A laser powder bed fusion hot scan layer tomography quality evaluation method and system

Abstract

The application discloses a kind of laser powder bed melting hot scanning tomography quality evaluation method, system.The laser powder bed melting hot scanning tomography quality evaluation method of the application uses signal processing technology and image fusion technology to the layer section radiation data obtained with gradient image mode fusion update, generates layer section fusion gradient image, carries out gain noise reduction processing to layer section fusion gradient image, and layer section fusion gradient image is converted into pseudo-color chart, gradient abnormal area extraction operation is carried out to pseudo-color chart, the correlation model between gradient anomaly and printing anomaly and printing defect is established, using the correlation model established, in combination with the layer section radiation data obtained in real time, online quality evaluation can be realized, gradient anomaly in printing process is monitored in real time, and printing defect possibly appearing can be predicted, so that timely measures can be taken to intervene and adjust, reduce the rate of defective products, to realize cost reduction and efficiency increase.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of additive manufacturing, in particular to a laser powder bed fusion heat scanning tomography quality evaluation method and system. BACKGROUND

[0002] Laser powder bed fusion (LPBF) is an additive manufacturing technology that uses a laser beam to melt and solidify metal powder layer by layer at extremely high speed to form the required parts. It has a wide range of materials, integrates design and manufacturing, and has obvious advantages in the fine manufacturing of complex structures. It has been widely used in the manufacturing process of complex components of various equipment.

[0003] However, the thermal physical and non-equilibrium metallurgical processes of materials in the LPBF process are very complex. There is a complex interaction between the laser and the metal powder, the molten pool and the substrate. The heat conduction of the powder and the heat transfer process inside the molten pool are extremely complex. When the laser energy density reaches a certain threshold, the molten pool fluctuates violently under the coupling effect of recoil pressure, surface tension and Marangoni effect, and is easy to form spoon-shaped porosity defects. In addition, abnormal thermal history caused by rapid heating and cooling may also produce other defects such as un-melted holes and micro-cracks. Due to the possible existence of these internal quality defects, aviation parts are often required to be subjected to non-destructive testing such as X-ray or industrial CT. However, industrial CT has many limitations in detection cost, detection efficiency and accessibility of complex structures. At the same time, this post-detection scheme can only be tested after forming, and cannot timely find problems and optimize during the forming process. Even if the problem is found, the high manufacturing cost cannot be recovered.

[0004] Therefore, it is urgent to develop a laser powder bed fusion heat scanning tomography quality evaluation method, which aims to timely find possible printing defects during the forming process, so as to timely take measures to intervene and adjust, and realize cost reduction and efficiency increase. SUMMARY

[0005] The technical problem to be solved by the present application is to provide a laser powder bed fusion heat scanning tomography quality evaluation method and system.

[0006] To solve the above technical problems, the technical scheme adopted by the present application is as follows:

[0007] A laser powder bed fused thermal scanning tomography (FLST) quality evaluation method includes the following steps: real-time acquisition of layer cross-sectional radiation data during the layer cross-section processing; fusion and updating of the acquired layer cross-sectional radiation data in the form of gradient images using signal processing and image fusion techniques to generate a layer cross-sectional fused gradient image; gain reduction and noise reduction processing of the layer cross-sectional fused gradient image, and conversion of the layer cross-sectional fused gradient image into a pseudo-color image; gradient anomaly region extraction operation of the pseudo-color image, establishing a correlation model between gradient anomalies, printing anomalies, and printing defects, thereby achieving online quality evaluation.

[0008] A laser powder bed fusion thermal scanning tomography quality evaluation system includes: a tomographic detection sensor mounted on the top plate of the forming cavity, which can perform photosensitive measurement on the entire forming cross section to collect cross-sectional radiation data during the processing of the cross section; a control module for acquiring the cross-sectional radiation data in real time; using signal processing and image fusion technology to fuse and update the acquired cross-sectional radiation data in a gradient image manner to generate a fused gradient image of the cross section; performing gain denoising processing on the fused gradient image of the cross section and converting the fused gradient image of the cross section into a pseudo-color image; performing gradient anomaly region extraction operation on the pseudo-color image, establishing a correlation model between gradient anomalies, printing anomalies, and printing defects, thereby realizing online quality evaluation.

[0009] The beneficial technical effects of this invention are as follows: The above-mentioned laser powder bed fused thermal scanning tomography quality evaluation method utilizes signal processing technology and image fusion technology to fuse and update the acquired layer cross-sectional radiation data, generating a layer cross-sectional fused gradient image. The layer cross-sectional fused gradient image after gain denoising is converted into a pseudo-color image. Gradient anomaly region extraction is performed on the pseudo-color image, and a correlation model between gradient anomalies, printing anomalies, and printing defects is established. Using the established correlation model, combined with the real-time acquired layer cross-sectional radiation data, online quality evaluation can be achieved, gradient anomalies during the printing process can be monitored in real time, and possible printing defects can be predicted. Thus, timely intervention and adjustment measures can be taken to reduce the product defect rate, thereby achieving cost reduction and efficiency improvement. Attached Figure Description

[0010] Figure 1 This is a schematic diagram of the laser powder bed fusion thermal scanning chromatography quality evaluation system of the present invention;

[0011] Figure 2 This is a schematic flowchart of the laser powder bed fusion thermal scanning chromatography quality evaluation method of the present invention;

[0012] Figure 3 This is a schematic diagram of chromatographic fusion.

[0013] Explanation of reference numerals in the attached figures:

[0014] 1-Laser, 2-Laser beam, 3-Scanning device, 4-Part forming surface, 5-Forming platform, 6-Cross-section radiation, 7-Observation window, 8-Optical components, 9-Tomographic detection sensor, 10-Top plate of forming cavity. Detailed Implementation

[0015] To enable those skilled in the art to more clearly understand the purpose, technical solution, and advantages of the present invention, the present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0016] This invention provides a laser powder bed fused thermal scanning chromatography quality evaluation system.

[0017] like Figure 1 As shown, in one embodiment of the present invention, the laser powder bed fusion thermal scanning tomography quality evaluation system includes a tomography detection sensor 9 and a control module (not shown). The tomography detection sensor 9 is installed on the top plate 10 of the forming cavity. The tomography detection sensor 9 can perform photosensitive measurement on the entire forming cross section and is used to collect the cross section radiation data during the processing of the cross section. The control module is used to acquire the cross section radiation data in real time, use signal processing technology and image fusion technology to fuse and update the acquired cross section radiation data in the form of a gradient image, generate a cross section fused gradient image, perform gain denoising processing on the cross section fused gradient image, convert the cross section fused gradient image into a pseudo-color image, perform gradient anomaly region extraction operation on the pseudo-color image, establish a correlation model between gradient anomalies and printing anomalies and printing defects, and thus realize online quality evaluation.

[0018] The tomographic detection sensor 9 employs any one or any combination of a CMOS camera, photodiode array, near-infrared thermal imager, or infrared thermal imager. Since the emission spectra of different materials differ when heated and melted by a high-energy beam, different sensors can be selected as the tomographic detection sensor 9 based on the spectral emission curves of different metal materials during the melting and solidification process: first, the emission spectrum of the metal melted by the high-energy beam is measured and analyzed to determine the spectral emission curve of the metal material during the melting and solidification process; then, the target wavelength is determined based on the spectral emission curve of the metal material during the melting and solidification process; finally, a suitable sensor is selected as the tomographic detection sensor 9 based on the target wavelength.

[0019] Among them, the target band can be selected from the spectral emission curve of the metal material during the melting and solidification process. Depending on the different processing materials, the target band can be a plasma band, a laser reflection band, a near-infrared band, an infrared band, or a narrow band or a wide-range band.

[0020] The target wavelength band is the wavelength band of the cross-sectional radiation 6 that the tomographic detection sensor 9 is to collect. The tomographic detection sensor 9 should be able to collect all the cross-sectional radiation 6 within the target wavelength band. Therefore, after determining the target wavelength band, a suitable sensor can be selected as the tomographic detection sensor 9 based on the target wavelength band. If the target wavelength band is within the visible and near-infrared wavelength bands, the tomographic detection sensor 9 can be a CMOS camera, a photodiode array, or a near-infrared thermal imager; if the target wavelength band exceeds the near-infrared wavelength band, the tomographic detection sensor 9 can be an infrared thermal imager, etc.

[0021] For example Figure 1 As shown, the top plate 10 of the forming cavity is provided with an observation window 7, and the front end of the tomographic detection sensor 9 is provided with an optical component 8. The optical component 8 is used to transmit the layer cross section radiation 6 of a specific wavelength band to the tomographic detection sensor 9.

[0022] The wavelength band that the observation window 7 can transmit is consistent with the wavelength band of the layer cross-section radiation 6. When the wavelength band of the layer cross-section radiation 6 is in the plasma band, laser reflection band, and near-infrared band, the observation window 7 can be made of optical windows made of materials such as fused silica and N-BK7. When the wavelength band of the layer cross-section radiation 6 includes a portion greater than 2000nm, optical windows made of materials such as germanium, zinc sulfide, and zinc selenide can be used.

[0023] The optical component 8 is used to determine the wavelength band of the cross-sectional radiation 6 transmitted to the tomographic detection sensor 9, thereby transmitting the cross-sectional radiation 6 of a specific wavelength band to the tomographic detection sensor 9. The optical component 8 determines the wavelength band of light that can pass through it by adjusting the internal filter combination. The wavelength band of light that can pass through the optical component 8 can be a narrowband band (such as 800nm±25nm, 950nm±25nm, etc.) or a wide range of wavelengths (such as 1200nm-1800nm, 1200nm-2600nm, 2000nm-5000nm, etc.).

[0024] In the laser powder bed melting process, laser 1 emits a laser beam 2, which is used by scanning device 3 to heat and melt metal powder on the forming surface 4 of the part using a specific scanning strategy. During the heating, melting, and cooling solidification processes, layer cross-sectional radiation 6 is formed on the layer cross-section. The layer cross-sectional radiation 6 is transmitted through spatial light to the observation window 7 located on the top plate 10 of the forming cavity, and then through optical component 8 to the tomographic detection sensor 9, where it is collected to monitor the changes in the layer cross-sectional radiation gradient. It should be noted that the laser beam 2 emitted by laser 1 is a high-energy laser beam in the 1064nm band or other bands. The band of the layer cross-sectional radiation 6 can be plasma band, laser reflection band, near-infrared band, infrared band, etc.

[0025] This invention provides a method for evaluating the quality of laser powder bed fused thermal scanning chromatography.

[0026] like Figure 2 As shown, in one embodiment of the present invention, the laser powder bed fusion thermal scanning chromatography quality evaluation method includes steps S10 to S40:

[0027] S10. Real-time acquisition of layer cross-sectional radiation data during the layer cross-section processing.

[0028] In specific implementation, utilizing Figure 1 In the laser powder bed fusion thermal scanning chromatography quality evaluation system shown in the embodiment, the chromatography detection sensor 9 collects the cross-sectional radiation data of the cross-section in real time during the cross-sectional processing process, and then obtains the cross-sectional radiation data of the cross-section in real time through communication connection with the chromatography detection sensor 9.

[0029] S20. Using signal processing and image fusion techniques, the acquired cross-sectional radiation data is fused and updated in the form of gradient images to generate a cross-sectional fused gradient image.

[0030] like Figure 3 As shown, from left to right, the image frames acquired within a set unit time are fused and refreshed in units to gradually form a layer cross-section fused gradient image. The set unit time can be a short period (any value between 1-1000ms), a long period (1s-60s), or an ultra-long period (more than 1 minute, or even the sintering time of the entire layer).

[0031] In the image fusion process, if integral fusion is used, the fusion process continuously integrates each pixel within the frame image, generating a fused image that is an integral distribution image. If extremum fusion is used, the fusion process continuously calculates extrema for each pixel, generating a fused image that is an extremum distribution image. If weighted fusion is used, each pixel is multiplied by a weighting coefficient during integral fusion and extremum fusion, generating a weighted integral image and a weighted extremum image. If normalization distribution processing is applied to the image during fusion, normalizing the image gradient distribution to between 0 and 1, a normalized fused image is generated. When the layer cross-section processing is completed, the generated layer cross-section fused gradient image can be any one or any combination of integral distribution images, extremum distribution images, weighted integral images, weighted extremum images, and normalized fused images.

[0032] S30. Perform gain denoising processing on the fused gradient image of the layer cross sections, and convert the fused gradient image of the layer cross sections into a pseudo-color image.

[0033] Gain denoising methods, such as median filtering, Gaussian filtering, wavelet transform, or deep learning denoising, are employed to perform gain denoising on the fused gradient image of the layer cross sections. This aims to improve the contrast of the effective radiometric information in the image while ensuring that the gain algorithm does not introduce new noise or distortion. Simultaneously, the fused gradient image of the layer cross sections is converted into a pseudo-color image for analysis.

[0034] S40. Perform gradient anomaly region extraction on the pseudo-color image, establish a correlation model between gradient anomalies, printing anomalies, and printing defects, and then realize online quality evaluation.

[0035] Image algorithms such as pixel value thresholding, region segmentation, and morphological operations, or machine learning algorithms such as convolutional neural networks, are used to extract gradient anomaly regions from pseudo-color images. A correlation model between gradient anomalies and printing anomalies and defects is established, thereby enabling online quality assessment.

[0036] The pixel value thresholding method defines a threshold range for the overall grayscale distribution or gradient distribution of an image. Areas exceeding this range are considered gradient anomaly regions. Pixel value thresholding is a basic image segmentation method. Its fundamental principle is to classify pixels in an image into different categories (such as target and background) by setting one or more grayscale thresholds. In practice, each pixel in the image is traversed, and classification is performed based on the relationship between its grayscale value and the threshold. If a pixel's grayscale value is greater than the threshold, it is classified as a target; otherwise, it is classified as background.

[0037] The region segmentation method divides an image into sets of pixels corresponding to objects or object surfaces. There are various methods for region segmentation, including contour fitting, region growing, and split-merge methods. These methods typically divide regions based on the grayscale values ​​of the merged image. For example, region growing starts with a set of seed points and gradually "grows" regions based on the similarity between pixels (such as grayscale similarity) until a set stopping condition for abnormal regions (such as area, shape, etc.) is met.

[0038] The morphological operations described are image processing methods based on shape and structure, with the core being the definition and operation of structuring elements (SEs). Morphological operations include dilation, erosion, opening, and closing operations. Dilation can enlarge objects in an image, making them more connected; erosion, conversely, can shrink objects, making them more refined. Opening and closing operations are combinations of dilation and erosion, used for noise removal and edge smoothing, and filling holes, respectively. Morphological operations are widely used in image processing, especially in image segmentation, edge detection, and noise reduction, primarily for further precise processing of extracted anomalous regions.

[0039] Convolutional Neural Networks (CNNs) use structures such as convolutional layers, pooling layers, and fully connected layers to extract features and classify input images. In image processing, CNNs can automatically learn hierarchical feature representations in images, from low-level edge and texture features to high-level shape and structural features. Through training, CNNs can learn the mapping relationship between images and specific tasks (gradient anomaly detection), thereby achieving accurate extraction of gradient anomaly regions in images.

[0040] After extracting gradient anomaly regions, these regions are correlated with sample data such as anomalies in the actual printing process, or the internal quality or physicochemical properties of printed parts. Machine learning algorithms (such as support vector machines, random forests, and neural networks) are used to explore the relationship between gradient anomalies and printing quality, thereby establishing a correlation model between gradient anomalies, printing anomalies, and printing defects. Once the correlation model is established, online quality evaluation can be achieved, gradient anomalies during the printing process can be monitored in real time, and potential printing defects can be predicted, allowing for timely intervention and adjustments.

[0041] The laser powder bed fusion thermal scanning tomography quality evaluation method of the present invention utilizes signal processing technology and image fusion technology to fuse and update the acquired layer cross-sectional radiation data, generating a layer cross-sectional fused gradient image. The layer cross-sectional fused gradient image after gain reduction and noise reduction is converted into a pseudo-color image. Gradient anomaly region extraction is performed on the pseudo-color image, and a correlation model between gradient anomalies, printing anomalies, and printing defects is established. Using the established correlation model, combined with the real-time acquired layer cross-sectional radiation data, online quality evaluation can be achieved, gradient anomalies during the printing process can be monitored in real time, and potential printing defects can be predicted. This allows for timely intervention and adjustment, reducing the product defect rate and thus achieving cost reduction and efficiency improvement.

[0042] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention in any way. Those skilled in the art can make various equivalent changes and improvements based on the above embodiments, and all equivalent variations or modifications made within the scope of the claims should fall within the protection scope of the present invention.

Claims

1. A method for evaluating the quality of laser powder bed fused thermal scanning chromatography, characterized in that, Includes the following steps: S10. Real-time acquisition of layer cross-sectional radiation data during the layer cross-section processing; S20. Using signal processing and image fusion techniques, the acquired cross-sectional radiation data is fused and updated in the form of gradient images to generate a cross-sectional fused gradient image. S30. Perform gain denoising processing on the fused gradient image of the layer cross sections, and convert the fused gradient image of the layer cross sections into a pseudo-color image; S40. Perform gradient anomaly region extraction on the pseudo-color image, establish a correlation model between gradient anomalies, printing anomalies, and printing defects, and use the established correlation model in conjunction with the real-time acquired layer cross-sectional radiation data to achieve online quality evaluation. Step S20 further includes: fusing and refreshing image frames acquired within a set unit time period to gradually form a layer cross-sectional fusion gradient image, wherein the layer cross-sectional fusion gradient image is any one or any combination of integral distribution image, extreme value distribution image, weighted integral image, and weighted extreme value image.

2. The laser powder bed fused thermal scanning chromatography quality evaluation method as described in claim 1, characterized in that, In step S30, the layer cross-section fused gradient image is subjected to gain denoising processing using median filtering, Gaussian filtering, wavelet transform, or deep learning denoising methods.

3. The laser powder bed fused thermal scanning chromatography quality evaluation method as described in claim 1, characterized in that, In step S40, gradient anomaly region extraction is performed on the pseudo-color image using a pixel value thresholding algorithm, a region segmentation algorithm, a morphological operation algorithm, or a convolutional neural network algorithm.

4. A laser powder bed fused thermal scanning chromatography quality evaluation system, characterized in that, Including: A tomographic sensor is installed on the top plate of the forming cavity. This tomographic sensor can perform photosensitive measurements on the entire forming cross section and is used to collect the cross section radiation data during the processing of the cross section. The control module is used to acquire the cross-sectional radiation data of the layer in real time; to fuse and update the acquired cross-sectional radiation data in a gradient image manner using signal processing and image fusion techniques to generate a cross-sectional fused gradient image; to perform gain denoising processing on the cross-sectional fused gradient image and convert it into a pseudo-color image; to perform gradient anomaly region extraction on the pseudo-color image, and to establish a correlation model between gradient anomalies, printing anomalies, and printing defects. Using the established correlation model, combined with the real-time acquired cross-sectional radiation data, online quality evaluation is achieved. Specifically, when fusing and updating the acquired cross-sectional radiation data in a gradient image manner, fusion and refresh are performed on image frames acquired within a set unit time period to gradually form a cross-sectional fused gradient image. The cross-sectional fused gradient image can be any one or several of the following: integral distribution image, extreme value distribution image, weighted integral image, and weighted extreme value image.

5. The laser powder bed fused thermal scanning chromatography quality evaluation system as described in claim 4, characterized in that, The front end of the tomographic detection sensor is provided with an optical component, which is used to transmit the cross-sectional radiation of the layer to the tomographic detection sensor.

6. The laser powder bed fused thermal scanning chromatography quality evaluation system as described in claim 4, characterized in that, The tomographic detection sensor is any one or any combination of a CMOS camera, a photodiode array, a near-infrared thermal imager, or an infrared thermal imager.

Images