A method and system for predicting mechanical properties of a laser selective melting formed part

Abstract

The present disclosure provides a laser selective melting forming part mechanical property prediction method and system, relating to the technical field of laser selective melting online monitoring, comprising: obtaining the relationship between powder laying defects and porosity and the relationship between powder laying defects and tensile properties, and constructing a mechanical property prediction model according to the relationship between powder laying defects and porosity and the relationship between powder laying defects and tensile properties; obtaining the powder laying image of each layer of the current forming part, and pre-processing the powder laying image; identifying the powder laying defect layer number in the pre-processed powder laying image, and calculating the powder laying defect area; predicting by using the mechanical property prediction model according to the powder laying defect layer number and the powder laying defect area, to obtain the mechanical property prediction result of the current forming part; comparing the mechanical property prediction result with the set mechanical property reduction threshold value, to judge whether to perform remelting and repowdering operations.

Description

Technical Field

[0001] This disclosure relates to the field of laser selective melting online monitoring technology, specifically to a method and system for predicting the mechanical properties of laser selective melting formed parts. Background Technology

[0002] The statements in this section are merely background information relating to this disclosure and do not necessarily constitute prior art.

[0003] Selective laser melting (SLM) is an additive manufacturing technology that uses a laser beam to rapidly heat powder materials to their melting point and then quickly solidify them. Its advantages include high precision, non-contact processing, and programmability. However, current SLM technology uses open-loop control, making it difficult to guarantee powder spread quality. Repeatability and consistency of manufacturing quality have become major bottlenecks. Powder spread quality is a key factor determining the mechanical properties of the formed part. Furthermore, powder spread defects during processing can lead to porosity and incomplete fusion within certain layers of the part, indirectly affecting its mechanical properties.

[0004] Traditional empirical models and limited data-based methods for predicting the mechanical properties of metals in laser additive manufacturing face significant challenges in terms of efficiency and accuracy. In recent years, with the development of big data and artificial intelligence, machine learning (ML) methods have become an inevitable product. Combining laser selective melting equipment with machine learning allows for monitoring the powder spreading state and predicting the mechanical properties of parts, while also providing feedback on issues such as powder shortages. This is of great significance for improving the mechanical properties of parts.

[0005] However, most current machine learning techniques used for selective laser melting only involve the identification of single-layer powder bed defects. They cannot combine porosity with the state of adjacent powder beds to predict mechanical properties. They cannot predict situations where single-layer powder bed defects have a small impact on mechanical properties, but the accumulation of multiple layers leads to a significant decrease in mechanical properties. Summary of the Invention

[0006] To address the aforementioned issues, this disclosure proposes a method and system for predicting the mechanical properties of laser selective melting formed parts. By leveraging the relationship between multi-layer powder shortage defects, porosity, and tensile properties, and utilizing a mechanical property prediction algorithm, the method can predict the mechanical properties of parts during the printing process and provide timely feedback when mechanical properties decline.

[0007] According to some embodiments, the present disclosure adopts the following technical solutions:

[0008] A method for predicting the mechanical properties of laser selective melting formed parts includes:

[0009] Obtain the relationship between powder spreading defects and porosity, and between powder spreading defects and tensile properties. Based on the relationship between powder spreading defects and porosity, and between powder spreading defects and tensile properties, construct a mechanical property prediction model.

[0010] Obtain powder spreading images for each layer of the current molded part, and preprocess the powder spreading images;

[0011] Identify the number of powder-spreading defect layers in the preprocessed powder-spreading image and calculate the area of ​​powder-spreading defects;

[0012] Based on the number of powder spreading defect layers and the area of ​​powder spreading defect, the mechanical property prediction model is used to make a prediction and obtain the current mechanical property prediction result of the molded part.

[0013] The predicted mechanical properties are compared with the set mechanical property reduction threshold to determine whether remelting and re-coating operations should be performed.

[0014] According to some embodiments, the present disclosure adopts the following technical solutions:

[0015] A system for predicting the mechanical properties of laser selective melting formed parts includes:

[0016] The model building module is used to obtain the relationship between powder spreading defects and porosity, as well as the relationship between powder spreading defects and tensile properties. Based on the relationship between powder spreading defects and porosity, and between powder spreading defects and tensile properties, a mechanical property prediction model is constructed.

[0017] The image acquisition module is used to acquire images of each layer of powder spreading on the current molded part;

[0018] The image processing module is used to preprocess the powder spreading image, identify the number of powder spreading defect layers in the preprocessed powder spreading image, and calculate the area of ​​powder spreading defects.

[0019] The mechanical property prediction module is used to predict the mechanical properties of the current molded part by using the mechanical property prediction model based on the number of powder spreading defect layers and the area of ​​powder spreading defect.

[0020] According to some embodiments, the present disclosure adopts the following technical solutions:

[0021] A non-transitory computer-readable storage medium is provided for storing computer instructions, which, when executed by a processor, implement the aforementioned method for predicting the mechanical properties of laser selective melting formed parts.

[0022] According to some embodiments, the present disclosure adopts the following technical solutions:

[0023] An electronic device includes a processor, a memory, and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory to enable the electronic device to perform the aforementioned method for predicting the mechanical properties of laser selective melting formed parts.

[0024] Compared with the prior art, the beneficial effects of this disclosure are as follows:

[0025] This disclosure presents a method for predicting the mechanical properties of laser selective melting formed parts. A hardware system is built, and an image acquisition module within the system acquires powder-laying images. The images are preprocessed, and an algorithm is used for defect segmentation and defect area calculation. Then, based on the mechanical property prediction algorithm obtained from previous experiments, the predicted mechanical properties are output. This method solves the problem that single-layer powder-laying defects have a relatively small impact on mechanical properties, while accumulated multiple layers lead to a significant decrease in mechanical properties, making prediction impossible. By using machine learning methods, the complex nonlinear relationship between powder-laying defects and mechanical properties in metal laser additive manufacturing is effectively handled, providing a method for ensuring repeatability and quality consistency in the manufacturing of main load-bearing components for high-end equipment.

[0026] This disclosure presents a method for predicting the mechanical properties of laser selective melting formed parts. Based on the area and number of powder-laying defects, it predicts the current mechanical properties of the formed part, solving the problem that traditional methods only address the identification of single-layer powder-laying defects. During laser selective melting, powder shortage defects lead to a decrease in mechanical properties. Furthermore, the porosity and lack of fusion caused by powder shortage defects also increase the porosity of the formed part. Remelting and large layer thicknesses affect the microstructure, thus impacting mechanical properties. This disclosure combines the powder-laying defect-porosity relationship and the powder-laying defect-tensile property relationship to derive a mechanical property prediction algorithm. The defect segmentation algorithm achieves an identification accuracy of 97%, providing a new approach for obtaining formed parts with superior mechanical properties and a denser microstructure. Attached Figure Description

[0027] The accompanying drawings, which form part of this disclosure, are used to provide a further understanding of this disclosure. The illustrative embodiments of this disclosure and their descriptions are used to explain this disclosure and do not constitute an undue limitation of this disclosure.

[0028] Figure 1 This is a flowchart of the method for predicting the mechanical properties of laser selective melting formed parts according to an embodiment of the present disclosure;

[0029] Figure 2 This is a schematic diagram of the irregularly shaped part and the powder shortage location model according to an embodiment of this disclosure;

[0030] Figure 3 Metallographic images of powder-deficient regions with different numbers of powder-deficient layers in embodiments of this disclosure;

[0031] Figure 4 This is a graph showing the relationship between the number of powder-deficient layers and porosity in different embodiments of this disclosure;

[0032] Figure 5 This is a schematic diagram of the cutting of a tension member according to an embodiment of the present disclosure;

[0033] Figure 6 This is a graph showing the tensile properties of different powder-deficient layers in embodiments of this disclosure;

[0034] Figure 7 This is a schematic diagram of a laser selective melting forming part mechanical property prediction system according to an embodiment of the present disclosure;

[0035] Figure 8 This is a comparison image of the image segmentation module before and after defect segmentation in an embodiment of this disclosure;

[0036] The components include: 1. Laser, 2. Beam splitter, 3. Beam expander, 4. Galvanometer, 5. Protective gas, 6. Forming chamber, 7. Movable light source module, 8. Image acquisition module, 9. Interface data transmission, 10. Dust collector, 11. Visual monitoring module, 12. Image processing module, 13. Formed part, 14. Forming cylinder, 15. Powder cylinder, 16. Lifting platform, 17. Scraper, 18. Powder recovery tank, and 19. Fiber optic sensor. Detailed Implementation

[0037] The present disclosure will be further described below with reference to the accompanying drawings and embodiments.

[0038] It should be noted that the following detailed descriptions are illustrative and intended to provide further explanation of this disclosure. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure pertains.

[0039] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this disclosure. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms “comprising” and / or “including” are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0040] Example 1

[0041] One embodiment of this disclosure provides a method for predicting the mechanical properties of laser selective melting formed parts. By analyzing the relationship between multi-layer powder deficiency defects, porosity, and tensile properties, a mechanical property prediction model is designed to predict the mechanical properties of the parts during the printing process. The method also provides timely feedback when mechanical properties deteriorate. The steps include:

[0042] Step 1: Obtain the relationship between powder spreading defects and porosity, and between powder spreading defects and tensile properties. Based on the relationship between powder spreading defects and porosity, and between powder spreading defects and tensile properties, construct a mechanical property prediction model.

[0043] Step 2: Obtain powder spreading images for each layer of the current molded part, and preprocess the powder spreading images;

[0044] Step 3: Identify the number of powder spreading defect layers in the preprocessed powder spreading image and calculate the area of ​​powder spreading defects;

[0045] Step 4: Based on the number of powder spreading defect layers and the area of ​​powder spreading defect, use the mechanical property prediction model to make a prediction and obtain the current mechanical property prediction result of the molded part.

[0046] Step 5: Compare the predicted mechanical properties with the set mechanical property reduction threshold to determine whether to perform remelting and re-powdering operations.

[0047] As one embodiment, the specific implementation process of the laser selective melting forming part mechanical property prediction method disclosed herein is as follows:

[0048] 1) By printing irregularly shaped parts with different numbers of missing powder layers, the relationship formula between powder spreading defects and porosity was measured;

[0049] Specifically, the relationship between powder spreading defects and porosity, and the specific calculation steps include:

[0050] Metallographic images of molded parts with different powder-deficient layers were obtained. The porosity of powder-laying defects with different layers was calculated by metallographic method. The relationship between porosity defect area and powder-laying defect layer was obtained. A regression model was fitted to the relationship between different powder-deficient layers and porosity.

[0051] Furthermore, the relationship between the number of powder-spreading defect layers and porosity is as follows: no powder deficiency results in a porosity of 0.018%, one consecutive powder deficiency layer results in a porosity of 0.028%, two consecutive powder deficiency layers result in a porosity of 0.045%, three consecutive powder deficiency layers result in a porosity of 0.065%, four consecutive powder deficiency layers result in a porosity of 0.412%, five consecutive powder deficiency layers result in a porosity of 0.987%, and six consecutive powder deficiency layers result in a porosity of 2.275%.

[0052] Furthermore, data on the number of missing powder layers versus porosity were collected; based on the data, a cubic polynomial model was selected; the model was fitted using the data on the number of missing powder layers versus porosity to find the optimal parameter estimates that minimized the error between the model and the observed data. The nonlinear relationship between powder spreading defects and porosity was obtained through data fitting as follows:

[0053] Y1 = 0.025X1 3 -0.12X1 2+0.14X1+0.01

[0054] Where Y1 represents porosity; X1 represents the number of powder-deficient layers.

[0055] In one specific implementation, the process of calculating the porosity of powder-laying defects with different numbers of layers using metallographic methods to obtain the relationship between the porosity defect area and the number of powder-laying defect layers includes: preparing an 8mm×8mm×35mm irregularly shaped part with a powder-laying layer thickness of 30μm using a laser selective melting device, such as... Figure 2 The diagram shows a model of the irregularly shaped part and the location of the missing powder. Using optimal process parameters, the 0-5mm area (#1) is printed; the 5-10mm irregularly shaped notch area has one layer of missing powder, while other areas are printed normally, resulting in #2; the 10-15mm irregularly shaped notch area has two layers of missing powder, while other areas are printed normally, resulting in #3; the 15-20mm irregularly shaped notch area has three layers of missing powder, while other areas are printed normally, resulting in #4; the 10-25mm irregularly shaped notch area has four layers of missing powder, while other areas are printed normally, resulting in #5; the 25-30mm irregularly shaped notch area has five layers of missing powder, while other areas are printed normally, resulting in #6; and the 30-35mm irregularly shaped notch area has six layers of missing powder, while other areas are printed normally, resulting in #7. After obtaining the irregularly shaped part, wire cutting and polishing are performed.

[0056] Figure 3 Metallographic images of powder-deficient regions #1-#7 were used. Porosity was calculated for different numbers of powder-deficient layers using metallographic methods, yielding the relationship between porosity defect area and the number of powder-spreading defect layers, as follows: Figure 4 The relationship between the number of powder-spreading defect layers and porosity is as follows: no powder deficiency results in a porosity of 0.018%, one consecutive powder deficiency layer results in a porosity of 0.028%, two consecutive powder deficiency layers result in a porosity of 0.045%, three consecutive powder deficiency layers result in a porosity of 0.065%, four consecutive powder deficiency layers result in a porosity of 0.412%, five consecutive powder deficiency layers result in a porosity of 0.987%, and six consecutive powder deficiency layers result in a porosity of 2.275%.

[0057] 2) By printing tensile specimens with different numbers of missing powder layers, the relationship between powder spreading defects and tensile properties was measured;

[0058] Specifically, the relationship between powder spreading defects and tensile properties is calculated through the following steps: printing tensile specimens with different powder-deficient layers, obtaining tensile specimens through wire cutting, and conducting tensile tests using a room-temperature tensile testing machine to obtain the relationship between the number of powder spreading defect layers and tensile properties, and fitting a regression model.

[0059] The relationship between the number of powder-spreading defect layers and tensile properties is as follows: the tensile strength of the sample without powder deficiency is 573 MPa, the tensile strength of the sample with 1 consecutive powder deficiency layer is 550 MPa, the tensile strength of the sample with 2 consecutive powder deficiency layers is 550 MPa, the tensile strength of the sample with 3 consecutive powder deficiency layers is 559 MPa, the tensile strength of the sample with 4 consecutive powder deficiency layers is 545 MPa, the tensile strength of the sample with 5 consecutive powder deficiency layers is 548 MPa, and the tensile strength of the sample with 6 consecutive powder deficiency layers is 481 MPa.

[0060] Specifically, the powder layer number versus tensile properties data is obtained. Based on the data, an existing cubic polynomial model is selected, and the model is fitted using the powder layer number versus tensile properties data to find the optimal parameter estimates that minimize the error between the model and the observed data. The nonlinear relationship between powder layer defects and tensile properties is obtained through data fitting as follows:

[0061] Y2 = -2.36X2 3 +18.26X2 2 -39.47 x 2 + 573.48

[0062] Where Y2 represents porosity; X2 represents the number of powder-deficient layers.

[0063] In one specific implementation, the process of obtaining the relationship between the number of powder-laden defect layers and tensile properties by conducting tensile tests using a room-temperature tensile testing machine is as follows: Tensile templates with different powder-laden defect layers are printed using a laser selective melting device. Tensile template #1 is printed using optimal process parameters. A powder-laden defect layer of 1 is set in the middle of template #2, a powder-laden defect layer of 2 is set in the middle of template #3, a powder-laden defect layer of 3 is set in the middle of template #4, a powder-laden defect layer of 4 is set in the middle of template #5, a powder-laden defect layer of 5 is set in the middle of template #6, and a powder-laden defect layer of 6 is set in the middle of template #7. Through... Figure 5 Three tensile specimens were obtained in each group by wire cutting, and tensile tests were performed using a room temperature tensile testing machine. The average value was taken after multiple tensile tests.

[0064] Figure 6 The tensile property curves show different powder-deficient layers. The tensile strength of the sample without powder deficiency is 573 MPa, the tensile strength with 1 consecutive powder deficiency layer is 550 MPa, the tensile strength with 2 consecutive powder deficiency layers is 550 MPa, the tensile strength with 3 consecutive powder deficiency layers is 559 MPa, the tensile strength with 4 consecutive powder deficiency layers is 545 MPa, the tensile strength with 5 consecutive powder deficiency layers is 548 MPa, and the tensile strength with 6 consecutive powder deficiency layers is 481 MPa. It can be seen that 5 or fewer consecutive powder deficiency layers have little effect on the tensile strength of the formed part, while the tensile strength drops significantly when 6 consecutive powder deficiency layers are present.

[0065] 3) Based on the relationships between powder spreading defects and porosity, and between powder spreading defects and tensile properties, a mechanical property prediction model is obtained; specifically, the mechanical property model comprehensively considers the relationships between powder spreading defects and porosity, and between powder spreading defects and tensile properties.

[0066] Y1 = 0.025X1 3 -0.12X1 2 +0.14X1+0.01

[0067] Y2 = -2.36X2 3 +18.26X2 2 -39.47 x 2 + 573.48

[0068] The absence of powder in six consecutive layers has a significant impact on tensile strength. Based on the relationship between the degree of mechanical property degradation and the number of powder-laying defect layers, a threshold of 30 MPa is set. During the printing process, the predicted mechanical properties are compared with the set mechanical property degradation threshold. If the threshold is exceeded, remelting and re-laying of powder are performed. If the threshold is not exceeded, no intervention is performed.

[0069] Furthermore, as an example, in the actual prediction process, the prediction process is as follows:

[0070] 1) Print the image, capture the image of each layer of powder, and perform preprocessing such as perspective transformation and filtering to remove noise;

[0071] Specifically, the preprocessing steps for printing, acquiring images of each layer of powder, and performing perspective transformation, filtering, and noise reduction include:

[0072] Image acquisition module 8, installed on top of the forming chamber of the laser selective area device, is configured to acquire images after each powder application. It includes a light signal capture module, a photoelectric conversion module, and an optical module. The light signal capture module collects light after powder application using a CMOS sensor. The photoelectric conversion module converts the light collected by the light signal capture module into electrical signals, breaking down the object into individual pixels according to a specific arrangement. These pixels are then processed by an analog-to-digital converter and an image processor to obtain an image, which is then transmitted to the image processing module 12 via the Gige interface 7.

[0073] Furthermore, the image processing module 12 performs preprocessing on the image, including perspective transformation and filtering / denoising. The preprocessing includes:

[0074] Perspective transformation and connected component filtering are performed on the image. Because the image acquisition module in this system uses a rangefinder camera, different locations on the powder bed surface are not on the same plane, resulting in significant linear distortion in the acquired image. Correction of linear distortion involves performing a perspective transformation, mapping each pixel in the image sequentially to a new plane. Image acquisition is affected by various noises, such as sensor noise, ambient light variations, and electromagnetic interference. These noises degrade image quality, making the image unusable for subsequent analysis and processing. Filtering and denoising improve image clarity and contrast, making the image easier to interpret and use. Using connected component denoising, the pixel region near a pixel in a powder bed image is called the pixel's neighborhood.

[0075] In this embodiment, filtering out connected components with an area of ​​less than 10 pixels can achieve a good noise reduction effect.

[0076] The printing process is achieved through a laser selective melting forming device, which mainly consists of a laser system, a scanning system, a powder supply system, a forming platform, an inert gas device, and a control system. The laser system is the core component, responsible for generating a high-energy, highly focused laser beam. In this embodiment, the laser source used is a fiber laser 1. The scanning system controls the movement of the laser beam and includes a beam splitter 2, a beam expander 3, and a galvanometer 4. The beam splitter 2 and beam expander 3 are used to separate and expand the laser beam emitted by the laser 1, respectively. The galvanometer 4 then adjusts the direction of the laser beam, precisely scanning it onto the metal powder along a predetermined path. The powder supply system includes a powder cylinder 15, a scraper 17, and a dust collector 10, distributing the powder evenly on the forming platform. After each scan, the forming cylinder 14 descends one layer, the powder cylinder 15 rises one layer, and the scraper 17 evenly spreads the powder from the powder cylinder 15 onto the forming platform for the next laser scan. Excess powder falls into a powder recovery tank 18. The forming platform, consisting of a forming cylinder 14, a preheater, and a lifting platform 16, serves as the substrate for the part. Metal powder is melted and deposited layer by layer to form the shaped part 13. The forming platform can move up and down within the forming cylinder via the lifting platform 16 to control the height of the part. An inert gas protection device provides argon gas as a protective gas 5 to the part-making equipment, ensuring that the printing process is carried out in a low-oxygen environment, thereby improving part quality and safety.

[0077] 2) Perform defect segmentation on the powder-covered image using the U-Net target segmentation model and calculate the area of ​​missing powder;

[0078] Specifically, the training steps for the target segmentation model first include:

[0079] 3000 images of powder-spreading were collected through a printing experiment. After preprocessing such as perspective transformation, defective areas in the powder-spreading were labeled using LabelMe. The images were then divided into training, validation, and test sets in an 8:1:1 ratio. The U-Net algorithm was used for training with num_class = 2, input_shape = [512, 512], and 100 iterations using the Adam optimizer. The model's recognition accuracy was improved by adjusting hyperparameters, ultimately achieving a recognition accuracy of 97%. Model performance was measured by accuracy, precision, and mean squared error, as shown in the following formulas:

[0080] Accuracy = TP + TN / (TP + TN + FP + FN)

[0081] Precision = TP / (TP + FP)

[0082]

[0083] Where: TP is the number of true positives, i.e., the number of positive class samples correctly predicted as positive. TN is the number of true negatives, i.e., the number of negative class samples correctly predicted as negative. FP is the number of false positives, i.e., the number of negative class samples incorrectly predicted as positive. FN is the number of false negatives, i.e., the number of positive class samples incorrectly predicted as negative. n is the number of samples. y i It is the true value of the i-th sample. It is the predicted value of the i-th sample.

[0084] The preprocessed powder-spreading image is input into the trained U-Net target segmentation model to identify powder-spreading defects, including the number of powder-spreading defect layers, and the area of ​​powder-spreading defects is calculated.

[0085] Specifically, the U-Net object segmentation model includes an encoder, skip connection layers, a decoder, and an output layer, where:

[0086] 1) Encoder: An encoder consists of a series of convolutional and pooling layers, used to progressively reduce the spatial resolution of the input image and extract features.

[0087] 2) Skip Connections: There are skip connections between each layer of the encoder and the corresponding layer of the decoder, which connect the feature maps of the encoder layer and the feature maps of the decoder layer.

[0088] 3) Decoder: The decoder consists of a series of upsampling layers (usually deconvolutional or transposed convolutional layers) and convolutional layers corresponding to the encoder. ReLU activation functions are typically used in each layer of the decoder to add non-linearity.

[0089] 4) Output Layer: The output layer of UNet is typically a convolutional layer whose output is the same size as the input image and outputs a predicted label or segmentation mask for each pixel. The output layer usually uses an appropriate activation function (such as the sigmoid function or the softmax function) to produce pixel-level predictions.

[0090] After the UNet target model identifies defects and outputs their locations, the area of ​​the defects is calculated using the following steps:

[0091] 1. Obtain the prediction result: Use the UNet algorithm to obtain the segmentation mask of the defect region. The result is a binary image in which the defect region is marked as the foreground (white) and the non-defect region is marked as the background (black).

[0092] 2. Calculate the area of ​​the defect region: Perform pixel-level analysis on the prediction results and count the number of foreground pixels (defect region) to obtain the area of ​​the defect region. Use the following formula to calculate:

[0093]

[0094] Where N is the total number of pixels in the image, and Ii is the value of the i-th pixel.

[0095] 3. Output results: The calculated defect area is used as part of the identification results.

[0096] Furthermore, based on the number of powder-spreading defect layers and the area of ​​powder-spreading defects, mechanical properties are predicted using the relationships between powder-spreading defects and porosity, as well as between powder-spreading defects and tensile properties, thus obtaining the tensile property prediction results.

[0097] Finally, the predicted mechanical properties are compared with the set mechanical property reduction threshold. If the threshold is exceeded, remelting and re-powdering operations are performed. If the threshold is not exceeded, no intervention is performed.

[0098] Example 2

[0099] One embodiment of this disclosure provides a system for predicting the mechanical properties of laser selective melting formed parts, including:

[0100] The model building module is used to obtain the relationship between powder spreading defects and porosity, as well as the relationship between powder spreading defects and tensile properties. Based on the relationship between powder spreading defects and porosity, and between powder spreading defects and tensile properties, a mechanical property prediction model is constructed.

[0101] The image acquisition module is used to acquire images of each layer of powder spreading on the current molded part;

[0102] Furthermore, the image acquisition module enables positioning and imaging, ensuring that the scraper takes a timed picture after the powder is spread. An L-shaped diffuse reflection fiber optic sensor probe is installed inside the forming chamber. The sensor probe and fiber optic amplifier form a pair, connected by a single fiber optic cable. This pair of fiber optic sensors is installed at a predetermined position on the scraper to generate a signal as the scraper passes through, ensuring accurate detection of the scraper's position. The fiber optic amplifier is connected to the development board. After the sensor receives the signal, it is connected to the triggering system of the image acquisition module via the STM32 development board to trigger image acquisition.

[0103] Due to lens mounting and manufacturing processes, image acquisition modules are susceptible to various distortions, such as radial and tangential distortion. Accurate image information is crucial for controlling and monitoring the laser powder bed melting and spreading process. Camera calibration can correct these distortions, ensuring image accuracy.

[0104] Specifically, the camera calibration process is as follows: First, a 12×9 checkerboard calibration board is prepared, fixed on a molding platform, and its position is rotated and moved to take 16 photos. These images cover all possible working areas of the camera. For each calibration board image, a corner detection algorithm is used to detect corners on the calibration board, and the pixel coordinates of each corner are identified and recorded. For each detected corner, its pixel coordinates are correlated with the actual world coordinates of the calibration board. Using the collected pixel coordinate and world coordinate data, Zhang's calibration method is performed to calculate the camera's intrinsic and extrinsic parameters.

[0105] In addition, all the acquired data are obtained during the printing process, which is realized by a laser selective melting forming device. The device mainly consists of a laser system, a scanning system, a powder supply system, a forming platform, an inert gas device, and a control system. The laser system is the core component, responsible for generating a high-energy, highly focused laser beam. In this embodiment, the laser source used is a fiber laser 1. The scanning system is used to control the movement of the laser beam and includes a beam splitter 2, a beam expander 3, and a galvanometer 4. The beam splitter 2 and the beam expander 3 are used to separate and expand the laser beam emitted by the laser 1, respectively. The galvanometer 4 is then used to adjust the direction of the laser beam, precisely scanning the laser beam onto the metal powder along a predetermined path. The powder supply system includes a powder cylinder 15, a scraper 17, and a dust collector 10, which evenly distributes the powder on the forming platform. After each scan of the part, the forming cylinder 14 descends by one layer, the powder cylinder 15 rises by one layer, and the scraper 17 evenly spreads the powder in the powder cylinder 15 onto the forming platform for use in the next laser scan. Excess powder falls into the powder recovery tank 18. The forming platform, consisting of a forming cylinder 14, a preheater, and a lifting platform 16, serves as the substrate for the part. Metal powder is melted and deposited layer by layer to form the shaped part 13. The forming platform can move up and down within the forming cylinder via the lifting platform 16 to control the height of the part. An inert gas protection device provides argon gas as a protective gas 5 to the part-making equipment, ensuring that the printing process is carried out in a low-oxygen environment, thereby improving part quality and safety.

[0106] The image processing module is used to preprocess the powder spreading image, identify the number of powder spreading defect layers in the preprocessed powder spreading image, and calculate the area of ​​powder spreading defects.

[0107] The mechanical property prediction module is used to predict the mechanical properties of the current molded part based on the number of powder spreading defect layers and the area of ​​powder spreading defect using the mechanical property prediction model; and compares the predicted mechanical properties with the standard mechanical properties to monitor the degree of mechanical property degradation.

[0108] Also includes:

[0109] A movable light source module is configured for light source correction during the image acquisition process. Existing laser additive manufacturing equipment uses a fixed light source, and because the image acquisition module is off-axis mounted, and the environment is unstable during actual industrial production, uneven brightness in the acquired images occurs. By installing magnetic strips around the equipment's galvanometer and using a magnetically attached movable light source, the light source distribution within the chamber can be adjusted according to specific conditions.

[0110] The visualization monitoring module is configured to handle images acquired by the visualization image acquisition module, images processed by the visualization image processing module, and to generate operation commands such as re-coating and remelting.

[0111] Example 3

[0112] One embodiment of this disclosure provides a non-transitory computer-readable storage medium for storing computer instructions, which, when executed by a processor, implement the aforementioned method for predicting the mechanical properties of laser selective melting formed parts.

[0113] Example 4

[0114] One embodiment of this disclosure provides an electronic device, including a processor, a memory, and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory to enable the electronic device to perform the aforementioned method for predicting the mechanical properties of laser selective melting formed parts.

[0115] This disclosure is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create a machine for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0116] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0117] While the specific embodiments of this disclosure have been described above in conjunction with the accompanying drawings, this is not intended to limit the scope of protection of this disclosure. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art without creative effort based on the technical solutions of this disclosure are still within the scope of protection of this disclosure.

Claims

1. A method for predicting the mechanical properties of laser selective melting formed parts, characterized in that, include: Obtain the relationship between powder spreading defects and porosity, and between powder spreading defects and tensile properties. Based on the relationship between powder spreading defects and porosity, and between powder spreading defects and tensile properties, construct a mechanical property prediction model. Obtain powder spreading images for each layer of the current molded part, and preprocess the powder spreading images; Identify the number of powder-spreading defect layers in the preprocessed powder-spreading image and calculate the area of ​​powder-spreading defects; Based on the number of powder spreading defect layers and the area of ​​powder spreading defect, the mechanical property prediction model is used to make a prediction and obtain the current mechanical property prediction result of the molded part. The predicted mechanical properties are compared with the set mechanical property reduction threshold to determine whether remelting and re-coating operations should be performed.

2. The method for predicting the mechanical properties of laser selective melting formed parts as described in claim 1, characterized in that, The method for obtaining the relationship between powder spreading defects and porosity is as follows: Print irregular parts with different powder missing layers, obtain metallographic images of irregular parts formed with different defect layers, calculate the porosity of powder spreading defects with different layers using metallographic methods, obtain the relationship between porosity defect area and powder spreading defect layer number, fit a regression model to different powder missing layers and porosity, and obtain the formula for the relationship between powder spreading defects and porosity.

3. The method for predicting the mechanical properties of laser selective melting formed parts as described in claim 1, characterized in that, The method for obtaining the relationship between powder spreading defects and tensile properties is as follows: print tensile specimens with different powder missing layers, obtain tensile specimens by wire cutting, conduct tensile tests using a room temperature tensile testing machine, obtain the relationship between tensile properties and the number of powder spreading defect layers, fit a regression model, and determine the formula for the relationship between powder spreading defects and tensile properties.

4. The method for predicting the mechanical properties of laser selective melting formed parts as described in claim 1, characterized in that, The preprocessing of the powder-spreading image includes: a linear distortion correction process and a connected component filtering denoising operation. The linear distortion correction process involves performing a perspective transformation on the powder-spreading image, mapping each pixel in the powder-spreading image sequentially to a new plane, and then using connected component filtering denoising to take the pixel region near a pixel in the powder-spreading image as the connected neighborhood of that pixel, and then filtering out connected neighborhoods with an area of ​​less than 10 pixels.

5. The method for predicting the mechanical properties of laser selective melting formed parts as described in claim 1, characterized in that, The U-Net segmentation algorithm is used to identify powder spreading defects in powder spreading images, identify the number of powder spreading defect layers and calculate the area of ​​powder spreading defects. Based on the number of powder spreading defect layers and the area of ​​powder spreading defects, mechanical properties are predicted using the relationship between powder spreading defects and porosity and between powder spreading defects and tensile properties.

6. The method for predicting the mechanical properties of laser selective melting formed parts as described in claim 5, characterized in that, Based on the relationship between the degree of mechanical property degradation and the number of powder coating defect layers, a threshold of 30 MPa is set. During the printing process, the predicted mechanical property results are compared with the set mechanical property degradation threshold. If the threshold is exceeded, remelting and recoating operations are performed. If the threshold is not exceeded, no intervention is performed.

7. The method for predicting the mechanical properties of laser selective melting formed parts as described in claim 1, characterized in that, Acquiring images of each layer of powder coating on the current molded part includes: using an image acquisition module installed on the top of the forming chamber of the laser selective area equipment to acquire images of each layer of powder coating. The image acquisition module includes a light signal capture module, a photoelectric conversion module, and an optical module. The light signal capture module acquires the light after powder coating through a CMOS sensor. The photoelectric conversion module converts the light acquired by the light signal capture module into electrical signals and decomposes the photographed object into individual pixels according to a certain arrangement. These pixels are processed by an analog-to-digital converter and an image processor to obtain the powder coating image.

8. A system for predicting the mechanical properties of laser selective melting formed parts, characterized in that, include: The model building module is used to obtain the relationship between powder spreading defects and porosity, as well as the relationship between powder spreading defects and tensile properties. Based on the relationship between powder spreading defects and porosity, and between powder spreading defects and tensile properties, a mechanical property prediction model is constructed. The image acquisition module is used to acquire images of each layer of powder spreading on the current molded part; The image processing module is used to preprocess the powder spreading image, identify the number of powder spreading defect layers in the preprocessed powder spreading image, and calculate the area of ​​powder spreading defects. The mechanical property prediction module is used to predict the mechanical properties of the current molded part by using the mechanical property prediction model based on the number of powder spreading defect layers and the area of ​​powder spreading defect.

9. A non-transitory computer-readable storage medium, characterized in that, The non-transitory computer-readable storage medium is used to store computer instructions, which, when executed by a processor, implement a method for predicting the mechanical properties of laser selective melting formed parts as described in any one of claims 1-7.

10. An electronic device, characterized in that, include: The device includes a processor, a memory, and a computer program; wherein the processor is connected to the memory, the computer program is stored in the memory, and when the electronic device is running, the processor executes the computer program stored in the memory to enable the electronic device to perform a method for predicting the mechanical properties of laser selective melting forming parts as described in any one of claims 1-7.

Images