Molten pool behavior visualization method and system based on machine vision

Abstract

The invention provides a molten pool behavior visualization method and system based on machine vision, and the method comprises the steps: S1, obtaining the dynamic image data of the interior of a molten pool through synchrotron radiation imaging, and obtaining the dynamic image data of the surface of the molten pool through the shooting of a high-speed camera; s2, training and testing a YOLOv5 model by utilizing the preprocessed dynamic image data, and generating a weight file for detecting tungsten particles in a molten pool; and S3, inputting the dynamic image data of the interior of the molten pool and the dynamic image data of the surface of the molten pool into the trained YOLOv5 model, and tracking and responding to the movement of the tungsten particles in real time to obtain a visual image of the movement track of the tungsten particles. The tungsten particles are introduced to serve as tracers, the synchrotron radiation imaging technology and the high-speed camera are combined, and multi-level accurate observation of the flowing behaviors in the molten pool and on the surface can be achieved. The system utilizes a machine vision technology and a deep learning model to automatically detect and track the tungsten particle motion trail, so that the observation accuracy and efficiency are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the field of laser processing technology, and in particular to a method and system for visualizing molten pool behavior based on machine vision. Background Art

[0002] As one of the core technologies in the field of modern manufacturing, laser processing technology is widely used in high-tech fields such as aerospace, automobile manufacturing, and electronic devices. Due to its advantages such as high energy density, precise control, and non-contact processing, laser welding and additive manufacturing technology have become key means for the efficient manufacture of complex components. During the laser processing process, the behavior of the molten pool directly affects the microstructure and final performance of the material, so the study and characterization of the molten pool is particularly important.

[0003] The flow behavior of the molten pool, including its morphology, flow velocity, and flow direction, determines the microstructure of the workpiece, such as the cooling rate, grain growth, and defect distribution, which directly affects the mechanical properties of the final product. Therefore, studying the flow characteristics of the molten pool is extremely important for optimizing the laser processing technology and improving the performance of the product. However, due to the high temperature inside the molten pool, the high-speed flow of the molten metal, and the complex thermophysical reactions, the behavior of the molten pool is difficult to accurately observe and quantify through conventional experimental methods.

[0004] At present, numerical simulation technology is an important tool for studying the behavior of the molten pool. By establishing a physical model of the molten pool, numerical simulation can predict the temperature field, velocity field and flow behavior of the molten pool. However, numerical simulation has certain limitations, especially in complex actual working conditions, the physical model often needs to be simplified, which leads to a certain deviation between the simulation results and the flow behavior of the molten pool in actual working conditions. This deviation stems from the incompleteness of the physical model and the variability of the complex processing environment. Therefore, it is difficult to accurately characterize the true behavior of the molten pool by relying solely on numerical simulation.

[0005] In this context, an experimental method that can accurately characterize the molten pool flow is urgently needed to verify the numerical simulation results and further reveal the true dynamic behavior of the molten pool. Summary of the invention

[0006] In view of the defects in the prior art, the object of the present invention is to provide a method and system for visualizing molten pool behavior based on machine vision.

[0007] A method for visualizing molten pool behavior based on machine vision provided by the present invention includes:

[0008] Step S1: obtaining dynamic image data of the inside of the molten pool by synchrotron radiation imaging, and obtaining dynamic image data of the surface of the molten pool by high-speed camera photography;

[0009] Step S2: using the preprocessed dynamic image data to train and test the YOLOv5 model, and generating a weight file for detecting tungsten particles in the molten pool;

[0010] Step S3: The dynamic image data inside the molten pool and the dynamic image data on the surface of the molten pool are input into the trained YOLOv5 model to track and respond to the movement of the tungsten particles in real time to obtain a visualized image of the movement trajectory of the tungsten particles.

[0011] Preferably, step S1 includes: introducing tungsten particles as tracers to track the flow behavior inside the molten pool; obtaining X-ray images through synchrotron radiation imaging to show the dynamic changes inside the molten pool; and recording the flow behavior on the surface of the molten pool through high-speed camera photography.

[0012] Preferably, step S2 comprises:

[0013] Step S2.1: Based on the preprocessed dynamic image data, manually or automatically annotate the tungsten particles to generate label data of the tungsten particle positions for model training;

[0014] Step S2.2: Use the generated data set to train the YOLOv5 model for multiple iterations until the model can accurately identify tungsten particles;

[0015] Step S2.3: After multiple rounds of training and validation, the optimal weight file is generated and used for model application;

[0016] Step S2.4: Use data from different scenarios to test the trained YOLOv5 model to verify its accuracy in detecting tungsten particles.

[0017] Preferably, step S3 comprises:

[0018] Step S3.1: The dynamic image data inside the molten pool and the dynamic image data on the surface of the molten pool are imported into the trained YOLOv5 model, and the model will automatically identify and annotate the tungsten particles in the molten pool.

[0019] Step S3.2: The user selects a tungsten particle with obvious behavioral characteristics through the interactive page, and the system tracks the movement of the tungsten particle in real time;

[0020] Step S3.3: Based on the continuous frame data, the system uses the speed and direction calculation unit, combined with the multi-target tracking function of the SORT algorithm, to calculate and display the speed and direction of the tungsten particles in real time;

[0021] Step S3.4: the system automatically stores the position information of the selected tungsten particles into a log file and saves its trajectory data;

[0022] Step S3.5: Generate a visualization image of the motion trajectory of the tungsten particle by reading the position information of the tungsten particle in the log file.

[0023] A machine vision-based molten pool behavior visualization system provided according to the present invention includes: a laser welding and additive unit, a data acquisition unit, a computer vision unit, and a data processing unit;

[0024] The laser welding and additive manufacturing unit performs laser welding and additive manufacturing, and introduces tungsten particles as tracer particles during the processing;

[0025] The data acquisition unit is used to capture the flow image of the molten pool during welding or additive manufacturing;

[0026] The computer vision unit detects tungsten particles based on the YOLOv5 model trained by the deep learning algorithm, and analyzes their motion trajectory in real time in combination with the SORT multi-target tracking algorithm;

[0027] The data processing unit determines the flow state of the molten pool by calculating the speed, direction and position changes of the tungsten particles, and outputs a visualized real-time tracking result.

[0028] Preferably, the data acquisition unit includes a synchrotron radiation imaging unit and a high-speed camera imaging unit;

[0029] The synchrotron radiation imaging unit tracks the tracer particles inside the molten pool through X-ray imaging, and the high-speed camera imaging unit is used to track the tracer particles on the surface of the molten pool.

[0030] Preferably, the computer vision unit includes a deep learning algorithm YOLOv5 model and a SORT multi-target tracking algorithm;

[0031] The YOLOv5 model accurately locates and identifies tungsten particles and generates bounding box coordinates of their positions. The SORT multi-target tracking algorithm continuously tracks and analyzes the flow behavior inside or on the surface of the molten pool by tracking the identified tungsten particles and analyzing their motion trajectories in multiple frames of images.

[0032] Preferably, the data processing unit accurately calculates the movement speed and direction of the tungsten particles based on the tungsten particle position information provided by the SORT algorithm;

[0033] By comparing the displacement of the tungsten particles in consecutive frames, the displacement between two frames is calculated, and the movement speed of the tungsten particles is calculated by combining the time interval between the frames;

[0034] The direction of movement of the tungsten particles is determined based on the relative position changes of the tungsten particles in consecutive frames.

[0035] Preferably, the YOLOv5 model generates multiple candidate bounding boxes through an anchor box mechanism, and uses a convolutional neural network to classify and regress these candidate boxes to determine the position and size of the target in the image; when tungsten particles are detected, the bounding box generated by the system is determined by the coordinates of the upper left corner and the lower right corner, and the coordinates of the center point of the target are recorded for motion tracking;

[0036] For tungsten particles of different scales, the YOLOv5 model dynamically adjusts the size of the bounding box, classifies the candidate boxes and outputs the category and its confidence, uses the non-maximum suppression algorithm, and selects the optimal bounding box through the intersection-union ratio to avoid repeated detection.

[0037] Preferably, the SORT multi-target tracking algorithm predicts the position of the tungsten particle in the next frame by using Kalman filtering, and updates it using historical data such as velocity and acceleration of the previous frame;

[0038] The tungsten particle bounding box of each frame is generated by YOLOv5 model detection. The SORT multi-target tracking algorithm matches these detection boxes with the results predicted by the Kalman filter through the Hungarian algorithm. For the tungsten particles that are successfully matched, the system updates their trajectory information; if a tungsten particle is not detected in a certain frame, SORT will record the number of times the target is lost and remove it if it is not matched for multiple consecutive frames.

[0039] Compared with the prior art, the present invention has the following beneficial effects:

[0040] 1. The present invention introduces tungsten particles as tracers, combines synchrotron radiation imaging technology and high-speed cameras, and can achieve multi-level accurate observation of the flow behavior inside and on the surface of the molten pool. The system uses machine vision technology and deep learning models to automatically detect and track the movement trajectory of tungsten particles, significantly improving the accuracy and efficiency of observation.

[0041] 2. Compared with the traditional manual observation method, the automatic tracking of the present invention not only reduces human intervention, but also enhances the stability and consistency of the observation results, providing reliable data support for the study of molten pool behavior.

[0042] 3. The present invention uses the YOLOv5 deep learning model and the SORT multi-target tracking algorithm to identify and track the movement of tungsten particles inside and outside the molten pool in real time. The system can automatically calculate the speed and direction of tungsten particles and generate real-time visualization results to help users intuitively analyze the dynamic behavior of the molten pool. Through this method, researchers can more comprehensively understand the flow characteristics of the molten pool, thereby optimizing the laser processing process and improving the final quality and performance of the product. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments made with reference to the following drawings:

[0044] Figure 1 It is a flow chart of the method of the present invention and a system composition diagram;

[0045] Figure 2 It is a schematic diagram of the system structure of the present invention;

[0046] Figure 3 It is the interactive interface of the synchrotron radiation program of the present invention;

[0047] Figure 4 This is the trajectory diagram of tracer particle No. 20 in the embodiment of the present invention;

[0048] Figure 5 Graphs showing the trajectories of different tracer particles in the embodiments of the present invention;

[0049] Figure 6 The trajectories of different tracer particles are reproduced in an embodiment of the present invention. DETAILED DESCRIPTION

[0050] The present invention is described in detail below in conjunction with specific embodiments. The following embodiments will help those skilled in the art to further understand the present invention, but are not intended to limit the present invention in any form. It should be noted that, for those of ordinary skill in the art, several changes and improvements can also be made without departing from the concept of the present invention. These all belong to the protection scope of the present invention.

[0051] The present invention discloses a method for visualizing molten pool behavior based on machine vision. The method automatically identifies the position of tungsten particles through deep learning technology, and combines a multi-target tracking algorithm to achieve real-time tracking of the motion trajectory of tungsten particles and accurate calculation of the speed direction. The system greatly improves the observation accuracy and efficiency of molten pool behavior, provides a new experimental method for studying molten pool behavior in laser processing, and overcomes the limitations of traditional manual observation.

[0052] Specifically, refer to Figure 1 As shown in the figure, the visualization method of melt pool behavior based on machine vision includes:

[0053] Step S1: Data acquisition and preprocessing: dynamic image data inside the molten pool is acquired through laser processing and synchrotron radiation imaging, or dynamic image data on the surface of the molten pool is acquired through laser processing and high-speed camera photography, and contrast adjustment and format conversion processing are performed on these data.

[0054] Step S2: Model training and testing: The YOLOv5 model is trained and tested using the preprocessed dataset to generate an optimal weight file for accurate detection of tungsten particles.

[0055] Step S3: Specific usage of the model: input the preprocessed data into the model, automatically detect tungsten particles, select particles through the interactive interface for real-time tracking, record their coordinates, and finally visualize the particle trajectories.

[0056] In a specific implementation, data acquisition and preprocessing include:

[0057] Laser Processing (Welding / Additive Manufacturing): Tungsten particles are introduced as tracers during laser welding or additive manufacturing to track the flow behavior inside the molten pool. During processing, the system introduces tungsten particles as tracers to track the behavior of the molten pool. For laser additive manufacturing, the system feeds tungsten particles into the molten pool synchronously with the powder material through a powder feeder, and evenly distributes the tungsten particles without affecting the process. For laser welding processes, the system allows tungsten particles to enter the molten pool by laying a small amount of tungsten powder on the welding surface.

[0058] Synchrotron radiation imaging or high-speed camera photography: Through synchrotron radiation imaging technology, the flow process inside the molten pool can be captured in real time, high-resolution X-ray images can be obtained, and the dynamic changes inside the molten pool can be displayed. In addition, by recording the flow behavior on the surface of the molten pool with a high-speed camera, the motion trajectory of the surface tracer particles can be captured, providing visual data support for indirect inference of the overall flow behavior of the molten pool. The combination of these two imaging methods provides multi-level dynamic information image data inside and on the surface of the molten pool.

[0059] Preprocessing: Contrast adjustment: adjust the contrast of the original image to improve the image quality and make the movement of tungsten particles more clearly visible; Image to video conversion: export the image sequence to video format for subsequent model training and analysis.

[0060] In a specific embodiment, model training and testing includes:

[0061] Dataset generation: Based on the preprocessed images and videos, tungsten particles are manually or automatically annotated to generate label data of tungsten particle locations for model training.

[0062] YOLOv5 model training: Use the generated dataset to train the YOLOv5 model, adjust model parameters such as anchor boxes and learning rates, and iterate the training repeatedly until the model can accurately identify tungsten particles.

[0063] Weight file generation: After multiple rounds of training and verification, a weight file best.pt suitable for actual use is generated for further application of the model.

[0064] Model testing: Use data from different scenarios to test the trained YOLOv5 model to verify its accuracy in tungsten particle detection and ensure that the model can be applied in actual operations.

[0065] In a specific implementation, the specific use of the model includes:

[0066] Data input: Import the preprocessed video or image data into the trained YOLOv5 model, and the model will automatically identify and annotate the tungsten particles in the molten pool.

[0067] Interactive page operation: The user selects tungsten particles with obvious behavioral characteristics through the interactive page, and the system begins to track the movement of these particles in real time.

[0068] Speed ​​and direction calculation: Based on continuous frame data, the system uses the speed and direction calculation unit, combined with the multi-target tracking function of the SORT algorithm, to calculate and display the speed and direction of the tungsten particles in real time. The SORT algorithm matches and tracks multiple targets through Kalman filtering and the Hungarian algorithm to ensure the continuity of the tungsten particles in multiple frames and predict their future movement positions to enhance the stability of tracking.

[0069] Log Recording: The system automatically stores the position information of the selected tungsten particles in a log file, preserving their trajectory data. The data in the log file can be used for visual analysis later.

[0070] Trajectory visualization: By reading the position information of tungsten particles in the log file, a visualization image of its motion trajectory is generated to help study the flow behavior of the molten pool and provide data support for subsequent process optimization.

[0071] The present invention also discloses a molten pool behavior visualization system based on machine vision, comprising:

[0072] Laser welding / additive manufacturing unit: performs laser welding / additive manufacturing. This unit controls the process parameters of laser welding or additive manufacturing and introduces tungsten particles into the molten pool during the processing as tracers to track the flow inside the molten pool.

[0073] Data acquisition unit: used to capture the flow image of the molten pool during welding or additive manufacturing, including a synchrotron radiation imaging unit to capture the flow inside the molten pool with X-ray imaging technology, and a high-speed camera to record the flow behavior on the surface of the molten pool, and obtain different levels of molten pool flow information through a variety of imaging methods;

[0074] Computer Vision Unit: The YOLOv5 model trained with deep learning algorithm is used to detect tungsten particles. The SORT multi-target tracking algorithm is used to analyze their motion trajectory in real time to ensure accurate identification of tungsten particles in and on the surface of the molten pool.

[0075] Data processing unit: By calculating the speed, direction and position changes of tungsten particles, the flow state of the molten pool is determined, and visual real-time tracking results are output.

[0076] Specifically, the laser welding / additive manufacturing unit includes a series of industrial-grade equipment such as a laser, a laser head, a powder feeder, a water cooler and an electrical control box, which can perform additive manufacturing or welding processes on the surface of a substrate.

[0077] The laser transmits the laser beam to the laser head through an optical fiber, focusing the energy on the substrate surface to heat and melt the powder material to form a molten pool. The powder feeder evenly feeds the additive material (such as AlSi7Mg powder) and a certain proportion of tungsten tracer particles into the molten pool to ensure that the tungsten particles are distributed in the molten pool for subsequent flow tracking.

[0078] Specifically, the data acquisition unit includes synchrotron radiation imaging and high-speed camera photography, wherein synchrotron radiation imaging is mainly used to track tracer particles inside the molten pool, while high-speed camera photography is mainly used to track tracer particles on the surface of the molten pool.

[0079] The synchrotron radiation imaging uses polychromatic X-rays to provide optimal penetration and signal-to-noise ratio. The X-ray energy is 20 keV, which can achieve high-precision observation and recording of the flow behavior inside the molten pool. When the laser moves on the top surface of the substrate, the X-rays can penetrate the sample for imaging. The imaging system consists of a 200μm thick LuAg scintillator and a 4x magnifying glass. After passing through the sample, the X-rays are converted into visible light by the scintillator and finally recorded by a high-resolution camera. The imaging field of view is 3mm×1.6mm (width×height). The entire observation process is recorded at a speed of 20,000 frames per second, with an exposure time between 1 and 40μs, lasting for several seconds, fully capturing the entire process of laser turning on, moving, and shutting down.

[0080] When the data high-speed camera captures images during the processing, the frame rate range of the capture is 5 to 200 fps.

[0081] By adding a small amount of tungsten particles into the powder feeding material, the particles enter the molten pool along with the powder material during the additive process. The movement trajectory of the tungsten particles in the molten pool can clearly reflect the flow direction, speed and morphological changes. The high density of tungsten particles can be clearly displayed in X-ray imaging, providing a good marker for observing the fluid movement inside the molten pool.

[0082] Specifically, the computer vision unit includes the deep learning algorithm YOLOv5 model and the SORT multi-target tracking algorithm, which are used together for real-time analysis of the movement of tungsten particles. YOLOv5 is used to accurately locate and identify tungsten particles and generate bounding box coordinates of their positions; the SORT algorithm tracks these identified tungsten particles and analyzes their movement trajectories in multiple frames of images, thereby achieving continuous tracking and analysis of the flow behavior inside or on the surface of the molten pool.

[0083] The YOLOv5 target detection model is characterized in that YOLOv5 generates multiple candidate bounding boxes (bbox) through an anchor box mechanism, such as Figure 2 As shown in the figure, a convolutional neural network is used to classify and regress these candidate boxes to determine the location and size of the target in the image. In this process, the image is divided into multiple sub-regions, and each grid is responsible for predicting the target in its region. When a tungsten particle is detected, the bounding box generated by the system is determined by the coordinates of the upper left corner (x1, y1) and the lower right corner (x2, y2), and the center point of the target (x_center, y_center) is recorded to assist in subsequent motion tracking. For tungsten particles of different scales, YOLOv5 dynamically adjusts the size of the bounding box to ensure accurate coverage of the target. On this basis, YOLOv5 achieves detection by classifying the candidate boxes and outputting the category (tungsten particle) and its confidence. In addition, YOLOv5 uses the non-maximum suppression (NMS) algorithm to filter out the optimal bounding box through the intersection over union (IoU) to avoid repeated detection.

[0084] The SORT multi-target tracking algorithm is characterized in that the SORT algorithm predicts the position of the tungsten particle in the next frame through Kalman filtering, and updates it using historical data such as the speed and acceleration of the previous frame. The tungsten particle bounding box of each frame is generated by YOLOv5 detection, and SORT matches these detection boxes with the results predicted by the Kalman filter through the Hungarian algorithm to ensure that the trajectory of the tungsten particles in consecutive frames remains consistent. For tungsten particles that are successfully matched, the system updates its trajectory information; if a tungsten particle is not detected in a certain frame, SORT will record the number of times the target is lost and remove it if it is not matched for multiple consecutive frames.

[0085] Specifically, the data processing unit implements the speed and direction real-time calculation unit based on the tungsten particle position information (x1, y1, t) provided by the SORT algorithm, and accurately calculates the movement speed and direction of the tungsten particles through mathematical formulas. The system compares the displacement of the tungsten particles in consecutive frames and uses the Euclidean distance formula:

[0086]

[0087] Among them, (x1, y1) and (x2, y2) are the coordinates of the tungsten particle in two consecutive frames. The displacement between the two frames is calculated, and the movement speed of the tungsten particle is calculated in real time in combination with the time interval t between the frames.

[0088] In addition, the system determines the direction of movement of the tungsten particles using the azimuth angle calculation formula based on the relative position changes of the tungsten particles in consecutive frames:

[0089] θ=atan 2 (y2-y1, x2-x1)

[0090] The results are updated in real time and stored in a data structure for further analysis of the melt pool flow behavior.

[0091] The data processing unit allows the user to select the tungsten particles to be observed in the program and save their position information in a log file. Through the interactive interface, the user can double-click to select the tungsten particle, and its coordinate position will be recorded in real time and stored in the log file. The log file contains the time series data of the tracer particle during the entire tracking process, and the trajectory can be calculated and visualized based on this data later.

[0092] The method for visualizing the molten pool behavior based on machine vision provided by the present invention is further illustrated below with reference to a specific case.

[0093] 1. Laser additive: DED powder feeding additive mode is adopted, the substrate is 1060 aluminum substrate, the powder feeding powder is AlSi7Mg powder, the particle size is about 50 microns, and 5% tungsten tracer particles are doped into the powder. In the laser additive process, the laser parameters are selected as follows: laser parameters are 1500W, scanning speed is 1200mm / min, powder feeding speed is 0.7r / min, powder feeding gas flow rate is 20L / min, and shielding gas flow rate is 12L / min. In the laser welding process, the original data is recorded by the synchrotron radiation fast program device to obtain the original data.

[0094] 2. Data preprocessing: The acquired raw synchrotron radiation imaging data was imported into the image processing software ImageJ, and the image quality was optimized by adjusting the brightness and contrast to make the motion trajectory of the tungsten particles more clearly visible. After the processing was completed, the image sequence was exported into a video format.

[0095] 3. Program preprocessing: Import the preprocessed video file into the trained program folder and change the file name to "testvideo" so that the program can recognize and test it (the program only tests files named "testvideo"). Then, configure the program's operating environment in the terminal, make sure the required dependent libraries and settings have been correctly loaded, and then start the automatic detection and analysis process.

[0096] 4. Program running: After the program is started, a demo interface (interactive interface) will be displayed, in which multiple tracer particles have been automatically identified by the system (such as the attached Figure 3 In this interface, the user can select a tracer particle of interest, such as tracer particle No. 20, and double-click the particle to display its real-time motion trajectory in green (see Appendix Figure 4 ) for easy observation. At the same time, the speed and direction information of the selected particle will be displayed in real time in the upper left corner of the interface. After a certain period of observation, the complete motion trajectory of tracer particle No. 20 will be gradually drawn (see attached Figure 5 ), providing intuitive data support for the analysis of molten pool flow behavior.

[0097] 5. Data post-processing. During the program running, the user can select multiple tracer particles for tracking analysis, such as particles No. 2, No. 62, and No. 99 (see Appendix Figure 5 ). After selecting these particles, the system will automatically store the coordinate data of their motion trajectory in the log folder of the program folder. By reading the data in the log file, the user can accurately reproduce and analyze the motion trajectory of the selected particles (see Appendix Figure 6 ), so as to further understand the flow behavior characteristics of the molten pool and provide a basis for the subsequent optimization of the additive process.

[0098] Those skilled in the art know that, in addition to realizing the system and its various devices, modules, and units provided by the present invention in a purely computer-readable program code, it is entirely possible to realize the same functions in the form of logic gates, switches, application-specific integrated circuits, programmable logic controllers, and embedded microcontrollers by logically programming the method steps. Therefore, the system and its various devices, modules, and units provided by the present invention can be considered as a hardware component, and the devices, modules, and units included therein for realizing various functions can also be regarded as structures within the hardware component; the devices, modules, and units for realizing various functions can also be regarded as both software modules for realizing the method and structures within the hardware component.

[0099] In the description of the present application, it should be understood that the terms "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present application.

[0100] The above describes the specific embodiments of the present invention. It should be understood that the present invention is not limited to the above specific embodiments, and those skilled in the art can make various changes or modifications within the scope of the claims, which does not affect the essence of the present invention. In the absence of conflict, the embodiments of the present application and the features in the embodiments can be combined with each other arbitrarily.

Claims

1. A method for visualizing molten pool behavior based on machine vision, characterized in that: include: Step S1: obtaining dynamic image data of the inside of the molten pool by synchrotron radiation imaging, and obtaining dynamic image data of the surface of the molten pool by high-speed camera photography; Step S2: using the preprocessed dynamic image data to train and test the YOLOv5 model, and generating a weight file for detecting tungsten particles in the molten pool; Step S3: The dynamic image data inside the molten pool and the dynamic image data on the surface of the molten pool are input into the trained YOLOv5 model to track and respond to the movement of the tungsten particles in real time to obtain a visualized image of the movement trajectory of the tungsten particles.

2. The method for visualizing molten pool behavior based on machine vision according to claim 1, characterized in that: The step S1 includes: introducing tungsten particles as tracers to track the flow behavior inside the molten pool; obtaining X-ray images through synchrotron radiation imaging to show the dynamic changes inside the molten pool; and recording the flow behavior on the surface of the molten pool through high-speed camera photography.

3. The method for visualizing molten pool behavior based on machine vision according to claim 1, characterized in that: The step S2 comprises: Step S2.1: annotating tungsten particles based on the preprocessed dynamic image data to generate label data of the tungsten particle positions for model training; Step S2.2: Use the generated data set to train the YOLOv5 model for multiple iterations until the model can accurately identify tungsten particles; Step S2.3: After multiple rounds of training and validation, the optimal weight file is generated and used for model application; Step S2.4: Use data from different scenarios to test the trained YOLOv5 model to verify its accuracy in detecting tungsten particles.

4. The method for visualizing molten pool behavior based on machine vision according to claim 1, characterized in that: The step S3 comprises: Step S3.1: Import the dynamic image data of the molten pool into the trained YOLOv5 model, and the model will automatically identify and annotate the tungsten particles in the molten pool; Step S3.2: The user selects a tungsten particle with obvious behavioral characteristics through the interactive page, and the system tracks the movement of the tungsten particle in real time; Step S3.3: Based on the continuous frame data, the system uses the speed and direction calculation unit, combined with the multi-target tracking function of the SORT algorithm, to calculate and display the speed and direction of the tungsten particles in real time; Step S3.4: the system automatically stores the position information of the selected tungsten particles into a log file and saves its trajectory data; Step S3.5: Generate a visualization image of the motion trajectory of the tungsten particle by reading the position information of the tungsten particle in the log file.

5. A machine vision-based molten pool behavior visualization system, characterized in that: Includes: laser welding and additive unit, data acquisition unit, computer vision unit, and data processing unit; The laser welding and additive manufacturing unit performs laser welding and additive manufacturing, and introduces tungsten particles as tracer particles during the processing; The data acquisition unit is used to capture the flow image of the molten pool during welding or additive manufacturing; The computer vision unit detects tungsten particles based on the YOLOv5 model trained by the deep learning algorithm, and analyzes their motion trajectory in real time in combination with the SORT multi-target tracking algorithm; The data processing unit determines the flow state of the molten pool by calculating the speed, direction and position changes of the tungsten particles, and outputs a visualized real-time tracking result.

6. The machine vision-based molten pool behavior visualization system according to claim 5, characterized in that: The data acquisition unit includes a synchrotron radiation imaging unit and a high-speed camera imaging unit; The synchrotron radiation imaging unit tracks the tracer particles inside the molten pool through X-ray imaging, and the high-speed camera imaging unit is used to track the tracer particles on the surface of the molten pool.

7. The machine vision-based molten pool behavior visualization system according to claim 5, characterized in that: The computer vision unit includes a deep learning algorithm YOLOv5 model and a SORT multi-target tracking algorithm; The YOLOv5 model accurately locates and identifies tungsten particles and generates bounding box coordinates of their positions. The SORT multi-target tracking algorithm continuously tracks and analyzes the flow behavior inside or on the surface of the molten pool by tracking the identified tungsten particles and analyzing their motion trajectories in multiple frames of images.

8. The machine vision-based molten pool behavior visualization system according to claim 5, characterized in that: The data processing unit accurately calculates the movement speed and direction of the tungsten particles based on the tungsten particle position information provided by the SORT algorithm; By comparing the displacement of the tungsten particles in consecutive frames, the displacement between two frames is calculated, and the movement speed of the tungsten particles is calculated by combining the time interval between the frames; The direction of movement of the tungsten particles is determined based on the relative position changes of the tungsten particles in consecutive frames.

9. The machine vision-based molten pool behavior visualization system according to claim 7, characterized in that: The YOLOv5 model generates multiple candidate bounding boxes through an anchor box mechanism, and uses a convolutional neural network to classify and regress these candidate boxes to determine the position and size of the target in the image; when tungsten particles are detected, the bounding box generated by the system is determined by the upper left and lower right corner coordinates, and the center point coordinates of the target are recorded for motion tracking; For tungsten particles of different sizes, the YOLOv5 model dynamically adjusts the size of the bounding box, classifies the candidate boxes and outputs the category and its confidence, uses the non-maximum suppression algorithm, and selects the optimal bounding box through the intersection-union ratio to avoid repeated detection.

10. The machine vision-based molten pool behavior visualization system according to claim 7, characterized in that: The SORT multi-target tracking algorithm predicts the position of the tungsten particle in the next frame through Kalman filtering and updates it using historical data such as velocity and acceleration of the previous frame; The bounding box of the tungsten particle in each frame is generated by the YOLOv5 model detection. The SORT multi-target tracking algorithm matches these detection boxes with the results predicted by the Kalman filter through the Hungarian algorithm. For the tungsten particles that are successfully matched, the system updates their trajectory information. If a tungsten particle is not detected in a frame, SORT will record the number of times the target is lost and remove it if it is not matched for multiple consecutive frames.

Images