Method and system for predicting quality of molten forming part of laser powder bed

Abstract

The invention relates to the technical field of additive manufacturing quality control and process optimization, and particularly provides a method and system for predicting the quality of a molten forming part of a laser powder bed, and the method comprises the steps: obtaining the three-dimensional point cloud data of the molten forming part of the laser powder bed; dividing the aligned three-dimensional point cloud data into a plurality of cross section point sets along the central axis direction; performing independent local geometric fitting on each cross section point set to obtain local fitting parameters of each cross section; based on the three-dimensional point cloud data and the local fitting parameters of all the cross sections, the three-dimensional original surface of the formed part is unfolded into a two-dimensional surface deviation topological graph; and calculating the surface roughness based on the two-dimensional surface deviation topological graph, and carrying out statistical evaluation on the process-induced macroscopic geometric deviation of the formed part. According to the method, through a closed loop from process prediction to result evaluation, powerful data support is provided for accurate decoupling and integrated representation of two types of key quality characteristics of the LPBF forming part, and a new way is opened up for digital pre-optimization of the process.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of additive manufacturing quality control and process optimization, and particularly relates to a laser powder bed fusion forming part quality prediction method. BACKGROUND

[0002] As an advanced metal additive manufacturing technology, laser powder bed fusion (LPBF) has been widely used in the fields of aerospace, medical devices, etc. However, its layer-by-layer fusion characteristics also introduce complex quality problems, mainly manifested as the coupling of macro-geometric distortion and micro-surface topography. Macro-distortion is caused by residual stress due to large thermal gradient, leading to deformation such as bending and warping; micro-topography is affected by factors such as partial melting powder adhesion and molten pool dynamic behavior, and is usually manifested as high surface roughness. The coupling of multi-scale errors makes the traditional evaluation method based on global ideal model fitting invalid, and cannot independently and accurately quantify the geometric accuracy and surface quality. At the same time, the process optimization method relying on physical experiments is costly and time-consuming. SUMMARY

[0003] In view of the above problems, the present application provides a laser powder bed fusion forming part quality prediction method, which can accurately predict the geometric and surface quality of the LPBF forming part by segmenting and locally fitting the three-dimensional point cloud data of the aligned LPBF forming part, and then expanding the three-dimensional original surface of the forming part into a two-dimensional surface deviation topology map based on the three-dimensional point cloud data and the local fitting parameters of each cross section for integrated quality evaluation.

[0004] To achieve the purpose of the present application, the present application provides the following technical solutions: In a first aspect, the present application provides a laser powder bed fusion forming part quality prediction method, comprising: obtaining three-dimensional point cloud data of a laser powder bed fusion forming part; aligning the three-dimensional point cloud data along the center axis, and segmenting the aligned three-dimensional point cloud data into a plurality of cross-sectional point sets along the center axis direction; independently locally fitting each cross-sectional point set to obtain local fitting parameters of each cross section; expanding the three-dimensional original surface of the forming part into a two-dimensional surface deviation topology map based on the three-dimensional point cloud data and the local fitting parameters of each cross section; calculating the surface roughness of the forming part based on the two-dimensional surface deviation topology map, and statistically evaluating the process-induced macro-geometric deviation of the forming part based on the local fitting parameters of each cross section.

[0005] The further improvement of the present application is that the center axis alignment processing of the three-dimensional point cloud data comprises: determining a first principal component vector by using a principal component analysis algorithm, and determining the first principal component vector as an actual center main axis of the formed piece; and obtaining the aligned three-dimensional point cloud data by coinciding the actual center main axis with a preset axis of a coordinate system through calculation of a rotation matrix.

[0006] The further improvement of the present application is that the three-dimensional point cloud data after alignment is segmented into a plurality of cross-sectional point sets along the center axis direction, comprising: defining a series of equidistant virtual slice planes along the center axis direction; defining a thickness tolerance zone around each virtual slice plane, extracting three-dimensional points falling within the thickness tolerance zone for each virtual slice plane, and taking the coordinates perpendicular to the center axis to form the cross-sectional point set.

[0007] The further improvement of the present application is that the independent local geometric fitting is performed on each cross-sectional point set to obtain local fitting parameters of each cross section, comprising: for each cross-sectional point set, a group of inliers conforming to the main body contour is identified from the cross-sectional point set by using a random sample consensus algorithm to form an inlier set; and the inlier set is optimized by using a Levenberg-Marquardt algorithm to find optimal circle parameters to determine the local fitting parameters; wherein for each inlier in the inlier set, the residual is defined as the difference between the actual radial distance of the point to the fitted circle and the fitted radius.

[0008] The further improvement of the present application is that the local fitting parameters of each cross section form a local fitting parameter sequence; and the three-dimensional original surface of the formed piece is unfolded into a two-dimensional surface deviation topological map based on the three-dimensional point cloud data and the local fitting parameters of each cross section, comprising: for any point in the three-dimensional point cloud data, a dynamic local fitting reference of the axial position of the point is established by interpolating the discrete local fitting parameter sequence; the radial deviation value of any point in the three-dimensional point cloud data relative to the dynamic local fitting reference of the point and the circumferential angle coordinate of the point in the unfolded coordinate system are calculated; the circumferential angle coordinate, axial position coordinate and radial deviation value of all points in the three-dimensional point cloud data are mapped onto the created regular two-dimensional grid based on natural neighborhood interpolation to generate the two-dimensional surface deviation topological map.

[0009] The further improvement of the present application is that the surface roughness of the formed piece is calculated based on the two-dimensional surface deviation topological map, comprising: applying a two-dimensional Gaussian filter to the two-dimensional surface deviation topological map to separate and obtain a filtered roughness matrix; the standard deviation of the two-dimensional Gaussian filter is calculated from the cutoff wavelength set by the user and the pixel density of the grid, and the relationship is as follows: ; wherein, a standard deviation of the two-dimensional Gaussian filter; a cutoff wavelength; a pixel density; calculating a surface roughness based on the roughness matrix; the surface roughness including a face roughness parameter and an average axial profile roughness distributed along a circumference.

[0010] Further improvements of the present application are that the local fitting parameters of each cross section include a local fitting radius sequence and a local fitting center coordinate sequence of all cross sections; and the statistical evaluation of the process-induced macro-geometric deviations of the shaped part based on the local fitting parameters of each cross section includes: statistical analysis of the local fitting radius sequence to evaluate the dimensional deviations of the shaped part; and statistical analysis of the local fitting center coordinate sequence to evaluate the waviness or bending deformation of the shaped part caused by thermal stress; wherein the statistical analysis of the local fitting center coordinate sequence includes: calculating the distance from each circle center to the actual center main axis to form a distance sequence, and then quantitatively characterizing the overall bending deformation caused by thermal stress based on the maximum value or the root mean square value of the distance sequence.

[0011] Further improvements of the present application are that the shaped part is a predicted shaped part simulated by a multi-physical field numerical model of a laser powder bed fusion process; and the method further includes: establishing a multi-physical field numerical model of a laser powder bed fusion process, simulating a forming process based on the forming process parameters of the shaped part to be formed in the multi-physical field numerical model to obtain a predictive three-dimensional surface data set of the shaped part; and converting the predictive three-dimensional surface data set into a three-dimensional point cloud data.

[0012] Further improvements of the present application are that the laser powder bed fusion shaped part is an actual shaped part manufactured by physical manufacturing, and for the actual shaped part, a scanning technique is used to obtain an actual three-dimensional surface data set of the surface topography of the actual shaped part; and the method further includes: comparing a first evaluation result based on the predictive three-dimensional surface data set and a second evaluation result based on the actual three-dimensional surface data set to obtain a third evaluation result; and based on the third evaluation result, verifying and calibrating the multi-physical field numerical model, and / or feeding back and optimizing the laser powder bed fusion process parameters.

[0013] In a second aspect, the present application provides a laser powder bed fusion shaped part quality prediction system for implementing the laser powder bed fusion shaped part quality prediction method described above, comprising: an acquisition module configured to acquire three-dimensional point cloud data of a laser powder bed fusion shaped part; a processing module configured to perform center axis alignment processing on the three-dimensional point cloud data, and divide the aligned three-dimensional point cloud data into a plurality of cross-sectional point sets along the center axis direction; a fitting module, configured to perform independent local geometric fitting on each of the cross-sectional point sets to obtain local fitting parameters of each cross section; a developing module, configured to develop a three-dimensional original surface of the formed part into a two-dimensional surface deviation topology map based on the three-dimensional point cloud data and the local fitting parameters of each cross section; an evaluating module, configured to calculate surface roughness of the formed part based on the two-dimensional surface deviation topology map, and statistically evaluate process-induced macro-geometric deviation of the formed part based on the local fitting parameters of each cross section.

[0014] Compared with the prior art, the present application has the following beneficial effects: Through the laser powder bed fusion formed part quality prediction method provided in the present application, the three-dimensional point cloud data of the formed part is segmented into a plurality of cross-sectional point sets along the central axis, and independent local geometric fitting is performed on each cross section. Then, based on a series of discrete local fitting parameters, a dynamic local fitting reference varying along the axis is established by interpolation, the entire three-dimensional original surface is developed into a two-dimensional surface deviation topology map, and comprehensive evaluation of the formed part quality is realized. Through the technical path of "global alignment, local fitting, and overall development", the present application separates the complex coupling error, and finally realizes the synchronous, independent, and precise quantitative characterization of the two key quality characteristics of the laser powder bed fusion formed part. The present application provides strong data support for the precise decoupling and integrated characterization of the two key quality characteristics of the LPBF formed part, and opens up a new way for the digital front-end optimization of the process. BRIEF DESCRIPTION OF DRAWINGS

[0015] The accompanying drawings are included to provide a further understanding of the present application, and form a part of the specification, illustrate embodiments of the present application, and are used to explain the present application, and do not form a limitation on the present application; Figure 1 An optional flowchart of a laser powder bed fusion formed part quality prediction method provided by an embodiment of the present application; Figure 2 A structural schematic diagram of a multi-physical field simulation formed columnar part three-dimensional model provided by an embodiment of the present application; Figure 3 A deviation visualization schematic diagram of a prediction cylindrical part relative to a nominal cylindrical part provided by an embodiment of the present application; Figure 4 An optional flowchart of a laser powder bed fusion formed part quality prediction method provided by an embodiment of the present application; Figure 5 An optional flowchart of a laser powder bed fusion formed part quality prediction method provided by an embodiment of the present application; Figure 6An optional flowchart of a laser powder bed fusion formed part quality prediction method provided for an embodiment of the present application is shown in FIG. 1. Figure 7 A schematic diagram of a generated two-dimensional surface deviation topological map provided for an embodiment of the present application is shown in FIG. 2. Figure 8 A graph of a calculated average axial profile roughness Ra curve distributed along the circumference provided for an embodiment of the present application is shown in FIG. 3. Figure 9 A radius deviation distribution histogram of a macroscopic geometric deviation statistical analysis provided for an embodiment of the present application is shown in FIG. 4. Figure 10 A center offset distribution histogram of a macroscopic geometric deviation statistical analysis provided for an embodiment of the present application is shown in FIG. 5. DETAILED DESCRIPTION

[0016] To make the objectives, technical solutions and advantages of the present application clearer, the technical solutions of the present application will be described clearly and completely below with reference to the drawings in the present application. Obviously, the described embodiments are some embodiments of the present application, rather than all embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative labor fall within the scope of protection of the present application.

[0017] The terms "first", "second" are only for descriptive purposes, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of indicated technical features. Therefore, the features defined with "first", "second" can explicitly or implicitly include one or more of the features; in the description of the present application, unless otherwise specified, the meaning of "a plurality of" is two or more.

[0018] As one of the core technologies in the field of metal additive manufacturing, LPBF technology has been widely used in high-end fields such as aerospace and medical devices that require strict performance and precision of components, due to its unique advantages in forming complex and precise components. This technology realizes three-dimensional entity manufacturing through layer-by-layer laying of metal powder and selective laser melting and solidification, which can break through the process limitations of traditional subtractive manufacturing, efficiently form key components with lightweight topology structure and function-integrated characteristics, and provide important technical support for structural innovation and performance upgrade of high-end equipment.

[0019] In the layer-by-layer melting process, the local high-energy heating of the laser beam on the powder layer causes the temperature of the molten pool area to rise sharply, and the rapid heat dissipation of the surrounding non-melted area and the substrate forms a significant temperature difference. The thermal stress caused by this temperature difference accumulates continuously inside the component. After the forming is completed and cooled, the release of residual thermal stress will cause the component to appear macroscopic deformations such as bending, warping, and size contraction. For large-size, thin-walled components, such deformations are more pronounced, directly affecting the precision stability and application reliability of the component. At the same time, not all metal powders are completely melted under the action of the laser. Some powders with smaller particle sizes or far from the laser focus area only undergo partial melting, which easily adheres to the surface of the formed part after cooling to form a "powder sticking" defect, destroying the surface flatness. At the same time, the molten pool has complex behaviors such as dynamic flow, splashing, solidification shrinkage, etc. during laser scanning. The irregular solidification of the molten pool boundary forms uneven melt channel textures on the surface of the formed part, and the unevenness of the interlayer fusion interface further exacerbates the surface roughness. For components such as medical implants and precision transmission components that have very high requirements for surface quality, excessive surface roughness not only affects the appearance, but also may cause subsequent problems such as reduced biocompatibility and increased stress concentration.

[0020] Currently, traditional quality evaluation methods are mostly based on global ideal model fitting comparison, that is, the actual three-dimensional data of the formed part is matched with the designed ideal model as a whole. This method cannot effectively distinguish the independent influence of macroscopic geometric distortion and microscopic surface defects, and is easy to misjudge the local deviation caused by surface roughness as macroscopic geometric error, or to cover up the real surface quality problem due to macroscopic deformation, leading to distorted quantitative evaluation results of geometric precision and surface quality, and making it difficult to guide the targeted adjustment of process parameters. In the process optimization aspect, the existing method of designing multiple sets of different laser power, scanning speed, powder layer thickness, etc. process parameter combinations, preparing a large number of physical samples and conducting quality detection, and then screening the optimal parameters is costly and time-consuming, and is difficult to quickly respond to the process optimization needs of different materials and different structural components.

[0021] To solve the above technical problems, the present application proposes the following technical solutions and corresponding embodiments.

[0022] The technical solutions of the present application will be described below with reference to the embodiments shown in the accompanying drawings: Figures 1 to 10 Embodiment 1 A laser powder bed fusion formed part quality prediction method according to an embodiment of the present application is used for comprehensive quality control of metal formed parts manufactured by LPBF technology, as shown in Figure 1 The laser powder bed fusion formed part quality prediction method according to an embodiment of the present application includes the following steps S101 to S105: Step S101, acquiring three-dimensional point cloud data of a formed part fused by a laser powder bed. ​

[0023] In the embodiments of the present application, the surface topography of the laser powder bed fusion formed part is reconstructed into three-dimensional point cloud data , wherein . Here, the laser powder bed fusion formed part is a predicted formed part of a simulation forming or an actual formed part by physical manufacturing; for the predicted formed part of a simulation forming, the simulation forming process is simulated by a multi-physics field numerical model to obtain a predictive three-dimensional surface data set of the predicted formed part; for the actual formed part, a scanning technique is used to obtain an actual three-dimensional surface data set of the surface topography of the actual formed part.

[0024] In the embodiments of the present application, a multi-physics field numerical model of the laser powder bed fusion process is established, based on the forming process parameters of the part to be formed, the simulation forming process is simulated by the multi-physics field numerical model to obtain a predictive three-dimensional surface data set of the formed part, and the predictive three-dimensional surface data set is converted into three-dimensional point cloud data. In the embodiments, the laser powder bed fusion process is simulated by the multi-physics field numerical model to obtain a predictive three-dimensional surface data set of the formed part, and the predictive three-dimensional surface data set is converted into three-dimensional point cloud data. Figure 2 、 Figure 3 As shown in FIGS. 1 and 2, the object to be measured is a cylindrical sample prepared by using IN718 high-temperature alloy powder, with a nominal radius of 0.25 mm and a nominal height of 4 mm.

[0025] In the embodiments of the present application, the predictive three-dimensional surface data set refers to the surface topography and defect feature distribution information of the whole formed part obtained by simulating the simulation forming process by the multi-physics field numerical model. In the embodiments of the present application, the numerical model used is a melt pool scale multi-physics field model coupled based on the lattice Boltzmann method and the height function method, combined with the slice CLI (Common Layer Interface) file (laser scanning path) of the part to be formed model and the adaptive mesh refinement technology, and then the scale of the multi-physics field numerical model is extended to the component scale, to realize the simulation of the state of each laser scanning powder bed process in the forming process under any process parameter condition, including the melting and spattering of the powder, the dynamic solidification behavior of the melt pool and the temperature accumulation effect, etc. In the simulation process, the system records the surface morphology data after each layer deposition layer by layer, and finally integrates all the layer data after the simulation ends to generate a complete predictive three-dimensional surface data set. Based on the data set, three-dimensional point cloud data can be directly or indirectly generated; the indirect method includes: first converting the predictive three-dimensional surface data set into a visual model in STL (StereoLithography) format, and then further extracting the corresponding three-dimensional point cloud data from the STL file.

[0026] In the embodiments of the present application, the obtained three-dimensional point cloud data is cleaned to remove invalid points containing non-numerical NaN (Not a Number) values to obtain cleaned valid point cloud data .

[0027] Step S102, center axis alignment processing is performed on the three-dimensional point cloud data, and the aligned three-dimensional point cloud data is segmented into a plurality of cross-section point sets along the center axis direction.

[0028] In the embodiments of the present application, referring to Figure 4 , the content of the center axis alignment processing of the three-dimensional point cloud data by using the following steps S1021 to S1022 is described, including: Step S10211, a first principal component vector is determined by using a principal component analysis algorithm, and the first principal component vector is determined as the actual center main axis of the shaped part.

[0029] In the embodiments of the present application, the effective point cloud data after cleaning The principal component analysis (PCA) is applied to calculate the covariance matrix and solve the eigenvalue and eigenvector, and the eigenvector corresponding to the maximum eigenvalue is determined as the actual center main axis of the deformed sample. .

[0030] Step S10212, the actual center main axis is coincided with the preset axis of the coordinate system by calculating a rotation matrix, and the aligned three-dimensional point cloud data is obtained.

[0031] In the embodiments of the present application, in order to ensure that the direction of the actual center main axis (the main axis direction) is consistent with the original design forming direction (usually the +Z direction) of the shaped part, the correlation coefficient of the projection coordinates in the main axis direction and the original Z coordinates is calculated . If the correlation coefficient is positive, it indicates that the main axis direction is consistent with the original design forming direction, and no adjustment is needed; if the correlation coefficient is negative, the main axis vector is reversed, that is .

[0032] In the embodiments, the rotation matrix required for rotating the corrected main axis to coincide with the Z axis of the coordinate system is calculated , and the rotation matrix can be obtained based on the Rodrigues' rotation formula. Thus, the rotation transformation is performed on all point cloud coordinates to obtain the rotation-aligned point cloud : , wherein, is the rotation matrix, and the transpose matrix of the point cloud coordinates; for the rotated point cloud, the Z coordinates are translated as a whole, so that the minimum value of the Z coordinates is 0, that is, the lowest point Z coordinate is 0, and the Z coordinates of the remaining points are recalculated according to the relative height, and the final attitude correction is completed: .

[0033] In this way, the three-dimensional point cloud data of the formed part (including the point cloud data that may have been affected by the attitude deviation due to macroscopic deformation) is unified into a standard coordinate system with the actual central axis as the Z-axis, ensuring that subsequent geometric analysis is based on a consistent spatial standard.

[0034] In the embodiments of this application, reference is made to Figure 5 As shown, the process of segmenting the aligned 3D point cloud data into multiple cross-sectional point sets along the central axis using steps S10221 to SS10222 is explained, including: Step S10221: Define a series of equally spaced virtual slice planes along the central axis direction.

[0035] In this embodiment of the application, along the Z-axis direction, in Define within range There are three equally spaced virtual slice planes, each with a Z-coordinate of [missing value]. For example, It is 1500.

[0036] Step S10222: Define a thickness tolerance zone around each of the virtual slice planes. For each virtual slice plane, extract three-dimensional points that fall within the thickness tolerance zone and take the coordinates perpendicular to the central axis to form the cross-sectional point set.

[0037] In the embodiments of this application, around each A plane is defined as a thickness tolerance. Extract all that satisfy the condition. The three-dimensional point, and take its Coordinates, forming the first A set of two-dimensional cross-sectional points For example, thickness tolerance It can be 3 times the slice spacing.

[0038] Step S103: Perform independent local geometric fitting for each set of cross-sectional points to obtain the local fitting parameters for each cross-section.

[0039] In this embodiment, a two-stage robust fitting strategy is adopted: first, outliers are removed using the RANSAC (Random Sample Consensus) algorithm, and then the Levenberg-Marquardt nonlinear optimization algorithm is used to accurately fit the interior points.

[0040] Specifically, for each cross-section point set, a set of inliers conforming to the subject contour is identified from the cross-section point set by using a random sample consensus algorithm to form an inlier set; and the inlier set is optimized by using a Levenberg-Marquardt algorithm to perform a nonlinear least squares method, so as to find optimal circle parameters and determine local fitting parameters.

[0041] In the embodiments of the present application, for each two-dimensional cross-section point set , the following operations are iteratively performed by using a random sample consensus algorithm: three points are randomly sampled to fit a candidate circle; the radial distances of all points in the point set to the candidate circle are calculated; a threshold of inliers is dynamically set based on the distribution of the distances (such as median absolute deviation, MAD), and then the points with distances less than the threshold are marked as inliers; and finally, the inlier set corresponding to the model that can obtain the maximum number of inliers is selected .

[0042] For the inlier set , a Levenberg-Marquardt nonlinear least squares method is applied to iteratively optimize, and the goal is to find optimal circle parameters to minimize the sum of squared residuals . For each point in the inlier set, the residual of the point is defined as the difference between the actual radial distance of the point to the fitted circle and the fitted radius: .

[0043] The above steps are repeated to obtain a sequence of local fitting parameters and a sequence of local fitting center coordinates for all valid cross sections.

[0044] In step S104, a three-dimensional original surface of the formed piece is unfolded into a two-dimensional surface deviation topological map based on the three-dimensional point cloud data and the local fitting parameters of the cross sections.

[0045] In the embodiments of the present application, as shown in Figure 6 , the content of step S104 can be implemented by the content of steps S1041 to S1043 as follows: In step S1041, for any point in the three-dimensional point cloud data, a dynamic local fitting reference at the axial position of the point is established by interpolating the discrete local fitting parameter sequence.

[0046] In the embodiments of the present application, for any point in the original aligned point cloud , the dynamic local fitting reference at the axial position of the point is established by interpolating the parameter sequence and at the axial position of the point. One-dimensional linear interpolation is performed at the coordinates to calculate the dynamic local fitting reference at the height of that point, i.e., the center of the interpolation circle. and interpolation radius .

[0047] Step S1042: Calculate the radial deviation value of any point in the three-dimensional point cloud data relative to the dynamic local fitting reference of that point and the circumferential angular coordinate of that point in the unfolded coordinate system.

[0048] In this embodiment, the radial deviation value dev of any point relative to its dynamic local fitting datum is calculated, and its circumferential angular coordinate theta in the unfolded coordinate system is also calculated. That is, based on the dynamic datum, the point is calculated... radial deviation and circumferential angle : , ;in, For point x-coordinate For point The ordinate, The interpolation radius; The x-coordinate of the interpolation circle center. The ordinate of the interpolation circle's center; yes A function used to calculate the square root of a non-negative number.

[0049] Step S1043: Based on natural neighborhood interpolation, map the circumferential angular coordinates, axial position coordinates, and radial deviation values ​​of all points in the three-dimensional point cloud data onto the created regular two-dimensional grid to generate the two-dimensional surface deviation topology map.

[0050] In the embodiments of this application, reference is made to Figure 7 As shown, the circumferential angular coordinates, axial position coordinates, and radial deviation values ​​of all points are interpolated to generate the two-dimensional surface deviation topology map; that is, the unfolded coordinates and deviation values ​​of all points are interpolated to generate the two-dimensional topology map.

[0051] In this embodiment of the application, a regular two-dimensional grid is created, whose coordinate axes are the circumferential angles. and axial position Apply the scattered interpolation method to all points. Scattered data is mapped onto this regular grid to generate a two-dimensional surface deviation topology map matrix. In this embodiment, the scatter interpolation method is natural neighborhood interpolation.

[0052] Step S105, calculating the surface roughness of the shaped part based on the two-dimensional surface deviation topological map, and statistically evaluating the process-induced macro-geometric deviation of the shaped part based on the local fitting parameters of the cross sections.

[0053] In the embodiment of the present application, the integrated quality of the laser powder bed fusion shaped part is evaluated. After applying Gaussian filtering on the topological map, the surface roughness parameters such as Sa and Ra are calculated, and the macro-geometric deviation is statistically evaluated based on the local fitting parameter sequence.

[0054] For surface roughness evaluation, the following steps S10511 to S10513 are used to explain: Step S10511, the two-dimensional surface deviation topological map matrix A two-dimensional Gaussian filter is applied to separate the surface roughness, and the standard deviation of the filter is According to the user's setting of the cut-off wavelength And the pixel density of the grid (i.e. the number of pixels per millimeter) The formula is calculated as: Thus, the filtered roughness matrix is obtained As a feasible implementation, .

[0055] Step S10512, on the roughness matrix The surface roughness parameters (Arithmetic mean height) and (Root mean square height) are calculated: , .

[0056] In this embodiment, the surface roughness parameters Sa are calculated according to the ISO 25178 standard on the filtered data.

[0057] Step S10513, for each circumferential angle (i.e. each row of the matrix) in the topological map, which represents a profile line along the Z axis; the roughness of the profile line is calculated according to the ISO 25178 standard, i.e. , and then the values of of all angles are plotted into a curve, as shown in Figure 8 , which can be used to analyze the surface quality in different directions. For each strut of different forming direction samples, at least 2000 cross sections are randomly selected for convergence analysis to verify the sufficiency and representativeness of the number of cross sections in capturing the strut geometric characteristics and surface morphology distribution.

[0058] In the embodiment of the present application, refer to Figure 9 , Figure 10As shown, for the evaluation of macroscopic geometric deviation, 1) analyze the radius sequence. The difference between its mean and nominal radius reflects the overall dimensional deviation, and its variation with... The changing trend can be used to quantify taper or bulge; 2) Analyze the center coordinate sequence Calculate the distance from the center of each circle to the Z-axis. The maximum or root mean square value of this distance sequence can be used to quantitatively characterize the overall bending deformation caused by thermal stress.

[0059] In the embodiments of this application, the multiphysics numerical model can be verified and calibrated by comparing the first evaluation result based on the predictive three-dimensional surface dataset and the second evaluation result based on the actual three-dimensional surface dataset, and / or the laser powder bed melting process parameters can be optimized by feedback.

[0060] The laser powder bed fusion forming part quality prediction method provided in this embodiment achieves precise decoupling and integrated evaluation of macroscopic geometric deviations and microscopic surface quality of the formed part through a series of feature combinations. First, by acquiring the three-dimensional point cloud data of the formed part and using principal component analysis (PCA) to determine and align its central axis, a unified and standardized coordinate reference is established for subsequent analysis, eliminating errors introduced by workpiece posture deviation during measurement. Next, the method divides the point cloud into multiple cross-sectional point sets along the central axis and performs independent local geometric fitting on each cross-section. This fitting uses a random sampling consensus algorithm to identify and remove outliers caused by defects such as "powder adhesion," and then uses the Levenberg-Marquardt algorithm to perform nonlinear least squares optimization on the internal points conforming to the main body contour, thereby accurately obtaining the dynamic fitting parameters for each cross-section. Based on this series of discrete local fitting parameters, a dynamic local fitting reference varying along the axis is established through interpolation. The entire three-dimensional original surface is then unfolded into a two-dimensional surface deviation topology map. The coordinate axes of this topology map are the axial position and circumferential angle of the part, and the values ​​are the radial deviation of each point relative to the dynamic reference. Finally, a comprehensive quality evaluation is achieved based on the topology graph: on the one hand, high-frequency components are separated by applying two-dimensional Gaussian filtering and other methods, thereby independently and accurately calculating surface roughness; on the other hand, by performing statistical analysis on the local fitting parameter sequence, macroscopic geometric deviations caused by process factors, such as dimensional deviations, waviness, or overall bending deformation, can be quantitatively evaluated.

[0061] In summary, the method separates the complex coupling errors through the technical path of "global alignment, local fitting, and overall development", and finally realizes the synchronous, independent and accurate quantitative characterization of the two key quality characteristics of the laser powder bed fusion formed part. Through the closed loop from process prediction to result evaluation, problems in the part manufacturing process can be predicted, potential manufacturing risks can be solved, the iteration optimization of the design and manufacturing of actual parts can be accelerated, effective protection can be provided for actual formed high-quality and performance parts, and the cost and unnecessary energy consumption of experiments can be greatly reduced.

[0062] The embodiments of the present application also provide a computer readable storage medium, and the computer readable storage medium stores a computer program. The computer program is executed by a processor to implement the method in any of the embodiments of the present application. Specifically, a system or device provided with a storage medium can be provided. The storage medium stores software program codes for implementing the functions of any of the above embodiments, and the computer of the system or device reads and executes the program codes stored in the storage medium.

[0063] In particular, according to the embodiments of the present application, the processes described above with reference to the flowcharts can be implemented as a computer software program. For example, the embodiments of the present application include a computer program product comprising a computer program carried on a computer readable storage medium, the computer program comprising program codes for executing the method shown in the flowchart. In such embodiments, the computer program can be downloaded and installed from a network by a communication part, and / or installed from a detachable medium. When the computer program is executed by a central processing unit, the above-mentioned functions defined in the system of the present application are executed.

[0064] It should be noted that the computer-readable storage medium shown in the present application can be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may, for example, be an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or component, or any combination of the above. More specific examples of computer-readable storage media can include, but are not limited to, electrical connections with one or more conductive wires, portable computer disks, hard disks, random access memories, read-only memories, erasable programmable read-only memories (flash memories), optical fibers, portable compact disk read-only memories, optical storage devices, magnetic storage devices, or any suitable combination of the above. In the present application, the computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, device or component. In the present application, the computer-readable signal medium can include a data signal carrying computer-readable program code in a baseband or as part of a carrier wave. Such a propagated data signal can take various forms, including but not limited to electromagnetic signals, optical signals or any suitable combination of the above. The computer-readable signal medium can also be any computer-readable storage medium other than the computer-readable storage medium, which can send, propagate or transmit programs for use by or in conjunction with an instruction execution system, device or component. The program code contained on the computer-readable storage medium can be transmitted by any suitable medium, including but not limited to wireless, wire, optical cable, RF, etc., or any suitable combination of the above.

[0065] The flowcharts and block diagrams in the drawings illustrate the possible implementation architectures, functions and operations of the systems, methods and computer program products according to various embodiments of the present application. In this regard, each block in the flowcharts or block diagrams can represent a module, a program segment or a part of code containing one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions noted in the blocks can occur in different order than that shown in the drawings. For example, two blocks that are shown in succession can actually be executed substantially in parallel, and they can also be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams or flowcharts, and the combination of blocks in the block diagrams or flowcharts, can be implemented by a dedicated hardware-based system that performs the specified functions or operations, or can be implemented by a combination of special-purpose hardware and computer instructions.

[0066] The units described in the embodiments of the present application can be implemented by software, or by hardware, or by a combination of software and hardware. The units described can be located in one place, or distributed over several places. The names of the units in some cases do not limit the units themselves.

[0067] It should be noted that, although several modules or units of the device for action execution are mentioned in the foregoing detailed description, such a division is not mandatory. Indeed, according to embodiments of the application, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided into several modules or units.

[0068] In several embodiments provided in the present application, it should be understood that the disclosed system, modules and methods can be implemented in other ways. For example, the above-described module embodiments are only schematic, for example, the division of units is only a logical function division, and actual implementation can have another division manner, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interface, module or unit, and can be electrical, mechanical or other forms.

[0069] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than limit them. The present application is not limited to the exact structure as shown in the above description and the accompanying drawings, and the specific implementation of the present application should not be considered as limited to the above description. For those skilled in the art of the present application, various changes and modifications made without departing from the concept of the present application should be considered as falling within the scope of the present application.

Claims

1. A method of laser powder bed fusion part quality prediction, the method comprising: The method comprises: acquiring three-dimensional point cloud data of a formed part of laser powder bed melting; aligning the three-dimensional point cloud data along a central axis, and dividing the aligned three-dimensional point cloud data into a plurality of cross-sectional point sets along the central axis direction; performing independent local geometric fitting on each cross-sectional point set to obtain local fitting parameters of each cross section; based on the three-dimensional point cloud data and the local fitting parameters of each cross section, unfolding a three-dimensional original surface of the formed part into a two-dimensional surface deviation topological map; based on the two-dimensional surface deviation topological map, calculating the surface roughness of the formed part, and statistically evaluating the process-induced macro-geometric deviation of the formed part based on the local fitting parameters of each cross section.

2. The method of claim 1, wherein, The central axis alignment of the three-dimensional point cloud data comprises: determining a first principal component vector by principal component analysis algorithm, and determining the first principal component vector as the actual central principal axis of the formed part; aligning the actual central principal axis with the preset axis of the coordinate system by calculating a rotation matrix to obtain the aligned three-dimensional point cloud data.

3. The method of claim 2, wherein, The division of the aligned three-dimensional point cloud data into a plurality of cross-sectional point sets along the central axis direction comprises: defining a series of equidistant virtual slice planes along the central axis direction; defining a thickness tolerance zone around each virtual slice plane, extracting three-dimensional points falling within the thickness tolerance zone for each virtual slice plane, and taking the coordinates perpendicular to the central axis to form the cross-sectional point set.

4. The method of claim 3, wherein, The independent local geometric fitting on each cross-sectional point set to obtain the local fitting parameters of each cross section comprises: for each cross-sectional point set, identifying a group of inliers conforming to the main contour from the cross-sectional point set by random sample consensus algorithm to form an inlier set; and performing nonlinear least squares optimization on the inlier set by Levenberg-Marquardt algorithm to find the optimal circle parameters and determine the local fitting parameters; wherein for each inlier in the inlier set, the residual is defined as the difference between the actual radial distance of the point to the fitted circle and the fitted radius.

5. The method of claim 4, wherein, The local fitting parameters of each cross section form a local fitting parameter sequence; and the unfolding of the three-dimensional original surface of the formed part into a two-dimensional surface deviation topological map based on the three-dimensional point cloud data and the local fitting parameters of each cross section comprises: for any point in the three-dimensional point cloud data, establishing a dynamic local fitting reference of the axial position of the point by interpolating the discrete local fitting parameter sequence; calculating the radial deviation value of any point in the three-dimensional point cloud data relative to the dynamic local fitting reference of the point and the circumferential angle coordinate of the point in the unfolded coordinate system; mapping the circumferential angle coordinate, axial position coordinate and radial deviation value of all points in the three-dimensional point cloud data to the created regular two-dimensional grid based on natural neighborhood interpolation to generate the two-dimensional surface deviation topological map.

6. The method of claim 5, wherein, The calculation of the surface roughness of the formed part based on the two-dimensional surface deviation topological map comprises: applying a two-dimensional Gaussian filter to the two-dimensional surface deviation topological map to obtain a filtered roughness matrix; a standard deviation of the two-dimensional Gaussian filter is calculated based on a cutoff wavelength set by a user and a pixel density of a grid, and a relationship between the standard deviation and the cutoff wavelength is: ； wherein, is a standard deviation of the two-dimensional Gaussian filter; is a cut-off wavelength; is a pixel density; is is a function to calculate a square root of a non-negative number; calculating surface roughness based on the roughness matrix; the surface roughness includes a face roughness parameter and an average axial profile roughness distributed along a circumference.

7. The method of claim 6, wherein, the local fitting parameters of each cross section include a local radius fitting sequence and a local fitting center coordinate sequence of all cross sections; the statistical evaluation of the process-induced macro-geometric deviation of the formed part based on the local fitting parameters of each cross section includes: statistical analysis of the local fitting radius sequence to evaluate the dimensional deviation of the formed part; statistical analysis of the local fitting center coordinate sequence to evaluate the waviness or bending deformation of the formed part caused by thermal stress; wherein the statistical analysis of the local fitting center coordinate sequence includes: calculating the distance from each center to the actual center main axis to form a distance sequence, and then quantitatively representing the overall bending deformation caused by thermal stress based on the maximum value or root mean square value of the distance sequence.

8. The method of claim 1-7, wherein, The laser powder bed fusion formed part is a predicted formed part simulated by a multi-physical field numerical model based on a laser powder bed fusion process; the method further includes: establishing a multi-physical field numerical model of the laser powder bed fusion process, simulating the forming process based on the forming process parameters of the to-be-formed part in the multi-physical field numerical model to obtain a predictive three-dimensional surface data set of the formed part; converting the predictive three-dimensional surface data set into a three-dimensional point cloud data.

9. The method of claim 8, wherein, The laser powder bed fusion formed part is an actual formed part manufactured by physical manufacturing, and for the actual formed part, a scanning technique is used to obtain an actual three-dimensional surface data set of the surface topography of the actual formed part; the method further includes: comparing a first evaluation result based on the predictive three-dimensional surface data set and a second evaluation result based on the actual three-dimensional surface data set to obtain a third evaluation result; based on the third evaluation result, verifying and calibrating the multi-physical field numerical model, and / or feeding back and optimizing the laser powder bed fusion process parameters.

10. A laser powder bed fusion part quality prediction system for implementing the laser powder bed fusion part quality prediction method of any one of claims 1-9, wherein, including: an acquisition module for acquiring three-dimensional point cloud data of a laser powder bed fusion formed part; a processing module for center axis alignment processing of the three-dimensional point cloud data, and for dividing the aligned three-dimensional point cloud data into a plurality of cross-sectional point sets along the center axis direction; a fitting module for independent local geometric fitting of each cross-sectional point set to obtain local fitting parameters of each cross section; an unfolding module for unfolding a three-dimensional original surface of the formed part into a two-dimensional surface deviation topological map based on the three-dimensional point cloud data and the local fitting parameters of each cross section; an evaluation module for calculating surface roughness of the formed part based on the two-dimensional surface deviation topological map, and for statistically evaluating process-induced macro-geometric deviation of the formed part based on the local fitting parameters of each cross section.

Images