Space engine 3D printing deformation compensation method

Abstract

The invention discloses an airspace engine 3D printing deformation compensation method. The method comprises the following steps that S1, a three-dimensional CAD model is constructed, and a printing layer set is generated; s2, obtaining residual stress tensor fields of the current printing layer and the historical printing layer; s3, constructing a stress shadow mapping tensor; s4, inputting the stress shadow mapping tensor and the current printing layer parameter into the reverse stress propagation neural network model, and outputting a deformation prediction vector field; s5, performing geometric compensation processing on the geometric area of the printing layer according to the deformation prediction vector field; s6, generating a printing code according to the compensated geometric model; and S7, carrying out error comparison on the actual deformation vector field and the overall prediction deformation vector field, if the error is greater than a preset tolerance threshold value, feeding back error information to the step S5, and repeatedly executing the steps S5 to S7 until the error vector field does not exceed the tolerance threshold value. According to the method, residual stress modeling and the reverse stress propagation neural network are fused, and 3D printing deformation compensation of the airspace engine is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of additive manufacturing and geometric modeling compensation, and particularly to a method for compensating the deformation of 3D printing of aero-engine. Background Art

[0002] With the continuous growth of the demand for high-performance and high-precision components in the aerospace field, additive manufacturing (AM) technology, especially metal 3D printing technology, plays an increasingly important role in the manufacturing of complex structures of key components of aero-engines. This technology manufactures integral structural parts by layer-by-layer stacking of molten metal powder, and has significant advantages such as high design freedom, high material utilization rate, and short manufacturing cycle. Typical processes such as selective laser melting (SLM) and electron beam melting (EBM) have been widely applied in complex geometric structures such as combustion chambers, nozzles, and turbine components.

[0003] However, since the metal 3D printing process is essentially a strongly non-linear thermo-mechanical coupling process, affected by factors such as laser energy input, scanning path, and material thermal properties, severe thermal gradients and non-uniform cooling phenomena will occur during the printing process, thereby inducing a complex residual stress field distribution. These residual stress fields will cause different degrees of warping, shrinkage, stress concentration, and micro-cracks in the structural parts, thus significantly affecting the geometric accuracy and mechanical properties of the components, and even leading to part scrapping in severe cases. In order to suppress this printing deformation, the existing technologies mainly adopt two types of solutions: one is to reduce stress concentration by optimizing process parameters (such as adjusting scanning speed, preheating temperature, etc.); the other is to perform geometric compensation on the CAD model before printing, that is, "reverse design" the pre-printing model on the premise of known deformation trend, so that the natural springback after printing tends to the target shape.

[0004] At present, some studies have attempted to introduce finite element analysis (FEA) means to simulate the printing residual stress to predict the deformation behavior, and then perform geometric compensation through the linear offset of the model points. Although these methods can improve the forming accuracy to a certain extent, they generally have three core problems: First, the finite element simulation process consumes a large amount of computing resources, is difficult to achieve rapid response, and usually only models the stress field of the current layer, ignoring the cumulative effect of the historical printing layer on the current layer, and it is difficult to accurately capture the stress propagation and non-linear response behavior between layers; Second, the geometric compensation method is mostly the simple linear displacement of control points, lacking a non-linear geometric reconstruction mechanism closely coupled with the stress prediction accuracy, and it is unable to effectively cope with the multi-scale deformation of complex boundary surfaces or high-order curvature regions; Third, the compensation strategy is usually a one-way process, lacking a closed-loop feedback mechanism. Once the printing error deviates from the initial assumption, subsequent adaptive iterative correction cannot be realized according to the physical feedback.

[0005] In addition, with the rise of the application of deep learning in engineering problems, some studies have also attempted to build data-driven deformation prediction models. For example, convolutional neural networks or recurrent neural networks are used to predict deformation trends. However, most network structures are still relatively shallow, failing to fully model the propagation path of the stress tensor in the spatial and temporal dimensions, and lacking interpretability and physical consistency that match the physical characteristics of actual residual stresses, resulting in unstable prediction accuracy and weak generalization ability.

[0006] Therefore, how to provide a 3D printing deformation compensation method for aerospace engines is an urgent problem to be solved by those skilled in the art. Summary of the Invention

[0007] An object of the present invention is to propose a 3D printing deformation compensation method for aerospace engines. The present invention integrates residual stress tensor modeling, stress propagation modeling and reverse stress propagation neural network prediction method to construct a complete 3D printing deformation compensation process for aerospace engines, introduces stress shadow mapping tensor and control point compensation function, realizes multi-layer residual stress perception, prediction and feedback compensation during the printing process, has high precision, adaptability and closed-loop optimization ability, and significantly improves the forming quality and printing stability of complex structural parts.

[0008] A 3D printing deformation compensation method for aerospace engines according to an embodiment of the present invention includes the following steps: S1. Construct a three-dimensional CAD model of an aerospace engine component, and perform slicing processing on the three-dimensional CAD model to generate a set of printing layers; S2. Obtain the printing parameters of each printing layer, and based on finite element simulation, obtain the residual stress tensor fields of the current printing layer and several previous historical printing layers, and construct a residual stress input data set; S3. Perform propagation modeling on the residual stress input data set in the time and space directions to construct a stress shadow mapping tensor, and the stress shadow mapping tensor is used to describe the residual stress influence area formed by the projection of the historical printing layer on the current layer; S4. Input the stress shadow mapping tensor and the printing parameters of the current printing layer into a reverse stress propagation neural network model, and output the deformation prediction vector field of the current printing layer; S5. Perform geometric compensation processing on the geometric region of the corresponding printing layer in the three-dimensional CAD model according to the deformation prediction vector field, adjust the control points of the geometric region of the corresponding printing layer in the three-dimensional CAD model by using a non-linear geometric reparameterization method, and generate a compensated geometric model; S6. Generate printing code according to the compensated geometric model, and control a printing device to perform printing; S7. After printing is completed, perform three-dimensional scanning on the printed product to obtain the actual deformation vector field, and compare the error between the actual deformation vector field and the overall predicted deformation vector field formed by superimposing the predicted deformation vector fields of each printed layer. If the error vector field is greater than the preset tolerance threshold, feedback the error information to step S5, re-perform geometric compensation processing on the geometric region of the corresponding printed layer in the three-dimensional CAD model, and repeat steps S5 to S7 until the error vector field is less than or equal to the preset tolerance threshold.

[0009] Optionally, the printing parameters include laser power, scanning speed, scanning path, scanning spacing, powder preheating temperature, laser spot diameter, and the thermal conductivity, specific heat capacity, density, thermal expansion coefficient, and Young's modulus of the printing material.

[0010] Optionally, S2 specifically includes: S21. Establish a thermo-mechanical coupled finite element simulation model using the printing parameters and the geometric information of the corresponding printed layer. The construction of the thermo-mechanical coupled finite element simulation model includes geometric discretization, mesh generation, material property assignment, and boundary condition setting; S22. Based on the thermo-mechanical coupled finite element simulation model, layer by layer simulate the heat input process of the laser scanning path during the actual 3D printing process, and calculate the dynamic change process of the temperature field of each printed layer over time to obtain the instantaneous temperature distribution and temperature gradient information; S23. Use the instantaneous temperature distribution and temperature gradient information to calculate the transient thermal stress field distribution of the current printed layer during the printing process, and after each layer of laser scanning is completed and cooled to room temperature, calculate and obtain the residual stress tensor field after the printed layer is printed; S24. Perform cross-layer stress accumulation calculation on the residual stress tensor fields of several printed layers before the current printed layer, and establish a residual stress tensor field historical database. The residual stress tensor field of each printed layer in the residual stress tensor field historical database is associated with the current layer in a layer-by-layer cumulative manner; S25. Based on the residual stress tensor field of the current printed layer and the residual stress tensor fields of several previous historical printed layers in the residual stress tensor field historical database, construct a residual stress input data set.

[0011] Optionally, S3 specifically includes: S31. Based on the residual stress tensor field of each printed layer in the residual stress input data set, establish a three-dimensional space coordinate system and a printing time axis with the current printed layer as the reference. The three-dimensional coordinate points are denoted as , where represents the spatial coordinate of the printed part in the width direction, represents the spatial coordinate of the printed part in the length direction, represents the spatial coordinate of the printed part in the height direction. The direction corresponds to the stacking direction of layer-by-layer printing; S32. Under the three-dimensional space coordinate system, perform spatial registration processing on the residual stress tensor fields of each historical printing layer through three-dimensional bilinear interpolation to obtain a tensor expression with the same spatial resolution and coordinate frame as the current printing layer; S33. Based on the heat-elastoplastic mechanics propagation theory, construct a stress propagation function model: ; where, represents the propagation influence tensor of the th layer on the th layer, represents the residual stress tensor of the th layer at the three-dimensional coordinate point , represents the material coupling coefficient of the th layer, represents the exponential decay factor, represents the th layer to the th layer in the direction of the vertical interlayer distance; S34. Through the time-weighted decay algorithm, perform temporal weighting on the propagation influence tensors of the historical printing layers to construct a stress shadow mapping tensor , used to describe the residual stress influence area formed by the projection of the historical printing layers at the spatial position of the current printing layer: ; where, represents the time decay control parameter, represents the printing time interval from the th layer to the current printing layer, represents the printing time interval from the th layer to the current printing layer, takes a value of , represents the total number of printing layers; S35. Output the stress shadow mapping tensor .

[0012] Optionally, the S4 specifically includes: S41. Construct a reverse stress propagation neural network model, which includes an input layer, a spatio-temporal tensor feature fusion layer, a bidirectional gated recurrent unit layer, a tensor attention mechanism layer, and an output prediction layer; S42. Input the stress shadow mapping tensor and the printing parameters of the current printing layer into the input layer of the reverse stress propagation neural network model. The input layer converts the three-dimensional tensor data into a high-dimensional feature vector and performs data preprocessing through linear normalization; S43. The spatio-temporal tensor feature fusion layer performs multi-scale convolution operations on the high-dimensional feature vector, extracts spatial local features and inter-layer temporal features through multiple convolutional kernels respectively, and generates a spatio-temporal coupling feature representation using feature splicing operations; S44. The bidirectional gated recurrent unit layer processes the spatio-temporal coupling feature representation, uses the gated unit mechanism to simultaneously capture the forward propagation of the influence of the historical printing layer on the current printing layer and the backward tracing of the stress state of the current layer on the historical layer, and obtains a bidirectional stress propagation feature representation that is coupled forward and backward; S45. The tensor attention mechanism layer, based on the bidirectional stress propagation feature representation, adopts an attention calculation method that combines the spatial dimension and the channel dimension to generate a spatial-channel joint attention weight mask tensor , where represents a three-dimensional coordinate point, represents the feature channel number; S46. After the output prediction layer performs tensor multiplication fusion on the spatial-channel joint attention weight mask tensor and the bidirectional stress propagation feature, it uses activation function for non-linear mapping, and outputs the deformation prediction vector field of the current printing layer. The deformation prediction vector field is the spatial deformation prediction vector corresponding to each three-dimensional coordinate point , representing the deformation prediction quantity of the current printing layer.

[0013] Optionally, the S5 specifically includes: S51. Based on the deformation prediction vector field, extract the corresponding geometric region of the current printing layer in the 3D CAD model, and locate the CAD surface fragments and boundary ranges included in the geometric region; S52. Perform parametric modeling on the geometric region of the printing layer, convert the original CAD surface fragment expression into a reparameterized surface form based on non-uniform rational B-splines, and construct a two-dimensional control point array , where represents the spatial coordinate vector of the control point in the th row and th column of the original CAD surface fragment; S53. According to the distribution of the deformation prediction vector field at the spatial positions of the control points, construct a control point compensation function: ; where, represents the spatial coordinate vector of the control point after geometric compensation, represents the global compensation adjustment coefficient, represents the direction weight matrix, represents the deformation prediction vector at the position, represents the deformation adjustment coefficient, represents the total number of multi - scale layers, represents the scale weighting factor of, represents the average deformation prediction vector of the adjacent area at the position under the scale ; S54. Perform surface reconstruction based on the spatial coordinate vectors of the control points after geometric compensation to form an updated expression of the corresponding area of the current printing layer in the three - dimensional CAD model after geometric compensation; S55. Perform geometric legality verification, and the geometric legality verification includes constraint judgments on boundary closure, surface continuity, and that the maximum compensation amplitude does not exceed the corresponding preset tolerance. If the constraint judgment conditions are met, output the geometric model after compensation.

[0014] Optionally, the specific steps of S6 include: S61. Convert the geometric model after compensation into an intermediate model file for the processing format. The intermediate model uses the patch - based meshing format for geometric approximation, and the file format is STL; S62. Perform path planning on the intermediate model, adjust the scanning pitch and path density according to the curvature distribution and structural thickness of the geometric compensation area, and generate a set of printing paths; S63. Based on the set of printing paths, generate a printing control parameter table through printing parameter mapping; S64. Generate printing control codes recognized by the standard printing device according to the printing control parameter table. The printing control codes are additive manufacturing control programs based on the G - code structure, including layer - by - layer control instructions and compensation correction amount annotation information; S65. Transmit the printing control codes to the printing device control module to control the printing device to perform additive manufacturing operations layer by layer according to the printing control codes.

[0015] Optionally, the specific steps of S7 include: S71. After printing is completed, use a laser scanner to perform three - dimensional full - field scanning on the printed product to obtain point cloud data of the complete surface and construct a scanning model; S72. Align the coordinates of the scanning model and the three - dimensional CAD model after compensation, use the iterative closest point algorithm to achieve rigid registration, and establish a point - to - point spatial mapping relationship; S73. Calculate the actual deformation vector field based on the registration result, where the actual deformation vector field is the spatial difference between corresponding points of the scanned model and the compensated 3D CAD model in a unified coordinate system; S74. Cumulatively add the deformation prediction vector fields of each layer during the printing process layer by layer to form an overall predicted deformation vector field, and compare it with the actual deformation vector field to obtain an error vector field; S75. According to the distribution characteristics of the error vector field in three-dimensional space, extract the error peak region, and obtain the control point indices within the corresponding printing layer and geometric region; S76. Use the sampled value of the error vector field at the control point position as an error correction term, introduce it into the control point update mechanism, and perform secondary iterative compensation on the control points based on the previous round of compensation results: ; where, represents the spatial coordinate vector of the control point after geometric compensation, represents the spatial coordinate vector of the control point after error feedback adjustment, is the error correction term at the position, and is the error adjustment coefficient;

[0016] The beneficial effects of the present invention are as follows: First, the present invention introduces a residual stress tensor field simulation mechanism based on thermo-mechanical coupling finite element simulation, and constructs a residual stress input data set for historical printing layers, establishing a high-dimensional and continuous physical data basis for subsequent stress propagation and deformation prediction. This residual stress tensor not only considers the thermal stress distribution of the current printing layer, but also fully models the coupling effect of historical layers on the current layer in both spatial and temporal directions, solving the problem that traditional methods cannot capture the non-linear stress transfer law between multi-layer printing.

[0017] Second, the present invention first proposes the concept of stress shadow mapping tensor, and combines the thermo-elastic-plastic propagation function and the time decay algorithm to map and weight the residual stress of historical printing layers layer by layer in three-dimensional space, realizing the visual expression of cross-layer stress influence and enhancing the structural perception ability. This stress shadow mapping tensor, as an important input of the reverse stress propagation neural network model, enables the model to establish a high-dimensional expression of the spatio-temporal evolution process of the historical stress field, improving the accuracy and robustness of deformation prediction.

[0018] In addition, the present invention designs an improved neural network structure with multi-level coupling, integrating spatio-temporal convolution, bidirectional gated recurrent unit and tensor attention mechanism, which significantly enhances the network's modeling ability for complex non-linear stress-deformation relationships. In cooperation with the non-linear activation function and the spatial channel joint attention mechanism, the deformation prediction resolution of the key area is improved, and the output deformation prediction vector field has stronger local expression ability and global coordination ability, providing high-confidence data support for subsequent compensation.

[0019] Furthermore, the present invention adopts a geometric reparameterization compensation method based on non-uniform rational B-spline, combines the multi-scale deformation prediction response of the control point positions, constructs a control point compensation function, and breaks through the traditional linear displacement compensation method. Through the directional weight adjustment of local control points, fine compensation for complex boundaries and high-curvature regions is achieved, making the local geometric response of the 3D CAD model more conform to the actual deformation trend.

[0020] Finally, by constructing a feedback mechanism based on the error vector field, the present invention establishes a closed-loop control process for fine-tuning the geometric error after printing to the control points. After each round of printing, the error vector is sampled, mapped and fed back to the control point update link, realizing the adaptive optimization of the deformation prediction and geometric compensation strategies, significantly improving the consistency and accuracy of the printing of complex structural parts, and reducing the number of iterations and resource waste. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The drawings are used to provide a further understanding of the present invention, and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention, and do not constitute a limitation to the present invention. In the drawings: Figure 1 is the overall flow chart of a 3D printing deformation compensation method for an aerospace engine proposed by the present invention; Figure 2 is the structural schematic diagram of a reverse stress propagation neural network model of a 3D printing deformation compensation method for an aerospace engine proposed by the present invention; Figure 3 is the flow schematic diagram of a control point error feedback compensation mechanism of a 3D printing deformation compensation method for an aerospace engine proposed by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0022] Now, the present invention will be further described in detail with reference to the drawings. These drawings are all simplified schematic diagrams, only illustrating the basic structure of the present invention in a schematic way, so they only show the components related to the present invention.

[0023] Refer to Figures 1-3 , a 3D printing deformation compensation method for an aerospace engine, comprising the following steps: S1. constructing a three-dimensional CAD model of aerospace engine parts, and slicing the three-dimensional CAD model to generate a set of printing layers; S2, obtaining the printing parameters of each printing layer, and obtaining the residual stress tensor field of the current printing layer and several previous historical printing layers based on finite element simulation, and constructing a residual stress input data set; S3, inputting the residual stress into the data set to perform propagation modeling in the time and space directions, and constructing a stress shadow mapping tensor, wherein the stress shadow mapping tensor is used to describe the residual stress influence area formed by the projection of the historical printing layer on the current layer; S4, inputting the stress shadow mapping tensor and the printing parameters of the current printing layer into the reverse stress propagation neural network model, and outputting the deformation prediction vector field of the current printing layer; S5, performing geometric compensation processing on the geometric area of the corresponding printing layer in the three-dimensional CAD model according to the deformation prediction vector field, adjusting the control points of the geometric area of the corresponding printing layer in the three-dimensional CAD model by using a nonlinear geometric reparameterization method, and generating a compensated geometric model; S6, generating a printing code according to the compensated geometric model, and controlling the printing device to print; S7. After printing is completed, the printed product is three-dimensionally scanned to obtain the actual deformation vector field, and the actual deformation vector field is compared with the overall predicted deformation vector field formed by superimposing the deformation prediction vector fields of each printing layer. If the error vector field is greater than the preset tolerance threshold, the error information is fed back to step S5, and the corresponding printing layer geometric area in the three-dimensional CAD model is re-geometrically compensated, and steps S5 to S7 are repeated until the error vector field is less than or equal to the preset tolerance threshold.

[0024] By establishing a full-process 3D printing deformation compensation system from CAD modeling, stress analysis, predictive modeling, geometric compensation to printing verification and feedback closed loop, active control of the printing accuracy of complex components of aerospace engines is achieved. The entire process takes into account the coupling characteristics of physical modeling and intelligent prediction. It not only fully considers the cross-layer propagation effect of residual stress during the printing process, but also can feedback printing errors in real time and perform iterative geometric adjustments. This method has opened up the three key links of prediction, compensation and feedback verification, has good engineering applicability, systematicness and scalability, and significantly improves the dimensional consistency and forming stability of metal 3D printed structures. It is particularly suitable for the manufacturing of high-precision and high-complexity aerospace parts.

[0025] In this embodiment, the printing parameters include laser power, scanning speed, scanning path, scanning spacing, powder preheating temperature, laser spot diameter, and thermal conductivity, specific heat capacity, density, thermal expansion coefficient and Young's modulus of the printing material.

[0026] In this embodiment, S2 specifically includes: S21. Establish a thermo-mechanical coupled finite element simulation model by using the printing parameters and the geometric information of the corresponding printing layer. The construction of the thermo-mechanical coupled finite element simulation model includes geometric discretization, mesh generation, material property assignment, and boundary condition setting; S22. Based on the thermo-mechanical coupled finite element simulation model, layer by layer simulate the heat input process of the laser scanning path during the actual 3D printing process, and calculate the dynamic change process of the temperature field of each printing layer over time to obtain the instantaneous temperature distribution and temperature gradient information; S23. Use the instantaneous temperature distribution and temperature gradient information to calculate the transient thermal stress field distribution of the current printing layer during the printing process, and after each layer of laser scanning is completed and cooled to room temperature, calculate and obtain the residual stress tensor field after the printing layer is printed; S24. Perform cross-layer stress accumulation calculation on the residual stress tensor fields of several printing layers before the current printing layer, and establish a residual stress tensor field historical database. The residual stress tensor field of each printing layer in the residual stress tensor field historical database is associated with the current layer in a layer-by-layer cumulative manner; S25. Based on the residual stress tensor field of the current printing layer and the residual stress tensor fields of several previous historical printing layers in the residual stress tensor field historical database, construct a residual stress input data set.

[0027] By constructing a thermo-mechanical coupled finite element model and simulating the heat conduction, heat gradient, and thermal stress evolution paths during the printing process, the multi-layer dynamic acquisition of the printing residual stress tensor field is realized. The residual stresses of the historical printing layers are not only accurately quantified but also retained through accumulation and database methods, providing hierarchical data support for subsequent stress propagation modeling and deformation prediction. This modeling mechanism from local to global and from transient to steady state effectively solves the problems of rough stress estimation and cross-layer information breakage in traditional methods, making the prediction results closer to the actual behavior of the component and greatly improving the reliability of the geometric compensation strategy.

[0028] In this embodiment, S3 specifically includes: S31. Based on the residual stress tensor field of each printing layer in the residual stress input data set, establish a three-dimensional space coordinate system and a printing time axis with the current printing layer as the reference. The three-dimensional coordinate points are denoted as , where represents the spatial coordinate of the printed part in the width direction, represents the spatial coordinate of the printed part in the length direction, represents the spatial coordinate of the printed part in the height direction, The direction corresponds to the stacking direction of layer-by-layer printing; S32. Under the three-dimensional space coordinate system, perform spatial registration processing on the residual stress tensor fields of each historical printing layer through three-dimensional bilinear interpolation to obtain a tensor expression with the same spatial resolution and coordinate framework as the current printing layer; S33. Based on the heat-elastoplastic mechanics propagation theory, construct a stress propagation function model: ; where, represents the propagation influence tensor of the -th layer on the -th layer, represents the residual stress tensor of the -th layer at the three-dimensional coordinate point , represents the material coupling coefficient of the -th layer, represents the exponential decay factor, represents the -th layer to the -th layer in the direction of the vertical interlayer distance; S34. Through the time-weighted decay algorithm, perform temporal weighting on the propagation influence tensors of the historical printing layers to construct a stress shadow mapping tensor , which is used to describe the residual stress influence area formed by the projection of the historical printing layers at the spatial position of the current printing layer: ; where, represents the time decay control parameter, represents the printing time interval from the -th layer to the current printing layer, represents the printing time interval from the -th layer to the current printing layer, takes the value of , represents the total number of printing layers; S35. Output the stress shadow mapping tensor .

[0029] By establishing a stress propagation function and introducing a time decay algorithm, for the first time, a three-dimensional mapping expression of the stress influence of historical printing layers on the current layer is realized, the concept of a stress shadow mapping tensor is proposed, and the influence distribution of the cross-layer coupling effect in spatial position is effectively described. This modeling method not only considers the non-linear relationship between spatial coordinates and interlayer distance, but also incorporates the weight decay in the time dimension, making the modeling result have time causality and physical rationality. The finally generated tensor provides a structured input for the neural network, making subsequent predictions more accurate, especially in complex components with significant multi-layer superposition effects.

[0030] In this embodiment, the S4 specifically includes: S41. Construct a reverse stress propagation neural network model, where the reverse stress propagation neural network model includes an input layer, a spatio-temporal tensor feature fusion layer, a bidirectional gated recurrent unit layer, a tensor attention mechanism layer, and an output prediction layer; S42. Input the stress shadow mapping tensor and the printing parameters of the current printing layer into the input layer of the reverse stress propagation neural network model together. The input layer converts the three-dimensional tensor data into a high-dimensional feature vector and performs data preprocessing through linear normalization; S43. The spatio-temporal tensor feature fusion layer performs multi-scale convolution operations on the high-dimensional feature vector, extracts spatial local features and inter-layer temporal features respectively through multiple convolutional kernels, and generates a spatio-temporal coupling feature representation by using a feature splicing operation; S44. The bidirectional gated recurrent unit layer processes the spatio-temporal coupling feature representation, uses the gated unit mechanism to simultaneously capture the forward propagation of the influence of the historical printing layer on the current printing layer and the reverse trace of the stress state of the current layer on the historical layer, and obtains a bidirectional stress propagation feature representation that is forward and reverse coupled; S45. The tensor attention mechanism layer generates a spatial-channel joint attention weight mask tensor based on the bidirectional stress propagation feature representation, using an attention calculation method that combines the spatial dimension and the channel dimension , where represents the three-dimensional coordinate point, represents the feature channel number; S46. After the output prediction layer performs tensor multiplication fusion on the spatial-channel joint attention weight mask tensor and the bidirectional stress propagation feature, it uses the activation function for non-linear mapping and outputs the deformation prediction vector field of the current printing layer. The deformation prediction vector field is the spatial deformation prediction vector corresponding to each three-dimensional coordinate point , representing the deformation prediction quantity of the current printing layer.

[0031] A reverse stress propagation neural network including an input layer, a spatio-temporal tensor feature fusion layer, a bidirectional gated recurrent unit layer, a tensor attention mechanism layer, and an output prediction layer is constructed to realize a deep learning-based coupled modeling of the historical stress field and printing parameters. The adopted bidirectional GRU structure can simultaneously capture the forward propagation and reverse feedback of stress, and the tensor attention mechanism further improves the feature attention degree of key regions. The deformation prediction vector field output by the model not only has the advantages of high spatial resolution and sufficient channel expression, but also has good adjustability and interpretability, significantly superior to the neural network prediction method of traditional single-layer perception.

[0032] In this embodiment, the S5 specifically includes: S51. Extract the corresponding geometric region of the current printing layer in the 3D CAD model based on the deformation prediction vector field, and locate the CAD surface fragments and boundary ranges included in the geometric region; S52. Perform parametric modeling on the geometric region of the printing layer, convert the original CAD surface fragment expression into a reparameterized surface form based on non-uniform rational B-splines, and construct a two-dimensional control point array , where represents the spatial coordinate vector of the control point in the -th row and -th column of the original CAD surface fragment; S53. Construct a control point compensation function according to the distribution of the deformation prediction vector field at the spatial positions of the control points: ; where, represents the spatial coordinate vector of the control point after geometric compensation, represents the global compensation adjustment coefficient, represents the direction weight matrix, represents the deformation prediction vector at the position of represents the deformation adjustment coefficient, represents the total number of multi-scale layers, represents the scale of the weighting factor, represents the average deformation prediction vector of the adjacent region at the position of under the scale S54. Perform surface reconstruction based on the spatial coordinate vector of the control point after geometric compensation to form an updated expression of the corresponding region of the current printing layer in the 3D CAD model after geometric compensation; S55. Perform geometric legality verification. The geometric legality verification includes constraint judgments on boundary closure, surface continuity, and that the maximum compensation amplitude does not exceed the corresponding preset tolerance. If the constraint judgment conditions are met, the compensated geometric model is output.

[0033] The proposed geometric compensation method is based on the control point mechanism of non-uniform rational B-spline surfaces. By constructing a compensation function strongly coupled with the deformation prediction field, it realizes refined compensation of local regions. Multi-scale adjustment and directional weighting enable the control point adjustment to not only conform to the deformation trend but also suppress the geometric mutation problem caused by local over-compensation. This compensation strategy can adapt to complex CAD geometric fragments of any shape, and the compensation result is smooth and stable, with the boundary of the compensated region transitioning naturally, fundamentally solving the problems of easy out-of-control and discontinuity in traditional displacement method compensation.

[0034] In this embodiment, the specific content of S6 includes: S61. Convert the compensated geometric model into an intermediate model file in a processing format. The intermediate model uses a facet meshing format for geometric approximation, and the file format is STL. S62. Perform path planning on the intermediate model, adjust the scanning spacing and path density according to the curvature distribution and structural thickness of the geometric compensation area, and generate a set of printing paths. S63. Based on the set of printing paths, generate a printing control parameter table through printing parameter mapping. S64. Generate printing control codes recognized by a standard printing device according to the printing control parameter table. The printing control codes are additive manufacturing control programs based on the G-code structure, including layer-by-layer control instructions and compensation correction amount annotation information. S65. Transmit the printing control codes to the printing device control module, and control the printing device to perform additive manufacturing operations layer by layer according to the printing control codes.

[0035] By constructing a fully automated printing path planning and parameter mapping process from CAD geometry to G-code, it is ensured that the compensated model can be losslessly converted into a printing control program recognizable by the device. During the printing path generation process, the scanning density is dynamically adjusted according to the local complexity of the geometric area, effectively improving the printing quality and material utilization efficiency. In addition, the compensation information is embedded in the printing instruction stream in an annotation manner, which helps with subsequent traceability and parameter correction.

[0036] In this embodiment, the specific steps of S7 are as follows: S71. After printing is completed, use a laser scanner to perform three-dimensional full-field scanning on the printed product, obtain point cloud data of the complete surface, and construct a scanning model. S72. Align the coordinates of the scanning model with the compensated three-dimensional CAD model, use the iterative closest point algorithm to achieve rigid registration, and establish a point-to-point spatial mapping relationship. S73. Calculate the actual deformation vector field based on the registration result. The actual deformation vector field is the spatial difference between the corresponding points of the scanning model and the compensated three-dimensional CAD model in the unified coordinate system. S74. Accumulate the deformation prediction vector fields of each layer during the printing process layer by layer to form an overall prediction deformation vector field, and compare it with the actual deformation vector field to obtain an error vector field. S75. According to the distribution characteristics of the error vector field in three-dimensional space, extract the error peak regions, and obtain the control point indexes within the corresponding printing layers and geometric regions. S76. Use the sampling values of the error vector field at the control point positions as error correction terms, introduce them into the control point update mechanism, and perform secondary iterative compensation on the control points based on the previous round of compensation results. ; Wherein, represents the spatial coordinate vector of the control point after geometric compensation, represents the spatial coordinate vector of the control point after being adjusted by error feedback, is the error correction term at the position of, is the error adjustment coefficient; S77. Use the control point adjusted by error feedback to reconstruct the geometric region of the corresponding printing layer in the 3D CAD model to generate a new geometric model, and repeat steps S5 to S7 until the error vector field is less than or equal to the preset tolerance threshold.

[0037] A geometric model re-compensation mechanism based on error vector field feedback is proposed, forming a closed-loop control path of 3D scanning - error recognition - control point re-optimization. The point-by-point comparison between the actual deformation field and the predicted deformation field provides an accurate basis for error positioning, while the extraction of the error peak region and the control point index mapping greatly improve the compensation efficiency. On this basis, the secondary iterative update of the control point is driven by the error correction term, enabling the model compensation ability to have the characteristics of self-adaptation and progressive optimization. This mechanism is particularly suitable for the multi-printing correction task of complex curved surfaces, significantly improving the compensation accuracy and system convergence.

[0038] Example 1: To verify the feasibility of the present invention in implementation, the present invention is applied to the inner liner component of a certain type of aerospace engine nozzle as the test object of this example. This component is prone to significant deformation due to temperature gradient and thermal stress accumulation during the actual printing process, affecting the precision control of key dimensions.

[0039] First, a 3D CAD model of the target nozzle inner liner is constructed by a 3D CAD modeling tool, and slicing operation is performed using an industrial slicing engine to generate a set of 120 printing layers. The thickness of each layer is set to 0.1 mm, and the total printing height is 12 mm.

[0040] In each layer, the printing parameters related to it are respectively extracted. The printing parameters include: laser power 350 W, scanning speed 800 mm / s, the scanning path adopts a rotating partition strategy, the scanning spacing is 0.08 mm, the powder preheating temperature is 150 °C, and the laser spot diameter is 70 μm. The printing material is a superalloy, and its thermal conductivity is 18 W / (m·K), specific heat capacity is 435 J / (kg·K), density is 8250 kg / m³, and thermal expansion coefficient is , and the Young's modulus is 135 GPa.

[0041] Based on the above printing parameters, a thermo-mechanical coupled finite element model is constructed using multi-physics simulation software. Considering the laser scanning path and heat input effects, the instantaneous temperature field and its gradient changes of each printing layer are obtained by advancing the dynamic time step. Furthermore, the residual stress tensor field of each layer after cooling to room temperature is extracted through the thermo-elasto-plastic stress calculation module, and the stress results of the historical printing layers are accumulated to construct a residual stress history database.

[0042] Subsequently, based on the spatial coordinates A unified tensor field framework is established, and all residual stress tensor fields are spatially registered using three-dimensional bilinear interpolation. Then, a stress propagation model and a stress shadow mapping tensor are constructed using an exponential decay function and a time weighting function . By introducing a time decay control parameter , all layers are weighted and superimposed to obtain the stress projection influence area for the current printing layer.

[0043] Taking and the printing parameters as the combined input, it is input into the proposed improved reverse stress propagation neural network. The network structure consists of five parts, which respectively complete feature extraction, spatio-temporal fusion, gated state tracking, attention weighting, and output prediction. During the training process, about 2000 sets of historical printing data are used, and the Adam optimizer is used to train for 30 rounds, and the final prediction accuracy reaches 94.7%.

[0044] The output deformation prediction vector field is used to compensate the control points of the CAD model area corresponding to the printing layer. The CAD model is represented by NURBS, and each two-dimensional control point is adjusted according to the following compensation function: ; where, represents the spatial coordinate vector of the compensated control point, represents the global compensation adjustment coefficient, with a value of 0.7, represents the multi-scale adjustment coefficient, with a value of 0.3, represents the direction weight matrix; represents the deformation prediction vector, represents the th scale's weighting factor, represents the scale's average deformation prediction vector within the neighborhood, represents the total number of scales, with a value of 3.

[0045] The compensated model is converted into an STL file, and a scanning path is generated using the printing path planning module. A G-code control program is automatically generated according to the compensated path and parameters and loaded into the printing device for printing.

[0046] After printing is completed, use a laser scanner to perform a full-surface three-dimensional scan of the sample, and perform iterative closest point registration with the CAD model before compensation to obtain the actual deformation vector field. The error vector field is mainly concentrated in the Z direction, with the maximum height deviation being +0.14 mm and the minimum being -0.08 mm. Based on the error vector field, secondary compensation iteration processing was performed on the control points. After reprinting, the maximum error of the sample was reduced to ±0.035 mm, meeting the tolerance requirement of ±0.05 mm.

[0047] Table 1 Comparative data table of 3D printing deformation compensation for aerospace engines

[0048] From the analysis of Table 1 above, it can be seen that during the initial printing process without compensation, there are obvious spatial deformation errors in the component. The maximum positive deviation reaches +0.14 mm, the maximum negative deviation is -0.08 mm, and the average prediction error is 0.067 mm, significantly exceeding the set tolerance threshold of ±0.05 mm, indicating that traditional printing methods are difficult to effectively control local warping and dimensional deviation caused by residual stress. Through the aerospace engine 3D printing deformation compensation method proposed in the present invention, combined with the stress shadow mapping tensor modeling and the reverse stress propagation neural network prediction mechanism, the overall spatial deviation of the component is significantly reduced during the second printing after one compensation. The data in Table 1 shows that after compensation, the maximum positive deviation drops to +0.032 mm, the maximum negative deviation is -0.035 mm, and the overall deformation is controlled within the tolerance range of ±0.05 mm. At the same time, the average prediction error is reduced to 0.021 mm, and the prediction accuracy is significantly improved, verifying the synergistic correction effect between the deformation prediction vector field and the non-linear reparameterization of the control points. From the adjustment of the compensation parameters in Table 1, it can be seen that by introducing the error vector field to feedback correct the spatial position of the control points and reasonably adjusting the global compensation adjustment coefficient (from 0.7 to 0.6) and the multi-scale adjustment coefficient (from 0.3 to 0.4), while enhancing the local adaptability of the geometric response, the sudden change or continuity breakage of the geometric morphology is avoided, reflecting the stability and self-adaptive ability of the compensation model.

[0049] This embodiment fully demonstrates the actual application effect of the present invention in the additive manufacturing of key components of aerospace engines. By constructing a multi-physical field simulation model, establishing a stress propagation mechanism, introducing a reverse stress propagation neural network for deformation prediction, and combining with non-linear geometric reparameterization technology to finely compensate the control points of the 3D CAD model, the closed-loop control of deformation prediction before printing and error feedback after printing is effectively realized. Through actual tests, without changing the manufacturing process path, the maximum geometric error of the printed parts is significantly reduced from ±0.14 mm to ±0.035 mm, comprehensively improving the printing accuracy and the quality of the manufactured parts, and verifying the feasibility, advancement and engineering practical value of the method in the manufacturing of high-precision aerospace complex components.

[0050] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, making equivalent substitutions or changes, should be covered within the protection scope of the present invention.

Claims

1. A 3D printing deformation compensation method for a space engine, characterized in that It includes the following steps: S1. Construct a three-dimensional CAD model of aerospace engine components, and perform slicing processing on the three-dimensional CAD model to generate a set of printing layers; S2. Obtain the printing parameters of each printing layer, and based on finite element simulation, obtain the residual stress tensor field of the current printing layer and several previous historical printing layers, and construct a residual stress input data set; S3. Perform propagation modeling on the residual stress input data set in the time and space directions to construct a stress shadow mapping tensor, which is used to describe the residual stress influence area formed by the projection of historical printing layers on the current layer; S4. Input the stress shadow mapping tensor and the printing parameters of the current printing layer into a reverse stress propagation neural network model, and output the deformation prediction vector field of the current printing layer; S5. Perform geometric compensation processing on the geometric area of the corresponding printing layer in the three-dimensional CAD model according to the deformation prediction vector field, adjust the control points of the geometric area of the corresponding printing layer in the three-dimensional CAD model by using a non-linear geometric reparameterization method, and generate a compensated geometric model; S6. Generate printing codes according to the compensated geometric model, and control the printing device to perform printing; S7. After printing is completed, perform three-dimensional scanning on the printed product to obtain the actual deformation vector field, and compare the error between the actual deformation vector field and the overall predicted deformation vector field formed by superimposing the deformation prediction vector fields of each printing layer. If the error vector field is greater than the preset tolerance threshold, feedback the error information to step S5, re-perform geometric compensation processing on the geometric area of the corresponding printing layer in the three-dimensional CAD model, and repeat steps S5 to S7 until the error vector field is less than or equal to the preset tolerance threshold.

2. The 3D printing deformation compensation method for a space engine according to claim 1, wherein The printing parameters include laser power, scanning speed, scanning path, scanning spacing, powder preheating temperature, laser spot diameter, and the thermal conductivity, specific heat capacity, density, thermal expansion coefficient, and Young's modulus of the printing material.

3. A 3D printing deformation compensation method for a space engine according to claim 1, characterized in that The specific content of S2 includes: S21. Establish a thermal-mechanical coupling finite element simulation model by using the printing parameters and the geometric information of the corresponding printing layer. The construction of the thermal-mechanical coupling finite element simulation model includes geometric discretization, mesh generation, material property assignment, and boundary condition setting; S22. Based on the thermal-mechanical coupling finite element simulation model, layer by layer simulate the heat input process of the laser scanning path in the actual 3D printing process, and calculate the dynamic change process of the temperature field of each printing layer over time to obtain the instantaneous temperature distribution and temperature gradient information; S23. Use the instantaneous temperature distribution and temperature gradient information to calculate the transient thermal stress field distribution of the current printing layer during the printing process, and after each layer of laser scanning is completed and cooled to room temperature, calculate and obtain the residual stress tensor field after the printing layer is printed; S24. Perform cross-layer stress accumulation calculation on the residual stress tensor fields of several printing layers before the current printing layer, establish a residual stress tensor field historical database, and the residual stress tensor field of each printing layer in the residual stress tensor field historical database is associated with the current layer in a layer-by-layer cumulative manner; S25. Construct a residual stress input data set based on the residual stress tensor field of the current printing layer and the residual stress tensor fields of several previous historical printing layers in the residual stress tensor field historical database.

4. A 3D printing deformation compensation method for a space engine according to claim 1, characterized in that The specific steps of S3 are as follows: S31. Based on the residual stress tensor field of each printing layer in the residual stress input dataset, establish a three-dimensional space coordinate system and a printing time axis with the current printing layer as the reference. The three-dimensional coordinate points are denoted as , where represents the spatial coordinate of the printed part in the width direction, represents the spatial coordinate of the printed part in the length direction, represents the spatial coordinate of the printed part in the height direction, The direction corresponds to the stacking direction of layer-by-layer printing; S32. Under the three-dimensional space coordinate system, perform spatial registration processing on the residual stress tensor fields of each historical printing layer through three-dimensional bilinear interpolation to obtain a tensor expression with the same spatial resolution and coordinate frame as the current printing layer. S33. Based on the heat-elastoplastic mechanics propagation theory, construct a stress propagation function model: ； Among them, represents the propagation influence tensor of the th layer in the th layer, represents the residual stress tensor of the th layer at the three-dimensional coordinate point , represents the material coupling coefficient of the th layer, represents the exponential decay factor, represents the th layer to the th layer in the direction of the vertical interlayer distance; S34. Through the temporal weight decay algorithm, perform temporal weighting on the propagation influence tensor of the historical printing layer to construct a stress shadow mapping tensor , which is used to describe the spatial position of the historical printing layer in the current printing layer and the residual stress influence area formed by the projection thereon; S35. Output the stress shadow mapping tensor .

5. A 3D printing deformation compensation method for a space engine according to claim 1, characterized in that The specific steps of S4 are as follows: S41. Construct a reverse stress propagation neural network model, which includes an input layer, a spatio-temporal tensor feature fusion layer, a bidirectional gated recurrent unit layer, a tensor attention mechanism layer, and an output prediction layer. S42. Input the stress shadow mapping tensor and the printing parameters of the current printing layer into the input layer of the reverse stress propagation neural network model. The input layer converts the three-dimensional tensor data into a high-dimensional feature vector and performs data preprocessing through linear normalization. S43. The spatio-temporal tensor feature fusion layer performs multi-scale convolution operations on the high-dimensional feature vector, extracts spatial local features and inter-layer temporal features through multiple convolutional kernels respectively, and generates a spatio-temporal coupled feature expression using a feature splicing operation. S44. The bidirectional gated recurrent unit layer processes the spatio-temporal coupled feature expression, uses the gated unit mechanism to simultaneously capture the forward propagation of the influence of historical printing layers on the current printing layer and the reverse trace of the stress state of the current layer on historical layers, and obtains a bidirectional stress propagation feature representation of forward and reverse coupling. S45. The tensor attention mechanism layer generates a spatial-channel joint attention weight mask tensor based on the bidirectional stress propagation feature representation, using an attention calculation method that combines the spatial dimension and the channel dimension , where represents a three-dimensional coordinate point, represents the feature channel number; S46. After the output prediction layer fuses the spatial-channel joint attention weight mask tensor and the bidirectional stress propagation feature through tensor multiplication, it uses an activation function for non-linear mapping to output the deformation prediction vector field of the current printing layer. The deformation prediction vector field is the spatial deformation prediction vector corresponding to each three-dimensional coordinate point, representing the deformation prediction quantity of the current printing layer.

6. A 3D printing deformation compensation method for a space engine according to claim 1, characterized in that The specific steps of S5 are as follows: S51. Based on the deformation prediction vector field, extract the corresponding geometric region of the current printing layer in the three-dimensional CAD model, and locate the CAD surface fragments and boundary ranges included in the geometric region. S52. Parametrically model the geometric region of the printing layer, convert the original CAD surface segment expression into a reparameterized surface form based on non-uniform rational B-splines, and construct a two-dimensional control point array , where represents the spatial coordinate vector of the control point in the th row and th column of the original CAD surface segment; S53. According to the distribution of the deformation prediction vector field at the spatial positions of the control points, construct a control point compensation function: ； Among them, represents the spatial coordinate vector of the control points after geometric compensation, represents the global compensation adjustment coefficient, represents the direction weight matrix, represents the deformation prediction vector at the location of, represents the deformation adjustment coefficient, represents the total number of multi-scale layers, represents the scale weighting factor of, represents the average deformation prediction vector of the adjacent area at the location of at scale ; S54. Based on the spatial coordinate vectors of the control points after geometric compensation, perform surface reconstruction to form an updated expression of the corresponding region of the current printing layer in the three-dimensional CAD model after geometric compensation. S55. Perform geometric legality verification. The geometric legality verification includes constraint judgments on boundary closure, surface continuity, and that the maximum compensation amplitude does not exceed the corresponding preset tolerance. If the constraint judgment conditions are met, output the compensated geometric model.

7. A 3D printing deformation compensation method for a space engine according to claim 1, characterized in that The specific steps of S6 are as follows: S61. Convert the compensated geometric model into an intermediate model file for processing format. The intermediate model uses a patch meshing format for geometric approximation, and the file format is STL. S62. Perform path planning processing on the intermediate model, adjust the scanning spacing and path density according to the curvature distribution and structural thickness of the geometric compensation region, and generate a set of printing paths. S63. Based on the set of printing paths, generate a printing control parameter table through printing parameter mapping. S64. Generate a print control code recognizable by a standard printing device according to the print control parameter table. The print control code is an additive manufacturing control program based on the G-code structure, and includes layer-by-layer control instructions and compensation correction amount annotation information. S65. Transmit the print control code to the print device control module, and control the print device to perform additive manufacturing operations layer by layer according to the print control code.

8. A 3D printing deformation compensation method for a space engine according to claim 1, characterized in that, The specific steps of S7 are as follows: S71. After printing is completed, use a laser scanner to perform three-dimensional full-field scanning on the printed product to obtain point cloud data of the complete surface, and construct a scanning model. S72. Align the coordinates of the scanning model with the compensated three-dimensional CAD model, use the iterative closest point algorithm to achieve rigid registration, and establish a point-to-point spatial mapping relationship. S73. Calculate the actual deformation vector field based on the registration result. The actual deformation vector field is the spatial difference between corresponding points of the scanning model and the compensated three-dimensional CAD model in the unified coordinate system. S74. Accumulate the deformation prediction vector fields of each layer during the printing process layer by layer to form an overall predicted deformation vector field, and compare it with the actual deformation vector field to obtain an error vector field. S75. According to the distribution characteristics of the error vector field in the three-dimensional space, extract the error peak region, and obtain the control point indexes within the corresponding printing layer and geometric region. S76. Use the sampling value of the error vector field at the control point position as an error correction term, introduce it into the control point update mechanism, and perform secondary iterative compensation on the control points based on the previous round of compensation results. S77. Use the control points adjusted by error feedback to reconstruct the geometric region of the corresponding printing layer in the three-dimensional CAD model to generate a new geometric model, and repeat steps S5 to S7 until the error vector field is less than or equal to the preset tolerance threshold.

Images