A method, equipment and medium for visual measurement of abrasive waterjet flow field

Abstract

This application discloses a method, device, and medium for visually measuring the flow field of an abrasive waterjet, relating to the field of fluid measurement technology. The method includes: acquiring first and second tracer particles, representing the characteristics of the jet fluid and abrasive particles respectively, and injecting them into a stable jet to obtain the measured distribution of the instantaneous velocity vectors of the jet fluid and abrasive particles across the entire field; constructing a visual prediction model of the abrasive waterjet flow field containing trainable internal parameters; predicting the flow field based on the initial velocities of the jet fluid and abrasive particles to obtain the corresponding predicted distribution of the instantaneous velocity vectors across the entire field; updating the model's internal parameters through backpropagation based on the difference between the measured and predicted distributions to obtain a trained prediction model; and visually predicting the abrasive waterjet flow field to be predicted based on the trained prediction model. This enables low-cost, non-contact, and rapid analysis of the internal structure and dynamic characteristics of the flow field.

Description

Technical Field

[0001] This application relates to the field of fluid measurement technology, and in particular to a method, equipment and medium for visual measurement of abrasive waterjet flow field. Background Technology

[0002] Ultra-high pressure abrasive waterjet technology, as an advanced cold processing technique, fundamentally depends on the complex gas-liquid-solid three-phase transient flow characteristics within the jet for its material removal mechanism, energy transfer efficiency, and final processing quality. During the formation, development, and impact of the jet on the target, intense turbulent mixing, interphase slip, and energy dissipation occur. Therefore, achieving high spatiotemporal resolution, non-invasive, and separate quantitative measurement of its core flow field, especially the motion characteristics of the water and abrasive phases, is fundamental to revealing its processing mechanism, establishing accurate process models, and ultimately realizing intelligent dynamic parameter control. However, the extremely high velocity of the jet, the ultra-high internal pressure, the strong scattering and obstruction of light by abrasive particles, and the design of a safe and reliable optical observation window under extreme conditions constitute multiple severe technical challenges for flow field visualization. Overcoming these challenges and achieving direct observation of the dynamic characteristics inside the "black box" of ultra-high pressure abrasive waterjet has significant scientific and engineering implications.

[0003] Before the maturity of optical measurement technology, the relevant technology mainly relied on measuring the jet impact force to inversely deduce its average velocity, which is a classic and simple method. That is, by establishing the mechanical relationship between the momentum generated by the jet impacting the target surface and the average velocity of the jet, a precise force measuring device is used to measure the total impact momentum of the abrasive water jet at the nozzle exit. By combining the continuity equation and the momentum conservation equation, the average velocities of the solid, liquid and gas phases at the exit section are derived in reverse. The system includes a thin plate sample, a jetting system, a Kistler 9257A three-component dynamometer, a signal amplifier, an analog-to-digital converter and LabVIEW data acquisition software, which can quantitatively reveal the three-phase average velocity of the abrasive water jet at different nozzle distances.

[0004] With the maturity of optical measurement technology, non-invasive optical measurement techniques have become mainstream for directly observing the internal structure of flow fields. Among them, particle image velocimetry (PIV) acquires the velocity field by tracking the displacement of tracer particles, while laser-induced fluorescence (LIF) uses the fluorescence signal of excited specific substances to identify specific phases or components. Combining the two provides a possible technical path for separating and measuring the velocities of each phase in multiphase flow. Summary of the Invention

[0005] This application provides a method, device, and medium for visually measuring the flow field of abrasive waterjet, in order to solve the following technical problem: how to achieve low-cost, non-contact, and rapid analysis of the internal structure and dynamic characteristics of the flow field.

[0006] In a first aspect, embodiments of this application provide a method for visually measuring the flow field of an abrasive waterjet, the method comprising: A first tracer particle and a second tracer particle are obtained. The first tracer particle is used to characterize the motion characteristics of the jet fluid, and the second tracer particle is used to characterize the motion characteristics of the abrasive particles. The first tracer particle and the second tracer particle were injected into the abrasive water jet under stable working conditions, and the measured distribution of the instantaneous velocity vector of the jet fluid and the measured distribution of the instantaneous velocity vector of the abrasive particles were obtained. The abrasive water jet flow field visualization prediction model is used to predict the abrasive water jet flow field based on the first initial velocity of the jet fluid and the second initial velocity of the abrasive particles, thereby obtaining the predicted distribution of the instantaneous velocity vector of the jet fluid and the predicted distribution of the instantaneous velocity vector of the abrasive particles. Determine the first difference between the measured distribution of the instantaneous velocity vector of the jet fluid and the predicted distribution of the instantaneous velocity vector of the jet fluid, and determine the second difference between the measured distribution of the instantaneous velocity vector of the abrasive particles and the predicted distribution of the instantaneous velocity vector of the abrasive particles. Based on the first difference and the second difference, the internal parameters of the abrasive waterjet flow field visualization prediction model are updated to obtain a trained abrasive waterjet flow field visualization prediction model, which is then used to perform visualization prediction of the abrasive waterjet flow field to be predicted.

[0007] Secondly, embodiments of this application also provide an abrasive waterjet flow field visualization measurement device, the device comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform an abrasive waterjet flow field visualization measurement method as described above.

[0008] Thirdly, embodiments of this application also provide a computer storage medium storing computer-executable instructions, which, when executed, implement a visual measurement method for abrasive waterjet flow field as described above.

[0009] The present application provides a method, device, and medium for visually measuring the flow field of an abrasive waterjet, which has the following beneficial effects: First, by introducing first and second tracer particles, respectively characterizing the motion properties of the jet fluid and abrasive particles, the motion differences between the two phases (jet fluid and abrasive particles) can be specifically distinguished, avoiding the flow field aliasing error caused by traditional single tracer particles, thus laying a data foundation for subsequent high-precision measurements. Next, the two types of tracer particles are injected into the abrasive water jet under stable operating conditions, and their measured distributions are obtained. This effectively avoids signal crosstalk between the two phases in a high-speed mixed environment, achieving high-fidelity decoupled measurement of the continuous and discrete phase flow fields. Subsequently, based on the initial velocities of the jet fluid and abrasive particles, the flow field is predicted using a visual prediction model of the abrasive water jet flow field, outputting the full-field instantaneous velocity vector prediction distribution of the two-phase flow. This allows for the prediction of the evolution from initial conditions to the full-field characteristics. This approach employs a physical-driven prediction method to avoid the problems of poor generalization and lack of physical meaning in complex multiphase flow scenarios caused by purely data-driven models. Then, the first and second differences between the measured and predicted distributions of the jet fluid and abrasive particles are calculated separately. This allows for the quantitative separation of flow field prediction deviations for the continuous and discrete phases, accurately pinpointing the model's inadequate fitting of different phase flows. This method is more targeted than a single overall difference assessment. Finally, the internal parameters of the model are jointly updated based on these dual differences, enabling the trained model to simultaneously adapt to the motion laws of both the jet fluid and abrasive particles. This achieves high-precision, highly physically reliable, and visualized prediction of the abrasive waterjet flow field, providing reliable flow field evolution data support for optimizing abrasive waterjet processing parameters and improving nozzle structures. Attached Figure Description

[0010] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 A flowchart illustrating a method for visualizing and measuring the flow field of an abrasive waterjet, as provided in this application embodiment; Figure 2 This is a schematic diagram of the structure of the PIV speed measurement system provided in the embodiments of this application; Figure 3 This is a schematic diagram of the PIV speed measurement principle provided in the embodiments of this application; Figure 4 This is a schematic diagram of the internal structure of an abrasive waterjet flow field visualization measurement device provided in an embodiment of this application. Detailed Implementation

[0011] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions of this application will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0012] It is understood that in the embodiments of this application, data related to user information (such as user accounts) is involved. When the embodiments of this application are applied to specific products or technologies, user permission or consent is required, and the collection, use and processing of related data must comply with relevant laws, regulations and standards.

[0013] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing embodiments of this application only and is not intended to limit this application.

[0014] In the following description, the terms “first, second, ...” are used merely to distinguish similar objects and do not represent a specific ordering of objects. It is understood that “first, second, ...” may be interchanged in a specific order or sequence where permitted, so that the embodiments of this application described herein can be implemented in an order other than that illustrated or described herein.

[0015] Ultra-high pressure abrasive waterjet technology, as an advanced cold processing technique, fundamentally depends on the complex gas-liquid-solid three-phase transient flow characteristics within the jet for its material removal mechanism, energy transfer efficiency, and final processing quality. During the formation, development, and impact of the jet on the target, intense turbulent mixing, interphase slip, and energy dissipation occur. Therefore, achieving high spatiotemporal resolution, non-invasive, and separate quantitative measurement of its core flow field, especially the motion characteristics of the water and abrasive phases, is fundamental to revealing its processing mechanism, establishing accurate process models, and ultimately realizing intelligent dynamic parameter control. However, the extremely high velocity of the jet, the ultra-high internal pressure, the strong scattering and obstruction of light by abrasive particles, and the design of a safe and reliable optical observation window under extreme conditions constitute multiple severe technical challenges for flow field visualization. Overcoming these challenges and achieving direct observation of the dynamic characteristics inside the "black box" of ultra-high pressure abrasive waterjet has significant scientific and engineering implications.

[0016] Before the maturity of optical measurement technology, the relevant technology mainly relied on measuring the jet impact force to inversely deduce its average velocity, which is a classic and simple method. That is, by establishing the mechanical relationship between the momentum generated by the jet impacting the target surface and the average velocity of the jet, a precise force measuring device is used to measure the total impact momentum of the abrasive water jet at the nozzle exit. By combining the continuity equation and the momentum conservation equation, the average velocities of the solid, liquid and gas phases at the exit section are derived in reverse. The system includes a thin plate sample, a jetting system, a Kistler 9257A three-component dynamometer, a signal amplifier, an analog-to-digital converter and LabVIEW data acquisition software, which can quantitatively reveal the three-phase average velocity of the abrasive water jet at different nozzle distances.

[0017] With the maturity of optical measurement technology, non-invasive optical measurement techniques have become mainstream for directly observing the internal structure of flow fields. Among them, particle image velocimetry (PIV) acquires the velocity field by tracking the displacement of tracer particles, while laser-induced fluorescence (LIF) uses the fluorescence signal of excited specific substances to identify specific phases or components. Combining the two provides a possible technical path for separating and measuring the velocities of each phase in multiphase flow.

[0018] In the velocity measurement of ultra-high pressure abrasive water jets, although the force measurement method is intuitive in principle, it has inherent defects that are difficult to overcome. This method can only obtain the overall average force of the jet impacting the target and inversely deduce the average velocity of a single cross section. It cannot analyze the velocity distribution along the axial direction of the flow field at the same moment, nor can it distinguish the velocity difference between the core region and the edge region. Moreover, as a contact measurement, the intervention of the measuring device itself will disturb or even destroy the original flow field, making the measured "impact force" no longer the true state of the free jet, thus introducing systematic errors. In addition, the surface morphology of the force measurement target is easily changed by the severe erosion of ultra-high pressure abrasive jets, which seriously threatens the stability and accuracy of long-term measurements. To build a sensor platform that simultaneously meets the requirements of ultra-high erosion resistance and high-precision force measurement, there are huge challenges in terms of mechanical structure design and manufacturing cost.

[0019] On the other hand, while combining optical particle tracking velocimetry (PTV) with LIF technology can acquire the trajectory of a single tracer particle, it also faces significant limitations in practical applications. Compared to PIV, PTV requires tracer particles to be sparsely distributed within the field of view, which makes it difficult to obtain continuous and complete velocity data of the entire flow field, and it is prone to recognition errors when particle trajectories intersect. PTV requires the laser sheet to be thin enough, otherwise particles will overlap in the depth direction, producing false high or low speeds. Furthermore, the algorithm of this technology is complex, the data processing efficiency is low, and the degree of software automation is far inferior to that of mature particle image velocimetry technology.

[0020] Based on this, this application provides a visualization measurement method for abrasive waterjet flow field, which can achieve low-cost, non-contact, and rapid analysis of the internal structure and dynamic characteristics of the flow field.

[0021] The technical solutions proposed in the embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0022] Figure 1 This application provides a flowchart of a method for visualizing and measuring the flow field of an abrasive waterjet. This method can be applied to various fluid measurement scenarios. For example, in the online optimization of machining parameters in aerospace manufacturing, a 300MPa ultra-high pressure device is used in the laboratory to inject dual tracer particles, collect measured flow field data under different pressures, train a prediction model, and when a new batch of titanium alloy is being processed on a production line, the system only needs to read the currently set working pressure (e.g., 280MPa) and nozzle orifice diameter. Through model prediction, a full-field velocity cloud map of the abrasive particles under the current parameters can be obtained. Based on the velocity cloud map, problems can be identified and adjustments made in a timely manner without stopping the machine for PIV experiments. This allows for real-time prediction of flow field quality, ensuring machining accuracy. In the adaptive operation scenario of a nuclear power plant pipeline rust removal robot, the robot... Using a trained prediction model, based on the current nozzle height and pump pressure, the robot predicts the slip velocity distribution of the jet fluid and abrasive particles under the current posture. If the prediction indicates insufficient impact velocity of the abrasive against the pipe wall, the robot automatically reduces its movement speed to ensure thorough rust removal. This effectively solves the problem of not being able to install high-speed cameras and lasers in confined spaces, enabling intelligent perception. In low-cost process development scenarios for composite materials, different abrasive particle sizes and fluid pressures can be input into the trained prediction model. The model can directly generate full-field instantaneous velocity vector maps under different ratios, allowing for more intuitive selection of parameters that concentrate the abrasive and increase its speed, thus significantly reducing the development cycle. Certain input parameters or intermediate results in the process can be manually adjusted to help improve accuracy.

[0023] This application provides a method for visualizing and measuring the flow field of abrasive waterjet. It should be noted that the execution entity in this embodiment can be a server or any terminal device with data processing capabilities. For example, the server can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms. The terminal device can be a smartphone, tablet, laptop, desktop computer, smart speaker, smartwatch, in-vehicle terminal, etc., but is not limited to these.

[0024] like Figure 1 As shown in the figure, the method for visualizing and measuring the flow field of an abrasive waterjet provided in this application embodiment specifically includes the following steps: Step 101: Obtain the first tracer particle and the second tracer particle.

[0025] Here, the first tracer particle is used to characterize the motion characteristics of the jet fluid, and the second tracer particle is used to characterize the running characteristics of the abrasive particles.

[0026] It should be noted that the first and second tracer particles can be obtained by purchasing or custom preparation, and no specific limitations are made here.

[0027] In some embodiments, the first tracer particle is a fluorescent particle with small diameter, light weight, and low density, which has good fluid following ability and can accurately reflect the velocity of the flow field.

[0028] Thus, by using fluorescent particles with small diameter, light weight, and low density as the first tracer particles, it is possible to ensure that the particles have excellent fluid following ability in ultra-high pressure jets, accurately capture and accurately reflect the microscopic motion trajectory of the water phase in the jet flow field, thereby solving the measurement distortion problem caused by inertial lag of traditional high-density particles. Compared with the traditional contact force measurement method, which can only obtain the cross-sectional average velocity, this method, combined with PIV technology, can realize non-invasive, high-fidelity measurement of the instantaneous velocity vector distribution of the jet fluid across the entire field. While eliminating the disturbance of the flow field by the physical probe, it can also avoid the erosion and wear of the measuring element by the high-pressure abrasive, thereby significantly improving the accuracy, spatial resolution, and long-term measurement stability of the flow field data acquisition.

[0029] As an example, the diameter of the first tracer particle can be 5-20 μm, and the density can be 0.8-1.2 g / cm³. 3(Approximately or slightly smaller than water), the material can be polyamide, polystyrene, or silica microspheres.

[0030] In some embodiments, the second tracer particle is a fluorescent particle having the same diameter and density as the abrasive particle.

[0031] Thus, by using fluorescent particles with the same diameter and density as real abrasive particles as the second tracer particles, it can be ensured that the kinematic characteristics of the tracer particles in the jet are completely consistent with those of the real abrasive particles, thereby truly reproducing the acceleration process, trajectory distribution, and energy transfer state of the abrasive particles. Compared with traditional surface coating technology, this solid doped structure can effectively avoid the risk of signal loss caused by the shedding of the fluorescent layer under ultra-high pressure and strong impact environments. Combined with PIV technology, it can achieve accurate and stable measurement of the instantaneous velocity vector of the abrasive phase across the entire field, providing reliable experimental evidence for revealing the dynamic behavior of the solid phase in the gas-liquid-solid three-phase flow field.

[0032] As an example, the diameter of the second tracer particle could be 180 μm, and the density could be 3.5-4.0 g / cm³. 3 (Same as garnet / corundum), it can be customized directly from special materials manufacturers, or you can directly purchase particles of the same material as real abrasives and use a fluidized bed coating machine to coat their surface with an extremely thin fluorescent polymer film.

[0033] Step 102: Inject the first tracer particle and the second tracer particle into the abrasive water jet under stable working conditions to obtain the measured distribution of the instantaneous velocity vector of the jet fluid and the measured distribution of the instantaneous velocity vector of the abrasive particles.

[0034] It should be noted that the measured instantaneous velocity vector distribution of the jet fluid refers to the velocity and direction of the fluid in each tiny region (pixel or query window) within the flow field region illuminated by the laser sheet; the measured instantaneous velocity vector distribution of the abrasive particles refers to the velocity and direction of the abrasive particles in each tiny region (pixel or query window) within the flow field region illuminated by the laser sheet.

[0035] In some embodiments, step 102 described above can be implemented as follows: injecting the first tracer particle into an abrasive water jet in a stable operating state, and irradiating the flow field region to be tested with laser light to obtain a fluorescence image of the first tracer particle; performing a first cross-correlation calculation on the fluorescence image of the first tracer particle to obtain the measured distribution of the instantaneous velocity vector of the jet fluid; injecting the second tracer particle into an abrasive water jet in a stable operating state, and irradiating the flow field region to be tested with laser light to obtain a fluorescence image of the second tracer particle; performing a second cross-correlation calculation on the fluorescence image of the second tracer particle to obtain the measured distribution of the instantaneous velocity vector of the abrasive particle.

[0036] Thus, by injecting the first and second tracer particles separately and performing laser irradiation and cross-correlation calculations respectively, independent and accurate measurement of the two-phase flow fields of the jet fluid and abrasive particles can be achieved. This effectively avoids the problem of kinetic characteristic confusion caused by the huge density difference between the two phases (jet fluid and abrasive particles) in traditional single-particle measurements. By combining the phase-separation measurement strategy with the cross-correlation algorithm, the measured distribution of the full-field instantaneous velocity vector with high spatiotemporal resolution can be calculated from the fluorescence image sequence. This not only accurately restores the high-speed flow characteristics of the water phase and the hysteresis acceleration trajectory of the abrasive phase, but also provides a high-fidelity reference data benchmark for subsequent visualization prediction models, ensuring the accuracy and reliability of model training.

[0037] It should be noted that the laser sheet light is shaped into an extremely thin, two-dimensional light plane by using cylindrical mirrors or Powell prisms; the cross-correlation solution is achieved through the PIV cross-correlation algorithm, which can statistically analyze the average motion of all particles within a region.

[0038] As an example, taking the optimization of waterjet cutting process for titanium alloy components in an aerospace manufacturing enterprise as an example, firstly, under the stable working condition of ultra-high pressure equipment in the laboratory, small-diameter, low-density fluorescent tracer particles are injected into a pure water jet. A laser sheet is then turned on to vertically irradiate the flow field area to be measured. A high-speed camera with a filter is used to collect particle fluorescence image pairs. The images are cross-correlated using PIV software, and the ratio of particle swarm displacement to time difference is calculated by dividing the query window, thus obtaining the measured instantaneous velocity vector distribution of the entire jet fluid field. Subsequently, the pipeline is cleaned, and fluorescent particles with the same diameter and density as the second tracer particles and the actual abrasive are injected into a stable abrasive water jet. Again, the jet is irradiated with laser light, and fluorescence images are collected. After independent cross-correlation processing, the measured instantaneous velocity vector distribution of the abrasive particles across the entire field is obtained. The two sets of high-precision measured distribution data obtained so far can intuitively reveal the velocity slip and energy transfer state between the water and solid phases, providing a reliable true value benchmark for subsequent training of a visualization prediction model, thereby ensuring the processing quality of complex components.

[0039] In some embodiments, the abrasive water jet includes multiple working pressures; step 102 described above can also be implemented in the following manner: for each working pressure, the following processing is performed respectively: the first tracer particle is injected into the abrasive water jet in a stable working state corresponding to the working pressure, and the flow field area to be tested is irradiated by a laser sheet to obtain a fluorescence image of the first tracer particle corresponding to the working pressure; a first cross-correlation calculation is performed on the fluorescence image of the first tracer particle corresponding to the working pressure to obtain the measured distribution of the instantaneous velocity vector of the jet fluid under the working pressure; the second tracer particle is injected into the abrasive water jet in a stable working state corresponding to the working pressure, and the flow field area to be tested is irradiated by a laser sheet to obtain a fluorescence image of the second tracer particle corresponding to the working pressure; a second cross-correlation calculation is performed on the fluorescence image of the second tracer particle corresponding to the working pressure to obtain the measured distribution of the instantaneous velocity vector of the abrasive particle under the working pressure.

[0040] Thus, by performing phase-separated injection, laser sheet illumination, and cross-correlation calculations under multiple working pressures, a systematic database of real-world instantaneous velocity vectors covering different working conditions can be constructed to effectively reveal the dynamic evolution of the velocity fields of the jet fluid and abrasive particles with pressure. Furthermore, this multi-pressure calibrated real-world data can provide rich and high-dimensional training samples for the visualization prediction model, significantly enhancing the model's ability to fit complex nonlinear mapping relationships of flow fields. It can also greatly improve the generalization accuracy and robustness of the trained model in visualization predictions under unknown working conditions.

[0041] As an example, taking the development of a multi-level intelligent processing system by an ultra-high pressure waterjet cutting equipment manufacturer as an example, experiments were conducted at multiple working pressures, including 150MPa, 200MPa, and 250MPa, as specified by the equipment. At each pressure, low-density first tracer particles were injected, and the flow field was irradiated with laser light to acquire fluorescence images. The instantaneous velocity vector distribution of the jet fluid at the current pressure was obtained through cross-correlation calculation (PIV cross-correlation algorithm). Subsequently, the pipeline was cleaned, and second tracer particles of the same specifications as the actual abrasive were injected. After sample irradiation and acquisition, independent cross-correlation calculation (PIV cross-correlation algorithm) is performed to obtain the measured distribution of the instantaneous velocity vector of abrasive particles in the whole field under the current pressure. Through the phase-separated measurement process of multiple pressures, a true database of flow field values ​​covering different energy levels can be established, so that the prediction model trained subsequently can accurately learn the dynamic influence of pressure changes on water phase acceleration and abrasive slip velocity. Thus, in the end application, when facing any set pressure, only the initial velocity needs to be input to visualize and predict the whole field distribution, which helps to quickly lock the optimal cutting parameters and avoid blind cutting trial and error.

[0042] Step 103: Using the abrasive water jet flow field visualization prediction model, based on the first initial velocity of the jet fluid and the second initial velocity of the abrasive particles, the abrasive water jet flow field is predicted to obtain the full-field instantaneous velocity vector prediction distribution of the jet fluid and the full-field instantaneous velocity vector prediction distribution of the abrasive particles.

[0043] It should be noted that the global instantaneous velocity vector prediction distribution of the jet fluid refers to the velocity and direction of the fluid in each tiny region (pixel or query window) of the test flow field region predicted by the abrasive water jet flow field visualization prediction model based on the initial velocity of the jet fluid; the global instantaneous velocity vector prediction distribution of the abrasive particles refers to the velocity and direction of the abrasive particles in each tiny region (pixel or query window) of the test flow field region predicted by the abrasive water jet flow field visualization prediction model based on the initial velocity of the abrasive particles.

[0044] In some embodiments, the step 103 above, which involves predicting the flow field of the abrasive water jet based on the first initial velocity of the jet fluid and the second initial velocity of the abrasive particles to obtain the predicted distribution of the instantaneous velocity vector of the jet fluid and the predicted distribution of the instantaneous velocity vector of the abrasive particles, can be achieved as follows: First, feature extraction is performed on the first initial velocity of the jet fluid to obtain a first feature; second, feature extraction is performed on the second initial velocity of the abrasive particles to obtain a second feature; the first feature and the second feature are fused to obtain a third feature; the third feature is then position-encoded to obtain a fourth feature; the fourth feature is then nonlinearly transformed to obtain a nonlinearly transformed fourth feature; and the nonlinearly transformed fourth feature is then decoded to obtain the predicted distribution of the instantaneous velocity vector of the jet fluid and the predicted distribution of the instantaneous velocity vector of the abrasive particles.

[0045] Thus, by extracting and fusing the initial velocities of the two phases (jet fluid and abrasive particles) separately, and by introducing spatial topological information through position encoding, the prediction model can simultaneously capture the dynamic correlation between the jet fluid and abrasive particles and the overall spatial evolution law, avoiding information silos in single-phase prediction. At the same time, by utilizing nonlinear transformation and decoding processing, the pressure and geometric physical constraints implicit in the model's internal parameters can be fully activated, enabling high-precision mapping from discrete initial conditions to a continuous high-dimensional vector field. This significantly improves the physical realism and structural detail restoration capabilities of the model's prediction of instantaneous velocity distribution across the entire field in complex multiphase flow scenarios.

[0046] As an example, taking the process optimization of a precision electronic glass cutting production line as an example, firstly, the initial velocity of the jet fluid is... With the second initial velocity of the abrasive particles Input the abrasive waterjet flow field visualization prediction model respectively. The model uses a fully connected layer to determine the first initial velocity. Second initial velocity Feature extraction is performed separately, transforming the features into high-dimensional feature vectors (first features) that contain dynamic properties. Second feature ); then, the model will use the first feature Second feature splicing and fusion at the channel dimension (third feature) ), and superimposed the position encoding of the flow field spatial coordinates (fourth feature) This endows the model with spatial awareness; subsequently, the fourth feature... The fluid flows through multiple nonlinear transformation layers (such as ReLU-activated convolutional layers), and the coupled acceleration process between the fluid and the abrasive is simulated using learned intrinsic parameters to obtain the fourth feature after nonlinear transformation. Finally, the fourth feature after nonlinear transformation. Through decoding processing by the decoder output head, the predicted distribution of the instantaneous velocity vector of the jet fluid and the predicted distribution of the instantaneous velocity vector of the abrasive particles in the whole field are obtained.

[0047] Step 104: Determine the first difference between the measured distribution of the instantaneous velocity vector of the jet fluid and the predicted distribution of the instantaneous velocity vector of the jet fluid.

[0048] In some embodiments, step 104 described above can be implemented as follows: spatial registration is performed on the measured distribution of the instantaneous velocity vector of the jet fluid across the entire field and the predicted distribution of the instantaneous velocity vector of the jet fluid across the entire field to obtain the aligned measured velocity field data and predicted velocity field data of the jet fluid; the velocity component deviation is calculated pixel by pixel on the aligned measured velocity field data and predicted velocity field data of the jet fluid to obtain a first velocity residual distribution map of the jet fluid across the entire field; a loss function is calculated on the first velocity residual distribution map to obtain the first difference.

[0049] Thus, by spatially registering the measured and predicted velocity fields and calculating the deviation pixel by pixel, refined error quantification can be achieved across the entire field, transforming macroscopic performance evaluation into microscopic pixel-level residual distribution maps. Simultaneously, based on the loss function calculation of the first velocity residual distribution map, the prediction deviation of local flow field structures (such as turbulent vortices and boundary layer thickness) can be keenly captured, enabling the model to focus on optimizing the fitting accuracy of the dynamic characteristics of the jet core and edge regions during backpropagation. This significantly improves the prediction model's ability to reproduce flow field details and its overall physical consistency.

[0050] It should be noted that the loss function can be selected from pixel-wise Euclidean distance loss (such as mean square error, absolute error), structural similarity loss, physical constraint loss, and vector angle loss, without specific limitations.

[0051] As an example, taking the process debugging of descaling operations in oil pipelines as an example, firstly, the measured velocity field of the jet fluid obtained from the actual measurement is spatially registered with the predicted velocity field output by the model to ensure that the origin of the physical coordinate system and the scaling ratio of the two are completely consistent, so as to eliminate the alignment error caused by different shooting angles or resolutions; then, the deviation of the measured and predicted velocity vectors in the axial and radial components is calculated pixel by pixel to generate a first velocity residual distribution map that can intuitively reflect the magnitude of the error in the entire field; finally, the loss function is calculated based on the first velocity residual distribution map (for example, combining mean square error and structural similarity loss), and the error of the entire map is quantified to obtain the first difference. The first difference can be used to measure the accuracy of the current model in restoring the shape of the water phase flow field. By feeding it back to the prediction model to correct the model parameters, the prediction accuracy of the model can be improved.

[0052] Step 105: Determine the second difference between the measured distribution of the instantaneous velocity vector of the abrasive particles in the whole field and the predicted distribution of the instantaneous velocity vector of the abrasive particles in the whole field.

[0053] In some embodiments, step 105 described above can be implemented as follows: spatial registration is performed on the measured distribution of the instantaneous velocity vector of the abrasive particles across the entire field and the predicted distribution of the instantaneous velocity vector of the abrasive particles across the entire field to obtain the aligned measured velocity field data and predicted velocity field data of the abrasive particles; the velocity component deviation is calculated pixel by pixel on the aligned measured velocity field data and predicted velocity field data of the abrasive particles to obtain a second velocity residual distribution map of the abrasive particles across the entire field; a loss function is calculated on the second velocity residual distribution map to obtain the second difference.

[0054] Thus, by spatially registering and calculating pixel-by-pixel deviations between the measured and predicted distributions of abrasive particles, a quantitative assessment of the prediction accuracy of the solid-phase flow field can be achieved. The generated second velocity residual distribution map can accurately expose the model's prediction shortcomings in the abrasive acceleration lag region and edge diffusion region. Furthermore, by combining the loss function calculation for solid-phase characteristics, the model's fitting process to the dynamic behavior of high-density particles can be effectively constrained, ensuring that the trained model can truly reflect the energy transfer efficiency and spatial aggregation morphology of abrasive particles, thereby significantly improving the reliability of the prediction of machining cutting capabilities.

[0055] It should be noted that the loss function can be selected from pixel-wise Euclidean distance loss (such as mean square error, absolute error), structural similarity loss, physical constraint loss, and vector angle loss, without specific limitations.

[0056] As an example, taking the optimization of precision machining process for cemented carbide cutting tools as an example, firstly, the measured full-field velocity vector distribution of abrasive particles is spatially registered with the distribution predicted by the model to ensure that the coordinate systems of the two are consistent, so as to eliminate the deviation caused by the shooting angle or image scaling; then, the deviation of the velocity components in the axial and radial directions is calculated pixel by pixel to generate a second velocity residual distribution map; finally, the loss function is calculated based on the second velocity residual map (for example, combining mean square error and structural similarity loss), and the full-field error is quantified to obtain the second difference. The second difference can be used to measure the accuracy of the current model in restoring the solid flow field morphology. By feeding it back to the prediction model to correct the model parameters, it can be ensured that the optimized process parameters can truly reflect the actual cutting ability and energy distribution of the abrasive on cemented carbide.

[0057] Step 106: Based on the first difference and the second difference, update the internal parameters of the abrasive waterjet flow field visualization prediction model to obtain a trained abrasive waterjet flow field visualization prediction model, and perform visualization prediction of the abrasive waterjet flow field to be predicted based on the trained abrasive waterjet flow field visualization prediction model.

[0058] It should be noted that the internal parameters of the abrasive waterjet flow field visualization prediction model may include the weight matrix and bias of the feature extraction layer, as well as implicit parameters of physical constraints (such as equivalent turbulent viscosity, drag coupling coefficient, and feature encoding vectors of corresponding working pressure and nozzle geometry, etc.), without specific limitations here.

[0059] In some embodiments, the step 106 above, which updates the internal parameters of the abrasive waterjet flow field visualization prediction model based on the first difference and the second difference to obtain a trained abrasive waterjet flow field visualization prediction model, can be achieved in the following way: based on a preset first weight, the first difference and the second difference are weighted and summed to obtain the total model loss; based on the total model loss, the internal parameters of the abrasive waterjet flow field visualization prediction model are updated to obtain the trained abrasive waterjet flow field visualization prediction model.

[0060] Thus, by pre-setting a first weight to weight and summing the first and second differences, a dynamic balance can be achieved for the two-phase losses of the jet fluid and abrasive particles. This effectively avoids the problem of the model overfitting the solid phase and ignoring the details of the aqueous phase due to the strong abrasive phase signal. Furthermore, by updating the internal parameters based on the backpropagation of the total model loss, the model can simultaneously take into account the physical characteristics of the jet fluid and abrasive particles during training. This significantly improves the overall convergence accuracy and generalization ability of the prediction model in multiphase coupled flow fields, ensuring that the final output visualization results have both the realism of the aqueous phase structure and the abrasive energy.

[0061] As an example, taking precision waterjet cutting of difficult-to-machine materials for aerospace as an example, firstly, according to the preset first weight (e.g., 4:6), the first difference of the jet fluid (water phase) and the second difference of the abrasive particles are weighted and summed to obtain the total model loss; then, based on the current total model loss, the weight parameters inside the prediction model are fine-tuned based on the backpropagation algorithm. This process is repeated until the model can accurately capture the acceleration lag characteristics of abrasive particles when cutting high-temperature alloys, and finally outputs a trained abrasive waterjet flow field visualization prediction model.

[0062] Here, we explain the backpropagation process. The first initial velocity of the jet fluid and the second initial velocity of the abrasive particles are input into the input layer of the neural network model (visual prediction model of abrasive water jet flow field). After passing through the hidden layer, the model finally reaches the output layer and outputs the result. This is the forward propagation process of the neural network model. Since there is an error between the output result of the neural network model and the actual result, the error between the calculated result and the actual value is propagated back from the output layer to the hidden layer until it reaches the input layer. During the backpropagation process, the values ​​of the model parameters are adjusted according to the error. The above process is iterated until convergence.

[0063] Taking the cross-entropy loss function as an example, the server determines the model loss based on the cross-entropy loss function, backpropagates the model loss from the output layer of the abrasive waterjet flow field visualization prediction model, and backpropagates the model loss layer by layer. When the model loss reaches each layer, the gradient (that is, the partial derivative of the loss function with respect to the parameters of each layer) is solved by combining the propagated model loss, and the corresponding gradient values ​​of the parameters of each layer are updated.

[0064] In some embodiments, the visualization prediction of the abrasive waterjet flow field based on the trained abrasive waterjet flow field visualization prediction model in step 106 above can be achieved by the following method: obtaining the third initial velocity of the jet fluid and the fourth initial velocity of the abrasive particles in the abrasive waterjet to be predicted; extracting features from the third initial velocity to obtain a fifth feature, and extracting features from the fourth initial velocity to obtain a sixth feature; fusing the fifth feature and the sixth feature to obtain a seventh feature, and performing position encoding on the seventh feature to obtain an eighth feature; performing a nonlinear transformation on the eighth feature to obtain a nonlinearly transformed eighth feature, and decoding the nonlinearly transformed eighth feature to obtain the full-field instantaneous velocity vector prediction distribution of the jet fluid and the full-field instantaneous velocity vector prediction distribution of the abrasive particles in the abrasive waterjet to be predicted; and performing visualization rendering on the full-field instantaneous velocity vector prediction distribution of the jet fluid and the full-field instantaneous velocity vector prediction distribution of the abrasive particles in the abrasive waterjet to be predicted to obtain a flow field visualization cloud map of the abrasive waterjet to be predicted.

[0065] Thus, by acquiring the initial velocity of the flow field to be predicted and performing feature fusion and position encoding, flow field prediction can be achieved without injecting tracer particles and laser irradiation, directly outputting a high-fidelity full-field instantaneous velocity vector prediction distribution; and combined with the flow field cloud map generated by visualization rendering, the complex multiphase fluid dynamics data is transformed into an intuitive color spectrum, enabling engineers to monitor and evaluate the energy distribution and abrasive aggregation state of the jet core region under different working conditions in real time and at low cost, thereby quickly locking the optimal processing parameters and significantly improving process debugging efficiency.

[0066] It should be noted that flow field visualization cloud maps refer to converting the digitized velocity vector data output by the model into pseudo-color images that can be directly recognized.

[0067] As an example, taking an automated waterjet cutting production line for high-strength steel plates in engineering machinery as an example, firstly, the system determines the third initial velocity of the jet fluid under the predicted working condition. The fourth initial velocity of the abrasive particles For the third initial velocity With the fourth initial velocity Feature extraction is performed separately, transforming the features into high-dimensional feature vectors (the fifth feature) that contain dynamic properties. and the sixth feature ); then, the fifth feature and the sixth feature Perform splicing and fusion along the channel dimension (seventh feature) The eighth feature is obtained by superimposing the positional encoding of the flow field spatial coordinates. Afterwards, through internally trained nonlinear transformation and decoding, the instantaneous velocity vector prediction distribution of the jet fluid and abrasive particles under the current working conditions is obtained. Finally, through the visualization rendering engine, the velocity amplitude is mapped into a red-blue gradient flow field visualization cloud map and directional streamlines are superimposed, so that on-site engineers can directly view the cloud map to determine whether the abrasive energy is concentrated in the cutting core area, thereby quickly locking the optimal feed speed and realizing intelligent process decision-making with zero trial cutting cost.

[0068] The following will describe an exemplary application of the embodiments of this application in a real-world application scenario.

[0069] Ultra-high pressure abrasive waterjet technology, as an advanced cold processing technique, fundamentally depends on the complex gas-liquid-solid three-phase transient flow characteristics within the jet for its material removal mechanism, energy transfer efficiency, and final processing quality. During the formation, development, and impact of the jet on the target, intense turbulent mixing, interphase slip, and energy dissipation occur. Therefore, achieving high spatiotemporal resolution, non-invasive, and separate quantitative measurement of its core flow field, especially the motion characteristics of the water and abrasive phases, is fundamental to revealing its processing mechanism, establishing accurate process models, and ultimately realizing intelligent dynamic parameter control. However, the extremely high velocity of the jet, the ultra-high internal pressure, the strong scattering and obstruction of light by abrasive particles, and the design of a safe and reliable optical observation window under extreme conditions constitute multiple severe technical challenges for flow field visualization. Overcoming these challenges and achieving direct observation of the dynamic characteristics inside the "black box" of ultra-high pressure abrasive waterjet has significant scientific and engineering implications.

[0070] Before the maturity of optical measurement technology, the relevant technology mainly relied on measuring the jet impact force to inversely deduce its average velocity, which is a classic and simple method. That is, by establishing the mechanical relationship between the momentum generated by the jet impacting the target surface and the average velocity of the jet, a precise force measuring device is used to measure the total impact momentum of the abrasive water jet at the nozzle exit. By combining the continuity equation and the momentum conservation equation, the average velocities of the solid, liquid and gas phases at the exit section are derived in reverse. The system includes a thin plate sample, a jetting system, a Kistler 9257A three-component dynamometer, a signal amplifier, an analog-to-digital converter and LabVIEW data acquisition software, which can quantitatively reveal the three-phase average velocity of the abrasive water jet at different nozzle distances.

[0071] With the maturity of optical measurement technology, non-invasive optical measurement techniques have become mainstream for directly observing the internal structure of flow fields. Among them, particle image velocimetry (PIV) acquires the velocity field by tracking the displacement of tracer particles, while laser-induced fluorescence (LIF) uses the fluorescence signal of excited specific substances to identify specific phases or components. Combining the two provides a possible technical path for separating and measuring the velocities of each phase in multiphase flow.

[0072] In the velocity measurement of ultra-high pressure abrasive water jets, although the force measurement method is intuitive in principle, it has inherent defects that are difficult to overcome. This method can only obtain the overall average force of the jet impacting the target and inversely deduce the average velocity of a single cross section. It cannot analyze the velocity distribution along the axial direction of the flow field at the same moment, nor can it distinguish the velocity difference between the core region and the edge region. Moreover, as a contact measurement, the intervention of the measuring device itself will disturb or even destroy the original flow field, making the measured "impact force" no longer the true state of the free jet, thus introducing systematic errors. In addition, the surface morphology of the force measurement target is easily changed by the severe erosion of ultra-high pressure abrasive jets, which seriously threatens the stability and accuracy of long-term measurements. To build a sensor platform that simultaneously meets the requirements of ultra-high erosion resistance and high-precision force measurement, there are huge challenges in terms of mechanical structure design and manufacturing cost.

[0073] On the other hand, while combining optical particle tracking velocimetry (PTV) with LIF technology can acquire the trajectory of a single tracer particle, it also faces significant limitations in practical applications. Compared to PIV, PTV requires tracer particles to be sparsely distributed within the field of view, which makes it difficult to obtain continuous and complete velocity data of the entire flow field, and it is prone to recognition errors when particle trajectories intersect. PTV requires the laser sheet to be thin enough, otherwise particles will overlap in the depth direction, producing false high or low speeds. Furthermore, the algorithm of this technology is complex, the data processing efficiency is low, and the degree of software automation is far inferior to that of mature particle image velocimetry technology.

[0074] Therefore, both contact force measurement based on momentum principle and non-contact particle tracking velocity measurement based on optical tracer have significant bottlenecks in terms of methodology and application when dealing with the velocity measurement challenge of ultra-high pressure abrasive water jet, a complex two-phase (jet fluid and abrasive particles) flow, which restricts the in-depth revelation of the fine structure and dynamic characteristics of the flow field.

[0075] To address the problems of limited measurement information, insufficient flow field analysis capability, significant measurement disturbances, and poor long-term measurement stability in ultra-high pressure abrasive waterjet velocity measurement, this application proposes a visualization measurement method for abrasive waterjet flow field. This method achieves high-precision, non-contact, full-field, and stable measurement of the jet velocity field, overcoming the limitations of traditional measurement methods that can only obtain the cross-sectional average velocity. It can obtain the full-field velocity distribution of ultra-high pressure abrasive waterjet at the same moment, including the instantaneous velocity fields in the axial and radial directions. It clearly presents the velocity gradient changes in the jet core region, shear layer, and edge region, providing more complete and refined data support for revealing the fine flow structure and dynamic characteristics inside the jet. Meanwhile, this application employs a fully optical, non-contact measurement method, placing both the laser sheet and the camera outside the flow field. This avoids interfering with the original flow field and prevents disturbances and damage to the flow field caused by contact measurement devices. This ensures that the measured velocity information accurately reflects the motion state of the free jet, effectively reducing systematic errors introduced by measurement device intervention. Regarding system stability, this application avoids erosion problems caused by physical contact through non-contact measurement, significantly improving the reliability and repeatability of the measurement system under long-term, high-intensity operating conditions. In terms of flow field resolution, this application, based on the statistical correlation principle of PIV technology, allows for a high-density distribution of tracer particles, overcoming the difficulty of obtaining continuous and complete flow data due to the reliance on sparse particle distribution in PTV technology. The limitations of field data can be overcome by obtaining the instantaneous velocity field of the entire measurement plane in a single measurement, thus effectively avoiding recognition errors caused by particle trajectory intersections, and significantly improving the continuity and completeness of data acquisition. At the same time, the PIV technology has relatively relaxed requirements on the thickness of the laser sheet, and a thicker sheet can be used for volume averaging calculation, reducing the impact of depth direction errors while also reducing the accuracy requirements of the laser system and equipment costs. In addition, this application adopts a mature PIV cross-correlation algorithm, which has a high degree of automation and a simplified data processing flow, avoiding the inefficiency problems caused by the complex PTV algorithm. Under ultra-high pressure abrasive waterjet conditions, it can quickly and stably process a large amount of image data, thereby significantly improving measurement efficiency and making it more suitable for practical experiments and engineering tests.

[0076] In some embodiments, this application first prepares tracer particles for characterizing the velocity of the jet fluid and tracer particles for characterizing the velocity of the abrasive particles. The first type of tracer particle (corresponding to the first tracer particle) is a fluorescent particle with a small diameter, light weight, and low density, giving it good fluid following ability and enabling it to accurately reflect the velocity of the jet flow field. The second type of tracer particle (corresponding to the second tracer particle) is a fluorescent particle with the same diameter and density as the abrasive particles, enabling it to accurately reflect the trajectory of the abrasive particles. Subsequently, the first type of tracer particle is introduced into an abrasive water jet under stable operating conditions. The flow field region to be measured is illuminated using a laser sheet, and fluorescence images of the first tracer particle are acquired using a high-speed camera. The acquired image sequence is analyzed and processed based on a cross-correlation algorithm using PIV technology to obtain the instantaneous velocity field distribution of the jet fluid (corresponding to the jet flow field). (The process begins with the measured distribution of the instantaneous velocity vector of the fluid across the entire field.) Then, a second type of tracer particle is introduced into the abrasive water jet under stable operating conditions. The flow field region to be measured is illuminated using a laser sheet, and fluorescence images of the second tracer particle are acquired using a high-speed camera. The acquired image sequence is analyzed and processed using a cross-correlation algorithm based on PIV technology to obtain the instantaneous velocity field distribution of the abrasive particles (corresponding to the measured distribution of the instantaneous velocity vector of the abrasive particles across the entire field). Finally, under multiple preset operating pressure conditions, the above process of acquiring the instantaneous velocity field distribution of the jet fluid and abrasive particles is repeated to obtain flow field data (i.e., the instantaneous velocity field distribution of the jet fluid and abrasive particles) under different operating conditions (i.e., different preset operating pressures). The acquired image data is then transmitted to a computer, and the velocity vector distribution of the flow field is reconstructed using PIV analysis software to achieve flow field visualization.

[0077] It should be noted that in this application, PIV technology is used instead of traditional PTV technology. By allowing tracer particles to be distributed in a high density within the flow field, the instantaneous velocity distribution of the entire flow field is obtained based on the principle of statistical correlation. This can overcome the limitations of force measurement methods, which can only measure the average velocity of a single cross section and cannot simultaneously obtain the velocity distribution of the entire flow field. It also overcomes the technical defects of traditional PTV technology, such as the difficulty in obtaining continuous and complete flow field data due to the sparse distribution of particles, the identification ambiguity caused by the intersection of particle trajectories, and the stringent requirements on the thickness of the laser sheet.

[0078] In some embodiments, this application utilizes a non-contact optical measurement method, PIV+LIF, to obtain the velocity of the abrasive water jet and the velocity of the abrasive particles, and visualizes the flow field using experimental data. Thus, the PIV+LIF measurement method effectively avoids the problems of traditional force measurement methods, which cannot obtain the spatial distribution of the flow field at the same moment when obtaining the average velocity of a cross-section of the abrasive water jet, and where the measurement target surface disturbs the flow field and the jet erodes the target surface. When measuring the spatial distribution of the jet velocity, extremely small fluorescent particles are used; these particles have small diameters, light weights, and low densities, so their velocities are very close to the jet flow field, effectively characterizing the jet velocity.

[0079] In some embodiments, see Figure 2 , Figure 2 A schematic diagram of the PIV speed measurement system provided in the application embodiment is shown below. Figure 2 As shown, the PIV velocimetry system includes a laser system, an image acquisition system, a synchronization control system, an abrasive waterjet system, fluorescent particles, and a particle image velocimetry software system. Here, the laser system includes a dual-pulse laser and a sheet light assembly system. The dual-pulse laser generates high-energy laser pulses to illuminate the tracer particles. The sheet light assembly system converts the point laser into sheet light to illuminate the flow field profile under test. The image acquisition system includes a high-speed camera and a lens. The high-speed camera captures two frames of particles. The lens is a pass filter that allows only the fluorescence wavelength to pass through, filtering out background light. The synchronization control system precisely controls the timing of the laser pulses and camera capture.

[0080] In some embodiments, see Figure 3 , Figure 3 This is a schematic diagram of the PIV speed measurement principle provided in the embodiments of this application, such as... Figure 3 As shown, (a) is a front view of PIV velocimetry and (b) is a side view of PIV velocimetry. The fluorescent particles in the core region of the jet are illuminated by the laser sheet light and captured by the high-speed camera. The data is processed according to the query interval, which is 32x32 or 64x64 pixels. Each query interval outputs a velocity result, and all velocity results are integrated together to obtain the velocity distribution of the entire flow field.

[0081] In some embodiments, related technologies typically utilize PTV+LIF. PTV requires tracer particles to be sparsely distributed within the field of view, making it difficult to obtain continuous and complete flow field velocity data, and it is prone to identification errors when particle trajectories intersect. Furthermore, it requires the laser sheet to be sufficiently thin, otherwise depth direction errors will occur. Simultaneously, its algorithm is complex, data processing efficiency is low, and software automation is not high. In addition, applying a Rhodamine B fluorescent coating to the abrasive particles in PTV carries the risk of coating peeling off under high-speed, high-pressure impact conditions, potentially leading to particle identification errors or signal loss. The PIV+LIF measurement method of this application solves these problems. This measurement method is based on statistical correlation principles, allowing for high-density distribution of tracer particles, and can obtain the instantaneous velocity field of the entire measurement plane in a single measurement, avoiding identification ambiguity caused by particle trajectory intersections. It can also use a relatively thick sheet to calculate volume averaging, reducing the influence of the depth direction and lowering costs. Its algorithm is mature, highly automated, and has high data processing efficiency. Furthermore, this application uses fluorescent particles with the same diameter and density as the abrasive particles instead of covering the abrasive particle surface with phosphor, resulting in stronger stability.

[0082] In some embodiments, 1g of fluorescent particles with a diameter of 8.5μm is weighed, the abrasive waterjet machine is started, and after the output pressure stabilizes, the fluorescent particles are sucked in using negative pressure while the camera starts taking pictures. Then, the jet velocity is measured at 150MPa, 200MPa, 250MPa, and 300MPa respectively. Three sets of data are measured under each pressure condition, and 100 pairs of pictures are taken for each set of data to ensure that the error of the data obtained between each set is not greater than the specified value. At the same time, it is compared with the theoretical calculation value to ensure the accuracy of the experiment. Finally, the obtained data is transferred to the computer for processing to obtain the jet flow field velocity (corresponding to the measured distribution of the instantaneous velocity vector of the jet fluid in the entire field).

[0083] In some embodiments, 1g of fluorescent particles with a diameter of 180μm is first weighed, the abrasive waterjet machine is started, and after the output pressure stabilizes, the fluorescent particles are sucked in using negative pressure while the camera starts taking pictures. Then, the jet velocity is measured at 150MPa, 200MPa, 250MPa, and 300MPa respectively. Three sets of data are measured under each pressure condition, and 100 pairs of pictures are taken for each set of data to ensure that the error of the data obtained between each set is not greater than the specified value. At the same time, it is compared with the theoretical calculation value to ensure the accuracy of the experiment. Finally, the obtained data is transmitted to a computer for processing to obtain the velocity distribution of abrasive particles in the flow field (corresponding to the measured distribution of the instantaneous velocity vector of abrasive particles in the entire field).

[0084] In some embodiments, to address the problems of limited measurement information, insufficient flow field analysis capability, significant measurement disturbances, and poor long-term measurement stability in the measurement of ultra-high pressure abrasive waterjet velocity, this application aims to propose a visualization measurement method for abrasive waterjet flow field, which can achieve high-precision, non-contact, full-field, and stable measurement of the jet velocity field, and solves the following technical problems: (1) In view of the problem that the relevant measurement methods can only obtain overall average information and are difficult to characterize the spatial distribution of the flow field, this application can realize the resolution of the velocity distribution of ultra-high pressure abrasive water jet along the axial and radial directions at the same time, and can distinguish the velocity difference between the core area and the edge area of ​​the jet, thereby obtaining more complete and finer flow field structure information; (2) In view of the problem that contact measurement methods cause disturbance or even damage to the original flow field, this application can avoid the measurement device directly intervening in the jet body, so that the measured velocity information can truly reflect the motion state of the free jet and reduce the systematic error introduced by measurement intervention. (3) In view of the problem that the measuring device is prone to failure and the measurement results are unstable under the strong erosion environment of high pressure abrasive jet, this application can avoid the performance degradation or surface morphology change of the sensing element caused by erosion during the measurement process, thereby improving the reliability and repeatability of the measuring system under long-term and high-intensity working conditions. (4) In view of the problem that PTV+LIF is difficult to achieve full flow field measurement under high pressure and high speed two-phase (jet fluid and abrasive particles) flow conditions, this application can obtain continuous and complete velocity field information without relying on sparse particle distribution, reduce the identification error caused by particle trajectory intersection, and improve the continuity and completeness of data acquisition. (5) PTV+LIF requires higher laser sheet light requirements, otherwise the particles will overlap in the depth direction, resulting in inaccurate velocity measurement. This application can reduce the cost of the laser system. (6) In view of the problem that the complex algorithm and low level of automation of PTV+LIF restrict the engineering application, this application can simplify the data processing process, improve the robustness and automation level of the algorithm, improve the measurement efficiency, and make the method applicable to actual experiments and engineering tests under ultra-high pressure abrasive water jet conditions.

[0085] By solving the above problems, this application can overcome the bottlenecks in the methods and applications of related technologies in the measurement of ultra-high pressure abrasive water jet velocity, and provide a reliable, stable and highly applicable technical means for revealing the fine flow structure and dynamic characteristics inside the jet.

[0086] The above are embodiments of the method proposed in this application. Based on the same inventive concept, embodiments of this application also provide a visualization and measurement device for abrasive waterjet flow fields, the structure of which is as follows: Figure 4 As shown.

[0087] Figure 4 This is a schematic diagram of the internal structure of an abrasive waterjet flow field visualization measurement device provided in an embodiment of this application. Figure 4 As shown, the device includes: At least one processor 201; And a memory 202 that is communicatively connected to at least one processor; The memory 202 stores instructions that can be executed by at least one processor. The instructions are executed by at least one processor 201 to enable at least one processor 201 to perform the steps of the method corresponding to any of the above embodiments.

[0088] Some embodiments of this application provide corresponding to Figure 1 A non-volatile computer storage medium stores computer-executable instructions configured to perform the steps of the method corresponding to any of the above embodiments.

[0089] The various embodiments in this application are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the embodiments for IoT devices and media are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0090] The systems, media, and methods provided in this application are one-to-one correspondences. Therefore, the systems and media also have similar beneficial technical effects as their corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the systems and media will not be repeated here.

[0091] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0092] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0093] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0094] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0095] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0096] Memory may include non-persistent storage in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0097] Computer-readable media include both permanent and non-permanent, removable and non-removable media that can store information by any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0098] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0099] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A method for visually measuring the flow field of an abrasive waterjet, characterized in that, The method includes: A first tracer particle and a second tracer particle are obtained. The first tracer particle is used to characterize the motion characteristics of the jet fluid, and the second tracer particle is used to characterize the motion characteristics of the abrasive particles. The first tracer particle and the second tracer particle were injected into the abrasive water jet under stable working conditions, and the measured distribution of the instantaneous velocity vector of the jet fluid and the measured distribution of the instantaneous velocity vector of the abrasive particles were obtained. The abrasive water jet flow field visualization prediction model is used to predict the abrasive water jet flow field based on the first initial velocity of the jet fluid and the second initial velocity of the abrasive particles, thereby obtaining the predicted distribution of the instantaneous velocity vector of the jet fluid and the predicted distribution of the instantaneous velocity vector of the abrasive particles. Determine the first difference between the measured distribution of the instantaneous velocity vector of the jet fluid and the predicted distribution of the instantaneous velocity vector of the jet fluid, and determine the second difference between the measured distribution of the instantaneous velocity vector of the abrasive particles and the predicted distribution of the instantaneous velocity vector of the abrasive particles. Based on the first difference and the second difference, the internal parameters of the abrasive waterjet flow field visualization prediction model are updated to obtain a trained abrasive waterjet flow field visualization prediction model, which is then used to perform visualization prediction of the abrasive waterjet flow field to be predicted.

2. The method according to claim 1, characterized in that, The second tracer particle is a fluorescent particle with the same diameter and density as the abrasive particle.

3. The method according to claim 1, characterized in that, The step of injecting the first tracer particle and the second tracer particle into an abrasive water jet under stable operating conditions to obtain the measured distribution of the instantaneous velocity vector of the jet fluid and the measured distribution of the instantaneous velocity vector of the abrasive particles in the entire field includes: The first tracer particle is injected into the abrasive water jet under stable working conditions, and the flow field area to be tested is irradiated by laser sheet light to obtain the fluorescence image of the first tracer particle. The fluorescence image of the first tracer particle is subjected to a first cross-correlation calculation to obtain the measured distribution of the instantaneous velocity vector of the jet fluid in the entire field; The second tracer particle is injected into the abrasive water jet under stable working conditions, and the flow field area to be tested is irradiated by laser sheet light to obtain the fluorescence image of the second tracer particle. A second cross-correlation calculation is performed on the fluorescence image of the second tracer particle to obtain the measured distribution of the instantaneous velocity vector of the abrasive particle across the entire field.

4. The method according to claim 1, characterized in that, The step of predicting the abrasive water jet flow field based on the first initial velocity of the jet fluid and the second initial velocity of the abrasive particles to obtain the predicted distribution of the instantaneous velocity vector of the jet fluid and the predicted distribution of the instantaneous velocity vector of the abrasive particles across the entire field includes: The first initial velocity of the jet fluid is extracted to obtain a first feature, and the second initial velocity of the abrasive particles is extracted to obtain a second feature; The first feature and the second feature are fused to obtain the third feature, and the third feature is then positionally encoded to obtain the fourth feature; The fourth feature is subjected to a nonlinear transformation to obtain the nonlinearly transformed fourth feature, and the nonlinearly transformed fourth feature is decoded to obtain the full-field instantaneous velocity vector prediction distribution of the jet fluid and the full-field instantaneous velocity vector prediction distribution of the abrasive particles.

5. The method according to claim 1, characterized in that, The determination of the first difference between the measured distribution of the full-field instantaneous velocity vector of the jet fluid and the predicted distribution of the full-field instantaneous velocity vector of the jet fluid includes: Spatial registration is performed on the measured distribution of the instantaneous velocity vector of the jet fluid and the predicted distribution of the instantaneous velocity vector of the jet fluid to obtain the aligned measured velocity field data and predicted velocity field data of the jet fluid. The velocity component deviation of the jet fluid is calculated pixel by pixel by comparing the measured velocity field data and the predicted velocity field data after alignment. The first velocity residual distribution map of the jet fluid in the whole field is obtained. The first difference is obtained by calculating the loss function on the first velocity residual distribution map.

6. The method according to claim 1, characterized in that, The determination of the second difference between the measured distribution of the full-field instantaneous velocity vector of the abrasive particles and the predicted distribution of the full-field instantaneous velocity vector of the abrasive particles includes: Spatial registration is performed on the measured distribution of the instantaneous velocity vector of the abrasive particles and the predicted distribution of the instantaneous velocity vector of the abrasive particles to obtain the aligned measured velocity field data and predicted velocity field data of the abrasive particles. The velocity component deviation of the abrasive particles in the measured velocity field data and the predicted velocity field data after alignment is calculated pixel by pixel to obtain the second velocity residual distribution map of the abrasive particles in the entire field range. The second difference is obtained by calculating the loss function on the second velocity residual distribution map.

7. The method according to claim 1, characterized in that, The step of updating the internal parameters of the abrasive waterjet flow field visualization prediction model based on the first difference and the second difference to obtain a trained abrasive waterjet flow field visualization prediction model includes: Based on the preset first weight, the first difference and the second difference are weighted and summed to obtain the total model loss; Based on the total loss of the model, the internal parameters of the abrasive waterjet flow field visualization prediction model are updated to obtain the trained abrasive waterjet flow field visualization prediction model.

8. The method according to claim 1, characterized in that, The visualization prediction of the abrasive waterjet flow field based on the trained abrasive waterjet flow field visualization prediction model includes: The third initial velocity of the jet fluid and the fourth initial velocity of the abrasive particles in the abrasive water jet to be predicted are obtained. Feature extraction is performed on the third initial velocity to obtain the fifth feature, and feature extraction is performed on the fourth initial velocity to obtain the sixth feature; The fifth feature and the sixth feature are fused to obtain the seventh feature, and the seventh feature is then positionally encoded to obtain the eighth feature; The eighth feature is subjected to a nonlinear transformation to obtain the nonlinearly transformed eighth feature, and the nonlinearly transformed eighth feature is decoded to obtain the full-field instantaneous velocity vector prediction distribution of the jet fluid and the full-field instantaneous velocity vector prediction distribution of the abrasive particles in the abrasive water jet to be predicted. The predicted distribution of the instantaneous velocity vector of the jet fluid and the predicted distribution of the instantaneous velocity vector of the abrasive particles in the abrasive water jet to be predicted are visualized and rendered to obtain a flow field visualization cloud map of the abrasive water jet to be predicted.

9. A visualization and measurement device for abrasive waterjet flow field, characterized in that, The device includes: At least one processor; And, a memory communicatively connected to the at least one processor; The memory stores instructions that can be executed by the at least one processor to enable the at least one processor to perform the method as described in any one of claims 1-8.

10. A computer storage medium storing computer-executable instructions, characterized in that, When the computer-executable instructions are executed, they implement the method as described in any one of claims 1-8.

Images