An improved inherent strain loading method based on the idea of partitioning

Abstract

The method of the present application relates to an improved inherent strain loading method based on the idea of partitioning, comprising the following steps: for the extracted inherent strain values which are non-uniformly distributed along the scanning path, clustering and partitioning processing is performed to obtain a collection of different inherent strain classes of the components of the small-scale thermal structure standard model under specific process parameters, i.e. the inherent strain clustering analysis result of the scale model; the small-scale model inherent strain clustering analysis result is used as the basis for partitioning the structure to be printed; the structure to be printed is meshed and the centroid coordinates of each unit are obtained, and the unit is determined according to the centroid coordinates of the unit; the thermal expansion coefficient of each feature area unit of the structure to be printed is changed according to the partitioning result of the structure to be printed, so as to complete the assignment of different inherent strain classes; the calculation of the layer-by-layer partitioning inherent strain method is completed according to the actual layer-by-layer printing strategy, the deformation and stress distribution of the structure to be printed are obtained, and the prediction of the residual deformation and stress of the structure to be printed is realized.

Description

Technical fields:

[0001] This invention belongs to the field of advanced additive manufacturing technology, and more specifically relates to a method for efficiently predicting residual stress and deformation in additive manufacturing, including a clustering method for inherent strain distribution based on the concept of partitioning and a loading method. Background technology:

[0002] Additive manufacturing (AM), often referred to as 3D printing, is a technology that creates objects by adding materials layer by layer. This revolutionary manufacturing technology has dramatically changed traditional manufacturing methods by providing design freedom, improving material utilization, enabling flexible production, reducing supply chain links, minimizing assembly steps, and enabling lightweight design. Due to its unique forming characteristics, it has broad application prospects and research value in fields such as prototyping, consumer electronics, biomedicine, automotive, and aerospace. GE Aviation uses additive manufacturing technology to produce fuel nozzles for gas turbine engines, achieving a 25% weight reduction compared to traditional manufacturing methods, while also improving part strength and performance.

[0003] However, additive manufacturing also faces challenges such as product consistency, molding accuracy, and quality control. During the deposition of each layer, inconsistent printing parameters, uneven powder distribution, and differences in the coefficients of thermal expansion and shrinkage rates of different materials during heating and cooling lead to uneven thermal stress distribution, resulting in a series of molding quality problems. These uneven stress distributions and molding quality issues can cause parts to fail to meet design requirements in terms of dimensional accuracy, reduce mechanical properties such as fatigue strength and toughness, and increase the risk of failure during use.

[0004] The aforementioned problems hinder the rapid and sustained development of additive manufacturing. Therefore, studying the evolution and mechanism of thermodynamic behavior during additive manufacturing is crucial for improving the precision and quality of the final product. However, due to the high cost of materials, equipment, and processes in advanced additive manufacturing technologies, experimental research is not widespread. Simulation methods, on the other hand, can simulate the manufacturing process in a virtual environment, predict the deformation and residual stress of parts during manufacturing, and provide a basis for optimizing manufacturing processes.

[0005] There are currently two main methods for simulating macroscopic thermodynamic problems in additive manufacturing. The first is the transient thermo-coupling method, which requires establishing a sequentially coupled finite element model of thermodynamics throughout the entire process, either element-wise or layer-wise, and then predicting stress and strain distribution through transient thermo-coupling analysis. This full-process transient thermo-coupling method needs to consider heat conduction, material phase transformation, and mechanical response simultaneously, resulting in unacceptable computational costs due to its highly nonlinear nature. The second method is the intrinsic strain method, which transforms the thermodynamic problem into a purely mechanical problem. It only requires layer-wise quasi-static analysis to predict residual stress and deformation during the structural printing process. The concept of intrinsic strain first appeared in the field of metal welding and was later extended to the field of additive manufacturing to solve the problem of predicting residual stress and deformation during additive manufacturing. The intrinsic strain calculated by a small-scale model under a specific process is used as an inherent property of the material and loaded into a large-scale model in the form of pre-strain. Only layer-wise static analysis is needed to obtain the structural residual stress and deformation field, saving a lot of computational costs. However, current intrinsic strain analysis methods lack consideration for the true distribution of intrinsic strain and use the intrinsic strain averaged at each point for loading the entire model, which fails to achieve good accuracy. Therefore, in order to efficiently predict residual stress and deformation in additive manufacturing, the method of the present invention is improved based on the second method described above to take into account the actual inherent strain distribution and achieve accurate prediction of stress and deformation. Summary of the Invention

[0006] This invention addresses the shortcomings of existing inherent strain analysis methods, which fail to consider the true distribution of inherent strain and rely on averaged inherent strain at various points for full model loading, resulting in poor accuracy. Therefore, to efficiently predict residual stress and deformation in additive manufacturing, the inherent strain calculation process needs improvement to consider the true inherent strain distribution, thereby enhancing calculation accuracy and addressing the low accuracy of inherent strain loading models. This invention discloses an improved inherent strain loading method based on a partitioning approach, comprising the following steps: An improved inherent strain loading method based on a partitioning approach, comprising the following steps:

[0007] S1: Design a small-scale thermal structure standard model of the structure to be printed, extract the inherent strain values ​​of each node of the small-scale thermal structure model along the scanning path under specific process parameters, and determine its distribution law.

[0008] S2: For the extracted inherent strain values ​​that are not uniformly distributed along the scanning path, clustering and partitioning are performed to obtain a set of different inherent strain classes of the components of the small-scale thermal structure standard model under specific process parameters, i.e., the clustering analysis results of the inherent strain of the scale model.

[0009] S3: Based on the results of the small-scale model inherent strain clustering analysis, the structure to be printed is partitioned; the structure to be printed is meshed and the centroid coordinates of each element are obtained; the partition to which the element belongs is determined based on the centroid coordinates of the element.

[0010] S4: Change the thermal expansion coefficient of each feature region unit of the structure to be printed according to the partitioning result of the structure to be printed, so as to complete the assignment of different inherent strain types;

[0011] S5: Perform layer-by-layer partitioning inherent strain method calculations according to the actual layer-by-layer printing strategy to obtain the deformation and stress distribution of the structure to be printed, and realize the prediction of the residual deformation and stress of the structure to be printed.

[0012] Furthermore, the standard model of the small-scale thermal structure is a finite element model of a three-layer deposition layer with a substrate.

[0013] Furthermore, the inherent strain values ​​of each node along the scanning path of the small-scale thermal structure model under specific process parameters are obtained by obtaining the inherent strain values ​​of each node along the scanning direction of the model through element-by-element transient thermo-mechanical coupling calculation.

[0014] Furthermore, the clustering partitioning is performed using the K-means method for cluster analysis, and the specific process is as follows:

[0015] Based on the magnitude of the inherent strain value, an appropriate number of class centers is selected to classify the extracted inherent strain at each point along the scanning direction, resulting in inherent strain values ​​for different feature regions. The classification results are as follows: the inherent strain of each unit in the region within 6 mm on both sides of the scanning direction path is classified as contour inherent strain; the inherent strain of each unit in the middle region of the scanning direction path is classified as center inherent strain.

[0016] Furthermore: the process of partitioning the structure to be printed based on the results of small-scale model inherent strain clustering analysis is as follows:

[0017] First, the structure to be printed is divided into mesh elements. Then, the distance between the centroid coordinates of the element and the boundary of the structure is calculated. If the distance is ≤6mm, it belongs to the contour element; otherwise, it is the center element.

[0018] Furthermore: the process of changing the coefficient of thermal expansion of each feature region unit of the structure to be printed according to the partitioning result of the structure to be printed, in order to complete the assignment of different inherent strain types, is as follows:

[0019] Different types of inherent strain are equivalent to the inherent properties of materials under specific process parameters, and inherent strain is introduced in the form of changing the coefficient of thermal expansion of the material.

[0020] Furthermore: the actual calculation is performed using the layer-by-layer partitioning inherent strain method according to the actual layer-by-layer printing strategy to obtain the deformation and stress distribution of the structure to be printed, and the prediction process of the residual deformation and stress of the structure to be printed is as follows:

[0021] First, after imparting inherent strain, all units of the structure to be printed are deactivated to simulate the state where the material does not exist at the beginning of actual additive manufacturing. Then, the units of the structure to be printed are activated layer by layer, and a quasi-static mechanical analysis is performed until the printing process ends, so as to obtain the residual stress and deformation field distribution of the structure to be printed.

[0022] This invention improves the loading of inherent strain based on a partitioning approach. It considers the true inherent strain distribution of additively manufactured structural components and clusters the non-uniformly distributed inherent strain along the laser scanning path in a small-scale standard model according to its numerical value, obtaining a set of inherent strain classes for characteristic regions along the scanning path. Then, based on the small-scale characteristic region standard, the structural units to be printed are partitioned, and corresponding inherent strains are assigned to the partitioned characteristic region units. This improves the prediction accuracy of the inherent strain model without significantly increasing computational costs. Because it is a method based on the application of clustering and partitioning ideas from the field of data processing, it is called an improved inherent strain loading method based on the partitioning approach.

[0023] The beneficial effects of this invention are:

[0024] (1) The present invention describes the design of a small-scale thermal structure model for observing the inherent strain distribution characteristics and extracting data for application to the structure to be printed, which considers a more realistic inherent strain distribution. Traditional structural models, in order to simplify the calculation process and reduce the calculation cost, usually simply assume that each node of the small-scale thermal structure model experiences the same thermal history, and thus consider the inherent strain to be a uniform and fixed value. Although this assumption is simple to implement, it greatly reduces the accuracy of the calculation. The present invention improves this by increasing the accuracy of predicting residual stress and deformation in additive manufacturing.

[0025] (2) The present invention uses clustering and partitioning of non-uniform inherent strain data, rather than directly applying the extracted data to the structure to be printed. This avoids the problem that the method of the present invention is not applicable to other different structural models. While loading according to the extracted inherent strain distribution can improve accuracy to a certain extent, it is only applicable to a specific model designed.

[0026] (3) The data clustering method described in this invention effectively classifies large amounts of unordered data based on the characteristics of interest, in order to achieve a specific need. After clustering the data, it can not only better restore the characteristics of the true inherent strain distribution, but also simplify the calculation procedure, making its function universal.

[0027] (4) The partitioning of the structural unit to be printed based on the above data clustering results described in this invention is achieved by calculating the distance between the centroid coordinates of the current unit and the structural outline, and determining which type of inherent strain the current unit belongs to. Attached Figure Description

[0028] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0029] Figure 1 This is a flowchart illustrating an embodiment of the present invention;

[0030] Figure 2 Schematic diagram of a three-layer small-scale standard model mesh;

[0031] Figure 3 This is a diagram showing the inherent strain distribution along the laser scanning path of a small-scale model.

[0032] Figure 4 A schematic diagram of the structure to be printed;

[0033] Figure 5 The deformation distribution results are for a ten-layer thin-walled structure;

[0034] Figure 6 The stress distribution results are for a ten-layer thin-walled structure.

[0035] Figure 7 Comparison of residual deformation between transient thermo-coupling model and partitioned intrinsic strain model; Detailed Implementation

[0036] It should be noted that, unless otherwise specified, the embodiments and features in the embodiments of the present invention can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and embodiments.

[0037] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The following description of at least one exemplary embodiment is merely illustrative and is in no way intended to limit the present invention or its application or use. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0038] As described in the background section, the inherent strain method transforms thermodynamic problems into purely mechanical problems, requiring only layer-by-layer quasi-static analysis to predict residual stress and deformation during the structural printing process, thus saving significant computational costs. It uses the inherent strain calculated by a small-scale model under a specific process as an inherent property of the material, which is then loaded into a large-scale model in the form of pre-strain. To achieve the desired technical effect, this application designs a novel inherent strain loading mode and summarizes its partitioned loading method.

[0039] Figure 1 This is a flowchart of an improved inherent strain loading method based on the partitioning concept of the present invention.

[0040] S1: Design a small-scale thermal structure standard model of the structure to be printed, extract the inherent strain values ​​of each node of the small-scale thermal structure model along the scanning path under specific process parameters, and determine its distribution law.

[0041] S2: For the extracted inherent strain values ​​that are not uniformly distributed along the scanning path, clustering and partitioning are performed to obtain a set of different inherent strain classes of the components of the small-scale thermal structure standard model under specific process parameters, i.e., the clustering analysis results of the inherent strain of the scale model.

[0042] S3: Based on the results of the small-scale model inherent strain clustering analysis, the structure to be printed is partitioned; the structure to be printed is meshed and the centroid coordinates of each element are obtained; the partition to which the element belongs is determined based on the centroid coordinates of the element.

[0043] S4: Change the thermal expansion coefficient of each feature region unit of the structure to be printed according to the partitioning result of the structure to be printed, so as to complete the assignment of different inherent strain types;

[0044] S5: Perform layer-by-layer partitioning inherent strain method calculations according to the actual layer-by-layer printing strategy to obtain the deformation and stress distribution of the structure to be printed, and realize the prediction of the residual deformation and stress of the structure to be printed.

[0045] The steps S1 / S2 / S3 / S4 / S5 are executed sequentially. The inherent strain method analysis process is as follows: the inherent strain under specific process parameters of the node is obtained from the small-scale standard model. Then, the inherent strain is equivalent to the inherent property of the material (thermal expansion coefficient) and introduced into the structure to be printed. By applying temperature load, the residual stress and deformation field are calculated in the form of thermal strain.

[0046] A set of different inherent strain types is introduced into the structure to be printed to predict stress and deformation. Inherent strain is introduced by changing the material's coefficient of thermal expansion. The specific process is as follows:

[0047] First, the structure to be printed is divided into mesh cells. The cells are traversed to calculate the distance from the centroid coordinates of the cells to the boundary, and the partition to which the cells belong is determined.

[0048] Secondly, units within 6mm of the boundary of the structure to be printed are given contour-specific strain; otherwise, they are given center-specific strain. Applying inherent strain here means changing the coefficient of thermal expansion of the material in each feature region.

[0049] Finally, each element is activated layer by layer and a unit temperature load is applied. The resulting thermal strain is the equivalent inherent strain. After introducing the inherent strain, a quasi-static analysis is performed to predict the residual stress and deformation of the structure.

[0050] Furthermore, for S1: the small-scale standard model described is a typical three-layer deposition structure containing a substrate, such as... Figure 2 As shown. This transient thermo-mechanical coupling calculation is generally performed element by element, activating and completing the coupling calculation of the temperature field and stress field one by one until the printing simulation ends.

[0051] Furthermore, the inherent strain values ​​of each node along the scanning path of the small-scale thermal structure model under specific process parameters are obtained by obtaining the inherent strain values ​​of each node along the scanning direction of the model through transient thermo-mechanical coupling calculation of each element.

[0052] After obtaining the residual strain field, the inherent strain is calculated. The residual strain field obtained after completing the thermo-mechanical coupling analysis of the standard model is not the final strain value, but rather needs to consider the elastic strain that cannot be fully recovered due to the mutual constraints between layers. Therefore, the formula for calculating the inherent strain value is as follows:

[0053]

[0054] Where t1 and t2 represent the time corresponding to the intermediate state and steady state of the point of interest during additive manufacturing, respectively. Let these represent the plastic strain and elastic strain at the location of interest in the intermediate state, respectively. This represents the elastic strain at the point of interest in steady state. The intermediate state is the state at which the point first reaches its temperature peak, and the steady state is the state at which the point cools down to ambient temperature.

[0055] Furthermore, for S3: after extracting the data of each node along the scanning path and calculating the inherent strain, the following is obtained: Figure 3 The distribution shown is by Figure 3It is known that the distribution of inherent strain along the scanning path is not uniform. For this non-uniformly distributed inherent strain value, K-means cluster analysis is used to obtain the distribution characteristics of the inherent strain at each node along the scanning path and the set of inherent strain classes in the characteristic regions. The K-means cluster analysis process is as follows: based on the magnitude of the inherent strain value, an appropriate number of class centers are selected to classify the extracted inherent strain at each point along the scanning direction, obtaining inherent strain values ​​for different characteristic regions. The classification results are: the inherent strain class of each unit within 6 mm on both sides of the scanning path is the contour inherent strain; the inherent strain class of each unit in the middle region of the scanning path is the central inherent strain.

[0056] The goal of K-means clustering is to make each sample closer to the centroid of its own cluster than to the centroids of other clusters. The following formula is the objective function of the program:

[0057]

[0058] Here: k is the number of cluster centers, C i Let x be the set of data belonging to each class, and μ be the data point. i At the center of each class, ||x-μ i ‖ represents the intersection of point x and point μ i The Euclidean distance between them.

[0059] Initially, we select K points from the sample to serve as initial cluster centers. Then, we calculate the distance from each sample to each cluster center and select the cluster center with the smallest distance as the cluster center for that data point. This process is repeated until all data points are classified. However, the classification performance is easily affected by the initial cluster center values. Therefore, we use the mean of each class as the new cluster center and repeat the above steps to reduce the influence of the initial value selection.

[0060] In K-means clustering, determining the appropriate number of clusters k is an important issue. The commonly used method is the "elbow rule".

[0061] The elbow rule finds the point where the rate of decrease in SSE is significantly slowed down by plotting the total sum of squared errors (SSE) of clustering as a function of k, and uses this point as a reasonable k.

[0062] The formula for calculating the total sum of squared errors (SSE) is:

[0063]

[0064] Here: k is the number of cluster centers, x j It is the j-th data point, C i For each class, there is a set of data. ‖x-μ i‖ represents the intersection of point x and point μ i The Euclidean distance between them.

[0065] The process of partitioning the structure to be printed based on the results of small-scale model inherent strain clustering analysis is as follows:

[0066] First, the structure to be printed is divided into mesh elements. Then, the distance between the centroid coordinates of the element and the boundary of the structure is calculated. If the distance is ≤6mm, it belongs to the contour element; otherwise, it is the center element.

[0067] Furthermore, for the data that has already been clustered, we can obtain the classification results of the inherent strain values ​​as follows: the inherent strain of each unit in the region within 6 mm on both sides of the scanning direction path is the profile inherent strain, which is generally positive; the inherent strain of each unit in the middle region of the scanning direction path is the center inherent strain.

[0068] Further: The structure to be printed is divided into mesh cells, and the distance of the centroid coordinates of each cell from the boundary is calculated by traversing the cells to determine the partition to which the cell belongs. Cells within 6mm of the boundary of the structure to be printed are assigned profile inherent strain; otherwise, they are assigned center inherent strain.

[0069] The process of changing the coefficient of thermal expansion of each feature region unit of the structure to be printed according to the partitioning result of the structure to be printed, so as to complete the assignment of different inherent strain types, is as follows:

[0070] Different inherent strain types are equated to the inherent properties of materials under specific process parameters, and inherent strain is introduced by changing the material's coefficient of thermal expansion. Further, the inherent strain value is equated to the material's coefficient of thermal expansion, and this inherent strain field is introduced by increasing the unit temperature of each element. Elements are activated layer by layer and a unit thermal load is applied, and quasi-static calculations are performed layer by layer to obtain the final residual stress and deformation field.

[0071] For pre-strain loading, in practice, the inherent strain value is generally set as the thermal expansion coefficient of the material, and the inherent strain is applied by generating thermal strain through a unit temperature load on the active element.

[0072] ε I =ε

[0073] ε=αΔT

[0074] In the formula ε I ε is the inherent strain, α is the equivalent thermal expansion coefficient, ε is the equivalent thermal strain, and ΔT is the temperature change.

[0075] The actual process involves performing layer-by-layer partitioning inherent strain method calculations according to the actual layer-by-layer printing strategy to obtain the deformation and stress distribution of the structure to be printed, and to predict the residual deformation and stress of the structure to be printed. The process is as follows:

[0076] First, after imparting inherent strain, all units of the structure to be printed are deactivated to simulate the state where the material does not exist at the beginning of actual additive manufacturing. Then, the units of the structure to be printed are activated layer by layer, and a quasi-static mechanical analysis is performed until the printing process ends, so as to obtain the residual stress and deformation field distribution of the structure to be printed.

[0077] Furthermore, to make this application more convincing, a specific case is provided here, and the calculation results of the case applying the method of the present invention are explained and compared with the detailed thermal structure method.

[0078] Example 1:

[0079] In this embodiment of the invention, the model to be printed using this method is a ten-layer thin-walled structure with dimensions of 100mm * 6.4mm * 10mm, and the mesh element size in the finite element model is 2mm * 3.2mm * 1mm. The standard model at a smaller scale is a three-layer depositional structure, and the finite element mesh model is as follows... Figure 2 As shown. The grid cell size is also 2mm*3.2mm*1mm.

[0080] First, the inherent strain values ​​at each point need to be obtained through element-by-element small-scale transient thermo-mechanical coupling calculations. The distribution of inherent strain along the scanning path is as follows: Figure 3 As shown.

[0081] For these non-uniformly distributed inherent strain values, K-means cluster analysis was performed. Using the magnitude of the inherent strain value as the classification criterion, an appropriate number of cluster centers were selected to classify the extracted inherent strain values ​​at each point along the scanning direction, resulting in inherent strain values ​​for different characteristic regions. The classification results showed that the inherent strain of each unit within 6 mm on both sides of the scanning path was classified as contour inherent strain, generally positive; the inherent strain of each unit in the middle region of the scanning path was classified as central inherent strain.

[0082] After obtaining the inherent strain of the partition, it is applied to the structure to be printed to achieve stress-deformation prediction. The structure to be printed is as follows: Figure 4 As shown, the structure to be printed is meshed, and the distance from the centroid coordinates of each cell to the boundary is calculated by traversing the cells to determine the partition to which the cell belongs. Cells within 6mm of the boundary of the structure to be printed are assigned profile inherent strain; otherwise, they are assigned center inherent strain.

[0083] Finally, the inherent strain value is equivalent to the thermal expansion coefficient of the material, and this inherent strain field is introduced by increasing the unit temperature of each element. The elements are activated layer by layer and a unit thermal load is applied. Quasi-static calculations are performed layer by layer to obtain the residual stress and deformation field of the structure to be printed.

[0084] The deformation and stress of the structure to be printed are as follows: Figure 3 , Figure 4 As shown, the maximum deformation occurs in the top region of the structure's side and is symmetrically distributed at both ends of the structure. In this example, the scanning direction is the X direction, and the results show that the stress in the X direction is greater compressive stress on both sides, and greater tensile stress in the top and middle regions.

[0085] Figure 5 The deformation distribution results are for a ten-layer thin-walled structure;

[0086] Figure 6 The stress distribution results are for a ten-layer thin-walled structure.

[0087] To illustrate the effectiveness of the improved inherent strain loading method, the printed structure was analyzed using both the full-process transient thermo-mechanical coupling method and the traditional inherent strain method. Along... Figure 4 The paths extract the X-direction deformation values ​​at each point under the three calculation methods: thermo-coupling, inherent strain, etc., as follows: Figure 7 As shown, the deformation results obtained by the two methods are basically consistent. However, the calculation results of the traditional inherent strain method differ significantly from those of the full-process transient thermo-mechanical coupling calculation. The improved inherent strain loading method improves the calculation accuracy by 20.15% compared to the traditional calculation method. Furthermore, the calculation time of the improved inherent strain method is much shorter than that of the full-process transient thermo-mechanical coupling calculation. This demonstrates that the method of the present invention has the advantages of high accuracy and low computational cost.

[0088] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the embodiments have described the present invention in detail, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. An improved inherent strain loading method based on the partitioning concept, characterized in that: Includes the following steps: S1: Design a small-scale thermal structure standard model of the structure to be printed, extract the inherent strain values ​​of each node of the small-scale thermal structure standard model along the scanning path under specific process parameters, and determine its distribution law. S2: For the extracted inherent strain values ​​that are not uniformly distributed along the scanning path, clustering and partitioning are performed to obtain a set of different inherent strain classes of the components of the small-scale thermal structure standard model under specific process parameters, which is the clustering analysis result of the inherent strain of the small-scale model. S3: The structure to be printed is partitioned based on the results of small-scale model inherent strain clustering analysis. The structure to be printed is meshed and the centroid coordinates of each element are obtained. The partition to which the element belongs is determined based on the centroid coordinates. S4: Change the thermal expansion coefficient of each feature region unit of the structure to be printed according to the partitioning result of the structure to be printed, so as to complete the assignment of different inherent strain types; S5: Perform layer-by-layer partitioning inherent strain method calculations according to the actual layer-by-layer printing strategy to obtain the deformation and stress distribution of the structure to be printed, and realize the prediction of the residual deformation and stress of the structure to be printed.

2. The improved inherent strain loading method based on the partitioning concept according to claim 1, characterized in that: The standard model of the small-scale thermal structure is a finite element model of a three-layer deposition layer with a substrate.

3. The improved inherent strain loading method based on the partitioning concept according to claim 1, characterized in that: The inherent strain values ​​of each node along the scanning path of the standard model of small-scale thermal structure under specific process parameters are obtained by obtaining the inherent strain values ​​of each node along the scanning direction of the model through transient thermo-mechanical coupling calculation of each element.

4. The improved inherent strain loading method based on the partitioning concept according to claim 1, characterized in that: The clustering partitioning was performed using the K-means method, and the specific process is as follows: Based on the magnitude of the inherent strain value, an appropriate number of class centers is selected to classify the extracted inherent strain at each point along the scanning direction, resulting in inherent strain values ​​for different feature regions. The classification results are as follows: the inherent strain of each unit in the region within 6 mm on both sides of the scanning direction path is classified as contour inherent strain; the inherent strain of each unit in the middle region of the scanning direction path is classified as center inherent strain.

5. The improved inherent strain loading method based on the partitioning concept as described in claim 1, characterized in that: The process of partitioning the structure to be printed based on the results of small-scale model inherent strain clustering analysis is as follows: First, the structure to be printed is divided into mesh cells. Then, the distance between the centroid coordinates of the cells and the boundary of the structure to be printed is calculated. If the distance is ≤6mm, it belongs to the contour cell; otherwise, it is the center cell.

6. The improved inherent strain loading method based on the partitioning concept as described in claim 1, characterized in that: The process of changing the coefficient of thermal expansion of each feature region unit of the structure to be printed according to the partitioning result of the structure to be printed, so as to complete the assignment of different inherent strain types, is as follows: Different types of inherent strain are equivalent to the inherent properties of materials under specific process parameters, and inherent strain is introduced in the form of changing the coefficient of thermal expansion of the material.

7. The improved inherent strain loading method based on the partitioning concept as described in claim 1, characterized in that: The process of calculating the inherent strain method layer by layer according to the actual layer-by-layer printing strategy, obtaining the deformation and stress distribution of the structure to be printed, and predicting the residual deformation and stress of the structure to be printed is as follows: First, after imparting inherent strain, all units of the structure to be printed are deactivated to simulate the state where the material does not exist at the beginning of actual additive manufacturing. Then, the units of the structure to be printed are activated layer by layer, and a quasi-static mechanical analysis is performed until the printing process ends, so as to obtain the residual stress and deformation field distribution of the structure to be printed.

Images