On-machine detection method for surface roughness of laser polished 3D printed workpieces

Abstract

The present invention provides an on-machine detection method for the surface roughness of a 3D-printed workpiece laser polished. The method comprises the following steps: placing a 3D-printed workpiece to be processed on a workbench; adjusting the height of the laser by a lifting module and adjusting the position of the workbench by a mobile platform; performing surface polishing by the laser; capturing an image of the polished surface of the 3D-printed workpiece; extracting texture features of the grayscale co-occurrence matrix of the image; combining the texture feature vector with laser polishing processing parameters to form a feature parameter vector; inputting the feature parameter vector into a trained detection model to output a roughness value; completing the processing if the roughness value reaches the expected roughness value; otherwise, performing surface polishing again. The method can realize on-machine detection of the surface roughness of a 3D-printed workpiece laser polished, can timely rework 3D-printed workpieces that do not meet processing expectations, and can achieve surface roughness requirements with a single on-machine process, thereby improving polishing efficiency and polishing effect.

Description

Technical Field

[0001] The present invention relates to the technical field of laser polishing processing, and more specifically to an on-machine detection method for the roughness of the laser-polished surface of a 3D-printed workpiece. Background Art

[0002] TC4 titanium alloy has the advantages of low density, high strength, and good toughness, and is widely used in fields such as biomedicine, aerospace, and marine engineering. Titanium alloys are expensive and difficult to process into complex parts. To improve material utilization and machinability, 3D printing technology is used to manufacture parts. The surface roughness of 3D-printed workpieces is greater than that of traditionally machined workpieces, so 3D-printed workpieces require processing to improve surface quality. Laser polishing can complete the finishing of special surfaces of complex parts and is suitable for the surface treatment of 3D-printed TC4 titanium alloy workpieces. Roughness is the standard for evaluating the quality of workpiece surface processing. To ensure that the surface quality of the workpiece processing meets the requirements, it is particularly important to test the surface roughness of the workpiece.

[0003] Roughness detection can be divided into contact and non-contact methods. The traditional contact roughness detection method is represented by the stylus profilometer. When the stylus probe comes into direct contact with the surface of the workpiece being measured, scratches will be caused, which may affect the surface accuracy. In addition, the detection efficiency is low and the measurement range is small. Most of them are suitable for measuring the roughness of part planes and cannot meet the detection needs of the surface roughness of complex parts. Non-contact roughness detection methods include light sectioning, speckle method, machine vision method, etc., which can avoid problems such as scratching the workpiece surface. Roughness detection can determine whether the surface processing quality of the workpiece is qualified. Unqualified workpieces need to be reworked and polished again. In actual production, the workpiece needs to be polished repeatedly to meet the requirements. If the roughness is not up to standard after polishing, it needs to be placed under the laser again and the processing parameters must be manually adjusted for polishing, which to a certain extent affects the part processing efficiency. Therefore, designing an on-machine detection device and method for the polished surface roughness of 3D printed workpieces is a technical problem that needs to be solved urgently. Summary of the Invention

[0004] In order to overcome the shortcomings and deficiencies in the prior art, the purpose of the present invention is to provide a method for on-machine detection of the surface roughness of a 3D-printed workpiece laser polished. This method can improve the detection accuracy and stability of the roughness, realize on-machine detection of the surface roughness of a 3D-printed workpiece laser polished, and promptly rework 3D-printed workpieces that do not meet processing expectations, so that the surface roughness requirements can be met in one on-machine processing, thereby improving the polishing efficiency and polishing effect.

[0005] In order to achieve the above object, the present invention is implemented by the following technical solutions: a method for detecting the surface roughness of a 3D printed workpiece by laser polishing, characterized in that: it is realized by a detection device;

[0006] The detection device includes a workbench, a laser, a mobile platform, a lifting module, a machine vision device for capturing images of the 3D printed workpiece surface, and a distance sensor for detecting the distance between the laser and the 3D printed workpiece surface; the workbench is connected to the mobile platform to achieve movement in the x-axis and y-axis directions; the laser is connected to the lifting module to achieve lifting; the distance sensor is connected to the laser; the machine vision device includes an industrial camera with a telecentric lens and a perforated plate light source;

[0007] The on-machine detection method includes the following steps:

[0008] Step 1: Place the 3D printed workpiece to be processed on the workbench;

[0009] Step 2: Set the defocus amount and laser processing parameters; adjust the height of the laser through the lifting module and the position of the workbench through the moving platform according to the defocus amount, thereby adjusting the distance between the laser focus of the laser and the surface of the 3D printed workpiece; according to the laser processing parameters, the laser polishes the surface of the 3D printed workpiece;

[0010] Step 3: Move the workbench into the field of view of the industrial camera via the mobile platform;

[0011] Step 4: Turn on the light source of the hole plate, use an industrial camera to capture the image of the polished surface of the 3D printed workpiece, and pre-process the acquired image;

[0012] Step 5: Convert the image into a gray-level co-occurrence matrix; extract the texture features of the gray-level co-occurrence matrix to form a texture feature vector; the texture features include: energy ASM, which is used to reflect the uniformity of the image grayscale distribution and the coarseness of the texture; contrast CON, which is used to reflect the brightness contrast between pixel values ​​and area pixel values, thereby reflecting the image clarity and the depth of the texture grooves; entropy ENT, which is used to reflect the non-uniformity of the image texture; and inverse difference moment IDM, which is used to reflect the homogeneity of the image texture, thereby measuring the local change of the image texture;

[0013] Step 6: Combine the texture feature vector and the laser polishing processing parameters to form a feature parameter vector; input the feature parameter vector into the trained detection model, and the detection model outputs the roughness value of the polished surface of the 3D printed workpiece;

[0014] In step seven, the output roughness value is compared with the expected roughness value. If the expected roughness value is reached, the processing is completed; if the expected roughness value is not reached, the 3D printed workpiece is moved into the laser processing range through the mobile platform, and the surface is polished again in step two.

[0015] The method of the present invention can measure the surface roughness of a 3D printed workpiece after laser polishing, and simultaneously realize on-machine detection of the surface roughness, and can automatically return unqualified workpieces for reprocessing immediately.

[0016] The key to the present invention's use of machine vision methods is to extract image texture features and obtain a mapping model between feature indicators and roughness values; considering that the extracted image features may lack sufficient correlation with the surface roughness of the 3D printed workpiece, the measurement accuracy of the roughness model is affected. Since the processing parameters of laser polishing determine the processing energy, processing speed, processing frequency, etc., different parameters act on different surface textures on the 3D printed workpiece, thereby affecting the roughness value, that is, when processed with different combinations of parameters, the resulting roughness values ​​will also be different. Therefore, the present invention uses the texture feature vector extracted from the image and the laser polishing processing parameters as input parameters of the detection model to improve the detection accuracy and stability of the roughness model. By realizing on-machine detection of the laser polished surface roughness of 3D printed workpieces, workpieces that do not meet processing expectations can be reworked in a timely manner until the surface roughness of the workpiece reaches the expected level, thereby improving the processing efficiency and accuracy of laser polishing.

[0017] The method of the present invention can achieve the surface roughness requirement of the 3D printed workpiece in one machining operation, avoid the 3D printed workpiece from being repeatedly loaded and unloaded from the polishing equipment, reduce the number of operators required, and improve the polishing efficiency and polishing effect.

[0018] Preferably, the laser processing parameters include laser power, line spacing, processing times, repetition frequency and scanning speed.

[0019] Preferably, in step 4, the preprocessing includes contrast enhancement and median filtering denoising on the image.

[0020] Preferably, in step 5, the energy ASM, contrast CON, entropy ENT, and inverse moment IDM are respectively:

[0021]

[0022]

[0023]

[0024]

[0025] Where k is the number of rows and columns in the gray-level co-occurrence matrix; G(i,j) is the matrix value of the i-th row and j-th column in the gray-level co-occurrence matrix.

[0026] Preferably, in step six, the detection model is a BP neural network model; the detection model includes an input layer, a hidden layer, and an output layer; the detection model training method is: prepare a 3D printed workpiece sample, and measure the polished surface roughness value of the 3D printed workpiece sample through a three-dimensional profilometer; collect the polished surface image of the 3D printed workpiece sample, preprocess the image, and convert the image into a grayscale co-occurrence matrix; extract the texture features of the grayscale co-occurrence matrix to form a texture feature vector; combine the texture feature vector and the laser polishing processing parameters to form a feature parameter vector, which is input into the input layer of the detection model as a training parameter, and set the output result to the roughness value obtained by the three-dimensional profilometer measurement; by training the detection model, fit the function of the feature parameter vector and the roughness value in the hidden layer to obtain the mapping relationship between the feature parameter vector and the roughness value.

[0027] Preferably, in step seven, if the expected roughness value is not reached, the defocus amount is adjusted according to the difference between the output roughness value and the expected roughness value, and then the 3D printed workpiece is moved into the laser processing range by the mobile platform, and the process jumps to step two for surface polishing again.

[0028] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0029] 1. The present invention uses texture feature vectors extracted from images and laser polishing parameters as input parameters for the detection model, thereby improving the detection accuracy and stability of the roughness model. This enables on-machine detection of the surface roughness of laser-polished 3D-printed workpieces, enabling timely rework of workpieces that do not meet processing expectations until the surface roughness reaches the expected level.

[0030] 2. The method of the present invention can achieve the surface roughness requirements of 3D printed workpieces in one machining process, avoid repeatedly loading and unloading 3D printed workpieces from the polishing equipment, reduce the number of operators required, and improve polishing efficiency and polishing effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] Figure 1 This is a flow chart of the on-machine detection method for the laser polished surface roughness of a 3D printed workpiece according to the present invention;

[0032] Figure 2 It is a schematic structural diagram of a detection device in the on-machine detection method for the laser polished surface roughness of a 3D printed workpiece according to the present invention;

[0033] Figure 3 It is a network structure diagram of the detection model of the machine detection method for the laser polishing surface roughness of a 3D printed workpiece according to the present invention;

[0034] Among them, 1 is the laser, 2 is the x-axis moving platform, 3 is the y-axis moving platform, 4 is the lifting module, 5 is the ranging sensor, 6 is the industrial camera, 7 is the telecentric lens, 8 is the hole plate light source, and 9 is the fixed bracket. DETAILED DESCRIPTION

[0035] The present invention will be described in further detail below with reference to the accompanying drawings and specific embodiments.

[0036] Example

[0037] This embodiment provides a method for detecting the surface roughness of a 3D printed workpiece by laser polishing. Figure 1 As shown, it is achieved through a detection device.

[0038] Detection devices, such as Figure 2 As shown, the system includes a workbench, a laser 1, a distance sensor 5, a mobile platform, a lifting module 4, and a machine vision device. The mobile platform includes an x-axis mobile platform 2 and a y-axis mobile platform 3. The workbench is used to load the 3D printed workpiece. The workbench is connected to the mobile platform to enable movement in the x- and y-axis directions. The laser 1 is preferably an infrared nanosecond fiber laser. The laser beam emitted from the laser 1 polishes the surface of the 3D printed workpiece. The distance sensor 5 is used to detect the distance between the laser 1 and the surface of the 3D printed workpiece. The laser 1 is connected to the lifting module 4 to achieve lifting. The distance sensor 5 is connected to the laser 1 and adjusts the height of the laser 1 based on the distance feedback from the distance sensor 5.

[0039] The machine vision device includes an industrial camera 6 with a telecentric lens 7, a perforated plate light source 8, and a light source controller. The industrial camera 6 is secured by a mounting bracket 9. The industrial camera 6 captures images of the 3D-printed workpiece surface. The perforated plate light source 8 evenly illuminates the surface of the 3D-printed workpiece, with the reflected light entering the lens to facilitate imaging.

[0040] The on-machine detection method includes the following steps:

[0041] Step 1: Place the 3D printed workpiece to be processed on the workbench.

[0042] Step 2: Set the defocus and laser processing parameters. These parameters include laser power, line spacing, number of processing cycles, repetition rate, and scanning speed. Based on the defocus, the height of laser 1 is adjusted via the lifting module 4, and the position of the worktable is adjusted via the mobile platform. This adjusts the distance between the laser focus of laser 1 and the surface of the 3D printed workpiece. Laser 1 then polishes the surface of the 3D printed workpiece according to the laser processing parameters.

[0043] Step 3: Move the workbench into the field of view of the industrial camera 6 via the mobile platform.

[0044] Step 4: Turn on the hole plate light source, use the industrial camera 6 to capture the image of the polished surface of the 3D printed workpiece, and pre-process the acquired image; the pre-processing includes contrast enhancement and median filtering denoising of the image.

[0045] Step 5: Convert the image into a gray-level co-occurrence matrix; extract the texture features of the gray-level co-occurrence matrix to form a texture feature vector; the texture features include: energy ASM, which is used to reflect the uniformity of the image grayscale distribution and the coarseness of the texture; contrast CON, which is used to reflect the brightness contrast of the pixel value and the field pixel value, thereby reflecting the image clarity and the depth of the texture groove; entropy ENT, which is used to reflect the non-uniformity of the image texture; and inverse moment IDM, which is used to reflect the homogeneity of the image texture, thereby measuring the local change of the image texture; energy ASM, contrast CON, entropy ENT, and inverse moment IDM are:

[0046]

[0047]

[0048]

[0049]

[0050] Where k is the number of rows and columns in the gray-level co-occurrence matrix; G(i,j) is the matrix value of the i-th row and j-th column in the gray-level co-occurrence matrix.

[0051] Step 6: Combine the texture feature vector and the laser polishing processing parameters to form a feature parameter vector; input the feature parameter vector into the trained detection model, and the detection model outputs the roughness value of the polished surface of the 3D printed workpiece;

[0052] The detection model is a BP neural network model; the detection model includes an input layer, a hidden layer, and an output layer, such as Figure 3 As shown; the detection model training method is: prepare 3D printed workpiece samples, measure the roughness value of the polished surface of the 3D printed workpiece samples by a 3D profilometer; collect the polished surface image of the 3D printed workpiece samples, preprocess the images, and convert the images into a grayscale co-occurrence matrix; extract the texture features of the grayscale co-occurrence matrix to form a texture feature vector; combine the texture feature vector with the laser polishing processing parameters to form a feature parameter vector, which is input into the input layer of the detection model as a training parameter, and set the output result to the roughness value measured by the 3D profilometer; through training the detection model, fit the function of the feature parameter vector and the roughness value in the hidden layer to obtain the mapping relationship between the feature parameter vector and the roughness value.

[0053] In step seven, the output roughness value is compared with the expected roughness value. If the expected roughness value is reached, the processing is completed. If the expected roughness value is not reached, the 3D printed workpiece is moved into the processing range of the laser 1 via the mobile platform, and the process jumps to step two for surface polishing again. Preferably, if the expected roughness value is not reached, the defocus amount is adjusted based on the difference between the output roughness value and the expected roughness value, and the 3D printed workpiece is then moved into the processing range of the laser 1 via the mobile platform, and the process jumps to step two for surface polishing again.

[0054] The method of the present invention can measure the surface roughness of a 3D printed workpiece after laser polishing, and simultaneously realize on-machine detection of the surface roughness, and can automatically return unqualified workpieces for reprocessing immediately.

[0055] The key to the present invention's use of machine vision methods is to extract image texture features and obtain a mapping model between feature indicators and roughness values; considering that the extracted image features may lack sufficient correlation with the surface roughness of the 3D printed workpiece, the measurement accuracy of the roughness model is affected. Since the processing parameters of laser polishing determine the processing energy, processing speed, processing frequency, etc., different parameters act on different surface textures on the 3D printed workpiece, thereby affecting the roughness value, that is, when processed with different combinations of parameters, the resulting roughness values ​​will also be different. Therefore, the present invention uses the texture feature vector extracted from the image and the laser polishing processing parameters as input parameters of the detection model to improve the detection accuracy and stability of the roughness model. By realizing on-machine detection of the laser polished surface roughness of 3D printed workpieces, workpieces that do not meet processing expectations can be reworked in a timely manner until the surface roughness of the workpiece reaches the expected level, thereby improving the processing efficiency and accuracy of laser polishing.

[0056] The method of the present invention can achieve the surface roughness requirement of the 3D printed workpiece in one machining operation, avoid the 3D printed workpiece from being repeatedly loaded and unloaded from the polishing equipment, reduce the number of operators required, and improve the polishing efficiency and polishing effect.

[0057] The above embodiments are preferred implementation modes of the present invention, but the implementation modes of the present invention are not limited to the above embodiments. Any other changes, modifications, substitutions, combinations, and simplifications that do not deviate from the spirit and principles of the present invention should be considered as equivalent replacement methods and are included in the scope of protection of the present invention.

Claims

1. A method for detecting the surface roughness of a 3D printed workpiece by laser polishing, characterized by: This is achieved through a detection device; The detection device includes a workbench, a laser, a mobile platform, a lifting module, a machine vision device for capturing images of the 3D printed workpiece surface, and a distance sensor for detecting the distance between the laser and the 3D printed workpiece surface; the workbench is connected to the mobile platform to achieve movement in the x-axis and y-axis directions; the laser is connected to the lifting module to achieve lifting; the distance sensor is connected to the laser; the machine vision device includes an industrial camera with a telecentric lens and a perforated plate light source; The on-machine detection method includes the following steps: Step 1: Place the 3D printed workpiece to be processed on the workbench; Step 2: Set the defocus amount and laser polishing parameters; adjust the height of the laser through the lifting module and the position of the workbench through the moving platform according to the defocus amount, thereby adjusting the distance between the laser focus of the laser and the surface of the 3D printed workpiece; according to the laser polishing parameters, the laser polishes the surface of the 3D printed workpiece; Step 3: Move the workbench into the field of view of the industrial camera via the mobile platform; Step 4: Turn on the light source of the hole plate, use an industrial camera to capture the image of the polished surface of the 3D printed workpiece, and pre-process the acquired image; Step 5: Convert the image into a gray-level co-occurrence matrix; Extract the texture features of the gray-level co-occurrence matrix and form a texture feature vector; Texture features include: energy ASM, which reflects the uniformity of image grayscale distribution and texture coarseness; contrast CON, which reflects the brightness contrast of pixel values ​​and area pixel values, thereby reflecting image clarity and the depth of texture grooves; entropy ENT, which reflects the non-uniformity of image texture; and inverse difference moment IDM, which reflects the homogeneity of image texture and measures the local changes of image texture. Step 6: Combine the texture feature vector and the laser polishing processing parameters to form a feature parameter vector; input the feature parameter vector into the trained detection model, and the detection model outputs the roughness value of the polished surface of the 3D printed workpiece; In step seven, the output roughness value is compared with the expected roughness value. If the expected roughness value is reached, the processing is completed; if the expected roughness value is not reached, the 3D printed workpiece is moved into the laser processing range through the mobile platform, and the surface is polished again in step two.

2. The on-machine detection method for the surface roughness of a 3D printed workpiece laser polished according to claim 1, characterized in that: The laser polishing processing parameters include laser power, line spacing, processing times, repetition frequency and scanning speed.

3. The on-machine detection method for the surface roughness of a 3D printed workpiece laser polished according to claim 1, characterized in that: In the step 4, the preprocessing includes contrast enhancement and median filtering denoising on the image.

4. The on-machine detection method for the surface roughness of a 3D printed workpiece laser polished according to claim 1, characterized in that: In step 5, the energy ASM, contrast CON, entropy ENT, and inverse moment IDM are respectively: Where k is the number of rows and columns in the gray-level co-occurrence matrix; G(i,j) is the matrix value of the i-th row and j-th column in the gray-level co-occurrence matrix.

5. The on-machine detection method for the surface roughness of a 3D printed workpiece laser polished according to claim 1, characterized in that: In step six, the detection model is a BP neural network model; the detection model includes an input layer, a hidden layer, and an output layer; the detection model training method is as follows: preparing a 3D printed workpiece sample, measuring the polished surface roughness value of the 3D printed workpiece sample by a three-dimensional profilometer; collecting an image of the polished surface of the 3D printed workpiece sample, preprocessing the image, and converting the image into a grayscale co-occurrence matrix; extracting texture features of the grayscale co-occurrence matrix to form a texture feature vector; combining the texture feature vector with laser polishing processing parameters, The characteristic parameter vector is composed and input into the input layer of the detection model as a training parameter, and the output result is set to the roughness value measured by the three-dimensional profilometer; by training the detection model, the function of the characteristic parameter vector and the roughness value is fitted in the hidden layer to obtain the mapping relationship between the characteristic parameter vector and the roughness value.

6. The on-machine detection method for the surface roughness of a 3D printed workpiece laser polished according to claim 1, characterized in that: In step seven, if the expected roughness value is not reached, the defocus amount is adjusted according to the difference between the output roughness value and the expected roughness value, and then the 3D printed workpiece is moved into the laser processing range by the mobile platform, and the process jumps to step two for surface polishing again.

Images