Metal 3D printing quality prediction method and system

Abstract

The application relates to the technical field of 3D printing quality prediction, and particularly discloses a metal 3D printing quality prediction method and system, wherein the method comprises the following steps: collecting real-time state information of a 3D printer and state data of a current printing piece; calculating thermal stress of each layer of the printing piece in a printing process; generating a spatial geometric precision grade map by combining a convolutional neural network with a time series neural network and regional clustering; compared with the prior art, the method can analyze the quality by means of single temperature monitoring or post-processing detection, and in particular, under printing conditions with complex structures and significant interlayer stress coupling, it is difficult to realize the technical problems of precision prediction preposition and real-time spatial error analysis; due to the introduction of two-dimensional error modeling driven by thermal stress, three-dimensional error distribution prediction based on deep learning and clustering type spatial grade division, the method realizes process-level prediction and structure-level expression of the quality of the printing piece, and improves the prediction accuracy and spatial expression capability.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of 3D printing quality prediction, and particularly relates to a metal 3D printing quality prediction method and system. BACKGROUND

[0002] At present, metal 3D printing, as an important branch of additive manufacturing technology, has been widely used in the manufacturing field of aerospace, medical devices, high-end equipment and other fields with extremely high requirements for structural performance and precision. In the printing process, geometric accuracy is one of the key indicators to measure the quality of the printed part, which not only relates to whether the printed part meets the design size and structural tolerance, but also directly determines the feasibility of subsequent assembly and the reliability of function implementation. Especially in precise parts containing complex internal cavities, free-form surfaces and multi-layer detailed structures, slight spatial error accumulation may lead to assembly failure or abnormal performance. However, the existing metal 3D printing quality evaluation methods mostly rely on size measurement or mechanical testing in the post-processing stage, and cannot realize the accuracy trend prediction and regional error perception in the printing process. For example, some methods only make rough judgments based on single-point temperature monitoring or single-layer stress analysis, and do not fully consider important factors affecting geometric accuracy such as multi-layer thermal coupling, material thermal deformation behavior and printing path disturbance, which is difficult to meet the demand for spatial accuracy predictability and explainability in actual application. Therefore, there is an urgent need for a quality prediction method that can predict and grade the geometric error of the printed part at each position in space in advance without relying on additional sensors or interrupting the printing process, so as to improve the geometric accuracy guarantee capability and overall manufacturing reliability of metal 3D printing. SUMMARY

[0003] In view of the above technical deficiencies, the purpose of the present application is to provide a metal 3D printing quality prediction method, which aims to solve the technical problems that the existing technology mainly analyzes the quality through single temperature monitoring or post-processing detection, and it is difficult to realize the accuracy prediction and real-time spatial error analysis under the printing conditions of complex structure and significant interlayer stress coupling.

[0004] To solve the above technical problems, the technical scheme adopted by the present application is as follows: the present application provides a metal 3D printing quality prediction method,

[0005] The metal 3D printing quality prediction method comprises:

[0006] Step S10: collecting real-time state information of the 3D printer and current printed part state data, and calculating the current printed part thermal stress according to the real-time state information and the current printed part state data;

[0007] Step S20: obtaining a target design size of a target printed part, and constructing a two-dimensional error map according to the target design size and the current printed part thermal stress;

[0008] Step S30: After standardizing the two-dimensional error map, real-time status information and current printed part status data, a fused input vector is obtained. An error distribution prediction model is preset, and a convolutional neural network and a time series network are used to fuse and train the error distribution prediction model. The fused input vector is then input into the trained error distribution prediction model to obtain the three-dimensional error distribution.

[0009] Step S40: Perform region clustering and error level classification based on the three-dimensional error distribution to generate a spatial geometric accuracy level map of the target printed part;

[0010] Step S50: Generate a print quality prediction report based on the spatial geometric accuracy level map and the target design dimensions of the target print part.

[0011] Preferably, step S10, which involves collecting real-time status information of the 3D printer and current printed part status data, and calculating the thermal stress of the current printed part based on the real-time status information and current printed part status data, specifically includes:

[0012] Collect real-time status information of the 3D printer and current printed part status data, and obtain the melting zone temperature T at time t and position (x, y) from the current printed part status data. melt (t,x,y), the coefficient of thermal expansion of the metal material α and the elastic modulus of the metal material E, and obtain the ambient temperature of the printer cavity at time t (x,y) from the real-time status information.

[0013] Based on the melting zone temperature T melt The relative temperature difference is calculated from (t,x,y) and the ambient temperature of the printer cavity.

[0014] A global adjustment factor for the thermal stress of the current printed part is introduced. The thermal stress of the current printed part is calculated based on the relative temperature difference, the global adjustment factor for the thermal stress of the current printed part, the thermal expansion coefficient α of the metal material, and the elastic modulus E of the metal material.

[0015] Preferably, in step S10, the real-time status information includes printer temperature, printer laser power, and printer scanning speed; the current printed part status data includes metal material type, metal material thermal expansion coefficient, metal material elastic modulus, melting zone temperature, and current printed part thermal stress.

[0016] Preferably, in step S30, the fusion input vector is obtained by standardizing the two-dimensional error map, real-time status information, and current printed part status data. An error distribution prediction model is preset, and a convolutional neural network and a time series network are used to fuse and train the error distribution prediction model. The fusion input vector is then input into the trained error distribution prediction model to obtain the three-dimensional error distribution. Specifically, this step includes:

[0017] Step S301: Normalizing the two-dimensional error map, real-time state information and current printing piece state data to construct a fusion input vector in a unified numerical space;

[0018] Step S302: Extracting spatial correlation pattern features between local stress change features and geometric deformation features from the fusion input vector using a convolutional neural network;

[0019] Step S303: Extracting time series correlation pattern features in the printing process from the fusion input vector using a time series network;

[0020] Step S304: Fusing the spatial correlation pattern features between the local stress change features and the geometric deformation features extracted by the convolutional neural network, and the time series correlation pattern features extracted by the time series network to obtain a feature fusion vector;

[0021] Step S305: Pre-setting an error distribution prediction model, obtaining historical error distribution label data and historical feature fusion vector data, and pre-training the error distribution prediction model by taking the historical error distribution label data and the historical feature fusion vector data as inputs of the error distribution prediction model. Finally, the feature fusion vector is output through the inter-layer fusion layer and the full connection layer of the pre-trained error distribution prediction model to output a three-dimensional error distribution.

[0022] Preferably, in step S30, the inter-layer fusion layer is used to realize the time series fusion from the layered two-dimensional error map to the overall three-dimensional error distribution, wherein the formula of the inter-layer fusion layer is:

[0023]

[0024] wherein, is the predicted geometric error of the kth layer position (x, y), wherein z k is the height coordinate value of the kth layer position (x, y); ΔL (k) (x, y) is the two-dimensional error map of the kth layer position (x, y), h k-1 is the hidden state passed through the k-1th layer of the error distribution prediction model, s k is the process parameter feature vector corresponding to the kth layer, f fusion (·) is the fusion function of the error distribution prediction model.

[0025] Preferably, in step S40, the three-dimensional error distribution is used for region clustering and error level division to generate a spatial geometric precision level map of the target printing piece, which specifically includes:

[0026] Step S401: voxelizing and segmenting the three-dimensional error distribution to obtain a plurality of spatial sub-regions, the spatial sub-regions being used to represent geometric deviations of each point of the target printed product in a three-dimensional space;

[0027] Step S402: extracting local error statistical features of each spatial sub-region by using a mathematical statistical method to form an error feature vector;

[0028] Step S403: performing unsupervised clustering analysis on the plurality of spatial sub-regions based on the error feature vector to obtain a plurality of clustered sub-regions;

[0029] Step S404: obtaining a preset error level division standard, performing level mapping on the plurality of clustered sub-regions according to the preset error level division standard, and generating a spatial geometric precision level map of the target printed product.

[0030] Preferably, in step S20, the formula of the two-dimensional error map is:

[0031]

[0032] wherein σ(t, x, y) is a current printed product thermal stress at time t and position (x, y), ΔL(x, y) is a two-dimensional error map used to represent linear geometric deformation at position (x, y), L0(x, y) is a target design size at position (x, y), E is an elastic modulus of a metal material, and δ is a linear geometric deformation global adjustment factor.

[0033] The application also provides a metal 3D printing quality prediction system, which comprises:

[0034] a state stress calculation module, configured to collect real-time state information of a 3D printer and current printed product state data, and calculate a current printed product thermal stress according to the real-time state information and the current printed product state data;

[0035] an error mapping module, configured to obtain a target design size of a target printed product, and construct a two-dimensional error map according to the target design size and the current printed product thermal stress;

[0036] an error prediction modeling module, configured to obtain a fusion input vector by performing standardization processing on the two-dimensional error map, the real-time state information and the current printed product state data, predefine an error distribution prediction model, train the error distribution prediction model by using a convolutional neural network and a time series network, and input the fusion input vector into the trained error distribution prediction model to obtain a three-dimensional error distribution;

[0037] a level determination module, configured to perform region clustering and error level division according to the three-dimensional error distribution, and generate a spatial geometric precision level map of the target printed product;

[0038] The quality prediction output module is configured to generate a printing quality prediction report according to the spatial geometric accuracy grade map and in combination with the target design size of the target printed part.

[0039] The application further provides a computer program product comprising a metal 3D printing quality prediction program, which, when executed by a processor, implements the metal 3D printing quality prediction method.

[0040] The application has the beneficial effect that, compared with the prior art which mainly analyzes the quality by single temperature monitoring or post-processing detection, especially under the printing conditions of complex structure and significant interlayer stress coupling, it is difficult to realize the technical problem of preposition of precision prediction and real-time spatial error analysis. Since the application introduces a two-dimensional error modeling driven by thermal stress, a three-dimensional error distribution prediction based on deep learning, and a clustering type spatial grade division, the process-level prediction and structure-level expression of the printed part quality are realized, and the prediction accuracy and spatial expression capability are improved. BRIEF DESCRIPTION OF DRAWINGS

[0041] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description only constitute some embodiments of the application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0042] Figure 1 The figure is a flowchart of the first embodiment of the metal 3D printing quality prediction method of the application.

[0043] Figure 2 The figure is a device schematic diagram of the metal 3D printing quality prediction method of the application. DETAILED DESCRIPTION

[0044] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments only constitute some embodiments of the application, rather than all the embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the application.

[0045] Embodiment one: as shown in the figure, the figure is a flowchart of the first embodiment of the metal 3D printing quality prediction method of the application, which proposes the first embodiment of the metal 3D printing quality prediction method of the application. Figure 1

[0046] In the first embodiment, the metal 3D printing quality prediction method comprises:

[0047] ​Step S10: Collect real-time state information of the 3D printer and current printing piece state data, and calculate the current printing piece thermal stress according to the real-time state information and the current printing piece state data;

[0048] It should be noted that in step S10, the formula for calculating the current printing piece thermal stress is:

[0049] σ(t, x, y) = γ·E·α·(T melt (t, x, y) - T chamber (t, x, y))

[0050] Wherein, γ is the global adjustment factor of the current printing piece thermal stress, σ(t, x, y) is the current printing piece thermal stress at time t and position (x, y); E is the elastic modulus of the metal material; α is the thermal expansion coefficient of the metal material; T melt (t, x, y) is the temperature of the molten zone at time t and position (x, y), which is determined according to the current printing piece state data; T chamber (t, x, y) is the environmental temperature of the printing cavity at time t and position (x, y), which is determined according to the real-time state information.

[0051] It can be understood that this formula introduces the temperature difference term of two variables of time and space, effectively captures the local thermal stress change trend under the premise of keeping the material properties unchanged, and can pre-estimate the potential stress concentration phenomenon in different regions.

[0052] It should be understood that compared with the traditional method which only considers the average heat input or static temperature difference estimation, the present method introduces the dynamic difference term between the printer environmental temperature and the melting temperature in the thermal stress prediction, which can more truly reflect the stress response behavior of the material under the non-uniform temperature field, and is particularly suitable for the metal 3D printing scene with complex path scanning and non-constant temperature cavity conditions.

[0053] For example, for example, when printing a titanium alloy structural piece with significant wall thickness variation, by setting the material parameters E = 110 GPa, α = 8.5 × 10 -6 / ℃, when the local molten zone temperature is 1550℃, the cavity temperature is 100℃, and the global adjustment factor is set to 0.9, the predicted thermal stress is: σ = 0.9·110×10 9 ·8.5×10 -6 ·(1550-100)≈135MPa, which is consistent with the subsequent thermal deformation experiment measured residual stress field change trend, verifying the effectiveness of the constructed thermal stress model.

[0054] Step S20: Obtain the target design size of the target printing piece, and construct a two-dimensional error map according to the target design size and the current printing piece thermal stress;

[0055] It should be noted that in step S20, the two-dimensional error map is expressed as:

[0056]

[0057] Where σ(t, x, y) is the current thermal stress of the printed part at time t and position (x, y), ΔL(x, y) is the two-dimensional error map, which represents the linear geometric deformation at position (x, y), L0(x, y) is the target design size at position (x, y), E is the elastic modulus of the metal material, and δ is the global adjustment factor for linear geometric deformation.

[0058] It can be understood that the above error map construction method is based on the linear strain response relationship in elasticity, and under the assumption that the material is in the elastic deformation stage, the micro-variation of the target size is estimated by the stress field to form the regional error field. The error map still retains the mapping relationship with the CAD design model in space, which is convenient for subsequent spatial splicing and prediction modeling.

[0059] It should be understood that compared with the traditional estimation method based on scanning path superposition or statistical average error, the present application starts from the micro-field response of thermal stress, establishes the deformation mapping relationship combined with the target size, can realize the error prediction of spatial continuity, and has higher physical interpretation and structural adaptability. Especially in complex printed parts with porous structure and variable thickness layer design, the prediction ability of boundary deformation and local deviation can be significantly improved.

[0060] For example, the thermal stress prediction value of a certain point is σ = 120 MPa, the material elastic modulus is E = 200 GPa, the adjustment factor is δ = 0.95, and L0(x, y) = 40 mm. Substituting the formula, the error of the point is:

[0061]

[0062] The predicted error value is basically consistent with the actual error (23.1 μm) detected by the three-coordinate measuring instrument after printing, which verifies the effectiveness of the error map construction method at the micron level.

[0063] Step S30: After standardizing the two-dimensional error map, real-time state information and current printed part state data, a fusion input vector is obtained, a preset error distribution prediction model is set, and a convolutional neural network and a time series network are used to fuse and train the error distribution prediction model. The three-dimensional error distribution is obtained by inputting the fusion input vector into the trained error distribution prediction model.

[0064] It should be noted that in step S30, the two-dimensional error map, real-time state information and current printing piece state data are standardized to obtain a fusion input vector, a preset error distribution prediction model is obtained, a convolutional neural network and a time series network are used to fuse and train the error distribution prediction model, and the fusion input vector is input into the trained error distribution prediction model to obtain a three-dimensional error distribution. Specifically, the steps include:

[0065] Step S301: Normalizing the two-dimensional error map, real-time state information and current printing piece state data to construct a fusion input vector in a unified numerical space;

[0066] Step S302: Using a convolutional neural network to extract local stress change features and spatial correlation pattern features between geometric deformation from the fusion input vector;

[0067] Step S303: Using a time series network to extract time sequence correlation pattern features in the printing process from the fusion input vector;

[0068] Step S304: Fusing the local stress change features and the spatial correlation pattern features between geometric deformation extracted by the convolutional neural network, and the time sequence correlation pattern features extracted by the time series network to obtain a feature fusion vector;

[0069] Step S305: Pre-setting an error distribution prediction model, obtaining historical error distribution label data and historical feature fusion vector data, and pre-training the error distribution prediction model by taking the historical error distribution label data and the historical feature fusion vector data as inputs of the error distribution prediction model. Finally, the feature fusion vector is output through the interlayer fusion layer and the full connection layer of the pre-trained error distribution prediction model to output a three-dimensional error distribution.

[0070] It can be understood that the modeling scheme solves the problems of spatial and process decoupling and local prediction discontinuity in traditional models through a spatial-time joint feature extraction mechanism of deep learning. The fused three-dimensional error distribution not only represents the local thermal stress response of the current layer, but also reflects the influence of the thermal history of the upper and lower layers.

[0071] It should be understood that, compared with the prior art which only relies on surface point error regression or layer-by-layer error stacking, the present application introduces an interlayer fusion mechanism of convolutional neural network and time series network, realizes continuous prediction of three-dimensional error, and supports end-to-end data-driven training. The error evolution law is automatically obtained without manual setting of rules, and the adaptability and prediction accuracy are significantly improved.

[0072] For example, in the error prediction task of a high complexity topology print, after training the model with 800 historical printing samples, the new two-dimensional error map (average layer stress 90 MPa) and process data (laser power, scanning speed, material type) are used as fusion vectors to input the model, and the output three-dimensional error distribution has an average error of only 21 μm with the true measured value on the test set. Compared with the traditional error stacking method (error about 41 μm), the prediction error is reduced by about 48.8%, the spatial continuity is stronger, and the precision distribution is closer to the actual thermal stress deformation trend.

[0073] Step S40: performing regional clustering and error level division according to the three-dimensional error distribution to generate a spatial geometric precision level map of the target print;

[0074] It should be noted that step S401: voxelizing and segmenting the three-dimensional error distribution to obtain a plurality of spatial sub-regions, and the spatial sub-regions are used to represent the geometric deviation of each point of the target print in the three-dimensional space;

[0075] Step S402: extracting local error statistical features of each spatial sub-region by using a mathematical statistical method to form an error feature vector;

[0076] Step S403: based on the error feature vector, performing unsupervised clustering analysis on the plurality of spatial sub-regions to obtain a plurality of clustered sub-regions;

[0077] Step S404: obtaining a preset error level division standard, and performing level mapping on the plurality of clustered sub-regions according to the preset error level division standard to generate a spatial geometric precision level map of the target print.

[0078] It should be noted that the voxelization segmentation of the three-dimensional error distribution is specifically uniformly divided into a plurality of equal-size cubic sub-blocks along three spatial dimensions, each sub-block containing the local error value of a plurality of points as the spatial minimum unit for subsequent statistical analysis, ensuring consistent resolution in space and locality of error expression; the mathematical statistical method is used to extract the local error statistical characteristics of each spatial sub-region, specifically the mathematical characteristic extraction of average error, maximum error, standard deviation and local gradient in each sub-block; unsupervised clustering analysis specifically includes K-means clustering analysis and DBSCAN clustering analysis; the pre-set error level division standard defines the error index threshold range corresponding to each level, such as the classification standard for average error, maximum error, error stability, etc., and then the level mapping of multiple clustering sub-regions is specifically including comparing and judging the error characteristics, according to whether the characteristic value falls within the range corresponding to a certain level standard, marking it as the corresponding level label, which is used to represent the geometric accuracy risk level of the spatial sub-region, such as "high precision area", "general area", "warning area", etc. After all the clustering sub-regions complete the level mapping, the accuracy level map of the entire printed part in the three-dimensional space can be constructed, realizing the visualization and regional expression of the printing error.

[0079] It can be understood that the modeling scheme solves the problems of spatial and process decoupling and discontinuous local prediction in traditional models through the space-time joint feature extraction mechanism of deep learning. The fused three-dimensional error distribution not only represents the local thermal stress response of the current layer, but also reflects the influence of the thermal history of the upper and lower layers, and is a prediction result driven by structural consistency and physical rationality.

[0080] It should be understood that, compared with the prior art which only relies on surface point error regression or layer-by-layer error stacking, the present application realizes continuous prediction of three-dimensional error by introducing deep neural architecture and inter-layer fusion mechanism, automatically obtains error evolution law without manual rule setting, and significantly improves adaptability and prediction accuracy.

[0081] For example, in the error prediction task of a printed part with high complexity of topological structure, after training the model using 800 groups of historical printing samples, the new two-dimensional error map (average layer stress 90 MPa) and process data (laser power, scanning speed, material type) are used as fusion vectors to input the model, and the output three-dimensional error distribution has an average error of only 21 μm on the test set compared with the true measured value. Compared with the traditional error stacking method (error about 41 μm), the prediction error is reduced by about 48.8%, the spatial coherence is stronger, and the accuracy distribution is closer to the actual thermal stress deformation trend.

[0082] Step S50: generating a printing quality prediction report according to the spatial geometric accuracy level map and in combination with the target design size of the target printed part.

[0083] It should be noted that the generation process of the printing quality prediction report includes: global accuracy level evaluation, statistics of the voxel proportion of each level area in the spatial geometry accuracy level map; key structure area qualification determination, extracting the key structure area (such as hole position, connecting surface, precision curved surface, etc.) in the design drawing of the target printed part, comparing the corresponding error level, and determining whether it falls within the design tolerance; report index generation: including maximum predicted error, root mean square error (RMSE), spatial error heat map, qualified rate index, predicted confidence score, etc.; the report supports export as a structured file (such as PDF, JSON), which can be used as an important basis for print task evaluation, process parameter review or quality traceability.

[0084] It should be understood that the printing quality prediction report is not only a numerical statistical result, but also a structural combination result based on spatial level information and design target size, which has high application relevance. By matching with the key dimensions and tolerances marked in the design model, the "pre-judgment" of printing feasibility and expected achievement can be realized, helping users to predict potential failure areas before printing.

[0085] Embodiment two: In addition, the present application provides a metal 3D printing quality prediction system, which adopts the metal 3D printing quality prediction method in the above embodiment, and can solve the technical problem of metal 3D printing quality prediction. Compared with the prior art, the beneficial effects of the metal 3D printing quality prediction system provided by the present application are the same as those of the metal 3D printing quality prediction method provided by the above embodiment, and other technical features in the metal 3D printing quality prediction system are the same as those disclosed in the above embodiment method, which will not be repeated here.

[0086] Embodiment three: The present application provides a metal 3D printing quality prediction device, please refer to Figure 2A metal 3D printing quality prediction device includes at least one processor; and a memory communicatively connected with the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform a metal 3D printing quality prediction method of the above-mentioned embodiment one. The metal 3D printing quality prediction device of the embodiments of the present application can include, but is not limited to, mobile terminals such as mobile phones, notebook computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), and car terminals (e.g., car navigation terminals), and fixed terminals such as digital TVs and desktop computers. The metal 3D printing quality prediction device is only one example, and should not limit the functions and use range of the embodiments of the present application. The metal 3D printing quality prediction device can include a processing device 1001 (e.g., a central processing unit, a graphic processing unit, etc.), which can perform various appropriate actions and processes according to programs stored in a read-only memory (ROM) 1002 or loaded from a storage device 1003 to a random access memory (RAM) 1004. In the RAM 1004, various programs and data required for the operation of the metal 3D printing quality prediction device are also stored. The processing device 1001, the ROM 1002, and the RAM 1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. In general, the following systems can be connected to the I / O interface 1006: input devices 1007 including, for example, touch screens, touch pads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices 1008 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 1003 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1009. The communication device 1009 can allow the metal 3D printing quality prediction device to communicate with other devices wirelessly or by wire to exchange data. Although the metal 3D printing quality prediction device having various systems is shown in the figure, it should be understood that all the shown systems are not required to be implemented or provided. More or less systems can be alternatively implemented or provided.

[0087] Embodiment four: the application also provides a computer program product, comprising a computer program, which, when executed by a processor, implements the steps of a metal 3D printing quality prediction method as described above. The computer program product provided by the application can solve the technical problem of metal 3D printing quality prediction. Compared with the prior art, the beneficial effects of the computer program product provided by the application are the same as those of the metal 3D printing quality prediction method provided by the above-mentioned embodiment, and are not described here.

[0088] In particular, according to the embodiments disclosed by the application, the processes described above with reference to the flowcharts can be implemented as a computer software program. For example, the embodiments disclosed by the application include a computer program product comprising a computer program carried on a computer readable medium, the computer program comprising program code for executing the method shown in the flowchart. In such embodiments, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by a processing device 1001, the above-mentioned functions defined in the method of the embodiments disclosed by the application are executed.

[0089] It should be understood that various parts of the application disclosed can be realized by hardware, software, firmware or a combination thereof. In the description of the above-mentioned embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0090] Obviously, those skilled in the art can make various modifications and variations to the application without departing from the spirit and scope of the application. Thus, if these modifications and variations of the application fall within the scope of the claims of the application and their equivalent technologies, the application also intends to include these modifications and variations.

Claims

1. A method for predicting the quality of metal 3D printing, characterized in that, The methods include: Step S10: Collect the real-time status information of the 3D printer and the current status data of the printed part, and calculate the thermal stress of the current printed part based on the real-time status information and the current status data of the printed part; Step S20: Obtain the target design dimensions of the target printed part, and construct a two-dimensional error map based on the target design dimensions and the current thermal stress of the printed part; Step S30: After standardizing the two-dimensional error map, real-time status information and current printed part status data, a fused input vector is obtained. An error distribution prediction model is preset, and a convolutional neural network and a time series network are used to fuse and train the error distribution prediction model. The fused input vector is then input into the trained error distribution prediction model to obtain the three-dimensional error distribution. Step S40: Perform region clustering and error level classification based on the three-dimensional error distribution to generate a spatial geometric accuracy level map of the target printed part; Step S50: Generate a print quality prediction report based on the spatial geometric accuracy level map and the target design dimensions of the target printed part; Step S10 involves collecting real-time status information of the 3D printer and current printed part status data, and calculating the thermal stress of the printed part based on the real-time status information and current printed part status data. This specifically includes: Collect real-time status information of the 3D printer and current printed part status data, and obtain the melting zone temperature at time t (x, y) from the current printed part status data. Coefficient of thermal expansion of metallic materials and the elastic modulus of metallic materials The ambient temperature of the printer cavity at time t (x, y) is obtained from the real-time status information. Based on the temperature of the melting zone The relative temperature difference between the printer cavity ambient temperature and the actual temperature is calculated. Introducing a global adjustment factor for thermal stress of the current printed part, based on the relative temperature difference, the global adjustment factor for thermal stress of the current printed part, and the coefficient of thermal expansion of the metallic material. and the elastic modulus of metallic materials The thermal stress of the current printed part is calculated; In step S10, the real-time status information includes printer temperature, printer laser power, and printer scanning speed; the current printed part status data includes metal material type, metal material thermal expansion coefficient, metal material elastic modulus, melting zone temperature, and current printed part thermal stress. In step S30, the two-dimensional error map, real-time status information, and current printed part status data are standardized to obtain a fused input vector. An error distribution prediction model is preset, and a convolutional neural network and a time series network are used to fuse and train the error distribution prediction model. The fused input vector is then input into the trained error distribution prediction model to obtain the three-dimensional error distribution. Specifically, this step includes: Step S301: Normalize the two-dimensional error map, real-time status information and current printout status data to construct a fusion input vector in a unified numerical space; Step S302: Use a convolutional neural network to extract the spatial correlation pattern features between local stress change features and geometric deformation from the fused input vector; Step S303: Use a time series network to extract temporal correlation pattern features from the fused input vector during the printing process; Step S304: Fuse the spatial correlation pattern features between the local stress change features and geometric deformation extracted by the convolutional neural network and the temporal correlation pattern features extracted by the time series network to obtain a feature fusion vector; Step S305: Preset the error distribution prediction model, obtain historical error distribution labeled data and historical feature fusion vector data, and use the historical error distribution labeled data and historical feature fusion vector data as input to the error distribution prediction model to pre-train the error distribution prediction model. Finally, pass the feature fusion vector through the inter-layer fusion layer and fully connected layer of the pre-trained error distribution prediction model to output the three-dimensional error distribution. In step S30, the inter-layer fusion layer is used to achieve temporal fusion from the hierarchical two-dimensional error map to the overall three-dimensional error distribution. The formula for the inter-layer fusion layer is expressed as follows: ； in, For the first The predicted geometric error of the layer location (x,y), where For the first The height coordinates of the layer location (x, y); For the first Two-dimensional error plot of layer position (x,y) To predict the model using the error distribution Hidden states passed from layer to layer For the first The process parameter feature vector corresponding to the layer, This is the fusion function for the error distribution prediction model.

2. The method for predicting the quality of metal 3D printing as described in claim 1, characterized in that, Step S40, which involves performing region clustering and error level classification based on the three-dimensional error distribution to generate a spatial geometric accuracy level map of the target printed part, specifically includes: Step S401: Voxelize the three-dimensional error distribution to obtain multiple spatial sub-regions. The spatial sub-regions are used to represent the geometric deviation of the target printed part at each point in three-dimensional space. Step S402: Extract the local error statistical features of each spatial sub-region using mathematical statistical methods to form an error feature vector; Step S403: Based on the error feature vector, perform unsupervised clustering analysis on multiple spatial sub-regions to obtain multiple clustered sub-regions; Step S404: Obtain the preset error level classification standard, and perform level mapping on multiple clustered sub-regions according to the preset error level classification standard to generate a spatial geometric accuracy level map of the target printed part.

3. The method for predicting the quality of metal 3D printing as described in claim 1, characterized in that, In step S20, the formula for the two-dimensional error map is expressed as follows: ； in, Let (x, y) be the current thermal stress of the printed part at time t and position (x, y). This is a two-dimensional error map used to represent linear geometric deformation at position (x, y). Design dimensions for the target location (x,y) The elastic modulus of metallic materials. This is the global adjustment factor for linear geometric deformation.

4. A metal 3D printing quality prediction system, applied to the metal 3D printing quality prediction method according to any one of claims 1-3, characterized in that, The metal 3D printing quality prediction system includes: The state stress calculation module is used to collect real-time state information of the 3D printer and current printed part state data, and calculate the thermal stress of the current printed part based on the real-time state information and current printed part state data. The error mapping module is used to obtain the target design dimensions of the target printed part and construct a two-dimensional error map based on the target design dimensions and the current thermal stress of the printed part. The error prediction modeling module is used to obtain a fused input vector by standardizing the two-dimensional error map, real-time status information and current print status data. It presets an error distribution prediction model and trains the error distribution prediction model using a convolutional neural network and a time series network. The fused input vector is then input into the trained error distribution prediction model to obtain the three-dimensional error distribution. The grade determination module is used to perform regional clustering and error grade classification based on the three-dimensional error distribution, and generate a spatial geometric accuracy grade map of the target printed part. The quality prediction output module is used to generate a print quality prediction report based on the spatial geometric accuracy level map and the target design dimensions of the target print part.

5. A metal 3D printing quality prediction device, characterized in that, The metal 3D printing quality prediction device includes: a memory, a processor, and a metal 3D printing quality prediction program stored in the memory and executable on the processor. When the metal 3D printing quality prediction program is executed by the processor, it implements a metal 3D printing quality prediction method according to any one of claims 1 to 3.

6. A computer program product, characterized in that, The computer program product includes a metal 3D printing quality prediction program, which, when executed by a processor, implements a metal 3D printing quality prediction method according to any one of claims 1 to 3.

Images