Online powder beam quality detection method and system for laser cladding head

Abstract

The invention discloses an online powder beam quality detection method for a laser cladding head. The online powder beam quality detection method comprises the following steps that S1, internal parameters and external parameters of a camera are obtained; s2, acquiring gt by using a coaxial CCD (Charge Coupled Device) camera in a normal powder outlet state; three images of the flowing powder bundle are overlapped, and a circumcircle of the powder spot is obtained through fitting; s3, calculating the duty ratio of the powder particles in the circumcircle range, shrinking the circle, and when the duty ratio reaches 85%, the diameter of the circle is the effective powder spot diameter d pixel; s4, performing quantitative calibration on the diameter of the effective powder spot in the normal state by using a result of the step S1 to obtain a length D mm corresponding to the d pixel; s5, in the printing process, when only powder is emitted and no light is emitted, the diameters of effective powder spots are measured in real time for multiple times, and the mean value DAVG of the diameters of the effective powder spots is obtained; s6, delta D is calculated, if delta D is larger than a set threshold value, it is judged that the powder bundle is abnormal, and the number of abnormal points is accumulated; s7, if the number of the abnormal points is larger than or equal to the set threshold value, an alarm is given out, and the method is simple, easy to implement and good in practicability.

Description

Technical Field

[0001] This invention relates to the field of additive manufacturing technology, and in particular to an online powder beam quality inspection method and system for laser cladding heads. Background Technology

[0002] Directed Energy Deposition (DED) is an advanced additive manufacturing (3D printing) technology specifically designed for manufacturing or repairing high-value metal parts. During DED, multiple powder streams need to converge uniformly and stably to the cladding work plane to achieve higher powder utilization and better forming quality. The effective powder spot diameter on the powder stream convergence plane is one of the important indicators for evaluating powder stream quality.

[0003] In additive manufacturing metal powder directional energy deposition equipment, the quality of the laser cladding head powder bundle is not always stable due to various factors, which affects product quality. Traditionally, the on-site operator mainly relies on visual observation of the powder bundle convergence to determine whether the components need to be adjusted.

[0004] In existing technologies, a high-efficiency coaxial powder feeding nozzle convergence detection device disclosed in China (application number CN202120889399.9) requires offline qualitative testing, resulting in low detection efficiency. Another example is a monitoring and early warning method for replacing powder feeding tubes in 3D printing, disclosed in China (application number CN202510664632.6). This method utilizes the principle that powder convergence affects the molten pool deposition size, employing a laser coaxial measurement system and a lateral imaging data acquisition system for real-time measurement. It uses the melt width and melt height obtained from the laser coaxial measurement system for indirect judgment, and the lateral imaging data acquisition system for direct judgment. This two-pronged monitoring approach enables timely early warning and reconfirmation, efficiently and accurately determining the service life of the powder feeding tube automatically, reducing manual observation and judgment, and demonstrating good practicality. However, it has the following drawbacks:

[0005] First, the cause of changes in the toner overlap length d is not unique. Inaccurate settings in the printing program's lift amount may also cause the overlap length to change beyond expectations. In this case, there is no need to replace the toner tube. Second, the deposition height D can vary significantly when the printed workpiece collapses around its edges. Additionally, production efficiency is affected. When the program is paused and the lateral imaging data acquisition system takes a picture of the powder aggregation in the current toner tube, production stops.

[0006] Therefore, in view of the above problems, it is necessary for the present invention to provide a method for detecting powder quality in a highly efficient and simple manner during directional energy deposition. Summary of the Invention

[0007] The purpose of this invention is to overcome the shortcomings of the existing technology and provide an online powder beam quality detection method and system for laser cladding heads. This method quantitatively tests the powder spot diameter online and provides software alerts when the requirements are not met, thus achieving an efficient and simple way to detect powder beam quality during directional energy deposition.

[0008] To achieve the above objectives, the present invention provides the following technical solution:

[0009] This invention discloses an online powder beam quality detection method for laser cladding heads, comprising the following steps:

[0010] Step S1: Place the machine vision calibration plate on the powder convergence surface for calibration to obtain the camera's intrinsic and extrinsic parameters;

[0011] Step S2: Under normal powder output conditions, use a coaxial CCD camera to acquire more than 3 images of the flowing powder bundles. Use machine vision to overlay these images and fit them to obtain the circumcircle of the powder spot.

[0012] Step S3: Calculate the duty cycle of the powder particles within the circumcircle. Shrink the circle until it reaches an 85% duty cycle. The diameter of the circle is the effective powder spot diameter d pixels.

[0013] Step S4: Using the results of step S1, the effective powder spot diameter in the normal state is quantified and calibrated to obtain the length D millimeters corresponding to d pixels;

[0014] Step S5: During the printing process, when only toner is being emitted and no light is being emitted, the effective toner spot diameter is measured multiple times in real time, and its average value D is obtained. AVG ;

[0015] Step S6: Calculate ΔD = |DD AVG If ΔD is greater than the set threshold, it is judged as a powder bundle abnormality, and the number of abnormal points is accumulated.

[0016] Step S7: If the number of abnormal points is greater than or equal to the set threshold, an alarm is issued to prompt the operator to check.

[0017] Furthermore, step S1 includes the following steps:

[0018] Step S11: Turn on the lateral red laser illumination, with the laser beam positioned near the powder beam convergence surface;

[0019] Step S12: Place the prepared machine vision calibration plate under the cladding head on the powder convergence surface;

[0020] Step S13: Perform calibration to obtain the camera's intrinsic and extrinsic parameters for subsequent pixel and millimeter unit conversion.

[0021] Furthermore, step S2 includes the following steps:

[0022] Step S21: In normal powder dispensing mode, adjust the camera focus until a clear powder image can be seen;

[0023] Step S22: Multi-image reading and overlay fusion: Read multiple target images with the same resolution, accumulate the multiple images pixel by pixel, and generate an overlay image;

[0024] Step S23: Image normalization processing. Calculate the maximum and minimum global pixel gray values ​​of the superimposed image. Based on the maximum and minimum values, perform linear normalization processing on the superimposed image to map its pixel gray values ​​to a preset standard range.

[0025] Step S24: Region of Interest (ROI) extraction. The image processing range is limited to the ROI to obtain the ROI sub-image.

[0026] Step S25: Region segmentation and filtering. Threshold segmentation is performed on the ROI sub-image. A grayscale threshold range is set, and pixels with grayscale values ​​within the range are extracted to obtain an initial line bundle region. Connectivity analysis is performed on the initial line bundle region to obtain several independent connected regions. Regions that meet the requirements are filtered from all connected regions, and these regions are considered to be the effective regions of the target spot.

[0027] Step S26: Calculate the center and radius of the powder spot. Calculate the minimum circumcircle of the screened powder spot area. The coordinates of the center of the minimum circumcircle are the center position of the located wire harness, and the radius of the circle is the radius of the located wire harness.

[0028] Furthermore, in step S22, the multiple images are images continuously acquired in the same scene that contain the light spot to be detected.

[0029] 5. The method for online powder beam quality detection of a laser cladding head according to claim 1, characterized in that:

[0030] In step S23, the normalization algorithm is as follows: For each pixel value P_orig in the overlay image, its corresponding normalized pixel value P_norm is calculated using the following formula:

[0031] P_norm=(P_orig-hv_Min)*(255.0 / (hv_Max-hv_Min))

[0032] hv_Min is the minimum global pixel grayscale value;

[0033] hv_Max is the maximum grayscale value of all pixels globally.

[0034] Furthermore, step S3 includes the following steps:

[0035] Step S31: Calculate the duty cycle of the smallest circumcircle, that is, calculate the ratio of the area of ​​the powder spot to the smallest circumcircle;

[0036] Step S32: Repeat the operation, shrinking the smallest circumcircle each time, and calculate the percentage of the area of ​​the powder spot in the new circle until it is ≥85%, then exit the loop. The diameter of the circle at this time is taken as d.

[0037] Further, in step S51: during normal printing, take advantage of the gap when only toner is emitted but no light is emitted, repeat the operations of S2-S4, and accumulate the obtained diameters.

[0038] Step S52: Calculate the mean of the accumulated diameters to obtain the mean value D. AVG .

[0039] Furthermore, in step S6, if ΔD>0.03D occurs, the powder bundle quality is judged to be abnormal.

[0040] The detection system for implementing the above-described online powder beam quality detection method for laser cladding heads includes: a coaxial CCD camera, a lateral red laser illumination system, and a data processing system. The data processing system is connected to the coaxial CCD camera and includes an image processing unit, an anomaly statistics unit, and an external alarm unit.

[0041] A coaxial CCD is used to acquire image information of the light spot on the powder beam convergence surface;

[0042] Lateral red laser illumination is used to provide illumination conditions for the spot on the powder beam convergence surface;

[0043] The image processing unit is used to process image information acquired by the coaxial CCD to obtain the effective powder spot diameter D;

[0044] The anomaly statistics unit is used to calculate ΔD, compare it with the set threshold corresponding to ΔD, and count outliers;

[0045] The external alarm unit is used to count the number of anomalies. If the number of anomalies is greater than or equal to the set threshold, an early warning signal is sent out.

[0046] In the above technical solution, the present invention provides an online powder beam quality detection method and system for laser cladding heads, which has the following advantages:

[0047] This method utilizes coaxial imaging, whereby the radiation from particles on the powder beam convergence surface is projected onto a photosensitive unit via an optical imaging system. This converts the light intensity signal into a current / voltage signal, enabling powder spot image acquisition and storage. Image processing methods are then used to obtain powder spot diameter data, and early warning signals are sent out online in real time based on the frequency of anomalies. This method is simple to implement and highly practical, providing an efficient and straightforward way to measure powder beam quality during directional energy deposition. The efficient online detection method does not affect production efficiency; its simplicity requires only the addition of a lateral laser illumination to the cladding head containing a coaxial CCD camera, resulting in low equipment modification costs. Attached Figure Description

[0048] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0049] Figure 1 This is a schematic diagram of the state of the cladding head disclosed in this invention;

[0050] Figure 2 This is a schematic diagram of the powder spot on the powder clustering surface disclosed in this invention;

[0051] Figure 3 This is a flowchart of an online powder beam quality detection method for a laser cladding head disclosed in this invention;

[0052] Figure 4 This is a schematic diagram of the structure of an online powder beam quality inspection system for laser cladding heads disclosed in this invention.

[0053] Explanation of reference numerals in the attached figures:

[0054] 1. Lateral laser illumination; 2. Coaxial CCD camera; 3. Data processing system; 4. Cladding head; 5. Motion mechanism. Detailed Implementation

[0055] To enable those skilled in the art to better understand the technical solution of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings.

[0056] See Figure 1 As shown;

[0057] An online powder beam quality detection method for laser cladding heads is invented, comprising the following steps:

[0058] Step S1: With the device stationary, place the machine vision calibration plate on the powder convergence surface for calibration to obtain the camera's intrinsic and extrinsic parameters;

[0059] Step S2: Under normal powder output conditions, use a coaxial CCD camera to acquire more than 3 images of the flowing powder bundles. Use machine vision to overlay these images and fit them to obtain the circumcircle of the powder spot.

[0060] Step S3: Calculate the duty cycle of the powder particles within the circumcircle. Shrink the circle until it reaches an 85% duty cycle. The diameter of the circle is the effective powder spot diameter d pixels.

[0061] Step S4: Using the results of step S1, the effective powder spot diameter in the normal state is quantified and calibrated to obtain the length D millimeters corresponding to d pixels;

[0062] Step S5: During the printing process, when only toner is being emitted and no light is being emitted, the effective toner spot diameter is measured multiple times in real time, and its average value D is obtained. AVG ;

[0063] Step S6: Calculate ΔD = |DD AVG If ΔD is greater than the set threshold, it is judged as a powder bundle abnormality, and the number of abnormal points is accumulated.

[0064] Step S7: If the number of abnormal points is greater than or equal to the set threshold, an alarm is issued to prompt the operator to check.

[0065] To better implement the present invention, preferably, step S1 includes the following steps:

[0066] Step S11: Turn on the lateral red laser illumination, with the laser beam positioned near the powder beam convergence surface;

[0067] Step S12: Place the prepared machine vision calibration plate under the cladding head on the powder convergence surface;

[0068] Step S13: Perform calibration to obtain the camera's intrinsic and extrinsic parameters for subsequent pixel and millimeter unit conversion;

[0069] To better implement the present invention, preferably, step S2 includes the following steps:

[0070] Step S21: Put on a mask and take precautions. Under normal powder output conditions, adjust the camera focus until a clear powder image can be seen.

[0071] Step S22: Multi-image reading and overlay fusion;

[0072] Read multiple (e.g., 3) target images with the same resolution. The multiple images are images that contain the light spots to be detected, acquired consecutively in the same scene.

[0073] Multiple images are accumulated pixel by pixel to generate a superimposed image. The specific superimposition process is as follows: the pixel values ​​of the first image and the second image are added together to obtain the first intermediate superimposed image; then the pixel values ​​of the first intermediate superimposed image are added together with the third image, and so on, until all images are accumulated to obtain the final superimposed image ho_ImageSum.

[0074] This step effectively improves the signal-to-noise ratio of the image by superimposing multiple images, which enhances the grayscale value of target features such as line bundles, while random noise is relatively reduced due to mutual cancellation.

[0075] Step S23: Image normalization processing:

[0076] Calculate the global pixel grayscale maximum value hv_Max and minimum value hv_Min of the overlay image ho_ImageSum.

[0077] Based on the maximum and minimum values, the superimposed image is linearly normalized to map its pixel grayscale values ​​to a preset standard range (e.g., 0 to 255), resulting in a normalized image ho_normImage.

[0078] The specific normalization algorithm is as follows: For each pixel value P_orig in the overlay image, its corresponding normalized pixel value P_norm is calculated using the following formula:

[0079] P_norm=(P_orig-hv_Min)*(255.0 / (hv_Max-hv_Min))

[0080] This normalization step ensures that the problem of gray value overflow or uneven distribution that may occur after superposition is resolved, and the image is adjusted to the standard contrast range, which facilitates subsequent threshold segmentation processing.

[0081] Step S24: Region of Interest (ROI) extraction;

[0082] Based on the preset approximate position parameters of the light spot (e.g., center coordinates (vm.CenterY, vm.CenterX) and radius vm.Radius), a circular region of interest (ROI) ho_ROI_0 is generated on the normalized image ho_normImage;

[0083] By limiting the image processing range to the ROI region, the ROI sub-image ho_ImageReduced is obtained.

[0084] This step, by limiting the processing area, eliminates interference from irrelevant backgrounds in the image, significantly reducing the computational load of subsequent image processing and improving processing efficiency and positioning accuracy.

[0085] Step S25: Region segmentation and filtering;

[0086] Thresholding is performed on the ROI sub-image ho_ImageReduced. A grayscale threshold range [LowerValue, HigherValue] is set, and pixels with grayscale values ​​within this range are extracted to obtain an initial wireframe region ho_Region.

[0087] Connectivity analysis was performed on the initial harness region to obtain several independent connected regions ho_ConnectedRegions.

[0088] Based on preset area conditions (e.g., area between 5000 and 99999 pixels), select the region ho_SelectedRegion that meets the requirements from all connected regions. This region is considered to be the effective region of the target spot.

[0089] In this step, threshold segmentation and shape filtering can accurately separate the target line bundle from the background and remove noise interference areas that are too large or too small.

[0090] Step S26: Calculate the center and radius of the powder spot;

[0091] Calculate the minimum bounding circle of the selected powder spot region ho_SelectedRegion. The center coordinates of the minimum bounding circle (hv_RowCenter, hv_ColCenter) are the location of the center of the wire bundle. The radius of the circle hv_Radius is the radius of the located wire bundle. This step finally outputs the precise geometric parameters of the powder spot.

[0092] To better implement the present invention, preferably, step S3 includes the following steps:

[0093] Step S31: Calculate the duty cycle of the smallest circumcircle, i.e., calculate the ratio of the powder spot area to the smallest circumcircle. The powder spot area can be obtained by performing a region intersection operation on the connected component analysis results of step S22. Generally, the duty cycle at this time is <85%. If it is >85%, then step S32 is not necessary.

[0094] Step S32: Repeat the operation, shrinking the smallest circumcircle each time, and calculate the percentage of the area of ​​the powder spot in the new circle until it is ≥85%, then exit the loop. The diameter of the circle at this time is taken as d.

[0095] To better implement the present invention, preferably, step S5 includes the following steps:

[0096] Step S51: During normal printing, there will be gaps where only toner is emitted but no light is emitted. Use these gaps to repeat the operations of S2-S4 and sum up the obtained diameters.

[0097] Step S52: Calculate the mean of the accumulated diameters to obtain the mean value D. AVG ;

[0098] To better implement the present invention, preferably, in step S6, if ΔD>0.03D occurs, the powder bundle quality is judged to be abnormal.

[0099] An invention provides a detection system for online powder beam quality detection of a laser cladding head, comprising: a coaxial CCD camera, a lateral red laser illumination system, and a data processing system connected to the coaxial CCD camera; the data processing system includes an image processing unit, an anomaly statistics unit, and an external alarm unit.

[0100] A coaxial CCD is used to acquire image information of the light spot on the powder beam convergence surface;

[0101] Lateral red laser illumination is used to provide illumination conditions for the spot on the powder beam convergence surface;

[0102] The image processing unit is used to process the image information acquired by the coaxial CCD to obtain the effective powder spot diameter D;

[0103] Anomaly statistics unit, used to calculate ΔD = |DD AVG | Compare ΔD with the set threshold and count outliers;

[0104] The external alarm unit is used to count the number of anomalies. If the number of anomalies is greater than or equal to the set threshold, an early warning signal is sent to the host computer of the additive manufacturing metal powder directional energy deposition equipment.

[0105] See Figure 3 As shown, Example 1:

[0106] A method for online powder beam quality inspection of a laser cladding head involves measuring the spot size at the powder beam convergence surface of the cladding head 4 in a non-printing state to obtain the effective diameter, which is then used as a standard value D. During the printing process, when the cladding head only emits powder and not light, the spot size at the powder beam convergence surface is measured multiple times to obtain the average effective diameter D. AVG The system compares the absolute difference between the standard value and the mean value with a set threshold to determine whether the powder bundle quality is abnormal. If the powder bundle quality is abnormal, the number of abnormalities is accumulated. When the number of abnormalities reaches the set threshold, an alarm is issued.

[0107] The specific method includes the following steps:

[0108] Step S1: As Figure 1As shown, the powder convergence surface is used to place the machine vision calibration plate in its position, and the coaxial CCD camera 2 is used to calibrate it to obtain the camera's intrinsic and extrinsic parameters.

[0109] Step S2: Laser external multi-path powder feeding process. In order to obtain the image of the flowing powder spot, it is necessary to first turn on the side laser illumination 1, use a coaxial CCD camera to capture the state of the powder spot, rely on image processing technology to superimpose more than 3 images, and after morphological processing, fit to obtain the circumcircle of the powder spot.

[0110] Step S3: Calculate the duty cycle (the percentage of the area of ​​the powder spot to the area of ​​the circumscribed circle). If the duty cycle is <85%, shrink the circumscribed circle and repeat the duty cycle calculation until the duty cycle is >= 85%. At this point, the effective powder spot diameter d pixels is obtained. Figure 2 As shown;

[0111] Step S4: Based on the camera intrinsic and extrinsic parameters obtained in step S1, convert the effective powder spot diameter d pixels to D millimeters;

[0112] Step S5: The device starts printing;

[0113] Step S6: Laser illumination 1 remains constantly on. When the equipment is in a state of only emitting powder and not emitting light, the effective powder spot diameter D is obtained through multiple measurements. AVG Therefore, we obtain ΔD=|DD AVG |

[0114] Online monitoring of powder spot diameter does not affect production efficiency. This is because during the directional deposition of metal powder, according to the printing path planning, usually after one layer is completed and before the next layer is printed, only powder is emitted and no light is emitted during the movement of the cladding head 4. During this process, the effective powder spot diameter can be measured multiple times.

[0115] Step S7: Compare ΔD with the threshold 0.03D. If it is greater than ΔD, increment the number of abnormal points by one. Otherwise, wait for the next monitoring opportunity (only powder is emitted, no light).

[0116] Step S8: When the number of abnormal points exceeds the threshold for the number of abnormal points, an early warning signal is issued.

[0117] Example 2:

[0118] A method for online powder beam quality inspection of a laser cladding head includes the following steps:

[0119] Step S1: The coaxial CCD camera 2 takes a picture of the machine vision calibration board, and the data processing system 3 processes the image for subsequent pixel-to-distance conversion.

[0120] Step S2: Turn on the side laser illumination 1, feed powder into the 4-channel powder feeding and cladding head, use a coaxial CCD camera to capture the powder spot state, and the data processing system 3 processes the image to obtain an effective powder spot diameter D of 3.68mm.

[0121] Step S3: During the production process, after the first layer is printed, when only toner is emitted but no light is emitted, measure the effective toner spot diameter D multiple times;

[0122] Table 1:

[0123] Serial Number Effective powder spot diameter / mm 1 3.82 2 3.84 3 3.79 4 3.76

[0124] Data processing system 3 performs calculations on the data in Table 1, D AVG =3.80, thus obtaining ΔD = |DD AVG |=0.12, 0.12>0.03×D=0.03×3.68=0.11, so the number of abnormalities is increased by 1.

[0125] Step S4: If the number of anomalies is less than the anomaly count threshold (e.g., set to 3), continue printing.

[0126] Step S5: After printing the second and third layers, measure the effective toner diameter D multiple times.

[0127] Table 2:

[0128] Serial Number Effective powder spot diameter / mm 1 3.60 2 3.58 3 3.56 4 3.55

[0129] Table 3:

[0130] Serial Number Effective powder spot diameter / mm 1 3.86 2 3.81 3 3.87 4 3.79

[0131] The average effective spot size D for these two times AVG The values ​​are 3.57 and 3.83 respectively, and ΔD is 0.11 and 0.15 respectively.

[0132] Therefore, the measured values ​​were normal after the second layer, but abnormal after the third layer, and the cumulative number of abnormal values ​​increased to 2.

[0133] Step S6: If the number of anomalies is less than the anomaly count threshold (e.g., set to 3), continue printing.

[0134] Step S7: After printing the fourth layer, measure the effective toner diameter D multiple times.

[0135]

[0136]

[0137] D AVG=3.56, ΔD = 3.68 - 3.56 = 0.12, 0.12 > 0.03 × D, therefore the cumulative number of abnormalities increases to 3.

[0138] Step S8: If the number of anomalies 1 is greater than or equal to the number of anomalies threshold (e.g., set to 3), the data early warning system 3 sends out an early warning signal.

[0139] like Figure 4 As shown, Example 3:

[0140] An online powder beam quality inspection system for laser cladding head includes: lateral laser illumination 1, coaxial CCD camera 2, data processing system 3, cladding head 4, and motion mechanism 5;

[0141] Among them, the lateral laser illumination 1 is mainly used to illuminate the powder beam convergence surface, the coaxial CCD camera 2 is mainly used to collect image information of the powder spots, the cladding head 4 is the channel between the laser and the powder, and the motion mechanism 5 is mainly used to drive the cladding head 4, laser illumination 1, etc. to move.

[0142] Preferably, the data processing system 3 includes an image processing unit, an anomaly statistics unit, and an early warning unit;

[0143] The image processing unit is used to process the powder spot image captured by the coaxial CCD camera 2 to obtain the effective powder spot diameter D.

[0144] The exception handling unit is used to calculate D AVG ΔD, when ΔD>0.03D, accumulate the number of anomalies.

[0145] The early warning unit is used to compare the number of anomalies with the anomaly threshold and issue an early warning.

[0146] In the above technical solution, the present invention provides an online powder beam quality detection method and system for laser cladding heads. This method utilizes coaxial imaging, specifically by projecting the radiation light from particles on the powder beam convergence surface onto a photosensitive unit via an optical imaging system. This converts the light intensity signal into a current / voltage signal, enabling powder spot image acquisition and storage. Image processing methods are then used to obtain powder spot diameter data, and early warning signals are promptly sent online in real-time based on the frequency of anomalies. This method is simple, easy to implement, and has good practicality.

[0147] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A method for online powder beam quality inspection of a laser cladding head, characterized in that, Includes the following steps: Step S1: Place the machine vision calibration plate on the powder convergence surface for calibration to obtain the camera's intrinsic and extrinsic parameters; Step S2: Under normal powder output conditions, use a coaxial CCD camera to acquire more than 3 images of the flowing powder bundles. Use machine vision to overlay these images and fit them to obtain the circumcircle of the powder spot. Step S3: Calculate the duty cycle of the powder particles within the circumcircle. Shrink the circle until it reaches an 85% duty cycle. The diameter of the circle is the effective powder spot diameter d pixels. Step S4: Using the results of step S1, the effective powder spot diameter in the normal state is quantified and calibrated to obtain the length D millimeters corresponding to d pixels; Step S5: During the printing process, when only toner is being emitted and no light is being emitted, the effective toner spot diameter is measured multiple times in real time, and its average value D is obtained. AVG ; Step S6: Calculate ΔD = |DD AVG If ΔD is greater than the set threshold, it is judged as a powder bundle abnormality, and the number of abnormal points is accumulated. Step S7: If the number of abnormal points is greater than or equal to the set threshold, an alarm is issued to prompt the operator to check.

2. The method for online powder beam quality detection of a laser cladding head according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: Turn on the lateral red laser illumination, with the laser beam positioned near the powder beam convergence surface; Step S12: Place the prepared machine vision calibration plate under the cladding head on the powder convergence surface; Step S13: Perform calibration to obtain the camera's intrinsic and extrinsic parameters for subsequent pixel and millimeter unit conversion.

3. The method for online powder beam quality detection of a laser cladding head according to claim 1, characterized in that: Step S2 includes the following steps: Step S21: In normal powder dispensing mode, adjust the camera focus until a clear powder image can be seen; Step S22: Multi-image reading and overlay fusion: Read multiple target images with the same resolution, accumulate the multiple images pixel by pixel, and generate an overlay image; Step S23: Image normalization processing. Calculate the maximum and minimum global pixel gray values ​​of the superimposed image. Based on the maximum and minimum values, perform linear normalization processing on the superimposed image to map its pixel gray values ​​to a preset standard range. Step S24: Region of Interest (ROI) extraction. The image processing range is limited to the ROI to obtain the ROI sub-image. Step S25: Region segmentation and filtering. Threshold segmentation is performed on the ROI sub-image. A grayscale threshold range is set, and pixels with grayscale values ​​within the range are extracted to obtain an initial line bundle region. Connectivity analysis is performed on the initial line bundle region to obtain several independent connected regions. Regions that meet the requirements are filtered from all connected regions, and these regions are considered to be the effective regions of the target spot. Step S26: Calculate the center and radius of the powder spot. Calculate the minimum circumcircle of the screened powder spot area. The coordinates of the center of the minimum circumcircle are the center position of the located wire harness, and the radius of the circle is the radius of the located wire harness.

4. The method for online powder beam quality detection of a laser cladding head according to claim 1, characterized in that: In step S22, the multiple images are images continuously acquired in the same scene that contain the light spot to be detected.

5. The method for online powder beam quality detection of a laser cladding head according to claim 1, characterized in that: In step S23, the normalization algorithm is as follows: For each pixel value P_orig in the overlay image, its corresponding normalized pixel value P_norm is calculated using the following formula: P_norm=(P_orig-hv_Min)*(255.0 / (hv_Max-hv_Min)) hv_Min is the minimum global pixel grayscale value; hv_Max is the maximum grayscale value of all pixels globally.

6. The method for online powder beam quality detection of a laser cladding head according to claim 1, characterized in that: Step S3 Includes the following steps: Step S31: Calculate the duty cycle of the smallest circumcircle, that is, calculate the ratio of the area of ​​the powder spot to the smallest circumcircle; Step S32: Repeat the operation, shrinking the smallest circumcircle each time, and calculate the percentage of the area of ​​the powder spot in the new circle until it is ≥85%, then exit the loop. The diameter of the circle at this time is taken as d.

7. The method for online powder beam quality detection of a laser cladding head according to claim 1, characterized in that: Step S51: During normal printing, take advantage of the gap when only toner is emitted but no light is emitted, repeat the operations of S2-S4, and accumulate the obtained diameters. Step S52: Calculate the mean of the accumulated diameters to obtain the mean value D. AVG .

8. The method for online powder beam quality inspection of a laser cladding head according to claim 1, characterized in that: In step S6, if ΔD>0.03D, the powder cluster quality is judged to be abnormal.

9. A detection system for implementing the online powder beam quality detection method for laser cladding heads according to any one of claims 1-8, characterized in that, include: A coaxial CCD camera, a side-mounted red laser illumination system, and a data processing system are included. The data processing system is connected to the coaxial CCD camera and includes an image processing unit, an anomaly statistics unit, and an external alarm unit. A coaxial CCD is used to acquire image information of the light spot on the powder beam convergence surface; Lateral red laser illumination is used to provide illumination conditions for the spot on the powder beam convergence surface; The image processing unit is used to process image information acquired by the coaxial CCD to obtain the effective powder spot diameter D; The anomaly statistics unit is used to calculate ΔD, compare it with the set threshold corresponding to ΔD, and count outliers; The external alarm unit is used to count the number of anomalies. If the number of anomalies is greater than or equal to the set threshold, an early warning signal is sent out.

Images