Laser cladding molten pool classification method based on tracer particle trajectory statistics

Abstract

The present application relates to laser additive manufacturing and surface engineering technical field, provide a kind of laser cladding molten pool classification method based on tracer particle trajectory statistics, comprising: step S1, image acquisition: tracer particle is added to molten pool, and molten pool image sequence containing tracer particle is continuously collected by high-speed camera acquisition system in the process of laser cladding;Step S2, trajectory extraction: after the pre-processing of molten pool image sequence, the pixel position coordinates of tracer particle are extracted frame by frame based on pixel coordinate principle, and the motion trajectory of tracer particle on the surface of molten pool is generated;Step S3, trajectory condition analysis: calculate trajectory characteristic parameter, determine trajectory condition;Step S4, molten pool classification: trajectory condition is statistically analyzed, molten pool characteristic parameter is calculated, molten pool characteristic parameter is compared with preset classification criterion, and molten pool belonging to category is judged.The present application can preferably carry out laser cladding molten pool classification.

Description

Technical Field

[0001] This invention relates to the fields of laser additive manufacturing and surface engineering technology, and more specifically, to a method for classifying laser cladding pools based on tracer particle trajectory statistics. Background Technology

[0002] In laser cladding, the flow behavior of the molten pool directly determines the forming quality and defect formation of the cladding layer. Current technologies rely heavily on numerical simulation or subjective observation to determine the flow pattern of the molten pool, lacking objective and quantifiable classification methods based on experimental data. While tracer particle technology can visually display the flow on the molten pool surface, there is still no systematic solution for extracting features from a large number of random trajectories and establishing classification criteria. Summary of the Invention

[0003] The present invention provides a laser cladding molten pool classification method based on tracer particle trajectory statistics. This method can capture the cladding process with tracer particles added using a high-speed camera, extract the motion trajectory of the tracer particles, and classify the molten pool flow pattern according to the combination of trajectory types.

[0004] According to the present invention, a laser cladding molten pool classification method based on tracer particle trajectory statistics includes the following steps: Step S1, Image Acquisition: Tracer particles are added to the molten pool, and during the laser cladding process, a high-speed camera acquisition system continuously acquires a sequence of molten pool images containing tracer particles. Step S2, Trajectory Extraction: After preprocessing the molten pool image sequence, the pixel position coordinates of the tracer particles are extracted frame by frame based on the pixel coordinate principle to generate the motion trajectory of the tracer particles on the surface of the molten pool. Step S3, Trajectory Analysis: Calculate trajectory characteristic parameters and determine trajectory status; Step S4, Molten Pool Classification: Perform statistical analysis on the trajectory, calculate the molten pool characteristic parameters, compare the molten pool characteristic parameters with the preset classification criteria, and determine the category to which the molten pool belongs.

[0005] Preferably, in step S1, the high-speed camera acquisition system includes a high-speed camera, an auxiliary lighting source, and a narrowband filter. The high-speed camera is arranged at a 45° angle to the horizontal plane, and adopts an active illumination method, using an 808nm laser light source to illuminate the cladding area, and taking pictures in conjunction with a narrowband filter of the corresponding wavelength.

[0006] Preferably, in step S2, the trajectory extraction specifically involves: S2-1, Image preprocessing: Each frame of the acquired molten pool image is sequentially processed by grayscale conversion, Gaussian filtering for noise reduction, and contrast enhancement to highlight the edge and brightness features of the tracer particles. S2-2, Particle Recognition: An adaptive threshold segmentation algorithm is used to binarize the preprocessed image. Candidate particle regions are extracted using connected component analysis, and real tracer particles are selected based on particle area, circularity, and mean grayscale value. The centroid pixel coordinates of these particles in the image are then obtained. ; S2-3, Inter-frame matching: Based on the principle of the continuity of particle motion velocity between adjacent frames, the Hungarian algorithm or Kalman filter is used to perform correlation matching on the same particle in consecutive frames to establish the inter-frame correspondence of particles. S2-4. Trajectory Generation: Connect the pixel coordinates of the same particle in each frame in chronological order to generate the continuous motion trajectory of the particle, and record the timestamp corresponding to each frame to obtain the trajectory point set. t is the current frame number, and T is the total number of frames.

[0007] Preferably, in step S3, there are four possible trajectories for the tracer particles moving in the molten pool: (I) Moving from the center of the molten pool to all directions towards the edge; (II) It moves from the edge of the molten pool towards the central region; (III) Move along the edge of the molten pool without passing through the center of the molten pool; (IV) Located at the rear end of the molten pool, moving from a high point to a low point in the molten pool.

[0008] As a preferred option, the trajectory analysis in step S3 specifically includes: S3-1, Coordinate System and Feature Definition; The sequence of high-speed camera images of the molten pool containing tracer particles during the welding process was acquired and preprocessed to obtain: 3.11) Two-dimensional trajectory point sequence of tracer particles: , ,in Total number of frames; 3.12) Coordinates of the molten pool center for each frame: c represents the center; 3.13) The radius of the molten pool corresponding to each frame of the image: It is obtained from the radius of the equivalent circle of the molten pool area; 3.14) The height of the molten pool measured in real time during the cladding process: ; 3.15) Preset center and edge boundary values: This is used to divide the normalized radial distance into a central region and an edge region; Image coordinate system definition: The negative direction of the axis is behind the molten pool; S3-2, Calculation of characteristic quantities; 3.21) Calculate the normalized radial distance for each frame; For each frame Calculate the radial distance of the tracer particle relative to the center of the molten pool and divide it by the radius of the molten pool in that frame:

[0009] Obtain the sequence .

[0010] 3.22) Calculate the radial statistical characteristics; Radial distance from the starting point: ; Radial distance to the endpoint: Radial net change: Radial minimum: Radial maximum value: Radial fluctuation amplitude: ; 3.23) Calculate the backend flag ; Using only information from the trajectory start frame:

[0011] in The condition is the vertical offset of the starting point relative to the center of the molten pool. This indicates that the starting point is located in the rear end region of the molten pool; 3.24) Calculation of net displacement negative proportion ; Net displacement vector: , Represents the coordinates of the Nth frame; Displacement modulus: ; like :

[0012] like ,but ; Range of values A value close to 1 indicates that the net displacement of the trajectory is mainly oriented towards... Negative direction; S3-3, Classification Rules; The following conditions are checked in descending order of priority. Once a condition is met, the corresponding category is determined and subsequent checks are stopped. 3.31) Class IV Availability Determination; If the height of the molten pool If so, then Category IV is disabled; like Then Class IV is available; 3.32) Determine Class IV; Necessary and sufficient conditions: Class IV is available. ; backend flags ; net displacement negative proportion If the conditions are met, the classification is determined to be Category IV, and the process ends. 3.33) Classification: Class III; Necessary and sufficient conditions: Radial minimum Radial fluctuation amplitude If the conditions are met, the classification is determined to be Category III, and the process ends. 3.34) Determine whether it is Class II or Class I; Category II: and and

[0013] Then it is determined to be Category II, and the process ends; Category I: and and

[0014] Then it is determined to be of type I, and the process ends; If both of the above conditions are met due to numerical noise, it should be preferentially classified as Type II. 3.35) Boundary enforcement rules; If none of the outputs in steps 3.31) to 3.34) are triggered, then the following rules are applied sequentially, and an output is generated if a match is found: 3.351) According to the radial movement trend; Calculate the total radial inward amplitude and total outward amplitude Calculate the radial velocity for each time interval. , This refers to the inter-frame time interval. like Then it accumulates to ;like Then it accumulates to ; Compare and : like And a certain frame in the trajectory satisfies If so, it is classified as Category II; otherwise, if And a certain frame in the trajectory satisfies If so, it is classified as Class I; 3.352) By edge tangential features; like And the average tangential angle of the trajectory If so, it is classified as Category III; 3.353) By backend characteristics; Only when Class IV is available, i.e. :like and If so, it is classified as Category IV; 3.354) Guaranteed minimum rule; Compare the radial distances between the starting point and the ending point: like If the condition is met, it is classified as Class II; otherwise, it is classified as Class I.

[0015] Preferably, in step S4, the four types of tracer particle trajectories obtained from step S3 correspond to different molten pool driving mechanisms: Type I and Type III trajectories represent the Marangoni effect, Type II trajectories represent steam back pressure, and Type IV trajectories represent gravity.

[0016] Preferably, in step S4, after the molten pool reaches a stable stage, the number of various trajectories is counted, and the influence ratio of each driving mechanism is calculated using the following formula: (1) (2) (3) In the formula: P M P S P G N represents the percentage of influence from the Marangoni effect, steam back pressure, and gravity, respectively. I N II N III N IV , respectively, represent the number of trajectories that meet conditions (I), (II), (III), and (IV), and N represents the total number of trajectories counted.

[0017] The beneficial effects of this invention are as follows: This invention is based on the tracking particle trajectory acquired by high-speed cameras. By calculating quantifiable characteristic parameters such as normalized radial distance, radial fluctuation amplitude, rear marker and net displacement direction, and combining them with preset classification rules, the trajectory category is automatically determined. This overcomes the uncertainty of judgment caused by the reliance on numerical simulation or subjective human observation in traditional methods, and realizes objective and repeatable classification of molten pool flow patterns.

[0018] This invention classifies tracer particle trajectories into four typical cases (Type I: diffusion from center to edge; Type II: convergence from edge to center; Type III: tangential motion along edge; Type IV: motion from high to low at the rear end), corresponding to three physical driving mechanisms: the Marangoni effect, steam back pressure, and gravity. By statistically analyzing the proportion of each type of trajectory, the contribution of each driving mechanism to the molten pool flow can be quantitatively calculated, achieving quantitative decoupling of the coupling effects of multiple mechanisms and providing data support for a deeper understanding of molten pool flow behavior.

[0019] This invention designs a multi-level classification rule with clear priorities, and makes judgments on cases such as the trajectory being located at the edge throughout, having small radial fluctuations, radial convergence / diffusion, and directional movement at the rear end. It also sets boundary enforcement rules and minimum guarantee rules for ambiguous cases such as numerical noise and threshold boundaries, ensuring that any trajectory can be uniquely and unambiguously classified, avoiding classification blind spots.

[0020] This invention enables dynamic monitoring of the molten pool flow state through continuous high-speed camera acquisition and trajectory extraction. After the molten pool reaches a stable stage, it can statistically analyze the distribution of various trajectories and assess the influence ratio of each driving mechanism under current process parameters in real time. This provides feedback signals for online adjustment of laser power, scanning speed, or powder feeding rate, and has promising application prospects. Attached Figure Description

[0021] Figure 1 This is a flowchart illustrating a laser cladding pool classification method based on tracer particle trajectory statistics in an embodiment. Figure 2 This is a schematic diagram showing the location of the camera for acquiring images of the molten pool morphology in the embodiment; Figure 3 The graph shows the calculation results of Experiment 1-1 in the example (h<550). Figure 4 The calculation results for experiments 1-2 in the examples are shown in the figure (h<550). Figure 5 The calculation results for experiments 1-3 in the examples are shown in the figure (h<550). Figure 6 The calculation results for experiments 1-4 in the examples are shown in the figure (h<550). Figure 7 The calculation results for experiments 1-5 in the examples are shown in the figure (h<550). Figure 8 The graph shows the calculation results of experiments 1-6 in the example (550). Figure 9 The calculation results for experiments 1-7 in the examples are shown in the figure (h<550). Figure 10 The graph shows the calculation results of experiments 1-8 in the examples (550). Figure 11 The calculation results for experiments 1-9 in the examples are shown in the figure (h<550). Figure 12 The calculation results for experiments 1-10 in the examples are shown in the figure (h<550). Figure 13 The calculation results for experiments 1-11 in the examples are shown in the figure (h>850). Figure 14 The calculation results for experiments 1-12 in the examples are shown in the figure (h>850). Figure 15 The calculation results for experiments 1-13 in the examples are shown in the figure (h<550). Figure 16 The calculation results for experiments 1-14 in the examples are shown in the figure (h<550). Figure 17 The calculation results for experiments 1-15 in the examples are shown in the figure (h>850). Figure 18 The graph shows the calculation results of experiments 1-16 in the examples (h>850). Detailed Implementation

[0022] To further understand the content of this invention, a detailed description of the invention will be provided in conjunction with the accompanying drawings and embodiments. It should be understood that the embodiments are merely illustrative and not limiting of the invention.

[0023] Example like Figure 1 As shown, this embodiment provides a laser cladding pool classification method based on tracer particle trajectory statistics, which includes the following steps: Step S1, Image Acquisition: Tracer particles are added to the molten pool, and during the laser cladding process, a high-speed camera acquisition system continuously acquires a sequence of molten pool images containing the tracer particles.

[0024] The high-speed camera acquisition system includes a Thousand-Eye Wolf 5KF20 high-speed camera, an FC-W-808 auxiliary illumination source, and narrowband filters. The camera has a maximum resolution of 1920×1080 pixels and a maximum frame rate of 47,000 FPS (1920×64 image size), supporting the acquisition of regions of interest. To address interference caused by arc light during laser cladding, an active illumination method is employed, using an 808nm laser source to illuminate the cladding area, combined with a narrowband filter of the corresponding wavelength for imaging. This effectively filters out stray light and obtains clear images of the molten pool. The acquisition period is set to 150 µs, the exposure time to 50 µs, and the image size used during acquisition is 1472×336 pixels, with a scale of 185 pixels / mm to the actual size. The high-speed camera is deployed at... Figure 2The camera position shown is as follows: when capturing the height of the molten pool, the camera is placed horizontally; when capturing flow characteristics, the camera is at a 45° angle to the horizontal plane. The vertical distance between the camera lens and the substrate surface is set to approximately 20 mm.

[0025] Cladding was performed on a MobiMRO-2 mobile laser cladding device, and the experimental parameters are shown in Table 1. 45# steel was used as the laser cladding substrate material. The powder material used for acquiring the molten pool image was a mixture of AlN and Fe313 powder with a mass fraction of 1%. The Fe313 alloy powder was the main cladding material, with a particle size range of 45–106 µm; the AlN powder had a particle size of 50–80 µm.

[0026] Table 1 Orthogonal Experiment Parameter Table

[0027] Spherical AlN has a significantly higher melting point than Fe313 alloy, hardly melts during laser cladding, and has a lower density (approximately 3.26 g / cm³). 3 AlN particles float on the surface of the molten pool during the cladding process. Furthermore, aluminum has a high reflectivity to laser light, making AlN particles clearly visible and easily identifiable within the molten pool. Therefore, using AlN as a tracer particle and tracking its movement trajectory can effectively analyze the flow characteristics of the molten pool.

[0028] Since AlN powder particles enter the molten pool with the powder feed gas flow and their content is very low, the timing and location of tracer particles appearing in the molten pool are relatively random. However, under specific process parameters, whether in the molten pool evolution stage or the stage of reaching a stable state, the movement trajectory of tracer particles shows a distinguishable and certain regularity. The tracer particle trajectories at different times and locations characterize the transient flow field on the surface of the molten pool.

[0029] Step S2, Trajectory Extraction: After preprocessing the molten pool image sequence, the pixel position coordinates of the tracer particles are extracted frame by frame based on the pixel coordinate principle to generate the motion trajectory of the tracer particles on the surface of the molten pool.

[0030] Trajectory extraction specifically involves: S2-1, Image preprocessing: Each frame of the acquired molten pool image is sequentially processed by grayscale conversion, Gaussian filtering for noise reduction, and contrast enhancement to highlight the edge and brightness features of the tracer particles. S2-2, Particle Recognition: An adaptive threshold segmentation algorithm is used to binarize the preprocessed image. Candidate particle regions are extracted using connected component analysis, and real tracer particles are selected based on particle area, circularity, and mean grayscale value. The centroid pixel coordinates of these particles in the image are then obtained. ; S2-3, Inter-frame matching: Based on the principle of the continuity of particle motion velocity between adjacent frames, the Hungarian algorithm or Kalman filter is used to perform correlation matching on the same particle in consecutive frames to establish the inter-frame correspondence of particles. S2-4. Trajectory Generation: Connect the pixel coordinates of the same particle in each frame in chronological order to generate the continuous motion trajectory of the particle, and record the timestamp corresponding to each frame to obtain the trajectory point set. t is the current frame number, and T is the total number of frames.

[0031] Step S3, Trajectory Analysis: Calculate trajectory feature parameters and determine trajectory status.

[0032] Monitoring of 16 sets of orthogonal experimental molten pools revealed that the possible trajectories of tracer particles in the molten pool can be roughly categorized into four types: (I) Moving from the center of the molten pool to all directions towards the edge; (II) It moves from the edge of the molten pool towards the central region; (III) Move along the edge of the molten pool without passing through the center of the molten pool; (IV) Located at the rear end of the molten pool, moving from a high point to a low point in the molten pool.

[0033] The trajectory analysis is as follows: S3-1, Coordinate System and Feature Definition; The sequence of high-speed camera images of the molten pool containing tracer particles during the welding process was acquired and preprocessed to obtain: 3.11) Two-dimensional trajectory point sequence of tracer particles: , ,in Total number of frames; 3.12) Coordinates of the molten pool center for each frame: c represents the center; 3.13) The radius of the molten pool (in pixels) corresponding to each frame of the image: It is obtained from the radius of the equivalent circle of the molten pool area; 3.14) The height of the molten pool measured in real time during the cladding process: (Unit: μm); 3.15) Preset center and edge boundary values: Used to divide the normalized radial distance into a central region ( and edge areas ; Image coordinate system definition: The negative axis direction is behind the molten pool (needs to be calibrated according to the installation posture of the high-speed camera).

[0034] S3-2, Calculation of characteristic quantities; 3.21) Calculate the normalized radial distance for each frame; For each frame Calculate the radial distance of the tracer particle relative to the center of the molten pool and divide it by the radius of the molten pool in that frame:

[0035] Obtain the sequence .

[0036] 3.22) Calculate the radial statistical characteristics; Radial distance from the starting point: ; Radial distance to the endpoint: Radial net change: Radial minimum: Radial maximum value: Radial fluctuation amplitude: .

[0037] 3.23) Calculate the backend flag ; Use only the trajectory start frame Information:

[0038] in The condition is the vertical offset of the starting point relative to the center of the molten pool. This indicates that the starting point is located in the rear end region of the molten pool.

[0039] 3.24) Calculation of net displacement negative proportion ; Net displacement vector: , Represents the coordinates of the Nth frame; Displacement modulus: ; like :

[0040] like ,but ; Range of values A value close to 1 indicates that the net displacement of the trajectory is mainly oriented towards... Negative direction (moving downwards).

[0041] S3-3, Classification Rules; The following conditions are checked in descending order of priority. Once a condition is met, the corresponding category is determined and subsequent checks are stopped.

[0042] 3.31) Class IV Availability Determination; If the height of the molten pool If so, then Class IV is disabled (no trajectory will be classified as Class IV in subsequent steps); like Then Class IV is available.

[0043] 3.32) Determine Class IV (rear end of the molten pool, moving along the negative y direction); Necessary and sufficient conditions: Class IV is available. ; backend flags ; net displacement negative proportion If the conditions are met, the classification is determined to be Class IV, and the process ends.

[0044] 3.33) Class III (moving along the edge of the molten pool, without passing through the center); Necessary and sufficient conditions: Radial minimum (The entire trajectory lies within the edge region); Radial fluctuation amplitude (The trajectory is basically along an arc of equal radius with little radial variation); if this condition is met, it is classified as Class III, and the process ends.

[0045] 3.34) Determine whether it is Category II or Category I (radial convergence and diffusion); Type II (convergence from the edge to the center): and and

[0046] If it is determined to be Category II, then the process ends.

[0047] Type I (diffusion from center to edge): and and

[0048] If it is determined to be of type I, then the process ends.

[0049] If both of the above conditions are met due to numerical noise, it should be preferentially classified as Class II.

[0050] 3.35) Boundary enforcement rules (processing remaining trajectories); If none of the outputs in steps 3.31) to 3.34) are triggered, then the following rules are applied sequentially, and an output is generated if a match is found: 3.351) According to the radial movement trend; Calculate the total radial inward amplitude and total outward amplitude Calculate the radial velocity for each time interval (adjacent frames). , This refers to the inter-frame time interval. like Then it accumulates to (Take the absolute value); if Then it accumulates to ; Compare and : like And a certain frame in the trajectory satisfies If so, it is classified as Category II; otherwise, if And a certain frame in the trajectory satisfies If so, it is classified as Class I.

[0051] 3.352) By edge tangential features; like And the average tangential angle of the trajectory (The tangential angle is defined as the angle between the velocity direction and the radial direction, calculated by the displacement of adjacent frames), then it is judged as Class III.

[0052] 3.353) By backend characteristics (relaxed conditions); Only when Class IV is available, i.e. :like and If so, it is classified as Category IV.

[0053] 3.354) Guaranteed minimum rule; Compare the radial distances between the starting point and the ending point: like If the condition is met, it is classified as Class II; otherwise, it is classified as Class I.

[0054] Step S4, Molten Pool Classification: Perform statistical analysis on the trajectory, calculate the molten pool characteristic parameters, compare the molten pool characteristic parameters with the preset classification criteria, and determine the category to which the molten pool belongs.

[0055] Based on the four types of tracer particle trajectories obtained from step S3, they correspond to different molten pool driving mechanisms: Type I and Type III trajectories represent the Marangoni effect, Type II trajectories represent steam back pressure, and Type IV trajectories represent gravity.

[0056] After the molten pool reaches a stable stage, the number of various trajectories is counted, and the influence ratio of each driving mechanism is calculated using the following formula: (1) (2) (3) In the formula: P M P SP G N represents the percentage of influence from the Marangoni effect, steam back pressure, and gravity, respectively. I N II N III N IV These represent the number of trajectories that meet conditions (I), (II), (III), and (IV), respectively, and N is the total number of trajectories counted (usually no less than 40).

[0057] The calculation results of 16 orthogonal experiments are as follows Figure 3-18 As shown in the figure, h is the height of the molten pool, in µm.

[0058] The process parameters and molten pool characteristics of the three flow modes can be clearly defined: (1) Type I flow pattern: P M >90%, P S <10%, P G =0. This usually occurs under process parameters such as low laser power, high scanning speed, and high powder feeding rate. At this time, the energy input is insufficient, the molten pool temperature is low and the size is small, the liquid metal cannot evaporate violently, and the molten pool flow is dominated by the Marangoni effect, resulting in weak flow.

[0059] (2) Second type of flow pattern: P M <90%, P S >10%, P G =0. This condition often occurs under a combination of process parameters: high laser power, low scanning speed, and low powder feed rate. Under these conditions, the energy input is large, the molten pool temperature is high, the liquid metal volatilizes violently, and the steam back pressure has a significant effect; the temperature gradient is large, and the Marangoni effect is also strong. Conversely, insufficient material input means that most of the energy is used to melt the matrix, resulting in a large melt depth but a small melt height. The overall flow of the molten pool is violent, driven by both steam back pressure and the Marangoni effect.

[0060] (3) Third type of flow pattern: P M <90%, P S >10%, P G >0. This typically occurs under a combination of process parameters: high laser power, low scanning speed, and high powder feeding rate. Under these conditions, the energy and material input to the molten pool are abundant, resulting in strong lateral, longitudinal, and vertical expansion capabilities, a large three-dimensional size of the molten pool, and a pronounced gravitational effect. Simultaneously, the molten pool temperature is high, leading to intense evaporation of the liquid metal and significant vapor backpressure; the temperature gradient is large, and the Marangoni effect is equally strong. The overall flow of the molten pool is vigorous, driven primarily by three mechanisms.

[0061] This embodiment extracts the trajectory of tracer particles and quantifies it according to multi-level classification rules. It can objectively and unambiguously decouple the proportion of the effects of the Marangoni effect, steam back pressure and gravity in the molten pool, providing a reproducible data-driven basis for the optimization and control of laser cladding process parameters.

[0062] The present invention and its embodiments have been described above illustratively. This description is not restrictive, and the figures shown are only one embodiment of the present invention; the actual structure is not limited thereto. Therefore, if those skilled in the art are inspired by this description and design similar structures and embodiments without departing from the spirit of the present invention, such designs should fall within the protection scope of the present invention.

Claims

1. A laser cladding molten pool classification method based on tracer particle trajectory statistics, characterized in that, Includes the following steps: Step S1, Image Acquisition: Tracer particles are added to the molten pool, and during the laser cladding process, a high-speed camera acquisition system continuously acquires a sequence of molten pool images containing tracer particles. Step S2, Trajectory Extraction: After preprocessing the molten pool image sequence, the pixel position coordinates of the tracer particles are extracted frame by frame based on the pixel coordinate principle to generate the motion trajectory of the tracer particles on the surface of the molten pool. Step S3, Trajectory Analysis: Calculate trajectory characteristic parameters and determine trajectory status; In step S3, the trajectory analysis is as follows: S3-1, Coordinate System and Feature Definition; The sequence of high-speed camera images of the molten pool containing tracer particles during the welding process was acquired and preprocessed to obtain: 3.11) Two-dimensional trajectory point sequence of tracer particles: , ,in Total number of frames; 3.12) Coordinates of the molten pool center for each frame: , c Indicates the center; 3.13) The radius of the molten pool corresponding to each frame of the image: It is obtained from the radius of the equivalent circle of the molten pool area; 3.14) The height of the molten pool measured in real time during the cladding process: ; 3.15) Preset center and edge boundary values: This is used to divide the normalized radial distance into a central region and an edge region; Image coordinate system definition: The negative direction of the axis is behind the molten pool; S3-2, Calculation of characteristic quantities; 3.21) Calculate the normalized radial distance for each frame; For each frame Calculate the radial distance of the tracer particle relative to the center of the molten pool and divide it by the radius of the molten pool in that frame: ； Obtain the sequence ; 2.22) Calculate the radial statistical characteristics; Radial distance from the starting point: ; Radial distance to the endpoint: Radial net change: Radial minimum: Radial maximum value: Radial fluctuation amplitude: ; 3.23) Calculate the backend flag ; Using only information from the trajectory start frame: ； in The condition is the vertical offset of the starting point relative to the center of the molten pool. This indicates that the starting point is located in the rear end region of the molten pool; 3.24) Calculation of net displacement negative proportion ; Net displacement vector: , Indicates the first N Frame coordinates; Displacement modulus: ; like : ； like ,but ; Range of values A value close to 1 indicates that the net displacement of the trajectory is mainly oriented towards... Negative direction; S3-3, Classification Rules; The following conditions are checked in descending order of priority. Once a condition is met, the corresponding category is determined and subsequent checks are stopped. 3.31) Class IV Availability Determination; If the height of the molten pool Then Class IV is disabled; like Then Class IV is available; 3.32) Determine Class IV; Necessary and sufficient conditions: Class IV is available. ; backend flags ; net displacement negative proportion If the conditions are met, the classification is determined to be Category IV, and the process ends. 3.33) Classification: Class III; Necessary and sufficient conditions: Radial minimum Radial fluctuation amplitude If the conditions are met, the classification is determined to be Category III, and the process ends. 3.34) Determine whether it is Class II or Class I; Category II: and and ; Then it is determined to be Category II, and the process ends; Category I: and and ; Then it is determined to be of type I, and the process ends; If both of the above conditions are met due to numerical noise, it should be preferentially classified as Type II. 3.35) Boundary enforcement rules; If none of the outputs in steps 3.31) to 3.34) are triggered, then the following rules are applied sequentially, and an output is generated if a match is found: 3.351) According to the radial movement trend; Calculate the total radial inward amplitude and total outward amplitude Calculate the radial velocity for each time interval. , This refers to the inter-frame time interval. like Then it accumulates to ;like Then it accumulates to ; Compare and : like And a certain frame in the trajectory satisfies If so, it is classified as Category II; otherwise, if And a certain frame in the trajectory satisfies If so, it is classified as Class I; 3.352) By edge tangential features; like And the average tangential angle of the trajectory If so, it is classified as Category III; 3.353) By backend characteristics; Only when Class IV is available, i.e. :like and If so, it is classified as Category IV; 3.354) Guaranteed minimum rule; Compare the radial distances between the starting point and the ending point: like If yes, it is classified as Class II; otherwise, it is classified as Class I. Step S4, Molten Pool Classification: Perform statistical analysis on the trajectory, calculate the molten pool characteristic parameters, compare the molten pool characteristic parameters with the preset classification criteria, and determine the category to which the molten pool belongs.

2. The laser cladding pool classification method based on tracer particle trajectory statistics according to claim 1, characterized in that, In step S1, the high-speed camera acquisition system includes a high-speed camera, an auxiliary lighting source, and a narrowband filter. The high-speed camera is arranged at a 45° angle to the horizontal plane and uses active illumination. It illuminates the cladding area with an 808nm laser light source and takes pictures in conjunction with a narrowband filter of the corresponding wavelength.

3. The laser cladding pool classification method based on tracer particle trajectory statistics according to claim 2, characterized in that, In step S2, the trajectory extraction specifically involves: S2-1, Image preprocessing: Each frame of the acquired molten pool image is sequentially processed by grayscale conversion, Gaussian filtering for noise reduction, and contrast enhancement to highlight the edge and brightness features of the tracer particles. S2-2, Particle Recognition: An adaptive threshold segmentation algorithm is used to binarize the preprocessed image. Candidate particle regions are extracted using connected component analysis, and real tracer particles are selected based on particle area, circularity, and mean grayscale value. The centroid pixel coordinates of these particles in the image are then obtained. ; S2-3, Inter-frame matching: Based on the principle of the continuity of particle motion velocity between adjacent frames, the Hungarian algorithm or Kalman filter is used to perform correlation matching on the same particle in consecutive frames to establish the inter-frame correspondence of particles. S2-4. Trajectory Generation: Connect the pixel coordinates of the same particle in each frame in chronological order to generate the continuous motion trajectory of the particle, and record the timestamp corresponding to each frame to obtain the trajectory point set. , t For the current frame number, T This represents the total number of frames.

4. The laser cladding pool classification method based on tracer particle trajectory statistics according to claim 3, characterized in that, In step S3, there are four possible trajectories for the tracer particles moving in the molten pool: (I) Moving from the center of the molten pool to all directions towards the edge; (II) It moves from the edge of the molten pool towards the central region; (III) Move along the edge of the molten pool without passing through the center of the molten pool; (IV) Located at the rear end of the molten pool, moving from a high point to a low point in the molten pool.

5. The laser cladding pool classification method based on tracer particle trajectory statistics according to claim 4, characterized in that, In step S4, based on the four types of tracer particle trajectories obtained in step S3, they correspond to different molten pool driving mechanisms: Type I and Type III trajectories represent the Marangoni effect, Type II trajectories represent steam back pressure, and Type IV trajectories represent gravity.

6. The laser cladding pool classification method based on tracer particle trajectory statistics according to claim 5, characterized in that, In step S4, after the molten pool reaches a stable stage, the number of various trajectories is counted, and the influence ratio of each driving mechanism is calculated using the following formula: (1)； (2)； (3)； In the formula: P M , P S , P G These represent the percentages of influence from the Marangoni effect, steam back pressure, and gravity, respectively. N I , N II , N III , N IV The number of trajectories that meet conditions (I), (II), (III), and (IV) are respectively. N This represents the total number of trajectories counted.

Images