adaptability of the data-driven strategy to multi-physics coupling problems. Furthermore, the device effectively bridges the scale gap between experimental observations and numerical simulations through a real-time feedback mechanism, providing reliable technical support for model generalization in complex hydrodynamic scenarios.

[0140] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Thus, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable non-transitory storage media containing computer-usable program code.

[0141] The present invention can provide computer program instructions to a management platform of a general-purpose computer, a special-purpose computer, an embedded processor or other programmable data processing device to produce a machine, so that the instructions executed by the management platform of the computer or other programmable data processing device produce a device for implementing the system.

[0142] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture including an instruction device that implements the functions of the system.

[0143] These computer program instructions may also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, whereby the instructions executed on the computer or other programmable device provide steps for implementing the functions of the described system.

Claims

1. A test method for starting bed load particles, characterized in that: include: Sediment particles are laid in a wave tank, and the water flow rate is adjusted by flow control until the bed shear stress is less than the critical shear stress; The laser beam is used to excite the fluorescence of the tracer particles, and the camera array is used to synchronously collect the instantaneous flow velocity and sediment particle displacement data of the flow field; Based on the analysis of the spatiotemporal distribution characteristics of turbulent kinetic energy dissipation and pressure energy dissipation during turbulent bursts, a mathematical model for sediment particle initiation was established. This mathematical model is based on the correlation between the turbulent kinetic energy spectrum and the pressure energy dissipation rate. By synchronously collecting pressure sensor data and particle displacement data, a nonlinear relationship between the pressure energy fluctuation variance, skewness, and kurtosis and the particle motion trajectory was established. Combine experimental data with computational fluid dynamics numerical simulations to verify and optimize the mathematical model.

2. The method according to claim 1, wherein: By calculating the Reynolds stress and combining it with the quadrant analysis method to divide the contribution of the turbulent burst process to the drag and lift of sediment particles, a statistical method is used to quantify the influence of each quadrant on the average Reynolds stress.

3. The method according to claim 1, wherein: Through fluid-structure coupling numerical simulation and adjustment of bed roughness and particle shape parameters, the model is extended to low Reynolds number conditions to generate a general prediction model suitable for complex hydrodynamic conditions.

4. The method according to claim 1, wherein: The method also includes a real-time monitoring method for the microscopic strain of sediment particles based on fiber Bragg gratings. The microscopic strain is measured by the reflection wavelength shift caused by the change of the grating period. The dynamic stress distribution on the particle surface is inverted by the wavelength shift. The transient response mechanism of the particle stress is quantified by combining the spatiotemporal characteristics of the turbulent burst process. The starting criterion formula is: Where, is the effective shear stress; is the fluid density; is the time-averaged product of turbulent pulsating flow velocity; is the strain coupling coefficient; is the real-time strain value; is the critical strain threshold.

5. The method according to claim 1, wherein: It also includes a multi-parameter adaptive optimization control method based on reinforcement learning. It uses a deep reinforcement learning algorithm to build an interaction framework between the intelligent agent and the environment. The parameters are dynamically optimized through a reward function. The agent explores the parameter space through a Q-learning strategy. The reward function is defined as: Where, is the reward function output; || || is the Euclidean norm; is the particle displacement value; is the CFD predicted value of particle displacement; is the weight coefficient; is the effective shear stress value; is the shear stress of the theoretical model.

6. A test device for bed load particle initiation, characterized by: When the device is running, it executes the method according to any one of claims 1 to 5: The device comprises: A wave tank (1), a laser beam (2), sediment particles (3), a high-speed camera array (4), a water-filled tetrahedron (5), a USB data cable (6), an electronic computer (7), and a wave water flow (8); the sediment particles (3) are placed in the middle section of the wave tank (1), the laser beam (2) is vertically directed toward the sediment particles (3) and is perpendicular to the direction of the wave water flow (8), the high-speed camera array (4) is connected to the electronic computer (7) via the USB data cable (6), and the water-filled tetrahedron (5) is arranged on the side wall of the wave tank (1).

7. The device according to claim 6, characterized in that: The mud and sand particles (3) are a mixture of hollow glass balls and quartz sand, wherein a pressure sensor is embedded in the surface of the hollow glass ball, and the sensor circuit is hidden inside the glass ball and is connected to the data acquisition instrument through the side wall of the water tank.

8. The device according to claim 6, characterized in that: The high-speed camera array (4) includes three high-speed cameras, which are respectively arranged on the side and above the water tank. Two of the cameras are at a non-perpendicular angle to the water tank and are used for three-dimensional PIV flow velocity measurement, and the other camera is used for real-time monitoring of the spatial displacement of sediment particles (3).

9. A computer-readable storage medium storing a computer program, wherein: When the computer program is executed, the method according to any one of claims 1 to 5 is implemented.

Images