A method and system for online monitoring and real-time control of intelligent additive manufacturing

Abstract

The present invention discloses an intelligent additive manufacturing online monitoring and real-time control method and system, which relates to the field of additive manufacturing technology. It solves the technical problem that it is difficult to consider multiple processes of additive manufacturing, evaluate the quality of additive manufacturing from multiple angles, and avoid problems that may occur in subsequent processing, which affects the product qualification rate. Through the uniformity coefficient of the cladding layer thickness, uneven problems are discovered in time, thereby improving the consistency and reliability of the final product. Thermal infrared images and high-definition images are used for real-time monitoring to quickly identify abnormal situations. At the same time, it is convenient to help identify potential process problems and provide a basis for subsequent improvements. Through the comprehensive evaluation of multiple indicators, setting preset thresholds makes the evaluation process more objective and quantifiable, which helps to clarify the boundaries between qualified and unqualified. Marking grids with low cladding degrees can quickly discover abnormal situations in production, so that timely measures can be taken to adjust or correct them.

Description

Technical Field

[0001] The present invention belongs to the field of additive manufacturing, and specifically relates to an intelligent additive manufacturing online monitoring and real-time control method and system. Background Art

[0002] With the rapid development of Additive Manufacturing (AM) technology, especially in the fields of industry, aerospace, medical and automotive, it is particularly important to ensure the quality and efficiency of the production process. In the traditional additive manufacturing process, due to the influence of multiple variables such as material properties, equipment status and environmental factors, the product quality may be uncertain, which requires an effective method to monitor and control the entire production process in real time.

[0003] In recent years, the development of intelligent technologies, such as the Internet of Things (IoT), big data analysis, and artificial intelligence (AI), has provided new opportunities for online monitoring and real-time control in the additive manufacturing process. These technologies can help identify potential problems, optimize production parameters, and take corrective measures in a timely manner when abnormalities occur, thereby improving product consistency and reliability. The intelligent additive manufacturing online monitoring and real-time control solution refers to a solution that uses sensor networks, data acquisition and analysis tools, and advanced control algorithms to comprehensively monitor and dynamically adjust the additive manufacturing process.

[0004] The existing intelligent additive manufacturing online monitoring and real-time control solutions are difficult to take into account multiple additive manufacturing processes and conduct multi-angle evaluation of the quality of additive manufacturing. There is also a lack of solutions for timely repairing specific locations in product production during the next slicing manufacturing process, making it difficult to avoid problems that may arise in subsequent processing and affect the product qualification rate. Summary of the invention

[0005] The present invention aims to solve at least one of the technical problems existing in the prior art; to this end, the present invention proposes an intelligent additive manufacturing online monitoring and real-time control method and system, which is used to solve the technical problems that it is difficult to take into account multiple additive manufacturing processes, to evaluate the quality of additive manufacturing from multiple angles, and to avoid problems that may arise in subsequent processing, thereby affecting the product qualification rate.

[0006] To solve the above problems, the first aspect of the present invention provides an online monitoring and real-time control method for intelligent additive manufacturing, comprising:

[0007] According to the 3D model of additive manufacturing, the 3D model is sliced, and all the slice coverage areas of the printed part are meshed;

[0008] The laser displacement sensor is used to monitor the powder thickness and cladding thickness of each grid in real time during printing of each slice, ensuring that it is within the range of 30-60μm.

[0009] Monitor the temperature of the melting area and collect thermal infrared images and high-definition images of each grid slice in real time for the powder laying and cladding process;

[0010] According to the powder thickness and cladding layer thickness data of each grid, the cladding thickness uniformity coefficient of the cladding layer is evaluated, and the process quality coefficient of each grid is calculated based on the collected thermal infrared images and high-definition images;

[0011] According to the process quality coefficient, cladding thickness uniformity coefficient and cladding layer thickness data of each grid, the cladding degree of the grid is evaluated, the corresponding grids whose cladding degree is less than the preset threshold are marked, and the thickness difference of the marked grid is calculated according to the cladding layer thickness data of the marked grid;

[0012] According to the powder density and material data of the slices, a volume shrinkage function model based on material properties is established to adjust the powder thickness of the marked grid.

[0013] As a further solution of the present invention: according to the powder thickness and cladding layer thickness data of each grid, the cladding thickness uniformity coefficient of the cladding layer is evaluated, comprising the following steps:

[0014] According to the powder thickness and cladding layer thickness data of each grid, the average value and variance of the powder thickness and cladding layer thickness are calculated;

[0015] According to the average value and variance of the powder thickness and the cladding layer thickness, the cladding thickness uniformity coefficient of the cladding layer is calculated by the following formula:

[0016] ;

[0017] Among them, α is the cladding thickness uniformity coefficient of the cladding layer, S1 is the variance of the powder thickness of different grids of the same slice, μ1 is the average value of the powder thickness of different grids of the same slice, S2 is the variance of the cladding layer thickness of different grids of the same slice, and μ2 is the average value of the cladding layer thickness of different grids of the same slice.

[0018] As a further solution of the present invention: calculating the process quality coefficient of each grid according to the collected thermal infrared image and high-definition image, including the following steps:

[0019] Obtain historical data of thermal infrared images and high-definition images of the surface of the molten area and the powder spreading area when powder spreading defects and cladding defects occur during additive manufacturing of the same powder spreading material, and mark the locations of powder spreading defects and cladding defects;

[0020] Train the neural network model with labeled images to identify defect locations in thermal infrared images and high-definition images;

[0021] Acquire thermal infrared images and high-definition images collected in real time, and perform image enhancement on the thermal infrared images and high-definition images;

[0022] Input the acquired thermal infrared images and high-definition images into the trained neural network model to identify the defect location;

[0023] Count the number of grids covered by the defect position, perform grayscale processing on the HD image and thermal infrared image, count the grayscale value of each pixel, and calculate the process quality coefficient of each grid using the following formula:

[0024] ;

[0025] Among them, β is the process quality coefficient of the grid; Pa is the mean grayscale value of the pixel points in the grid of the powder laying HD image, Pa1 is the mean grayscale value of the pixel points in the grid of the powder laying HD image without defect coverage, fa is the ratio of the number of grids covered by the defect position in the powder laying HD image grid to the total number of grids; Pb is the mean grayscale value of the pixel points in the grid of the cladding HD image, Pb1 is the mean grayscale value of the pixel points in the grid of the cladding HD image without defect coverage, fb is the ratio of the number of grids covered by the defect position in the cladding HD image grid to the total number of grids. Pc is the mean gray value of the pixels in the grid of the powder laying thermal infrared image, Pc1 is the mean gray value of the pixels in the grid not covered by the defects in the powder laying thermal infrared image, fc is the ratio of the number of grids covered by the defective positions in the grid of the powder laying thermal infrared image to the total number of grids; Pd is the mean gray value of the pixels in the grid of the cladding thermal infrared image, Pd1 is the mean gray value of the pixels in the grid not covered by the defects in the cladding thermal infrared image, fd is the ratio of the number of grids covered by the defective positions in the grid of the cladding thermal infrared image.

[0026] As a further solution of the present invention: according to the process quality coefficient, cladding thickness uniformity coefficient and cladding layer thickness data of each grid, the cladding degree of the grid is evaluated, comprising the following steps:

[0027] Obtain the process quality coefficient β, cladding thickness uniformity coefficient α and cladding layer thickness data of each grid;

[0028] Calculate the difference C between the mean cladding thickness in the defect-free grid and the cladding thickness of the current grid;

[0029] The degree of mesh cladding is evaluated by the following formula:

[0030] ;

[0031] Among them, E is the evaluation value of the cladding degree of the grid, L0 is the average thickness of the cladding layer in the defect-free grid.

[0032] As a further solution of the present invention: corresponding grids whose cladding degree is less than a preset threshold are marked, and the thickness difference of the marked grids is calculated according to the cladding layer thickness data of the marked grids, including the following steps:

[0033] Obtain the cladding degree evaluation value of the grid, and mark the corresponding grid if the cladding degree evaluation value of the grid is less than the preset threshold;

[0034] According to the cladding layer thickness data of the marked grid, the difference between the mean cladding layer thickness in the defect-free grid and the cladding layer thickness of the current grid is taken as the thickness difference.

[0035] As a further solution of the present invention: according to the powder density and material data of the slice, a volume shrinkage function model based on material properties is established to adjust the powder thickness of the marked grid, including the following steps:

[0036] Collect the powder density and material data of the slices, obtain the historical data of the powder thickness and the layer thickness after cladding with the same powder density as the slices, and establish a volume shrinkage function model based on material properties;

[0037] According to the real-time collected data of powder thickness and cladding thickness, the established volume shrinkage function model based on material properties is parameterized.

[0038] Through the parameter-corrected volume shrinkage function model based on material properties, the thickness of the marked grid is adjusted according to the thickness difference of the marked grid when the powder is applied in the next slice.

[0039] As a further solution of the present invention, the powder density and material data of the slices are collected, the historical data of the powder thickness and the layer thickness after cladding with the same powder density as the slices are obtained, and a volume shrinkage function model based on material properties is established, including the following steps:

[0040] Collect the powder density and material data of the slices, obtain the historical data of the powder thickness and the layer thickness after cladding of the same powder material as the slices, and establish an initial volume shrinkage function model based on material properties;

[0041] ;

[0042] Among them, H initial H is the thickness of the powder coating, final is the layer thickness after cladding, k is the cladding shrinkage coefficient, and ρ is the powder density;

[0043] The powder density, material, layer thickness after cladding, and powder thickness are fitted to the equation of the initial volume shrinkage function model based on material properties using the least squares method to obtain the fitting parameter cladding shrinkage coefficient k as the volume shrinkage function model based on material properties.

[0044] As a further solution of the present invention: according to the real-time collected data of powder thickness and cladding thickness, the established volume shrinkage function model based on material properties is parameter corrected, including the following steps:

[0045] The expected cladding layer thickness is calculated using the established volume shrinkage function model based on material properties. The actual measured powder thickness is compared with the predicted value of powder thickness, and the error is calculated:

[0046] ;

[0047] Where: Hactual is the actual measured powder thickness, Hpredicted is the predicted value of powder thickness calculated based on existing parameters;

[0048] Calculate the derivative of the loss function with respect to the parameter cladding shrinkage coefficient k, and update the parameter using the following formula:

[0049] ;

[0050] Where: k' is the updated parameter cladding shrinkage coefficient, is the derivative of the loss function with respect to the parameter cladding shrinkage coefficient k, L = is the loss function, and ε is the prediction accuracy of the powder thickness prediction value

[0051] As a further solution of the present invention, according to the thickness difference of the marked grid, when the next slice is powdered, the powdering thickness of the marked grid is adjusted, comprising the following steps:

[0052] According to the thickness difference of the marked grid, when the next slice is clad, the planned layer thickness after cladding is added to the thickness difference of the marked grid to obtain the actual layer thickness after cladding of the next slice;

[0053] The thickness of the powder coating corresponding to the next slice of the marked grid is calculated through a volume shrinkage function model based on material properties;

[0054] According to the thickness of the powder coating corresponding to the next slice of the marked grid, the powder coating thickness of the marked grid is adjusted.

[0055] As another aspect of the present invention: the system adopts an intelligent additive manufacturing online monitoring and real-time control method as described above to achieve online monitoring and real-time control of additive manufacturing.

[0056] Compared with the prior art, the present invention has the following beneficial effects:

[0057] The present invention can timely detect uneven problems through the uniformity coefficient of the cladding layer thickness, thereby improving the consistency and reliability of the final product. At the same time, it is convenient to help identify potential process problems and provide a basis for subsequent improvements. Real-time monitoring using thermal infrared images and high-definition images can quickly identify abnormal situations, ensure the stability of the production process, facilitate the reasonable adjustment of the powder thickness and other parameters, effectively utilize materials, and improve overall production efficiency.

[0058] The present invention facilitates more accurate judgment of the cladding degree of each grid through comprehensive evaluation of multiple indicators; setting a preset threshold makes the evaluation process more objective and quantifiable, which helps to clarify the boundary between qualified and unqualified; at the same time, marking grids with low cladding degrees can quickly discover abnormal conditions in production, so that timely measures can be taken to adjust or correct them; by analyzing the thickness difference of the marked grids, it is convenient to improve relevant process parameters in a targeted manner in the later stage and improve the overall production efficiency; by identifying areas with insufficient cladding in advance, it is helpful to avoid problems that may occur in subsequent processing and reduce product defect rates; timely discovery and processing of unqualified grids can significantly reduce the waste of time and materials caused by rework, thereby saving costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] 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 of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0060] Figure 1 It is a schematic diagram of the process of the present invention;

[0061] Figure 2 The present invention is a schematic flow chart of a method for adjusting the powdering thickness of a marking grid. DETAILED DESCRIPTION

[0062] The technical scheme of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than 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 are within the scope of protection of the present invention.

[0063] See also Figure 1-Figure 2 The first aspect of the present invention provides an online monitoring and real-time control method for intelligent additive manufacturing, comprising:

[0064] According to the 3D model of additive manufacturing, the 3D model is sliced, and all the slice coverage areas of the printed part are meshed;

[0065] The powder thickness and cladding thickness data of each grid are monitored in real time by a laser displacement sensor during the printing of each slice. In this embodiment, it is necessary to ensure that the powder thickness is within the range of 30-100 μm.

[0066] Monitor the temperature of the melting area and collect thermal infrared images and high-definition images of each grid slice in real time for the powder laying and cladding process;

[0067] According to the powder thickness and cladding layer thickness data of each grid, the cladding thickness uniformity coefficient of the cladding layer is evaluated, and the process quality coefficient of each grid is calculated according to the collected thermal infrared image and high-definition image; in this embodiment, the cladding layer thickness uniformity coefficient is used to timely detect uneven problems, thereby improving the consistency and reliability of the final product. It is convenient to help identify potential process problems and provide a basis for subsequent improvements. Real-time monitoring using thermal infrared images and high-definition images can quickly identify abnormal situations, ensure the stability of the production process, facilitate the reasonable adjustment of powder thickness and other parameters, and effectively utilize materials to improve overall production efficiency.

[0068] According to the process quality coefficient, cladding thickness uniformity coefficient and cladding layer thickness data of each grid, the cladding degree of the grid is evaluated, and the corresponding grids whose cladding degree is less than the preset threshold are marked, and the thickness difference of the marked grid is calculated according to the cladding layer thickness data of the marked grid; in this embodiment, through the comprehensive evaluation of multiple indicators, it is convenient to more accurately judge the cladding degree of each grid; setting a preset threshold makes the evaluation process more objective and quantifiable, which helps to clarify the boundary between qualified and unqualified; at the same time, marking grids with low cladding degree can quickly discover abnormal situations in production, so as to take timely measures to adjust or correct them. By analyzing the thickness difference of the marked grid, the relevant process parameters can be improved in a targeted manner and the overall production efficiency can be improved. By identifying the areas with insufficient cladding in advance, it is helpful to avoid problems that may occur in subsequent processing and reduce the product defect rate; timely discovery and processing of unqualified grids can significantly reduce the time and material waste caused by rework, thereby saving costs.

[0069] According to the powder density and material data of the slices, a volume shrinkage function model based on material characteristics is established, and the powder thickness of the marked grid is adjusted. In this embodiment, by understanding the volume shrinkage characteristics of different materials, a corresponding powder thickness can be formulated for each material, thereby achieving a more accurate cladding effect. Adjusting the powder thickness of each material helps to improve the overall cladding quality. At the same time, adjusting the powder thickness helps to improve the surface quality of the cladding layer and reduce the need for subsequent processing. By calculating the required powder thickness, it is convenient to effectively compensate for defects in the previous powder spreading or cladding process.

[0070] By establishing a volume shrinkage function model, a standardized processing solution is provided for different materials, which helps to form a repeatable and stable production process. At the same time, during the production process, parameters can be adjusted in real time according to actual conditions to further improve process stability.

[0071] In one embodiment of the present invention, evaluating the cladding thickness uniformity coefficient of the cladding layer according to the powder thickness and cladding layer thickness data of each grid includes the following steps:

[0072] According to the powder thickness and cladding layer thickness data of each grid, the average value and variance of the powder thickness and cladding layer thickness are calculated;

[0073] According to the average value and variance of the powder thickness and the cladding layer thickness, the cladding thickness uniformity coefficient of the cladding layer is calculated by the following formula:

[0074] ;

[0075] Among them, α is the cladding thickness uniformity coefficient of the cladding layer, S1 is the variance of the powder thickness of different grids of the same slice, μ1 is the average value of the powder thickness of different grids of the same slice, S2 is the variance of the cladding layer thickness of different grids of the same slice, and μ2 is the average value of the cladding layer thickness of different grids of the same slice.

[0076] In this embodiment, when calculating the average values ​​of different grid powder thicknesses of the same slice, the different grid powder thicknesses are normalized and converted into the range of [0, 10], and then the cladding thickness uniformity coefficient of the cladding layer is calculated.

[0077] In one embodiment of the present invention, the process quality coefficient of each grid is calculated based on the collected thermal infrared image and high-definition image, including the following steps:

[0078] Obtain historical data of thermal infrared images and high-definition images of the surface of the molten area and the powder spreading area when powder spreading defects and cladding defects occur during additive manufacturing of the same powder spreading material, and mark the locations of powder spreading defects and cladding defects;

[0079] Train the neural network model with labeled images to identify defect locations in thermal infrared images and high-definition images;

[0080] Acquire thermal infrared images and high-definition images collected in real time, and perform image enhancement on the thermal infrared images and high-definition images;

[0081] Input the acquired thermal infrared images and high-definition images into the trained neural network model to identify the defect location;

[0082] Count the number of grids covered by the defect position, perform grayscale processing on the HD image and thermal infrared image, count the grayscale value of each pixel, and calculate the process quality coefficient of each grid using the following formula:

[0083] ;

[0084] Among them, β is the process quality coefficient of the grid; Pa is the mean grayscale value of the pixel points in the grid of the powder laying HD image, Pa1 is the mean grayscale value of the pixel points in the grid of the powder laying HD image without defect coverage, fa is the ratio of the number of grids covered by the defect position in the powder laying HD image grid to the total number of grids; Pb is the mean grayscale value of the pixel points in the grid of the cladding HD image, Pb1 is the mean grayscale value of the pixel points in the grid of the cladding HD image without defect coverage, fb is the ratio of the number of grids covered by the defect position in the cladding HD image grid to the total number of grids. Pc is the mean gray value of the pixels in the grid of the powder laying thermal infrared image, Pc1 is the mean gray value of the pixels in the grid not covered by the defects in the powder laying thermal infrared image, fc is the ratio of the number of grids covered by the defective positions in the grid of the powder laying thermal infrared image to the total number of grids; Pd is the mean gray value of the pixels in the grid of the cladding thermal infrared image, Pd1 is the mean gray value of the pixels in the grid not covered by the defects in the cladding thermal infrared image, fd is the ratio of the number of grids covered by the defective positions in the grid of the cladding thermal infrared image.

[0085] In one embodiment of the present invention, the cladding degree of the grid is evaluated according to the process quality coefficient, the cladding thickness uniformity coefficient and the cladding layer thickness data of each grid, comprising the following steps:

[0086] Obtain the process quality coefficient β, cladding thickness uniformity coefficient α and cladding layer thickness data of each grid;

[0087] Calculate the difference C between the mean cladding thickness in the defect-free grid and the cladding thickness of the current grid;

[0088] The degree of mesh cladding is evaluated by the following formula:

[0089] ;

[0090] Among them, E is the evaluation value of the cladding degree of the grid, L 0 is the average thickness of the cladding layer in the defect-free grid.

[0091] In one embodiment of the present invention, corresponding grids whose cladding degree is less than a preset threshold are marked, and thickness difference of the marked grids is calculated according to cladding layer thickness data of the marked grids, including the following steps:

[0092] Obtain the cladding degree evaluation value of the grid, and mark the corresponding grid if the cladding degree evaluation value of the grid is less than the preset threshold;

[0093] According to the cladding layer thickness data of the marked grid, the difference between the mean cladding layer thickness in the defect-free grid and the cladding layer thickness of the current grid is taken as the thickness difference.

[0094] Specifically, in this embodiment, the cladding degree evaluation value E of the grid is obtained, and if the cladding degree evaluation value E of the grid is less than 0.8, the corresponding grid is marked.

[0095] In one embodiment of the present invention, according to the powder density and material data of the slice, a volume shrinkage function model based on material properties is established to adjust the powder thickness of the marked grid, including the following steps:

[0096] Collect the powder density and material data of the slices, obtain the historical data of the powder thickness and the layer thickness after cladding with the same powder density as the slices, and establish a volume shrinkage function model based on material properties;

[0097] According to the real-time collected data of powder thickness and cladding thickness, the established volume shrinkage function model based on material properties is parameterized.

[0098] Through the parameter-corrected volume shrinkage function model based on material properties, the thickness of the marked grid is adjusted according to the thickness difference of the marked grid when the powder is applied in the next slice.

[0099] In one embodiment of the present invention, collecting the powder density and material data of the slice, obtaining the historical data of the powder thickness and the layer thickness after cladding with the same powder density as the slice, and establishing a volume shrinkage function model based on material properties, including the following steps:

[0100] Collect the powder density and material data of the slices, obtain the historical data of the powder thickness and the layer thickness after cladding of the same powder material as the slices, and establish an initial volume shrinkage function model based on material properties;

[0101] ;

[0102] Among them, H initial H is the thickness of the powder coating, final is the layer thickness after cladding, k is the cladding shrinkage coefficient, and ρ is the powder density;

[0103] The powder density, material, layer thickness after cladding, and powder thickness are fitted to the equation of the initial volume shrinkage function model based on material properties using the least squares method to obtain the fitting parameter cladding shrinkage coefficient k as the volume shrinkage function model based on material properties.

[0104] Specifically, in this embodiment, the specific steps of using the least squares method to fit the above equation to find the fitting parameter cladding shrinkage coefficient k are as follows:

[0105] Collect the powder density and material data of the slices, and obtain the historical data of the powder thickness and the layer thickness after cladding of the same powder material as the slices;

[0106] For example, a data table of powder density, layer thickness after cladding and powder thickness is constructed. Table 1 is a data table of powder density, layer thickness after cladding and powder thickness. The constructed data table of powder density, layer thickness after cladding and powder thickness is shown in Table 1:

[0107] Table 1 is the data table of powder density, layer thickness after cladding and powder thickness

[0108] ;

[0109] According to the initial volume shrinkage function model based on material properties:

[0110] ;

[0111] The least squares method is used to fit the cladding shrinkage coefficient k parameter through the Python scipy.optimize module.

[0112] In one embodiment of the present invention, according to the real-time collected powder thickness and cladding layer thickness data, the established volume shrinkage function model based on material properties is parameter corrected, including the following steps:

[0113] The expected cladding layer thickness is calculated using the established volume shrinkage function model based on material properties. The actual measured powder thickness is compared with the predicted value of powder thickness, and the error is calculated:

[0114] ;

[0115] Where: Hactual is the actual measured powder thickness, Hpredicted is the predicted value of powder thickness calculated based on existing parameters;

[0116] Calculate the derivative of the loss function with respect to the parameter cladding shrinkage coefficient k, and update the parameter using the following formula:

[0117] ;

[0118] Where: k' is the updated parameter cladding shrinkage coefficient, is the derivative of the loss function with respect to the parameter cladding shrinkage coefficient k, L = is the loss function, and ε is the prediction accuracy of the powder thickness prediction value.

[0119] In one embodiment of the present invention, according to the thickness difference of the marked grid, when the next slice is powdered, the powdering thickness of the marked grid is adjusted, comprising the following steps:

[0120] According to the thickness difference of the marked grid, when the next slice is clad, the planned layer thickness after cladding is added to the thickness difference of the marked grid to obtain the actual layer thickness after cladding of the next slice;

[0121] The thickness of the powder coating corresponding to the next slice of the marked grid is calculated through a volume shrinkage function model based on material properties;

[0122] According to the thickness of the powder coating corresponding to the next slice of the marked grid, the powder coating thickness of the marked grid is adjusted.

[0123] In this embodiment, when the thickness of the powder corresponding to the next slice of the marked grid is greater than 100 μm, the powder spreading operation is performed according to 100 μm to avoid excessively thick powder spreading, which affects subsequent cladding and makes it difficult to effectively repair the uneven cladding problem before.

[0124] In an embodiment of another aspect of the present invention, a system is provided that uses an intelligent additive manufacturing online monitoring and real-time control method as described above to achieve additive manufacturing online monitoring and real-time control.

[0125] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. A method for online monitoring and real-time control of intelligent additive manufacturing, characterized in that: include: According to the 3D model of additive manufacturing, the 3D model is sliced, and all the slice coverage areas of the printed part are meshed; The laser displacement sensor is used to monitor the powder thickness and cladding thickness of each grid in real time during the printing of each slice. Monitor the temperature of the melting area and collect thermal infrared images and high-definition images of each grid slice in real time for the powder laying and cladding process; According to the powder thickness and cladding layer thickness data of each grid, the cladding thickness uniformity coefficient of the cladding layer is evaluated, and the process quality coefficient of each grid is calculated based on the collected thermal infrared images and high-definition images; According to the process quality coefficient, cladding thickness uniformity coefficient and cladding layer thickness data of each grid, the cladding degree of the grid is evaluated, the corresponding grids whose cladding degree is less than the preset threshold are marked, and the thickness difference of the marked grid is calculated according to the cladding layer thickness data of the marked grid; According to the powder density and material data of the slice, a volume shrinkage function model based on material properties is established to adjust the powder thickness of the marked grid; According to the powder thickness and cladding layer thickness data of each grid, the cladding thickness uniformity coefficient of the cladding layer is evaluated, including the following steps: According to the powder thickness and cladding layer thickness data of each grid, the average value and variance of the powder thickness and cladding layer thickness are calculated; According to the average value and variance of the powder thickness and the cladding layer thickness, the cladding thickness uniformity coefficient of the cladding layer is calculated by the following formula: Among them, α is the cladding thickness uniformity coefficient of the cladding layer, S1 is the variance of the powder thickness of different grids of the same slice, μ1 is the average value of the powder thickness of different grids of the same slice, S2 is the variance of the cladding layer thickness of different grids of the same slice, and μ2 is the average value of the cladding layer thickness of different grids of the same slice; According to the collected thermal infrared images and high-definition images, the process quality coefficient of each grid is calculated, including the following steps: Obtain historical data of thermal infrared images and high-definition images of the surface of the molten area and the powder spreading area when powder spreading defects and cladding defects occur during additive manufacturing of the same powder spreading material, and mark the locations of powder spreading defects and cladding defects; Train the neural network model with labeled images to identify defect locations in thermal infrared images and high-definition images; Acquire thermal infrared images and high-definition images collected in real time, and perform image enhancement on the thermal infrared images and high-definition images; Input the acquired thermal infrared images and high-definition images into the trained neural network model to identify the defect location; Count the number of grids covered by the defect position, perform grayscale processing on the HD image and thermal infrared image, count the grayscale value of each pixel, and calculate the process quality coefficient of each grid using the following formula: Among them, β is the process quality coefficient of the grid; Pa is the mean grayscale value of the pixel points in the grid of the powder laying HD image, Pa1 is the mean grayscale value of the pixel points in the grid of the powder laying HD image without defect coverage, fa is the ratio of the number of grids covered by the defect position in the powder laying HD image grid to the total number of grids; Pb is the mean grayscale value of the pixel points in the grid of the cladding HD image, Pb1 is the mean grayscale value of the pixel points in the grid of the cladding HD image without defect coverage, fb is the ratio of the number of grids covered by the defect position in the cladding HD image grid to the total number of grids. Pc is the mean value of the grayscale value of the pixels in the grid of the powder laying thermal infrared image, Pc1 is the mean value of the grayscale value of the pixels in the grid of the powder laying thermal infrared image that is not covered by defects, fc is the ratio of the number of grids covered by defective positions in the grid of the powder laying thermal infrared image to the total number of grids; Pd is the mean value of the grayscale value of the pixels in the grid of the cladding thermal infrared image, Pd1 is the mean value of the grayscale value of the pixels in the grid of the cladding thermal infrared image that is not covered by defects, and fd is the ratio of the number of grids covered by defective positions in the grid of the cladding thermal infrared image to the total number of grids; According to the process quality coefficient, cladding thickness uniformity coefficient and cladding layer thickness data of each grid, the cladding degree of the grid is evaluated, including the following steps: Obtain the process quality coefficient β, cladding thickness uniformity coefficient α and cladding layer thickness data of each grid; Calculate the difference C between the mean cladding thickness in the defect-free grid and the cladding thickness of the current grid; The degree of mesh cladding is evaluated by the following formula: Among them, E is the evaluation value of the cladding degree of the grid, and L0 is the average thickness of the cladding layer in the defect-free grid; The corresponding grids whose cladding degree is less than a preset threshold are marked, and the thickness difference of the marked grids is calculated according to the cladding layer thickness data of the marked grids, including the following steps: Obtain the cladding degree evaluation value of the grid, and mark the corresponding grid if the cladding degree evaluation value of the grid is less than the preset threshold; According to the cladding layer thickness data of the marked grid, the difference between the average cladding layer thickness in the defect-free grid and the cladding layer thickness of the current grid is taken as the thickness difference; According to the powder density and material data of the slice, a volume shrinkage function model based on material properties is established to adjust the powder thickness of the marked grid, including the following steps: Collect the powder density and material data of the slices, obtain the historical data of the powder thickness and the layer thickness after cladding with the same powder density as the slices, and establish a volume shrinkage function model based on material properties; According to the real-time collected data of powder thickness and cladding thickness, the established volume shrinkage function model based on material properties is parameterized. Through the parameter-corrected volume shrinkage function model based on material properties, according to the thickness difference of the marked grid, when the next slice is powdered, the powdering thickness of the marked grid is adjusted; Collect the powder density and material data of the slice, obtain the historical data of the powder thickness and the layer thickness after cladding with the same powder density as the slice, and establish a volume shrinkage function model based on material properties, including the following steps: Collect the powder density and material data of the slices, obtain the historical data of the powder thickness and the layer thickness after cladding of the same powder material as the slices, and establish an initial volume shrinkage function model based on material properties; Among them, Hinitial is the thickness of the powder, Hfinal is the layer thickness after cladding, k is the cladding shrinkage coefficient, and ρ is the powder density; The powder density, material, layer thickness after cladding, and powder thickness are fitted to the equation of the initial volume shrinkage function model based on material properties using the least squares method to obtain the fitting parameter cladding shrinkage coefficient k as the volume shrinkage function model based on material properties.

2. The method for online monitoring and real-time control of intelligent additive manufacturing according to claim 1, characterized in that: According to the real-time collected data of powder thickness and cladding layer thickness, the established volume shrinkage function model based on material properties is parameter corrected, including the following steps: The expected cladding layer thickness is calculated using the established volume shrinkage function model based on material properties. The actual measured powder thickness is compared with the predicted value of powder thickness, and the error is calculated: ； Where: Hactual is the actual measured powder thickness, Hpredicted is the predicted value of powder thickness calculated based on existing parameters; Calculate the derivative of the loss function with respect to the parameter cladding shrinkage coefficient k, and update the parameter using the following formula: Where: k' is the updated parameter cladding shrinkage coefficient, is the derivative of the loss function with respect to the parameter cladding shrinkage coefficient k, L = is the loss function, and ε is the prediction accuracy of the powder thickness prediction value.

3. The method for online monitoring and real-time control of intelligent additive manufacturing according to claim 1, characterized in that: According to the thickness difference of the marked grid, when the next slice is powdered, the powdering thickness of the marked grid is adjusted, including the following steps: According to the thickness difference of the marked grid, when the next slice is clad, the planned layer thickness after cladding is added to the thickness difference of the marked grid to obtain the actual layer thickness after cladding of the next slice; The thickness of the powder coating corresponding to the next slice of the marked grid is calculated through a volume shrinkage function model based on material properties; According to the thickness of the powder coating corresponding to the next slice of the marked grid, the powder coating thickness of the marked grid is adjusted.

4. An intelligent additive manufacturing online monitoring and real-time control system, characterized in that: The system adopts an intelligent additive manufacturing online monitoring and real-time control method as described in any one of claims 1 to 3 to realize additive manufacturing online monitoring and real-time control.

Images