Training method, monitoring method and application of laser 3D printing defect monitoring model

Abstract

The invention discloses a training method, a monitoring method and application of a laser 3D printing defect monitoring model. The training method comprises the steps that an image sample set is established, the image sample set comprises an abnormal powder laying image and a melting channel defect image, the abnormal powder laying image is marked with at least one defect of scraper stripes, powder pile, insufficient powder supply, overhigh cladding layer and warping, and the melting channel defect image is marked with melting channel crack and / or hole defects; the image sample set is preprocessed, a training sample set is obtained, and preprocessing comprises at least one of image gray processing, histogram equalization processing, filtering processing, data enhancement processing and scarce defect sample synthesis processing; and training the defect monitoring model based on the training sample set, and determining model parameters of the defect monitoring model. Through the training method, the accuracy of defect monitoring model recognition can be improved, then intelligent decisions are made for defects in the printing process based on the model recognition result, and the printing quality is ensured.

Description

Technical Field

[0001] The present invention belongs to the technical field of laser printing, and in particular relates to a training method, a monitoring method and an application of a laser 3D printing defect monitoring model. Background Art

[0002] The laser selective melting 3D printing manufacturing process has the advantage of quickly manufacturing complex structural parts. It is based on the manufacturing concept of "layered manufacturing, layer-by-layer accumulation". The basic process of "powder laying-melting-solidification" is circulated during the manufacturing process to produce the corresponding structural parts. In these cycles, metallurgical defects may occur, which will significantly affect the final quality of the workpiece. Therefore, process control, process repeatability and quality monitoring of laser selective melting laser 3D printing are the focus of industrial mass production applications. In the 3D printing manufacturing process, if the powder laying quality does not meet the standards or the laser printing parameters do not meet the requirements, it may cause quality problems such as appearance defects and dimensional errors in the finished product.

[0003] The existing technology detects errors by taking full-area photographs of each layer of the printed structure and calculating the matching degree during the printing process, but it is unable to monitor the entire printing process in real time and make corresponding adjustments. In addition, it is difficult to monitor local defects in large-format scenes, which affects printing efficiency.

[0004] Therefore, in view of the above technical problems, it is necessary to provide an online intelligent monitoring method, device, electronic device and storage medium. Summary of the invention

[0005] The object of the present invention is to provide an online intelligent monitoring method, device, electronic device and storage medium, which can solve the above-mentioned problem that it is impossible to perform real-time defect monitoring for the entire printing process.

[0006] In order to achieve the above object, a specific embodiment of the present invention provides a training method for a laser 3D printing defect monitoring model, and the technical solution is as follows:

[0007] A training method for a laser 3D printing defect monitoring model, comprising:

[0008] Establishing an image sample set, the image sample set includes an abnormal powder spreading image and a melt path defect image, wherein the abnormal powder spreading image is annotated with at least one defect of scraper stripes, powder piles, insufficient powder supply, too high a cladding layer, and warpage, and the melt path defect image is annotated with melt path cracks and / or hole defects;

[0009] Preprocessing the image sample set to obtain a training sample set, wherein the preprocessing includes at least one of image grayscale processing, histogram equalization processing, filtering processing, data enhancement processing, and scarce defect sample synthesis processing;

[0010] The defect monitoring model is trained based on the training sample set to determine model parameters of the defect monitoring model.

[0011] In one or more embodiments of the present invention, the method specifically includes:

[0012] Increase the laser power and scanning speed in the 3D printing system to obtain images of melt crack defects;

[0013] Increase the laser power and reduce the scanning speed in the 3D printing system to obtain the image of the melt hole defect.

[0014] A specific embodiment of the present invention further provides a laser 3D printing defect monitoring method, and the technical solution is as follows:

[0015] A laser 3D printing defect monitoring method, comprising:

[0016] Acquire a powder spreading image during printing based on a camera device, and identify powder spreading abnormalities in the powder spreading image using a defect monitoring model;

[0017] When the powder spreading image meets the first preset condition, the temperature and shape of the molten pool during printing are monitored based on the photoelectric sensing device, thereby controlling the laser power and scanning speed;

[0018] When the molten pool meets the second preset condition, a solidified melt track image during printing is acquired in real time based on the camera device, and the melt track defects in the solidified melt track image are identified using the defect monitoring model;

[0019] Acquire a melt path image of the current layer after laser scanning is completed based on an imaging device, and identify melt path defects in the melt path image after laser scanning is completed using the defect monitoring model;

[0020] Wherein, the defect monitoring model is obtained by training based on the above method.

[0021] In one or more embodiments of the present invention, the method further comprises:

[0022] When a scraper stripe defect is identified in the powder spreading image, sending scraper replacement information; and / or,

[0023] When a powder pile defect is identified in the powder spreading image, sending re-powder spreading information; and / or,

[0024] When a powder supply shortage defect is identified in the powder spreading image, powder adding information or troubleshooting information is sent, and / or,

[0025] When it is identified that there is a defect of excessively high cladding layer in the powder spreading image, sending a further processing information; and / or,

[0026] When a warping defect is identified in the powder spreading image, sending a processing stop message; and / or,

[0027] The photoelectric sensor device monitors the temperature and shape of the melt pool during printing, and then controls the laser power and scanning speed, including:

[0028] When the photoelectric sensing device detects that the molten pool temperature and / or the molten pool size is less than the corresponding preset value, it controls to increase the current laser power and / or reduce the current laser scanning speed.

[0029] In one or more embodiments of the present invention, using the defect monitoring model to identify the melt path defect in the melt path image after the laser scanning is completed specifically includes:

[0030] Using the defect monitoring model to identify whether there are melt track cracks and / or holes greater than corresponding preset values ​​in the melt track image after the laser scanning is completed;

[0031] The method further comprises:

[0032] When the melt cracks and / or holes are larger than corresponding preset values, the current laser power is controlled to increase and / or the current laser scanning speed is controlled to decrease.

[0033] In one or more embodiments of the present invention, using the defect monitoring model to identify the melt path defect in the melt path image after the laser scanning is completed specifically includes:

[0034] Using the defect monitoring model to identify whether the number of melt track cracks and / or the area of ​​holes in the melt track image after the laser scanning is completed is greater than a corresponding preset value;

[0035] The method further comprises:

[0036] When the number of melt cracks and / or the hole area is greater than the corresponding preset value, the current layer of printed structure is remelted by laser scanning.

[0037] A specific embodiment of the present invention further provides a training device for a laser 3D printing defect monitoring model, and the technical solution is as follows:

[0038] A training device for a laser 3D printing defect monitoring model, comprising:

[0039] An acquisition module is used to establish an image sample set, wherein the image sample set includes an abnormal powder spreading image and a melt path defect image, wherein the abnormal powder spreading image is annotated with at least one defect of scraper stripes, powder piles, insufficient powder supply, excessive cladding layer, and warpage, and the melt path defect image is annotated with melt path cracks and / or hole defects;

[0040] A preprocessing module, used to preprocess the image sample set to obtain a training sample set, wherein the preprocessing includes at least one of image grayscale processing, histogram equalization processing, filtering processing, data enhancement processing, and rare defect sample synthesis processing;

[0041] The training module is used to train the defect monitoring model based on the training sample set to determine the model parameters of the defect monitoring model.

[0042] A specific embodiment of the present invention further provides a laser 3D printing defect monitoring device, and the technical solution is as follows:

[0043] A laser 3D printing defect monitoring device, comprising:

[0044] A first recognition module is used to obtain a powder spreading image during printing based on a camera device, and to identify powder spreading abnormalities in the powder spreading image using a defect monitoring model;

[0045] A monitoring module, used for monitoring the temperature and shape of the molten pool during printing based on a photoelectric sensor device when the powder spreading image meets the first preset condition, thereby controlling the laser power and scanning speed;

[0046] A second identification module is used to obtain a solidified melt track image during printing in real time based on a camera device when the melt pool meets a second preset condition, and identify a melt track defect in the solidified melt track image using the defect monitoring model;

[0047] A third identification module is used to obtain a melt path image of the current layer after the laser scanning is completed based on a camera device, and identify melt path defects in the melt path image after the laser scanning is completed using the defect monitoring model;

[0048] Wherein, the defect monitoring model is obtained by training based on the above method.

[0049] A specific embodiment of the present invention further provides an electronic device, and the technical solution is as follows:

[0050] An electronic device, comprising:

[0051] At least one processor; and a memory storing instructions, which, when executed by the at least one processor, enable the at least one processor to execute the training method of the laser 3D printing defect monitoring model or the laser 3D printing defect monitoring method as described above.

[0052] A specific embodiment of the present invention further provides a machine-readable storage medium, and the technical solution is as follows:

[0053] A machine-readable storage medium stores executable instructions, which, when executed, enable the machine to perform the training method of the laser 3D printing defect monitoring model or the laser 3D printing defect monitoring method as described above.

[0054] Compared with the existing technology, according to the training method of the laser 3D printing defect monitoring model, different abnormal powder spreading images and melt channel defect images can be obtained by establishing an image sample set. By preprocessing the image sample set, the standard can be unified and the number of training sample sets can be enriched to improve the generalization ability and robustness of the defect monitoring model, and enhance the accuracy of the model for defect monitoring. At the same time, after the defect monitoring model is applied to the laser 3D printing defect monitoring method, the entire laser 3D printing process can be monitored in real time. Among them, the powder spreading defects and melt channel defects are identified by the defect monitoring model, and intelligent decisions are made for different defects, which can significantly improve the stability and repeatability of the printing process, and greatly improve the mass production quality control of the 3D printing process. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0056] Figure 1 A schematic diagram of an implementation environment of a laser 3D printing defect monitoring model training method and a laser 3D printing defect monitoring method in one embodiment of the present invention;

[0057] Figure 2 It is a flowchart of a training method of a laser 3D printing defect monitoring model in one embodiment of the present invention;

[0058] Figure 3 Flow chart of a laser 3D printing defect monitoring method in one embodiment of the present invention;

[0059] Figure 4 It is a structural schematic diagram of a training device for a laser 3D printing defect monitoring model in one embodiment of the present invention;

[0060] Figure 5 It is a schematic diagram of the structure of a laser 3D printing defect monitoring device in one embodiment of the present invention;

[0061] Figure 6 is a hardware structure diagram of an electronic device in one embodiment of the present invention;

[0062] Figure 7An image of a molten pool in laser 3D printing in one embodiment of the present invention;

[0063] Figure 8 This is an image of a powder spreading defect in laser 3D printing according to an embodiment of the present invention.

[0064] Description of main reference numerals:

[0065] 11. Printing format; 12. f-θ mirror; 13. Galvanometer; 14. Laser; 15. Filter; 16. Industrial camera; 17. Photoelectric sensor; 18. Server; 19. Terminal;

[0066] 301, acquisition module; 302, preprocessing module; 303, training module;

[0067] 401, first identification module; 402, monitoring module; 403, second identification module; 404, third identification module. DETAILED DESCRIPTION

[0068] In order to enable those skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0069] The laser selective melting 3D printing manufacturing process has the advantage of rapidly manufacturing complex structural parts. It is based on the manufacturing concept of "layered manufacturing, layer-by-layer accumulation" and cycles the basic process of "powder laying-melting-solidification" during the manufacturing process to produce the corresponding structural parts. For the convenience of understanding, the terms involved in the embodiments of this application are first explained below.

[0070] Molten pool (such as Figure 7 The molten pool (shown in Figure 2) refers to the local melting area formed by the high-energy laser beam irradiating the powder bed during the laser selective melting process. The metal powder in this area is heated above the melting point to form liquid metal. The shape, size and stability of the molten pool directly affect the quality and microstructure of the printed part.

[0071] The melt path refers to the continuous melting area formed during a single laser scan. The formation of the melt path includes the melting of powder and the solidification of the melt. The surface quality and internal structure of the melt path have an important influence on the quality of the final product.

[0072] Reference Figure 1, showing a schematic diagram of an implementation environment provided by an exemplary embodiment of the present invention. The implementation environment includes a laser 3D printing device, a server 18 and a terminal 19. The laser 3D printing device, the server 18 and the terminal 19 communicate data through a communication network. Optionally, the communication network can be a wired network or a wireless network, and the communication network can be at least one of a local area network, a metropolitan area network and a wide area network.

[0073] The laser 3D printing device can complete the laser printing work, and can monitor and obtain the powder spreading image and the melting path image during printing. The terminal 19 can be an electronic device for receiving abnormal results in laser 3D printing, and the electronic device can be a smart phone, a tablet computer or a personal computer, etc. Figure 1 In the description, terminal 19 is taken as an example of a smart phone used by relevant technical personnel.

[0074] After receiving the powder spreading image and melt path image collected by the laser 3D printing device during printing, the server 18 can run the laser 3D printing defect monitoring method provided in the embodiment of the present application to identify them, and send the abnormal powder spreading information or abnormal melt path information to the terminal 19. Different decisions can be made for different defects, and the specific decisions can be sent to the laser 3D printing device.

[0075] In other possible implementations, the terminal 19 may also be an electronic device for identifying powder spreading defects or melt path defects, and the server 18 may also receive the defect identification result output by the terminal 19 .

[0076] The server can be an independent physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDNs), as well as big data and artificial intelligence platforms.

[0077] In this embodiment, a defect monitoring model is provided in the server 18, which is used to obtain the powder spreading image during printing, the solidified melt track image during printing, and the melt track image after the laser scanning of the current layer based on the camera device of the laser 3D printing device, and identify the powder spreading defects and melt track defects that may exist in the corresponding images. Optionally, the defect monitoring model is pre-trained in the server 18 based on the image sample set of abnormal powder spreading images and melt track defect images. Furthermore, the server 18 can also issue decision information to the laser 3D printing device based on the recognition result of the defect monitoring model to adjust the printing parameters of the laser printing device in real time.

[0078] Indicatively, Figure 1 As shown, after receiving the powder spreading image or the melt path image, the server 18 inputs the powder spreading image or the melt path image into the defect monitoring model to obtain the powder spreading defect or the melt path defect identified by the defect monitoring model. Then, the server 18 can issue decision information to the laser 3D printing device based on the identification result.

[0079] In other possible implementations, the defect monitoring model can also be deployed on the terminal 19 side, and the terminal 19 identifies the powder spreading defects or melt path defects and reports the identified powder spreading defects or melt path defects to the server 18 (to avoid the server 18 directly obtaining the powder spreading image or melt path image).

[0080] Reference Figure 1 The laser 3D printing device in this embodiment includes a printing surface 11, a laser 14, an industrial camera 16, and a photoelectric sensor 17; an f-θ mirror 12, a galvanometer 13, and a filter 15 may be correspondingly arranged above the printing surface 11. The laser beam emitted by the laser 14 can pass through the filter 15, the galvanometer 13, and the f-θ mirror 12 in sequence and reach the printing surface 11 to perform a 3D printing operation. The industrial camera 16 can obtain the powder spreading image and the melt path image in real time during printing, and the photoelectric sensor 17 can track the molten pool in real time. The filter 15 can filter the light signal of some wavelengths so that the photoelectric sensor 17 can capture the optical signal emitted by the molten pool in real time.

[0081] Reference Figure 2 , introduces a training method for a laser 3D printing defect monitoring model in an embodiment of the present invention, the training method comprising:

[0082] S101, establishing an image sample set, the image sample set including abnormal powder spreading images and melt path defect images, wherein the abnormal powder spreading images are marked with scraper stripes (such as Figure 8 As shown), at least one of powder pile, insufficient powder supply, too high cladding layer, and warping, and the melt channel defect image is annotated with melt channel cracks and / or hole defects. In this embodiment, the image sample set collects 591 scraper stripe defect images, 593 insufficient powder supply defect images, 577 powder pile defect images, 397 high cladding layer defect images, 300 warping defect images, 300 no powder laying defect images, more than 600 melt channel cracks and melt channel hole images, and 300 no melt channel defect images as an example to train the defect monitoring model. In other embodiments, the number of samples in the image sample set can be adjusted according to the actual situation.

[0083] Among them, the labeltool labeling tool can be used to label the abnormal powder spreading image and the melt channel defect image to provide accurate label information for the training of the defect monitoring model. Of course, in other embodiments, other labeling tools can also be used to label different defects.

[0084] For the melt defect image, the process parameters can be changed to deviate from the ideal process parameters to collect the melt crack and hole defect image data. Specifically, increasing the laser power and scanning speed in the 3D printing system can obtain the melt crack defect image; increasing the laser power and reducing the scanning speed in the 3D printing system can obtain the melt hole defect image.

[0085] S102, preprocessing the image sample set to obtain a training sample set, wherein the preprocessing includes at least one of image grayscale processing, histogram equalization processing, filtering processing, data enhancement processing, and scarce defective sample synthesis processing.

[0086] In this embodiment, the collected RGB image can be converted into a grayscale image to simplify the data processing process and highlight the characteristics of powder spreading defects and melt defects. The contrast of the image can be enhanced by histogram equalization processing, making the brightness distribution of the image more uniform. The median filtering technology can be used to remove salt and pepper noise in the image, while retaining edge information and improving image quality; in addition, the image can be convolved with a Gaussian filter kernel to smooth the image and reduce noise while retaining important defect features. The above-mentioned data enhancement processing includes image processing methods such as rotation, scaling, and flipping, which can increase the diversity of image data and improve the generalization ability of the model. For some scarce defect types, the image data is increased by artificial synthesis to balance the number of samples of each type of defect. Therefore, the above-mentioned preprocessing can improve image quality, highlight defect features, retain image details, and increase sample data, so it can be effectively applied to model training. Refer to Figure 8 , the image marked (a) in the figure is a powder defect image acquired by an industrial camera, and the image marked (b) in the figure is a powder defect image after preprocessing.

[0087] Furthermore, the image sample set can also be subjected to correction processing such as normalization, cropping, rotation correction, etc. to improve the training efficiency and recognition accuracy of the model.

[0088] S103: Train the defect monitoring model based on the training sample set to determine model parameters of the defect monitoring model.

[0089] In this embodiment, a convolutional neural network (CNN) is used to automatically extract features from the image, and these features can represent different types of defects. The convolutional neural network structure includes multiple convolutional layers, pooling layers, and fully connected layers to optimize the feature extraction process. Among them, the convolutional layer can contain multiple filters (also called convolution kernels), each of which is responsible for detecting specific features in the input data (such as edges, corners, etc. of the image). The pooling layer can reduce the feature dimension and enhance the invariance of the features. The extracted features can be mapped to the final output through the fully connected layer, and the classification labels corresponding to different defects are output in this embodiment.

[0090] Specifically, in this embodiment, the hyperparameters batch_size and learning_rate of the model training are set to 150 and 0.001, and the defect monitoring model is trained using the training sample set. At the same time, the network weights are optimized by the back propagation algorithm, and cross-validation and hyperparameter tuning are used to improve the generalization ability and robustness of the model. In other embodiments, the above parameters can be specifically adjusted according to actual conditions.

[0091] To further evaluate the training effect of the model, some image data can be selected from the training data set and input into the defect monitoring model to verify its recognition effect. In this embodiment, 15% of the images in the training data set were randomly selected to verify the defect monitoring model. After verification, the defect monitoring model training achieved good results, with mean_IoU of 0.615 and weighted IoU of 0.9533. There are 372 targets in the test set, 369 targets were correctly detected, the recall rate was 87.23%, the average precision rate of the test set was 88.14%, and the average detection time of a single image was 5.003ms, which meets the needs of online detection.

[0092] Furthermore, after the defect monitoring model is applied, the model can be further trained to perform iterative learning and model optimization, thereby improving the accuracy of defect identification.

[0093] Reference Figure 3 , a laser 3D printing defect monitoring method in an embodiment of the present invention is introduced. It should be noted in advance that the defect monitoring model involved in the monitoring method of this embodiment can be obtained based on the training method of the laser 3D printing defect monitoring model mentioned above.

[0094] The following specifically introduces a laser 3D printing defect monitoring method in an embodiment of the present invention, which specifically includes:

[0095] S201, obtaining a powder spreading image during printing based on a camera device, and identifying powder spreading anomalies in the powder spreading image using a defect monitoring model.

[0096] The camera device in this embodiment can be selected from the following Figure 1The industrial camera shown in the figure. To improve the accuracy and resolution of the powder spreading image, the CCD size of the industrial camera can be selected as 1 inch, the lens can be selected as 25-40mm, the working distance is in the range of 800-1000mm, and the full field of view of the 400mm side length format can be covered, and the resolution can reach greater than 0.01mm / pixel; the maximum shooting frame rate is 1kHz, and the shooting exposure time is set at 1-100ms to avoid reducing the printing efficiency.

[0097] It is understandable that for products of different sizes, if a larger printing format is required, such as 800mm or 1200mm side length, dual cameras or quad cameras can be selected for zone monitoring. By increasing the number of industrial cameras, high-resolution full coverage of the printing format can be guaranteed.

[0098] Before using the industrial camera, it needs to be calibrated. The calibration method includes: obtaining a calibration plate, obtaining at least 16 pictures at different angles by moving and rotating the calibration plate, and calibrating the linear distortion of different positions of the powder bed caused by the camera's paraxiality through perspective transformation.

[0099] After obtaining the powder spreading image during printing through a camera device, the defect monitoring model can be used to identify powder spreading anomalies in the powder spreading image.

[0100] In this embodiment, the monitoring method further includes:

[0101] When the scraper stripe defect is identified in the powder spreading image, the scraper replacement information is sent;

[0102] When powder pile defects are identified in the powder spreading image, a re-powder spreading information is sent;

[0103] When insufficient powder supply is detected in the powder spreading image, the system sends powder adding information or troubleshooting information.

[0104] When the over-cladding layer defect is identified in the powder spreading image, the information of continuing processing is sent;

[0105] When a warping defect is detected in the powder spreading image, a processing stop message is sent.

[0106] Among them, after receiving the corresponding identification information, the laser 3D printing device can make adjustments according to the corresponding strategy to ensure that the powder laying meets the requirements, thereby ensuring the printing quality of the product.

[0107] S202: When the powder-laying image meets the first preset condition, the temperature and shape of the molten pool during printing are monitored based on the photoelectric sensing device, thereby controlling the laser power and scanning speed.

[0108] The first preset condition is that the subsequent laser 3D printing work can be carried out after the identified powder spreading image does not have the above-mentioned scraper stripes, powder piles, insufficient powder supply, warping and other defects. That is, when the powder spreading quality meets the requirements, the next step of melting can be carried out.

[0109] Specifically, when the photoelectric sensor device detects that the molten pool temperature and / or the molten pool size is less than the corresponding preset value, the current laser power is controlled to increase and / or the current laser scanning speed is reduced. The preset value can be specifically set according to the printing requirements. In this embodiment, when the photoelectric sensor device detects that the molten pool temperature and / or the molten pool size is less than the corresponding preset value, the scanning speed can be reduced by 8-12%, and the laser power can be increased by 10-15%.

[0110] The photoelectric sensor can be a photoelectric sensor with a wavelength range of 600-900nm. The optical energy intensity emitted by the molten pool can be converted into the temperature of the molten pool. Specifically, it can be converted based on the Planck formula, and the formula is as follows:

[0111]

[0112] Among them, I λ represents the energy corresponding to the wavelength λ, T is the temperature value of Kelvin, h is the Planck constant, c is the speed of light; k B is the Boltzmann constant.

[0113] According to the Planck formula, the molten pool temperature can be obtained and the molten pool temperature distribution data can be obtained, so the geometric shape and size of the molten pool can be reconstructed according to the temperature distribution data. Furthermore, according to the laser scanning speed v, the sampling frequency of the photoelectric sensor is set to (8-12)v to ensure a balance between reducing the molten pool and data capacity.

[0114] S203: When the molten pool meets the second preset condition, a solidified melt track image during printing is acquired in real time based on an imaging device, and a melt track defect in the solidified melt track image is identified using a defect monitoring model.

[0115] The second preset condition is that the temperature and size of the molten pool meet the requirements. The camera device can be the industrial camera mentioned above. That is, when the melting work meets the requirements, the next step of solidification can be carried out.

[0116] Specifically, the defect monitoring model is used to identify whether there are melt cracks and / or holes greater than a corresponding preset value in the melt image after the laser scanning is completed. The above method also includes: when the melt cracks and / or holes are greater than the corresponding preset value, controlling to increase the current laser power and / or reduce the current laser scanning speed.

[0117] In this embodiment, the preset value corresponding to the melt crack is a length greater than 0.2mm, and the preset value corresponding to the melt hole is a diameter greater than 0.2mm. During the single-layer scanning printing process, the solidified melt is photographed and monitored by an industrial camera. When the melt crack and / or hole is greater than the corresponding preset value, the parameters can be dynamically adjusted to increase the scanning speed by 8-12% and reduce the laser power by 10-15%. It can be understood that in other embodiments, the preset values ​​of the melt crack and the hole can be adjusted and modified according to the actual printing quality requirements; at the same time, the laser power and the laser scanning speed can also be adjusted and modified according to the actual printing quality requirements.

[0118] S204, obtaining a melt path image of the current layer after laser scanning is completed based on an imaging device, and identifying melt path defects in the melt path image after laser scanning is completed using a defect monitoring model.

[0119] Specifically, the defect monitoring model is used to identify whether the number of melt track cracks and / or the area of ​​holes in the melt track image after the laser scanning is completed is greater than the corresponding preset value; the above method also includes: when the number of melt track cracks and / or the area of ​​holes is greater than the corresponding preset value, remelting the current layer of printed structure with laser scanning.

[0120] In this embodiment, the preset value of the number of melt cracks is 5, and the preset value of the hole area ratio is 1%. After the current layer scan is completed, the entire area is photographed. If the number of cracks is greater than 5 or the hole area ratio exceeds 1%, a second laser remelting scan is performed; if the corresponding number of defects still exist in the second scan, printing is interrupted. It can be understood that in other embodiments, the preset values ​​of the number of melt cracks and the hole area can be adjusted and modified according to the actual printing quality requirements.

[0121] The laser 3D printing defect monitoring method of this embodiment can monitor the printing process in real time, and accurately identify the powder spreading image and melt path image obtained by the camera device through the defect monitoring model, so as to make corresponding decisions based on the identified defects and control the laser 3D printing device to execute the corresponding decision. The laser 3D printing defect monitoring method of this embodiment can monitor the entire process of powder spreading, melting, and solidification online, and use the defect monitoring model to quickly identify defects, and make decisions based on the identification results, which can significantly improve the stability and repeatability of the printing process, and greatly improve the mass production quality control of the 3D printing process. At the same time, combined with Figure 1 , accidents that occur during the printing process can also be sent to the terminal, realizing unattended printing. Finally, since the laser 3D printing defect monitoring method of this embodiment can monitor the printing process in real time, it can obtain various data during the printing process, and also make the printing production process traceable.

[0122] Reference Figure 4, introduces a training device for a laser 3D printing defect monitoring model in an embodiment of the present invention. In this embodiment, the training device for a laser 3D printing defect monitoring model includes an acquisition module 301 , a preprocessing module 302 , and a training module 303 .

[0123] An acquisition module 301 is used to establish an image sample set, which includes an abnormal powder spreading image and a melt path defect image, wherein the abnormal powder spreading image is annotated with at least one defect of scraper stripes, powder piles, insufficient powder supply, excessive cladding layer, and warping, and the melt path defect image is annotated with melt path cracks and / or hole defects; a preprocessing module 302 is used to preprocess the image sample set to obtain a training sample set, wherein the preprocessing includes at least one of image grayscale processing, histogram equalization processing, filtering processing, data enhancement processing, and scarce defect sample synthesis processing; a training module 303 is used to train a defect monitoring model based on the training sample set to determine the model parameters of the defect monitoring model.

[0124] In an optional embodiment, the acquisition module 301 is also used to increase the laser power and scanning speed in the 3D printing system to obtain a melt crack defect image; increase the laser power and reduce the scanning speed in the 3D printing system to obtain a melt hole defect image.

[0125] Reference Figure 5 , introduces a laser 3D printing defect monitoring device in an embodiment of the present invention. In this embodiment, the laser 3D printing defect monitoring device includes a first recognition module 401, a monitoring module 402, a second recognition module 403 and a third recognition module 404.

[0126] The first recognition module 401 is used to obtain the powder spreading image during printing based on the camera device, and use the defect monitoring model to identify the powder spreading anomaly in the powder spreading image; the monitoring module 402 is used to monitor the temperature and shape of the molten pool during printing based on the photoelectric sensing device when the powder spreading image meets the first preset condition, and then control the laser power and the scanning speed; the second recognition module 403 is used to obtain the solidified melt track image during printing based on the camera device in real time when the molten pool meets the second preset condition, and use the defect monitoring model to identify the melt track defects in the solidified melt track image; the third recognition module 404 is used to obtain the melt track image after the laser scanning of the current layer is completed based on the camera device, and use the defect monitoring model to identify the melt track defects in the melt track image after the laser scanning is completed; wherein, the defect monitoring model is obtained by training based on the above-mentioned method.

[0127] In an optional embodiment, the first recognition module 401 is also used to send scraper replacement information when a scraper stripe defect is identified in the powder spreading image; and / or, the first recognition module 401 is also used to send re-powder spreading information when a powder pile defect is identified in the powder spreading image; and / or, the first recognition module 401 is also used to send powder adding information or troubleshooting information when a powder supply shortage defect is identified in the powder spreading image, and / or, the first recognition module 401 is also used to send continue processing information when a high cladding layer defect is identified in the powder spreading image; and / or, the first recognition module 401 is also used to send stop processing information when a warping defect is identified in the powder spreading image.

[0128] In an optional embodiment, the monitoring module 402 is also used to control increasing the current laser power and / or reducing the current laser scanning speed when the photoelectric sensing device detects that the molten pool temperature and / or the molten pool size is less than the corresponding preset value.

[0129] In an optional embodiment, the second identification module 403 is also used to use a defect monitoring model to identify whether there are melt track cracks and / or holes greater than corresponding preset values ​​in the melt track image after the laser scanning is completed; when the melt track cracks and / or holes are greater than the corresponding preset values, control to increase the current laser power and / or reduce the current laser scanning speed.

[0130] In an optional embodiment, the third identification module 404 is also used to use a defect monitoring model to identify whether the number of melt cracks and / or the area of ​​holes in the melt image after the laser scanning is completed is greater than the corresponding preset value; when the number of melt cracks and / or the area of ​​holes is greater than the corresponding preset value, the current layer of printed structure is remelted by laser scanning.

[0131] As above Figures 1 to 3 , a training method for a laser 3D printing defect monitoring model and a laser 3D printing defect monitoring method according to an embodiment of this specification are described. The details mentioned in the above description of the method embodiment are also applicable to the training device for the laser 3D printing defect monitoring model and the laser 3D printing defect monitoring device of the embodiment of this specification. The above training device for the laser 3D printing defect monitoring model and the laser 3D printing defect monitoring device can be implemented in hardware, or in software, or in a combination of hardware and software.

[0132] Reference Figure 6 , shows a hardware structure diagram of an electronic device according to an embodiment of this specification. Figure 6As shown, the electronic device 50 may include at least one processor 51, a memory 52 (e.g., a non-volatile memory), a memory 53, and a communication interface 54, and the at least one processor 51, the memory 52, the memory 53, and the communication interface 54 are connected together via an internal bus 55. At least one processor 51 executes at least one computer-readable instruction stored or encoded in the memory 52.

[0133] It should be understood that the computer executable instructions stored in the memory 52, when executed, cause at least one processor 51 to perform the above combined operations in various embodiments of the present specification. Figures 1 to 3 Describes the various operations and functions.

[0134] In the embodiments of the present specification, the electronic device 50 may include, but is not limited to: a personal computer, a server computer, a workstation, a desktop computer, a laptop computer, a notebook computer, a mobile electronic device, a smart phone, a tablet computer, a cellular phone, a personal digital assistant (PDA), a handheld device, a messaging device, a wearable electronic device, a consumer electronic device, and the like.

[0135] According to one embodiment, a program product such as a machine-readable medium is provided. The machine-readable medium may have instructions (i.e., the above-mentioned elements implemented in the form of software), which, when executed by a machine, causes the machine to perform the above-mentioned combination of various embodiments of this specification. Figure 1-Figure 3 Specifically, a system or device equipped with a readable storage medium may be provided, on which a software program code implementing the functions of any of the above-mentioned embodiments is stored, and a computer or processor of the system or device reads and executes the instructions stored in the readable storage medium.

[0136] In this case, the program code itself read from the machine-readable medium can realize the function of any one of the above embodiments, and thus the machine-readable code and the machine-readable storage medium storing the machine-readable code constitute part of this specification.

[0137] Examples of readable storage media include floppy disks, hard disks, magneto-optical disks, optical disks (such as CD-ROM, CD-R, CD-RW, DVD-ROM, DVD-RAM, DVD-RW, DVD-RW), magnetic tapes, non-volatile memory cards, and ROMs. Alternatively, the program code may be downloaded from a server computer or a cloud via a communication network.

[0138] Those skilled in the art should understand that the various embodiments disclosed above can be modified and altered in various ways without departing from the essence of the invention. Therefore, the protection scope of this specification should be defined by the appended claims.

[0139] It should be noted that not all steps and units in the above-mentioned processes and system structure diagrams are necessary, and some steps or units can be ignored according to actual needs. The execution order of each step is not fixed and can be determined as needed. The device structure described in the above-mentioned embodiments can be a physical structure or a logical structure, that is, some units may be implemented by the same physical client, or some units may be implemented by multiple physical clients, or some components in multiple independent devices may be implemented together.

[0140] In the above embodiments, the hardware unit or module can be realized by mechanical or electrical means. For example, a hardware unit, module or processor can include permanent dedicated circuit or logic (such as special processor, FPGA or ASIC) to complete the corresponding operation. The hardware unit or processor can also include programmable logic or circuit (such as general-purpose processor or other programmable processor), which can be temporarily set by software to complete the corresponding operation. Specific implementation (mechanical method or dedicated permanent circuit or temporary circuit) can be determined based on cost and time consideration.

[0141] The specific embodiments described above in conjunction with the accompanying drawings describe exemplary embodiments, but do not represent all embodiments that can be implemented or fall within the scope of protection of the claims. The term "exemplary" used throughout this specification means "used as an example, instance or illustration" and does not mean "preferred" or "having advantages" over other embodiments. For the purpose of providing an understanding of the described technology, the specific embodiments include specific details. However, these technologies can be implemented without these specific details. In some instances, in order to avoid making the concepts of the described embodiments difficult to understand, well-known structures and devices are shown in block diagram form.

[0142] The above description of the present disclosure is provided to enable any person of ordinary skill in the art to implement or use the present disclosure. Various modifications to the present disclosure will be apparent to those of ordinary skill in the art, and the general principles corresponding to the present disclosure may be applied to other variations without departing from the scope of protection of the present disclosure. Therefore, the present disclosure is not limited to the examples and designs described herein, but is consistent with the widest range of principles and novel features disclosed herein.

[0143] In addition, it should be understood that although the present specification is described according to implementation modes, not every implementation mode contains only one independent technical solution. This description of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment may also be appropriately combined to form other implementation modes that can be understood by those skilled in the art.

Claims

1. A training method for a laser 3D printing defect monitoring model, characterized in that: include: Establishing an image sample set, the image sample set includes an abnormal powder spreading image and a melt path defect image, wherein the abnormal powder spreading image is annotated with at least one defect of scraper stripes, powder piles, insufficient powder supply, too high a cladding layer, and warpage, and the melt path defect image is annotated with melt path cracks and / or hole defects; Preprocessing the image sample set to obtain a training sample set, wherein the preprocessing includes at least one of image grayscale processing, histogram equalization processing, filtering processing, data enhancement processing, and scarce defect sample synthesis processing; The defect monitoring model is trained based on the training sample set to determine model parameters of the defect monitoring model.

2. The training method of the laser 3D printing defect monitoring model according to claim 1, characterized in that: The method specifically comprises: Increase the laser power and scanning speed in the 3D printing system to obtain images of melt crack defects; Increase the laser power and reduce the scanning speed in the 3D printing system to obtain the image of the melt hole defect.

3. A laser 3D printing defect monitoring method, characterized in that: include: Acquire a powder spreading image during printing based on a camera device, and identify powder spreading abnormalities in the powder spreading image using a defect monitoring model; When the powder spreading image meets the first preset condition, the temperature and shape of the molten pool during printing are monitored based on the photoelectric sensing device, thereby controlling the laser power and scanning speed; When the molten pool meets the second preset condition, a solidified melt track image during printing is acquired in real time based on the camera device, and the melt track defects in the solidified melt track image are identified using the defect monitoring model; Acquire a melt path image of the current layer after laser scanning is completed based on an imaging device, and identify melt path defects in the melt path image after laser scanning is completed using the defect monitoring model; Wherein, the defect monitoring model is obtained by training based on the method described in claim 1 or 2.

4. The laser 3D printing defect monitoring method according to claim 3, characterized in that: The method further comprises: When a scraper stripe defect is identified in the powder spreading image, sending scraper replacement information; and / or, When a powder pile defect is identified in the powder spreading image, sending re-powder spreading information; and / or, When a powder supply shortage defect is identified in the powder spreading image, powder adding information or troubleshooting information is sent, and / or, When it is identified that there is a defect of excessively high cladding layer in the powder spreading image, sending a further processing information; and / or, When a warping defect is identified in the powder spreading image, sending a processing stop message; and / or, The photoelectric sensor device monitors the temperature and shape of the melt pool during printing, and then controls the laser power and scanning speed, including: When the photoelectric sensing device detects that the molten pool temperature and / or the molten pool size is less than the corresponding preset value, it controls to increase the current laser power and / or reduce the current laser scanning speed.

5. The laser 3D printing defect monitoring method according to claim 3, characterized in that: Using the defect monitoring model to identify the melt path defects in the melt path image after the laser scanning is completed specifically includes: Using the defect monitoring model to identify whether there are melt track cracks and / or holes greater than corresponding preset values ​​in the melt track image after the laser scanning is completed; The method further comprises: When the melt cracks and / or holes are larger than corresponding preset values, the current laser power is controlled to increase and / or the current laser scanning speed is controlled to decrease.

6. The laser 3D printing defect monitoring method according to claim 3, characterized in that: Using the defect monitoring model to identify the melt path defects in the melt path image after the laser scanning is completed specifically includes: Using the defect monitoring model to identify whether the number of melt track cracks and / or the area of ​​holes in the melt track image after the laser scanning is completed is greater than a corresponding preset value; The method further comprises: When the number of melt cracks and / or the hole area is greater than the corresponding preset value, the current layer of printed structure is remelted by laser scanning.

7. A training device for a laser 3D printing defect monitoring model, characterized in that: include: An acquisition module is used to establish an image sample set, wherein the image sample set includes an abnormal powder spreading image and a melt path defect image, wherein the abnormal powder spreading image is annotated with at least one defect of scraper stripes, powder piles, insufficient powder supply, excessive cladding layer, and warpage, and the melt path defect image is annotated with melt path cracks and / or hole defects; A preprocessing module, used to preprocess the image sample set to obtain a training sample set, wherein the preprocessing includes at least one of image grayscale processing, histogram equalization processing, filtering processing, data enhancement processing, and rare defect sample synthesis processing; The training module is used to train the defect monitoring model based on the training sample set to determine the model parameters of the defect monitoring model.

8. A laser 3D printing defect monitoring device, characterized in that: include: A first recognition module is used to obtain a powder spreading image during printing based on a camera device, and to identify powder spreading anomalies in the powder spreading image using a defect monitoring model; A monitoring module, used for monitoring the temperature and shape of the molten pool during printing based on a photoelectric sensor device when the powder spreading image meets the first preset condition, thereby controlling the laser power and scanning speed; A second identification module is used to obtain a solidified melt track image during printing in real time based on a camera device when the melt pool meets a second preset condition, and identify a melt track defect in the solidified melt track image using the defect monitoring model; A third identification module is used to obtain a melt path image of the current layer after the laser scanning is completed based on a camera device, and identify melt path defects in the melt path image after the laser scanning is completed using the defect monitoring model; Wherein, the defect monitoring model is obtained by training based on the method described in claim 1 or 2.

9. An electronic device, characterized in that: include: at least one processor; as well as A memory storing instructions, which, when executed by the at least one processor, causes the at least one processor to execute the training method for the laser 3D printing defect monitoring model as described in claim 1 or 2 or the laser 3D printing defect monitoring method as described in any one of claims 3 to 6.

10. A machine-readable storage medium, characterized in that: It stores executable instructions, which, when executed, enable the machine to execute the training method of the laser 3D printing defect monitoring model as described in claim 1 or 2 or the laser 3D printing defect monitoring method as described in any one of claims 3 to 6.

Images