Off-line analysis system based on molten pool morphology in SLM metal additive manufacturing

Abstract

The invention discloses an off-line analysis system based on molten pool morphology in SLM (selective laser melting) metal additive manufacturing, and relates to the technical field of laser additive manufacturing. The system comprises functions of intelligent molten pool image recognition, image processing, image storage, data visualization and the like, and the functions jointly form a powerful data analysis and visualization platform. The offline analysis system based on the molten pool morphology in SLM metal additive manufacturing aims at creating an efficient multi-format image analysis, processing, display and storage solution and focuses on molten pool morphology data analysis. According to the method, the heterogeneous image analysis architecture supporting multiple formats is constructed, so that automatic format recognition and lossless import of the engineering image are realized; a multi-thread parallel computing architecture is adopted, efficient feature data extraction, feature image conversion and feature data and feature image storage are achieved, and the processing efficiency is remarkably improved; an interactive visual interface is developed, and multi-view linkage of feature data and feature images is supported.

Description

Technical Field

[0001] The present invention relates to the technical field of laser processing, and particularly to an offline analysis system for melt pool morphology in SLM metal additive manufacturing. Background Art

[0002] Selective Laser Melting (SLM) is an advanced metal additive manufacturing technology that builds three-dimensional solid parts layer by layer by melting metal powder with a high-energy laser beam. SLM technology has broad application prospects in the fields of aerospace, medical devices, automotive manufacturing, etc. SLM technology can manufacture parts with complex shapes and internal features, and the manufactured parts have excellent density and mechanical properties.

[0003] However, during the manufacturing process of SLM technology, plume, spatter, and unstable laser beam output may occur. Through data analysis of melt pool images, melt pool, plume, and spatter characteristics can be extracted, and then the diagnostic performance based on the information fusion of the melt pool, plume, and spatter can be explored, which helps to understand the stability of SLM processing and in-situ monitoring.

[0004] Compared with real-time monitoring, offline data analysis can perform more in-depth data analysis, allowing the use of more complex algorithms and models to process image data to extract more subtle features and trends. At the same time, it can reduce the computational burden of real-time processing, separate computationally intensive tasks from the real-time system, which can reduce the computational resources required by the real-time monitoring system and improve the stability and response speed of the system. It can also improve the flexibility of data processing, reprocess image data as needed, use different parameters to optimize the results, and at the same time reverse deduce the optimal process parameter settings through the accumulated data, so as to optimize the SLM process and improve the forming quality. Summary of the Invention

[0005] (I) Technical Problems to be Solved Aiming at the deficiencies of the prior art, the present invention provides an offline analysis system for the molten pool morphology in SLM metal additive manufacturing. The system includes functions such as intelligent recognition of molten pool images, image processing, image storage, and data visualization. These functions together constitute a powerful data analysis and visualization platform, providing strong support for in-depth understanding and optimization of the SLM manufacturing process. The system aims to create an efficient multi-format image parsing, processing, and storage solution, focusing on molten pool morphology analysis. The present invention realizes the automatic format recognition and lossless import of engineering images by constructing a heterogeneous image parsing architecture that supports multiple formats such as BMP / RAW / JPG; adopts a multi-threaded parallel computing architecture to achieve efficient extraction of feature data (length, width, area, average gray level), conversion of feature images (temperature field map, contour map), and storage of feature data and feature images, significantly improving the processing efficiency; and supports the multi-view linkage of feature data and feature images by developing an interactive visualization interface. By deeply analyzing molten pool images, the system helps to reduce experimental errors, shorten the process development time, and thus improve production efficiency and reduce manufacturing costs.

[0006] (2) Technical solution To achieve the above objectives, the present invention is realized through the following technical solutions: An offline analysis system for the molten pool morphology in SLM metal additive manufacturing, including an engineering import module, an image processing module, an engineering parsing pause and resume module, and a data display module. The engineering import module is used to parse the image format under the project, select different channels to load and read the image information, display its project list, project size, and project analysis progress, and the image format is a common image format such as bmp, jpg, or a RAW image set format. The image processing module includes various image state displays, feature data displays, and storage. The various image state displays include the display of the original image, the binary image, and the contour image, and the feature data display includes the display of the length, width, area, and average gray level of the molten pool. The engineering parsing pause and resume module is used for the situation where abnormal exit occurs during the parsing process of large projects or continues to parse after actively stopping the analysis. The data display module is to display the image processing results, including feature data and feature images, and provide a comparison of image data results under different parameters.

[0007] Preferably, the engineering import module enables users to select the project layer to be analyzed through an intuitive graphical user interface (GUI).

[0008] Preferably, the project import module supports importing molten pool images in multiple formats, including RAW compressed images, BMP, and JPG, and can intelligently identify and import these images in different formats. The image data will be stored in a special data structure for subsequent parsing using specific algorithms.

[0009] Preferably, the project import module will perform a preliminary analysis on the imported project, extract key information such as the layer information list, the overall size of the project, and the parsing progress, and display this information intuitively on the software interface, enabling users to conveniently perform further operations and in-depth analysis.

[0010] Preferably, the image processing module includes image parsing, image processing, and image storage. The image parsing includes three stages: the unparsed stage, the partially parsed stage, and the fully parsed stage. These stages are identified based on the parsing progress marked by the layer information list after importing the project, and the progress is represented as a percentage, increasing from 0% to 100%.

[0011] Preferably, in the unparsed stage: for the RAW image format, first perform a decompression operation, which involves reading the information list in the RAW package, including the image pixel size and the number of images, and then extract the image data from the RAW package and store it in the data structure List for subsequent processing.

[0012] Preferably, in the partially parsed stage: based on the current parsing progress and the total number of project images, accurately locate the serial number of the image to continue parsing, and at the same time read the feature data of the parsed images into the memory and perform data parsing, processing, and storage according to the established process.

[0013] Preferably, in the parsing completed stage: when the parsing progress reaches 100%, it indicates that the project has been fully parsed. At this time, users can directly access and obtain the feature data of the project. The system will display the feature data curve and provide an intuitive display of the original image and the feature image.

[0014] Preferably, the image processing includes feature data extraction (length and width, area, average gray level) and feature image conversion (binary image, contour image), and the steps are as follows: S1. Image smoothing: First, perform smoothing processing on the image through a Gaussian filter to reduce the influence of noise and improve the accuracy of edge detection; S2. Apply the Sobel operator: Apply the Sobel operator to the grayscale image to obtain gradient images in the horizontal and vertical directions, and these two images respectively represent the brightness changes of the image in the horizontal and vertical directions; S3. Calculate the gradient magnitude: Combine the horizontal and vertical gradient images and calculate the gradient magnitude of each pixel point through the following formula: ; wherein, and are the horizontal and vertical gradients respectively; S4, Non-Maximum Suppression: To obtain finer edges, non-maximum suppression is performed on the gradient magnitude image. This process checks whether the gradient magnitude of each pixel is the maximum in its neighborhood. If not, it is set to 0; S5, Double Thresholding: To further remove noise and retain strong edges, double thresholding can be applied. First, a high threshold is used to determine strong edges, and then a low threshold is used to determine weak edges. Weak edges are only retained if they are connected to strong edges; S6, Edge Linking: After double thresholding, some isolated edge points are obtained. Through edge linking, these isolated points are connected into continuous edges; S7, Obtaining Contours: Contours are obtained based on contour connectivity and stored in the data structure List; S8, To avoid the problem of multiple contours caused by factors such as splashing in the molten pool, compare the areas of the figures enclosed by the contours in step S6, and the contour with the largest area is the molten pool contour; S9, Obtaining the Minimum Bounding Rectangle of the Molten Pool Based on the Largest Contour: First, the minimum convex polygon of the contour is obtained through the Sklansky algorithm. Secondly, through the Rotating Calipers algorithm, that is, by enumerating the sides of the polygon, the bounding rectangles are calculated for each side, and the areas of these bounding rectangles are compared to find the rectangle with the smallest area; S10, Real-Time Image Processing: Based on the minimum rectangle obtained in step S8, the length and width are the length and width of the molten pool, and the largest contour area obtained in step S7 is the area of the molten pool. The average gray value of the molten pool is obtained by calculating the gray value mean of the pixel points in the contour. The binary image is calculated based on the binary mode (fixed threshold or automatic threshold). On the basis of the binary image, the molten pool contour map is obtained through the edge detection algorithm Sobel operator.

[0015] (III) Beneficial Effects The present invention provides an offline analysis system for the morphology of the molten pool in SLM metal additive manufacturing. Compared with the prior art, it has the following beneficial effects: (1). The offline analysis system for melt pool morphology in SLM metal additive manufacturing has efficient image import and parsing: The system can intelligently identify and import melt pool images in various formats, including RAW compressed images, BMP, JPG, etc. A dedicated parsing algorithm is developed for the RAW multi-image qualified format, supporting batch loading and automatic extraction of metadata (layer number, timestamp), and storing the image data in a special data structure, providing a solid foundation for subsequent parsing and processing.

[0016] (2). The offline analysis system for melt pool morphology in SLM metal additive manufacturing has a flexible image processing flow: The system constructs a standardized process of Gaussian filtering, Sobel operator, gradient magnitude calculation, non-maximum suppression, double threshold processing, and edge connection based on OpenCV to achieve the extraction of melt pool feature data (length, width, area, average grayscale) and the conversion of feature images (binary image, contour image). These processing steps not only improve the accuracy of edge detection but also enhance the recognizability of image features.

[0017] (3). The offline analysis system for melt pool morphology in SLM metal additive manufacturing has an optimized image storage mechanism: By adopting the strategies of delayed writing, batch processing, and multi-threaded concurrent storage, the system significantly improves the performance of image processing and storage, especially when dealing with a large amount of image data. It supports the parsing, processing, and storage of 3000 melt pool images of 112x96 pixels per second under a mechanical hard disk.

[0018] (4). The offline analysis system for melt pool morphology in SLM metal additive manufacturing has flexible engineering parsing control: The pause and continue parsing functions provided by the system enable users to interrupt and resume processing at any time during the analysis process, improving the flexibility of processing and saving time and computing resources.

[0019] (5). The offline analysis system for melt pool morphology in SLM metal additive manufacturing has a powerful data visualization platform: The system provides a dual-view linkage interface to synchronously display the melt pool contour feature map, parameter curve, and comparison with the original image, and supports data export, enabling users to intuitively track and compare the changes in key parameters of the melt pool morphology. Description of the Drawings

[0020] Figure 1 It is the flow chart of the system of the present invention; Figure 2 It is the framework diagram of the system of the present invention; Figure 3 It is the effect diagram of the specific implementation of the system of the present invention; From left to right, they are the original melt pool morphology image, the binary melt pool morphology image, and the contour melt pool morphology image. Detailed Embodiment

[0021] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0022] Please refer to Figure 1 , the embodiments of the present invention provide two technical solutions: an offline analysis system for the molten pool morphology in SLM metal additive manufacturing, specifically including the following embodiments: Embodiment 1: An offline analysis system for the molten pool morphology in SLM metal additive manufacturing, characterized in that it includes an engineering import module, an image processing module, an engineering parsing pause and resume module, and a data display module. The engineering import module is used to parse the image format under the project, select different channels for image loading and information reading, display its project list, project size, and project analysis progress, and the image format is a common image format such as bmp, jpg, etc. or a RAW image set format; The image processing module includes various image state displays, feature data displays, and storage. The various image state displays include original image display, binary image display, contour image display, and pseudo-color image display, and the feature data display includes molten pool length and width, area, and average gray level display; The engineering parsing pause and resume module is used for the situation where abnormal exit or active stop of analysis occurs during the parsing process of large projects and then continues to parse; The data display module is to display the image processing results, including feature data and feature images, and provide comparison of image data results under different parameters.

[0023] In the embodiments of the present invention, the engineering import module enables users to select the project layer to be analyzed through an intuitive graphical user interface (GUI). The engineering import module supports importing molten pool images in multiple formats, including RAW compressed images, BMP, and JPG, and can intelligently identify and import these different format images. The image data will be stored in a dedicated data structure for subsequent parsing using specific algorithms. The engineering import module will perform a preliminary analysis on the imported project, extract key information such as the layer information list, overall project size, and parsing progress, and intuitively display this information on the software interface, enabling users to conveniently perform further operations and in-depth analysis.

[0024] In the embodiments of the present invention, the image processing module includes image parsing, image processing, and image storage. The image parsing includes three stages: the unparsed stage, the partially parsed stage, and the fully parsed stage. These stages are identified based on the parsing progress marked by the layer information list after importing the project. This progress is represented as a percentage and increases from 0% to 100%. Unparsed stage: For the RAW image format, first, a decompression operation is performed. This process involves reading the information list in the RAW packet, including the image pixel size and the number of images. Subsequently, the image data is extracted from the RAW packet and stored in the data structure List for subsequent processing. Partially parsed stage: According to the current parsing progress and the total number of project images, accurately locate the serial number of the image to continue parsing. At the same time, read the feature data of the parsed images into the memory and perform data parsing, processing, and storage according to the established process. Parsing completed stage: When the parsing progress reaches 100%, it indicates that the project has been fully parsed.

[0025] Retrieve the feature data of the project, and the system will display the feature data curve and provide an intuitive display of the original image and the feature image.

[0026] In the embodiments of the present invention, image processing includes feature data extraction (length and width, area, average grayscale) and feature image conversion (binary image, contour image), and the steps are as follows: S1. Image smoothing: First, perform smoothing processing on the image through a Gaussian filter to reduce the influence of noise and improve the accuracy of edge detection. S2. Apply the Sobel operator: Apply the Sobel operator to the grayscale image to obtain gradient images in the horizontal and vertical directions. These two images respectively represent the brightness changes of the image in the horizontal and vertical directions. S3. Calculate the gradient magnitude: Combine the horizontal and vertical gradient images and calculate the gradient magnitude of each pixel point through the following formula: ; where and are the horizontal and vertical gradients respectively; S4. Non-maximum suppression: To obtain finer edges, perform non-maximum suppression on the gradient magnitude image. This process checks whether the gradient magnitude of each pixel point is the maximum value in its neighborhood. If not, it is set to 0. S5. Double threshold processing: To further remove noise and retain strong edges, double threshold processing can be applied. First, use a high threshold to determine strong edges, and then use a low threshold to determine weak edges. Weak edges are only retained if they are connected to strong edges. S6. Edge connection: After double-threshold processing, some isolated edge points are obtained. Through edge connection, these isolated points are connected into continuous edges; S7. Obtain the contour: Obtain the contour according to contour connectivity and store it in the data structure List; S8. To avoid the problem of multiple contours caused by factors such as splash in the molten pool, compare the areas of the figures enclosed by the contours in step S6, and the contour with the largest area is the molten pool contour; S9. Obtain the minimum circumscribed rectangle of the molten pool based on the largest contour: First, obtain the minimum convex polygon of the contour through the Sklansky algorithm. Secondly, through the Rotating Calipers algorithm, that is, by enumerating the sides of the polygon, calculate the circumscribed rectangle for each side, and compare the areas of these circumscribed rectangles to find the rectangle with the smallest area; S10. Real-time image processing: Based on the minimum rectangle obtained in step S8, the length and width are the length and width of the molten pool, and the area of the largest contour obtained in step S7 is the area of the molten pool. Calculate the average gray value of the molten pool by calculating the gray value mean of the pixel points in the contour. Calculate the binary image according to the binary mode (fixed threshold or automatic threshold). Based on the binary image, obtain the molten pool contour map through the edge detection algorithm Sobel operator.

[0027] In the embodiment of the present invention, image storage is mainly used to store feature data and feature images. The RAW format project needs to store the parsed images. The storage part adopts a multi-threaded design to improve the storage speed.

[0028] Image storage mechanism: The molten pool morphology project usually contains a large amount of image data. Using ordinary image storage methods is relatively slow. This system adopts a delayed writing and batch processing strategy. The feature images processed by image processing are first stored in the cache memory. When a certain amount of image data accumulates in the cache, the system will batch transfer these image data from the cache to the disk. This strategy can effectively reduce the frequency of disk I / O operations, thereby significantly improving the performance of the image processing process. Especially in scenarios where a large amount of image data needs to be processed, it can significantly improve the overall processing speed and efficiency. In this way, the system can optimize resource utilization while ensuring the accuracy of data processing, achieving high-performance image processing and storage.

[0029] Multi-threaded concurrent storage: To improve storage performance, the storage part adopts a multi-threaded design. This design can make full use of the computing power of modern multi-core processors to achieve parallel data processing and storage, thereby significantly improving the storage speed and the overall throughput of the system.

[0030] Embodiment 2: The difference technical solution of the embodiment of the present invention compared with Embodiment 1 lies in: The project parsing pause and continued parsing part: Pause Function: When delving into the impact of the molten pool morphology on the SLM manufacturing process and the quality of its finished products, since the amount of data involved is usually extremely large, a comprehensive analysis of the entire project may take a long time. Additionally, it may be necessary to readjust the image processing parameters during the research process. In such cases, the pause function provides a flexible solution, allowing users to interrupt the processing at any time during the analysis for necessary adjustments or optimizations.

[0031] Continue Parsing: After completing the adjustment of the image processing parameters or in case of an unexpected interruption, the continue parsing function enables users to resume the analysis from the position where it was last paused and complete the remaining data processing tasks. This mechanism not only improves the flexibility of processing but also helps save time and computing resources, ensuring the continuity and efficiency of the research. With this design, the system can adapt to dynamically changing research requirements while maintaining the integrity and accuracy of data processing.

[0032] Image Display: This system provides a detailed engineering data visualization function, covering the key feature data of the molten pool morphology and their corresponding feature images. Specifically, the feature data includes the size (length and width), area, and average gray level of the molten pool, while the feature images include the original image and its contour map. In addition, the system also supports comparative analysis of images under different binary threshold modes, allowing users to adjust the threshold size as needed to observe the impact of different threshold settings on the image processing results. To further enhance the readability and depth of analysis of the data, the system also provides a function to display line charts of all feature data, enabling users to intuitively track and compare the changes in the key parameters of the molten pool morphology.

[0033] Meanwhile, the content not described in detail in this specification belongs to the prior art well-known to those skilled in the art.

[0034] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device.

[0035] Although embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An offline analysis system for molten pool morphology in SLM metal additive manufacturing, characterized in that: It includes an engineering import module, an image processing module, an engineering parsing pause and resume module, and a data display module. The engineering import module is used to parse the image format under the project, select different channels for image loading and information reading, display its project list, project size, and project analysis progress, and the image format is a common image format or a RAW image set format; The image processing module includes various image state displays, feature data displays, and storage. The various image state displays include original image display, binary image display, and contour image display, and the feature data displays include molten pool length and width, area, and average gray scale display; The engineering parsing pause and resume module is used for the situation where abnormal exit occurs during the parsing process of large projects or continues to parse after actively stopping the analysis; The data display module is to display the image processing results, including feature data and feature images, and provide comparison of image data results under different parameters.

2. The offline analysis system for molten pool morphology in SLM metal additive manufacturing according to claim 1, wherein: The engineering import module enables users to select the project layer to be analyzed through an intuitive graphical user interface (GUI).

3. An off-line analysis system for the molten pool morphology in SLM metal additive manufacturing according to claim 1, characterized in that: The engineering import module also includes an adaptive format recognition unit. The adaptive format recognition unit automatically distinguishes BMP / JPG standard images and RAW image sets through dual verification of file header feature codes and file extensions, selects different channels for image loading, and the image data will be stored in a dedicated data structure for subsequent parsing using specific algorithms.

4. An offline analysis system for the molten pool morphology in SLM metal additive manufacturing according to claim 1, characterized in that: The engineering import module will perform a preliminary analysis on the imported project, extract key information such as the layer information list, overall project size, and parsing progress, and intuitively display this information on the software interface, enabling users to conveniently perform further operations and in-depth analysis.

5. An offline analysis system for molten pool morphology in SLM metal additive manufacturing according to claim 1, characterized in that: The image processing module includes image parsing, image processing, and image storage. The image parsing includes three stages: the unparsed stage, the partially parsed stage, and the fully parsed stage. These stages are divided based on the parsing progress marked by the layer information list after importing the project, and this progress is represented in percentage form, increasing from 0% to 100%.

6. The offline analysis system for molten pool morphology in SLM metal additive manufacturing according to claim 5, wherein: The unparsed stage: For the RAW image format, first perform a decompression operation, which involves reading the information list in the RAW package, including the image pixel size and the number of images, and then extract the image data from the RAW package and store it in the data structure List for subsequent processing.

7. An offline analysis system for molten pool morphology in SLM metal additive manufacturing according to claim 5, characterized in that: The partially parsed stage: According to the current parsing progress and the total number of project images, accurately locate the serial number of the image to continue parsing. At the same time, read the feature data of the parsed images into the memory and perform data parsing, processing, and storage according to the established process.

8. An off-line analysis system for molten pool morphology in SLM metal additive manufacturing according to claim 5, characterized in that: The parsing completed stage: When the parsing progress reaches 100%, it indicates that the project has been fully parsed. At this time, users can directly access and obtain the feature data of the project, and the system will display the feature data curve and provide an intuitive display of the original image and the feature image.

9. An offline analysis system for molten pool morphology in SLM metal additive manufacturing according to claim 5, characterized in that: The image processing includes feature data extraction and feature image conversion, and has the following steps: S1. Image smoothing: First, perform smoothing processing on the image through a Gaussian filter to reduce the influence of noise and improve the accuracy of edge detection; S2. Apply the Sobel operator: Apply the Sobel operator to the grayscale image to obtain gradient images in the horizontal and vertical directions. These two images represent the brightness changes of the image in the horizontal and vertical directions respectively; S3. Calculate the gradient magnitude: Combine the horizontal and vertical gradient images and calculate the gradient magnitude of each pixel point, which is achieved by the following formula: ； wherein, and are the horizontal and vertical gradients respectively; S4. Non-maximum suppression: Check whether the gradient magnitude of each pixel point is the maximum value within its neighborhood. If not, set it to 0; S5. Double-threshold processing: First, use a high threshold to determine strong edges, and then use a low threshold to determine weak edges. Weak edges are only retained when they are connected to strong edges; S6. Edge connection: After double-threshold processing, some isolated edge points will be obtained. Through edge connection, these isolated points are connected into continuous edges; S7. Obtain the contour: Obtain the contour according to contour connectivity and store it in the data structure List; S8. Compare the areas of the figures enclosed by the contours in step S6, and the contour with the largest area is the molten pool contour; S9. Obtain the minimum bounding rectangle of the molten pool based on the largest contour: First, obtain the minimum convex polygon of the contour through the Sklansky algorithm. Secondly, through the Rotating Calipers algorithm, that is, by enumerating the sides of the polygon, calculate the bounding rectangle for each side and compare the areas of these bounding rectangles to find the rectangle with the smallest area; S10. Real-time image processing: Based on the minimum rectangle obtained in step S8, the length and width are the length and width of the molten pool, and the largest contour area obtained in step S7 is the area of the molten pool. Calculate the average gray value of the molten pool by calculating the gray value mean of the pixel points in the contour. Calculate the binary image according to the binary mode, and obtain the molten pool contour map through the edge detection algorithm Sobel operator on the basis of the binary image.

Images