Gradient lattice structure generation and optimization method and device for laser additive manufacturing

Abstract

The invention belongs to the technical field of additive manufacturing, and discloses a gradient lattice structure generation and optimization method and device for laser additive manufacturing. Comprising the steps of collecting design demand data; according to the design demand data, generating a gradient lattice structure by adopting an improved evolutionary algorithm; evaluating whether the generated gradient lattice structure reaches the standard or not; if the gradient lattice structure does not reach the standard, optimizing the gradient lattice structure; according to the method, the problem that an existing method depends on manual modeling is solved, multi-objective optimization of the gradient lattice structure is achieved, the excellent gradient lattice structure meeting the complex design requirement is effectively generated, the design cycle period can be shortened, the product performance can be improved, and more efficient gradient lattice structure design is achieved in the field of laser additive manufacturing.

Description

Technical Field

[0001] The present invention relates to the technical field of additive manufacturing, and more specifically, to a method and device for generating and optimizing a gradient lattice structure for laser additive manufacturing. Background Art

[0002] With the continuous advancement of manufacturing technology, laser additive manufacturing (LAM), as an emerging manufacturing method, has gradually attracted widespread attention; this technology has the characteristics of high manufacturing flexibility, high material utilization and large design freedom, and can effectively meet the needs of modern industry for complex gradient lattice structures and personalized products; in many applications, gradient lattice structures have become a research hotspot due to their excellent mechanical properties and lightweight characteristics; gradient lattice structures can effectively optimize product performance and improve usage efficiency by realizing different physical and mechanical properties in materials; however, how to efficiently generate and optimize these complex gradient lattice structures in the laser additive manufacturing process still faces many challenges; existing methods often lack flexibility and precision for specific applications, and it is difficult to achieve ideal gradient lattice structure performance and manufacturing efficiency.

[0003] For example, the Chinese patent with publication number CN111451505A discloses a laser selective melting preparation process for a metal lattice structure variable density gradient material; the process includes: selecting a metal spherical powder suitable for laser selective melting as a raw material; using a three-dimensional modeling software to construct the required variable density gradient lattice structure model, slicing the model with a dedicated software and then importing it into a laser selective melting forming device; setting appropriate laser selective melting forming process parameters, and the entire forming process is carried out under an argon or nitrogen atmosphere; after the processing is completed, the lattice structure and the substrate are annealed, and after annealing, the lattice structure is wire cut, the surface is cleaned, and sandblasted to obtain the material; the variable density gradient lattice structure obtained by this invention has good surface quality and high specific strength, and while ensuring the functionality and reliability of the gradient lattice structure, it also meets the demand for lightweight gradient lattice structure; although the above method can realize the generation of gradient lattice structure, the inventor has found through research and practical application of the above method and the prior art that the above method and the prior art have at least the following defects:

[0004] (1) It relies on manual modeling, which is limited in complexity and flexibility. For gradient lattice structures with complex gradient changes or nonlinear gradients, manual modeling is very difficult and time-consuming.

[0005] (2) It is only applicable to limited optimization objectives (such as strength and lightweight), but it is difficult to consider multiple complex performance indicators (such as mechanical properties, materials, shape, etc.) at the same time, making multi-objective optimization difficult;

[0006] (3) If the performance of the manually designed gradient lattice structure does not meet the standard, the design needs to be manually adjusted again, which increases the design cycle;

[0007] In view of this, the present invention proposes a method and device for generating and optimizing a gradient lattice structure for laser additive manufacturing to solve the above-mentioned problems. Summary of the invention

[0008] In order to overcome the above-mentioned defects of the prior art and achieve the above-mentioned purpose, the present invention provides the following technical solution: a method for generating and optimizing a gradient lattice structure for laser additive manufacturing, comprising:

[0009] Collect design requirement data;

[0010] According to the design requirement data, the improved evolutionary algorithm is used to generate the gradient lattice structure;

[0011] Evaluate whether the generated gradient lattice structure meets the standards;

[0012] If the gradient lattice structure does not meet the standard, the gradient lattice structure is discretized into multiple finite element units, and the density of each unit is initially set to the maximum value; the goal is to minimize the structural flexibility, and the volume of the unit is set not to exceed the maximum allowable volume; the unit density value is iteratively adjusted through analytical sensitivity, and the distribution is optimized by smoothing processing, and the volume constraint is repeatedly checked; if not, the density value is readjusted until the result converges, and the optimized gradient lattice structure is determined.

[0013] Furthermore, the steps of using the improved clonal selection algorithm to generate a gradient lattice structure include:

[0014] Step 1: Collect m structure samples, set different numerical labels for different structure samples, and mark them as structure labels, where m is an integer greater than 1;

[0015] Step 2: Randomly generate an initial antibody group, where the antibodies in the initial antibody group correspond one to one with the structural tags;

[0016] Step 3: Determine the fitness function;

[0017] Step 4: Calculate the fitness of each antibody in the initial antibody group, and perform antibody screening to generate a memory antibody group;

[0018] Step 5: Calculate the corresponding clone size for each antibody in the memory antibody group, and clone it to generate a clone antibody group;

[0019] Step 6: Calculate the selection probability of each antibody in the cloned antibody group, and determine whether to select the antibody based on the selection probability to generate a selected antibody group;

[0020] Step 7: Perform cloud adaptive mutation on each antibody in the selected antibody group using a cloud adaptive mutation operator to generate a mutant antibody group;

[0021] Step 8: Using an interpolation method to generate new antibodies for the antibodies in the variant antibody group to generate an interpolated antibody group;

[0022] Step 9: Recombining the antibodies in the variant antibody group to generate a recombinant antibody group;

[0023] Step 10: merging the interpolated antibody group and the recombinant antibody group to generate a merged antibody group;

[0024] Step 11: Determine whether the iteration is finished; if so, calculate the fitness of each antibody in the combined antibody group, and obtain the structural sample corresponding to the structural label of the antibody with the largest fitness; if not, perform antibody screening on the combined antibody group, regenerate the memory antibody group, and return to step 5.

[0025] Furthermore, in step 1, the structure sample is a gradient lattice structure sample;

[0026] In step 2, the initial antibody population , is the qth antibody, that is, the size of the initial antibody population is q; the range of antibodies is m structural samples; the number of iterations t of the initial antibody population is 0;

[0027] In step 4, the method for antibody screening is: screening out The antibody with the highest fitness, ;

[0028] In step 5, the clone size is calculated by combining the fitness of the antibody with the affinity of the antibody, and the calculation is as follows:

[0029] ;

[0030] ;

[0031] In the formula, is the clone size of the i-th antibody, Int is the upward rounding function, q is the initial antibody population size, is the affinity of the ith antibody, is the fitness of the ith antibody, min is the minimum function, exp is the natural exponential function, is the Euclidean distance between the ith antibody and the jth antibody, , .

[0032] Further, the design requirement data includes additive manufacturing objects, mechanical requirement data, material requirement data, shape requirement data and lightweight requirement data; the additive manufacturing objects include product types, functional requirements and use environments; the mechanical requirement data include stress distribution, load conditions and stiffness and flexibility requirements; the material requirement data include material types and material distribution; the shape requirement data include geometric dimensions, geometric shapes and local density gradients; the lightweight requirement data include density distribution and weight distribution;

[0033] In step 3, the expression of the fitness function is: ; In the formula, For fitness, is the demand satisfaction; the method for obtaining the demand satisfaction is: different digital labels are set for the data that are not numerical values ​​in the design demand data, and marked as demand labels, wherein the digital labels of different data are different; the data that are not numerical values ​​in the design demand data are replaced with corresponding demand labels, and the replaced design demand data are marked as replacement data; the replacement data and the structural label corresponding to the antibody are marked as analysis data, and the analysis data is input into the trained demand prediction model to predict the corresponding demand satisfaction.

[0034] Furthermore, in step 6, the method for generating the selected antibody group includes:

[0035] Calculate the fitness of each antibody in the cloned antibody group, and add them up in sequence to obtain the total antibody fitness, divide the fitness of each antibody by the total antibody fitness, and obtain the selection probability of each antibody; sort the selection probability of each antibody from large to small, and replace each selection probability in order according to the positive order with the sum of each selection probability and all selection probabilities before the corresponding selection probability; obtain all selection probabilities after replacement and mark them as replacement probabilities; the antibody corresponding to the replacement probability is consistent with the antibody corresponding to the selection probability before replacement; take every two adjacent replacement probabilities as an analysis set according to the positive order; mark the replacement probability with a large value in each analysis set as the first probability, and mark the replacement probability with a small value as the second probability; construct the probability screening range corresponding to each analysis set according to the first probability and the second probability of each analysis set, wherein the first probability is the maximum value of the probability screening range, the second probability is the minimum value of the probability screening range, and the minimum value is an open interval, and the maximum value is a closed interval; each analysis set corresponds to the antibody corresponding to the second probability; wherein the minimum value of the probability screening range corresponding to the first antibody is 0, and the maximum value is the corresponding selection probability; randomly generate indivual A random number between The probability screening range of the random numbers corresponds to the antibodies to generate a selection antibody group;

[0036] In step 8, the method for generating an interpolated antibody group includes:

[0037] Calculate the fitness of each antibody in the variant antibody group, and sort them from large to small to generate a sorting table; select half of the antibodies in the variant antibody group in positive order according to the sorting table, and mark them as screening antibodies; use the interpolation method to generate new antibodies for every two adjacent screening antibodies; calculate the fitness corresponding to each new antibody, and add it to the sorting table, retain the antibodies corresponding to the first y fitness in the sorting table, and generate an interpolated antibody group, where y is the number of antibodies in the variant antibody group;

[0038] The new antibody is obtained by weighted fusion of two adjacent screening antibodies with random weights. The expression of the new antibody is: ; In the formula, For new antibodies, , For two adjacent screening antibodies, is the interpolation coefficient, for A random number between .

[0039] Furthermore, in step 9, the method for antibody recombination comprises:

[0040] Randomly select Y antibodies from the mutant antibody group as parent antibodies for antibody recombination to generate progeny antibodies , until the number of unselected antibodies in the mutant antibody group is less than Y, the antibody recombination is completed;

[0041] Progeny Antibodies Generated by the weighted average of Y randomly selected parent antibodies, the calculation method includes:

[0042] ;

[0043] In the formula, is the Yth parent antibody, is the Yth scale factor, the scale factor is a randomly generated real number and not all Y scale factors are 0;

[0044] In step 11, the method for determining whether the iteration is completed includes:

[0045] Preset an iteration threshold; compare the number of iterations t with the iteration threshold; if the number of iterations t is greater than or equal to the iteration threshold, the iteration ends; if the number of iterations t is less than the iteration threshold, mark the maximum fitness in the combined antibody group as the single maximum fitness; subtract the single maximum fitness in the previous iteration from the single maximum fitness in this iteration and take the absolute value as the fitness change;

[0046] Preset variable threshold and quantity threshold; compare the fitness change with the variable threshold. If the fitness change is greater than or equal to the variable threshold, no smoothing instruction is generated; if the fitness change is less than the variable threshold, a smoothing instruction is generated; count the number of smoothing instructions generated and compare it with the quantity threshold; if the number of smoothing instructions is less than the quantity threshold, the iteration continues and the number of iterations is set to , and select the one with the largest fitness among the combined antibody groups antibodies form a new memory antibody group; if the number of smooth instructions is greater than or equal to the threshold, the iteration ends.

[0047] Furthermore, the method for evaluating whether the generated gradient lattice structure meets the standards includes:

[0048] Finite element analysis is performed on the generated gradient lattice structure using finite element software to obtain corresponding performance data; the performance data includes mechanical data and lightweight data, wherein the data in the mechanical data is consistent with the data in the mechanical demand data, and the data in the lightweight data is consistent with the data in the lightweight demand data; the performance data, mechanical demand data, and lightweight demand data are used as test data, the test data are input into the trained structural evaluation model, and evaluation labels are output, and corresponding evaluation results are obtained according to the evaluation labels, and whether the generated gradient lattice structure meets the standards is evaluated according to the evaluation results; the evaluation labels are digital labels corresponding to the evaluation results, and the evaluation results include meeting the standards and not meeting the standards, and different evaluation results correspond to different digital labels; the structural evaluation model is trained based on the test data and the evaluation labels;

[0049] The training process of the structure evaluation model includes:

[0050] Convert performance data, mechanical requirement data, and lightweight data into numerical features to ensure that the MLP can process continuous inputs, collect training samples, and the number should be ≥ 1000; convert the judgment results "meet the standard / not meet the standard" into binary classification labels (0 / 1); divide the training samples into training set and test set according to 7:3.

[0051] The training set is used to train the structural evaluation model. The structural evaluation model takes the evaluation label corresponding to each group of test data as output, and the actual evaluation label corresponding to each group of test data as the prediction target, and the actual evaluation label is the pre-collected evaluation label corresponding to the test data; the mean absolute percentage error MAPE is used to evaluate the model accuracy of the prediction results. When the calculated MAPE is less than the preset MAPE, the structural evaluation model training is completed, and a structural evaluation model that predicts the evaluation label according to the test data is generated; wherein the structural evaluation model is a deep belief network model.

[0052] Furthermore, the step of optimizing the gradient lattice structure includes:

[0053] Step a: Discretize the gradient lattice structure into finite element units, obtain the volume of each unit, and set the initial density value for each unit is 1; ;

[0054] Step b: define the objective function and volume constraints;

[0055] Step c: Calculate the sensitivity of the current gradient lattice structure;

[0056] Step d: Update the density value of each cell , and perform smoothing operations;

[0057] Step e: After calculating the density value update, the volume of each unit in the current gradient lattice structure is determined to determine whether the volume constraint is satisfied. If not, the density value of each unit is updated. Make adjustments and then go to step f. If satisfied, go directly to step f.

[0058] Step f: Calculate the sensitivity of the current gradient lattice structure and determine whether the optimization is completed. If not, return to step d; if yes, obtain the current gradient lattice structure as the optimized gradient lattice structure;

[0059] In step b, the expression of the objective function is: ; In the formula, For softness, is the stiffness matrix, is the displacement vector, is the transpose of the displacement vector;

[0060] The expression of volume constraint is: ; In the formula, is the volume of the gradient lattice structure, For the The volume of a unit, is the maximum permissible volume of the gradient lattice structure, , M is the number of units in the gradient lattice structure.

[0061] Furthermore, in the step c, the method for calculating the sensitivity of the current gradient lattice structure includes: using the finite element method to calculate the displacement vector and stiffness matrix of the current gradient lattice structure, substituting them into the objective function, calculating the corresponding compliance, and calculating the sensitivity according to the compliance;

[0062] The expression of sensitivity is: ; In the formula, is the sensitivity, is the penalty factor, For the The flexibility of each unit;

[0063] In step d, the density value of each unit is updated The method is: ; In the formula, For the During the optimization process The density value of each unit, is the gradient coefficient, For the The sensitivity of each unit, ;

[0064] In the step e, the method for determining whether the volume constraint is satisfied is: substituting the volume of each unit in the current gradient lattice structure after the density value is updated into the expression of the volume constraint, if the expression of the volume constraint is established, the volume constraint is satisfied; if the expression of the volume constraint is not established, the volume constraint is not satisfied;

[0065] The density value for each cell The adjustment is done by calculating the scaling factor , scaling factor The expression is: ; The density value of each unit Multiply by the scaling factor , get the adjusted density value of each unit ;

[0066] In step f, the method for determining whether the optimization is finished is: presetting a change threshold; subtracting the sensitivity change calculated in the previous optimization process from the sensitivity change calculated in the current optimization process to obtain the change difference; if the change difference is less than the change threshold, the optimization is finished; if the change difference is greater than or equal to the change threshold, the optimization continues.

[0067] Furthermore, the step of obtaining the gradient coefficient includes:

[0068] Step d1: Preset the initial coefficient search interval ; Set the split ratio ;

[0069] Step d2: According to the segmentation ratio , the initial coefficient search interval Create two new points and , , ;

[0070] Step d3: According to and Update the density value of each unit, obtain the displacement vector and stiffness matrix of the gradient lattice structure after the unit density value is updated, and calculate the corresponding compliance respectively; The flexibility calculated after updating the cell density value is marked as the first flexibility. The flexibility calculated after the unit density value is updated is marked as the second flexibility;

[0071] Step d4: If the first flexibility is greater than or equal to the second flexibility, the minimum value of the updated coefficient search interval is the point corresponding to the second flexibility, and the maximum value is the maximum value of the coefficient search interval before updating; if the first flexibility is less than the second flexibility, the minimum value of the updated coefficient search interval is the minimum value of the coefficient search interval before updating, and the maximum value is the point corresponding to the first flexibility;

[0072] Step d5: preset a width threshold, subtract the minimum value from the maximum value in the updated coefficient search interval to obtain the interval width; compare the interval width with the width threshold; if the interval width is greater than or equal to the width threshold, the updated coefficient search area is divided according to the segmentation ratio , re-dividing two new points and , and return to step d3; if the interval width is less than the width threshold, the average of the maximum and minimum values ​​in the updated coefficient search area is used as the gradient coefficient.

[0073] A device for generating and optimizing a gradient lattice structure for laser additive manufacturing, implementing the method for generating and optimizing a gradient lattice structure for laser additive manufacturing, comprising:

[0074] Data collection module, used to collect design requirement data;

[0075] The structure generation module uses an improved evolutionary algorithm to generate a gradient lattice structure based on the design requirement data;

[0076] The structure evaluation module is used to evaluate whether the generated gradient lattice structure meets the standards;

[0077] The structure optimization module is used to optimize the gradient lattice structure if the gradient lattice structure does not meet the standards.

[0078] The technical effects and advantages of the gradient lattice structure generation and optimization method and device for laser additive manufacturing of the present invention are as follows:

[0079] According to the collected design requirement data, the improved evolutionary algorithm is used to automatically generate the gradient lattice structure, realizing the automatic conversion of the design requirement data into the gradient lattice structure, solving the problem that the existing method relies on manual modeling; at the same time, multiple design indicators are taken into consideration to realize the multi-objective optimization of the gradient lattice structure, which can improve the performance of the gradient lattice structure from many aspects and effectively generate an excellent gradient lattice structure that meets complex design requirements; in addition, if the generated gradient lattice structure does not meet the standards, the gradient lattice structure can also be automatically optimized, which can shorten the design cycle and improve product performance, thereby realizing more efficient gradient lattice structure design in the field of laser additive manufacturing. BRIEF DESCRIPTION OF THE DRAWINGS

[0080] Figure 1 Schematic diagram of a device for generating and optimizing a gradient lattice structure for laser additive manufacturing according to Embodiment 1 of the present invention;

[0081] Figure 2 This is a flow chart of a method for generating a gradient lattice structure according to Embodiment 1 of the present invention;

[0082] Figure 3 This is a flow chart of a method for generating and optimizing a gradient lattice structure for laser additive manufacturing according to Embodiment 2 of the present invention. DETAILED DESCRIPTION

[0083] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0084] Example 1

[0085] See also Figure 1 As shown, the gradient lattice structure generation and optimization device for laser additive manufacturing described in this embodiment includes a data acquisition module, a structure generation module, a structure evaluation module and a structure optimization module; each module is connected by wired and / or wireless means to achieve data transmission between modules;

[0086] Data collection module, used to collect design requirement data.

[0087] The design requirement data at least includes additive manufacturing objects, mechanical requirement data, material requirement data, shape requirement data and lightweight requirement data; the design requirement data is manually input and obtained by relevant staff.

[0088] Additive manufacturing objects include at least product type, functional requirements, usage environment, etc.; product types include medical devices (such as implants, prostheses, dental restorations, etc.), aerospace components (such as engine components, brackets, casings, etc.), automotive parts (such as lightweight gradient lattice structures, complex-shaped components, etc.); functional requirements include load-bearing components (such as frames, supports, etc.), thermal management components (such as radiators, heat exchangers, etc.); usage environment includes high temperature environment, corrosive environment, impact environment, etc.

[0089] The mechanical demand data at least include stress distribution, load conditions, stiffness and flexibility requirements, etc.; among them, stress distribution is the stress field information that the gradient lattice structure is subjected to in the working environment, usually including stress concentration or stress gradient in different areas; load conditions are the external loads (such as static loads, dynamic loads, impact loads, etc.) that the gradient lattice structure is subjected to during use, including tension, compression, bending and other working conditions; stiffness and flexibility requirements are the requirements for stiffness or flexibility in different areas. For example, in lightweight design, some areas require high stiffness, while other areas may require better flexibility.

[0090] The material requirement data at least includes material type, material distribution, etc.; among them, the material type is the metal or alloy powder material used in the gradient lattice structure, such as aluminum alloy, stainless steel, titanium alloy, etc.; different material types have different melting points, thermal conductivity and mechanical properties, which directly affect the performance of the gradient lattice structure; material distribution is the regional distribution of different materials in the gradient lattice structure.

[0091] The shape requirement data includes at least geometric dimensions, geometric shapes, local density gradients, etc.; wherein the geometric dimensions are the overall dimensions of the gradient lattice structure; the geometric shapes are the unit geometric shapes of the gradient lattice structure, such as tetrahedron, hexahedron, honeycomb, etc. Different geometric shapes have different mechanical properties, which affect the stress transfer, stiffness and energy absorption capacity of the gradient lattice structure; the local density gradient is the degree of change of density in the local area of ​​the gradient lattice structure.

[0092] The lightweight demand data at least include density distribution and weight distribution, among which, density distribution refers to the density change of the material in the gradient lattice structure; weight distribution refers to the weight distribution in different areas of the gradient lattice structure; through lightweight design, density and weight distribution can be optimized, so that the gradient lattice structure can reduce material consumption while ensuring the load-bearing capacity, thereby reducing manufacturing costs.

[0093] It should be noted that the reason for collecting design requirement data is that the additive manufacturing object clarifies the type of gradient lattice structure of additive manufacturing, such as aviation components, medical implants or mechanical parts; the manufacturing object defines the overall design goal and affects the functional requirements of the gradient lattice structure (such as strength, stiffness, elasticity, etc.); the mechanical requirement data covers the mechanical stress, load, stress distribution and other information that the manufacturing object needs to withstand; according to the mechanical requirement data, the gradient lattice structure can be optimized to adapt to the load requirements of different areas; for example, areas with high stress require denser lattices, while areas with low stress can use lighter gradient lattice structures; material requirement data affects the mechanics of the gradient lattice structure. Performance, thermal conductivity, corrosion resistance, etc. Different materials behave differently under stress, temperature and other conditions. It is necessary to generate the optimal gradient lattice structure according to the characteristics of the material to ensure the stability and performance of the gradient lattice structure in a specific environment; the shape requirement data of the additive manufacturing object directly affects the design of the gradient lattice structure. If the additive manufacturing object has a complex geometric shape, the gradient lattice structure needs to be locally adjusted in different areas to ensure adaptation to the boundary. For example, the lattice units in the curved area may require specific arrangements and density changes; through lightweight requirement data, the gradient lattice structure can be optimized to minimize material usage while maintaining strength, thereby achieving the goal of lightweighting.

[0094] The structure generation module uses an improved evolutionary algorithm to generate a gradient lattice structure based on the design requirement data.

[0095] Evolutionary algorithms include genetic algorithms, clonal selection algorithms, etc. This embodiment takes clonal selection as an example and uses an improved clonal selection algorithm to generate a gradient lattice structure.

[0096] like Figure 2 As shown, the steps of generating a gradient lattice structure using the improved clone selection algorithm include:

[0097] Step 1: Collect m structure samples, set different numerical labels for different structure samples, and mark them as structure labels, where m is an integer greater than 1;

[0098] Step 2: Randomly generate an initial antibody group, where the antibodies in the initial antibody group correspond one to one with the structural tags;

[0099] Step 3: Determine the fitness function;

[0100] Step 4: Calculate the fitness of each antibody in the initial antibody group, and perform antibody screening to generate a memory antibody group;

[0101] Step 5: Calculate the corresponding clone size for each antibody in the memory antibody group, and clone it to generate a clone antibody group;

[0102] Step 6: Calculate the selection probability of each antibody in the cloned antibody group, and determine whether to select the antibody based on the selection probability to generate a selected antibody group;

[0103] Step 7: Perform cloud adaptive mutation on each antibody in the selected antibody group using a cloud adaptive mutation operator to generate a mutant antibody group; cloud adaptive mutation is an existing technology and will not be described in detail here;

[0104] Step 8: Using an interpolation method to generate new antibodies for the antibodies in the variant antibody group to generate an interpolated antibody group;

[0105] Step 9: Recombining the antibodies in the variant antibody group to generate a recombinant antibody group;

[0106] Step 10: merging the interpolated antibody group and the recombinant antibody group to generate a merged antibody group;

[0107] Step 11: Determine whether the iteration is finished; if so, calculate the fitness of each antibody in the combined antibody group, and obtain the structural sample corresponding to the structural label of the antibody with the largest fitness; if not, perform antibody screening on the combined antibody group, regenerate the memory antibody group, and return to step 5.

[0108] In the above step 1, the structure sample is a gradient lattice structure sample, and the structure sample is obtained by technicians in this field by collecting gradient lattice structures generated in historical laser additive manufacturing processes, or extracted from an existing laser additive manufacturing database.

[0109] In step 2 above, the initial antibody population , is the qth antibody, that is, the size of the initial antibody population is q; the range of antibodies is m structural samples; the number of iterations t of the initial antibody population is 0.

[0110] In the above step 3, the expression of the fitness function is: ; In the formula, For fitness, is the demand satisfaction; the method for obtaining the demand satisfaction is: for the data that are not numerical values ​​in the design demand data (such as product type, use environment, material type, etc.), different digital labels are set respectively, and marked as demand labels, wherein the digital labels of different data are different; illustratively, the digital label of the prosthesis in the product type is set to 1, the digital label of the bracket is set to 2, the digital label of the aluminum alloy in the material type is set to 100, and the digital label of the stainless steel is set to 101; the data that are not numerical values ​​in the design demand data are replaced with the corresponding demand labels, and the replaced design demand data are marked as replacement data; the replacement data and the structural label corresponding to the antibody are marked as analysis data, and the analysis data is input into the trained demand prediction model to predict the corresponding demand satisfaction.

[0111] The specific training process of the demand forecasting model includes:

[0112] Preliminarily collect b groups of analysis data corresponding to the demand satisfaction, b being an integer greater than 1, and convert the analysis data and the corresponding demand satisfaction into a corresponding set of feature vectors; the demand satisfaction corresponding to the analysis data is determined by a technician in this field during the generation process of a historical gradient lattice structure, by collecting b groups of different analysis data, and performing experimental analysis in turn under the conditions of each group of analysis data to analyze the demand satisfaction of the generated gradient lattice structure for the corresponding design demand data, that is, obtaining the demand satisfaction corresponding to each group of analysis data, and setting corresponding demand satisfaction for the b groups of analysis data.

[0113] The demand satisfaction corresponding to the analysis data of group b is used as the training data of the demand prediction model.

[0114] Data preprocessing, including normalization of replacement data (such as standardization) to ensure consistent feature distribution. Unique hot encoding or label encoding of categorical data of antibody structure labels. Demand satisfaction is numerical data and is directly used as the target value.

[0115] Model selection: choose linear regression, random forest, neural network, etc., taking neural network as an example.

[0116] Input layer: merge replacement data and antibody structure labels; hidden layer: use ReLU activation function to capture nonlinear relationships; output layer: single neuron outputs demand satisfaction.

[0117] Model training: Divide the training data into a training set (80%) and a validation set (20%); Use the mean square error (MSE) loss function and the Adam optimizer to train the model. Monitor the validation set performance to prevent overfitting.

[0118] Model evaluation, test set validation model, calculate mean square error (MSE), and stop training when the sum of mean square error reaches convergence.

[0119] In the above step 4, the method of antibody screening is: screening out from the initial antibody group The antibody with the highest fitness, .

[0120] In step 5 above, the clone size is calculated by combining the fitness of the antibody with the affinity of the antibody, and the calculation is as follows:

[0121] ;

[0122] ;

[0123] In the formula, is the clone size of the i-th antibody, Int is the rounding function to ensure that the clone size is an integer, which meets the actual application requirements, q is the initial antibody group size, The affinity of the ith antibody is calculated by calculating the minimum distance between antibodies to measure the diversity of the population. If an antibody is very different from other antibodies in the population ( If the antibodies are too similar ( If the size is small, cloning is suppressed to prevent redundancy; is the fitness of the ith antibody. Fitness dominates “utilization” (focusing on the current high-quality solution), and affinity dominates “exploration” (encouraging diversity). The combination of the two ensures that the algorithm strikes a balance between global search and local optimization. min is the minimum function, exp is the natural exponential function, is the Euclidean distance between the ith antibody and the jth antibody, , .

[0124] It should be noted that when calculating the Euclidean distance between the ith antibody and the jth antibody, normalization is performed so that .

[0125] The calculation of clone scale is based on two core logics. First, high-quality individuals are selected by the fitness ratio of antibodies to ensure that antibodies with higher fitness have a greater probability of being copied, thereby retaining the dominant gene; second, the concept of affinity is introduced to measure the difference between antibodies and other individuals in the population. Specifically, the Euclidean distance between an antibody and its nearest neighbor antibody is calculated and exponentially amplified. If the antibody is unique (large distance), the number of its clones is increased, otherwise redundancy is suppressed. This design combines the "utilization" and "exploration" mechanisms: fitness drives the algorithm to converge to the current high-quality solution, and affinity actively maintains population diversity to avoid falling into local optimality. At the same time, the clone scale is converted to an integer by rounding up to ensure the feasibility of actual operation. The overall strategy dynamically balances global search capabilities and local development efficiency during the optimization process.

[0126] In the above step 6, the method for generating the selected antibody group comprises:

[0127] Calculate the fitness of each antibody in the cloned antibody group, and add them up in sequence to obtain the total antibody fitness, divide the fitness of each antibody by the total antibody fitness, and obtain the selection probability of each antibody; sort the selection probability of each antibody from large to small, and replace each selection probability in order according to the positive order with the sum of each selection probability and all selection probabilities before the corresponding selection probability; obtain all selection probabilities after replacement and mark them as replacement probabilities; the antibody corresponding to the replacement probability is consistent with the antibody corresponding to the selection probability before replacement; take every two adjacent replacement probabilities as an analysis set according to the positive order; mark the replacement probability with a large value in each analysis set as the first probability, and mark the replacement probability with a small value as the second probability; construct the probability screening range corresponding to each analysis set according to the first probability and the second probability of each analysis set, wherein the first probability is the maximum value of the probability screening range, the second probability is the minimum value of the probability screening range, and the minimum value is an open interval, and the maximum value is a closed interval; each analysis set corresponds to the antibody corresponding to the second probability; wherein the minimum value of the probability screening range corresponding to the first antibody is 0, and the maximum value is the corresponding selection probability; randomly generate indivual A random number between The probability of the random number in the screening range corresponds to the antibody, and a selection antibody group is generated.

[0128] For example, the selection probabilities of the three antibodies are 0.5, 0.3, and 0.2, respectively; since there is no value before 0.5, 0.5 is still 0.5 after replacement; since 0.3 is preceded by 0.5, 0.3 is replaced by 0.3+0.5=0.8, and since 0.2 is preceded by 0.3 and 0.5, 0.2 is replaced by 0.2+0.3+0.5=1; therefore, the replacement probabilities are 0.5, 0.8, and 1, respectively; 0.5 and 0.8 are an analysis set, and the probability screening range corresponding to the second antibody is ; 0.8 and 1 are also an analysis set, then the probability screening range corresponding to the third antibody is ; The probability screening range corresponding to the first antibody is ; If the randomly generated random number is 0.7, since 0.7 is in Therefore, a second antibody was selected.

[0129] In the above step 8, the method for generating the interpolated antibody group includes:

[0130] Calculate the fitness of each antibody in the variant antibody group, and sort them from large to small to generate a sorting table; select half of the antibodies in the variant antibody group in positive order according to the sorting table, and mark them as screening antibodies; use the interpolation method to generate new antibodies for every two adjacent screening antibodies; calculate the fitness corresponding to each new antibody, and add it to the sorting table, retain the antibodies corresponding to the first y fitnesses in the sorting table, and generate an interpolated antibody group, where y is the number of antibodies in the variant antibody group.

[0131] The new antibody is obtained by weighted fusion of two adjacent screening antibodies with random weights. The expression of the new antibody is: ; In the formula, For new antibodies, , For two adjacent screening antibodies, is the interpolation coefficient, for A random number between .

[0132] when =0, the new antibody completely inherits characteristics; when =1, the new antibody completely inherits The intermediate values ​​mix the characteristics of both in proportion.

[0133] The high-quality antibodies screened , ) Generate new antibodies to ensure that the algorithm performs fine searches near high-quality solutions and improve local optimization capabilities. Uncertainty is introduced to randomly distribute the generation locations of new antibodies between the two, thus preventing the population from converging to a single area too early. By mixing the characteristics of different antibodies, the new antibodies retain the dominant genes of the parent generation while producing certain mutations, increasing population diversity and preventing falling into local optimality. The linear interpolation operation is simple and has low computational cost, which is suitable for large-scale iterative optimization scenarios while ensuring generation efficiency. On the basis of retaining the local development capability of high-quality solutions, the search range is expanded through random interpolation, taking into account both the algorithm convergence speed and the global search potential.

[0134] In the above step 9, the method for antibody recombination comprises:

[0135] Randomly select Y antibodies from the mutant antibody group as parent antibodies for antibody recombination to generate progeny antibodies , until the number of unselected antibodies in the variant antibody group is less than Y, the antibody recombination is completed; in this embodiment, Y is preferably 3; the recombinant antibody group includes multiple progeny antibodies .

[0136] Progeny Antibodies Generated by the weighted average of Y randomly selected parent antibodies, as follows:

[0137] ;

[0138] In the formula, is the Yth parent antibody, is the Yth proportional factor, which is a randomly generated real number and not all Y proportional factors are 0, to ensure that the generation process of offspring antibodies is random, to avoid the population from falling into the local optimum, and to enhance the global search capability; by weighted averaging the characteristics of multiple parent antibodies, the offspring inherits the advantages of different parent antibodies, balancing exploration (new characteristics) and utilization (parent advantages); the denominator is the sum of the proportional factors to prevent the weight from being too large or too small, resulting in numerical instability, and to ensure that the calculation result is a reasonable weighted average.

[0139] In the above step 11, the method for determining whether the iteration is finished includes:

[0140] Preset the iteration threshold, which is preset by those skilled in the art according to the algorithm accuracy;

[0141] Compare the number of iterations t with the iteration threshold;

[0142] If the number of iterations t is greater than or equal to the iteration threshold, the iteration ends;

[0143] If the number of iterations t is less than the iteration threshold, the maximum fitness in the combined antibody group is marked as the single maximum fitness; the single maximum fitness in the current iteration is subtracted from the single maximum fitness in the previous iteration and the absolute value is taken as the fitness change;

[0144] Preset variable threshold and quantity threshold, both of which are preset by those skilled in the art according to actual conditions; compare the fitness change with the variable threshold, if the fitness change is greater than or equal to the variable threshold, no smoothing instruction is generated; if the fitness change is less than the variable threshold, a smoothing instruction is generated; count the number of smoothing instructions generated and compare it with the quantity threshold;

[0145] If the number of smooth instructions is less than the threshold, the iteration continues and the number of iterations is set to , and select the one with the largest fitness among the combined antibody groups antibodies form a new memory antibody group; among them, For computer languages, the value of the number of iterations t is added by 1 and then assigned to the number of iterations t;

[0146] If the number of smoothing instructions is greater than or equal to the number threshold, the iteration ends.

[0147] It should be noted that the reason for using the improved clonal selection algorithm to generate the gradient lattice structure is:

[0148] 1. Optimization performance: The improved clone selection algorithm improves the efficiency of the optimization process by simulating the principle of the biological immune system; it can effectively find the optimal solution and is suitable for complex gradient lattice structure optimization problems;

[0149] 2. Handling multi-objective optimization: In laser additive manufacturing, design requirements often involve multiple performance indicators (such as mechanical properties, materials, lightweight, etc.); the improved clone selection algorithm can consider multiple objectives at the same time and adapt to complex design requirements;

[0150] 3. Enhanced flexibility: The improved clone selection algorithm can achieve more flexible exploration in the search space and adapt to different gradient changes and nonlinear gradient structural designs by introducing individual clones and mutation mechanisms;

[0151] 4. Reduce the need for manual modeling: The improved clone selection algorithm can automatically generate and optimize structures, reducing the complexity and time cost caused by human factors;

[0152] 5. Strong adaptability: The clone selection algorithm can dynamically adjust parameters according to the design requirement data, so that the generated gradient lattice structure is more in line with actual use requirements, improving the performance and reliability of the product.

[0153] 6. Accelerate the iteration process: Through adaptive mutation and antibody recombination, the improved clone selection algorithm can speed up the convergence speed, improve the optimization efficiency and shorten the design cycle.

[0154] The structure evaluation module is used to evaluate whether the generated gradient lattice structure meets the standards.

[0155] Methods for evaluating whether the generated gradient lattice structure meets the standards include:

[0156] Finite element analysis is performed on the generated gradient lattice structure using finite element software (such as ANSYS, Abaqus, etc.) to obtain corresponding performance data; the performance data at least includes mechanical data and lightweight data, wherein the data in the mechanical data is consistent with the data in the mechanical demand data, and the data in the lightweight data is consistent with the data in the lightweight demand data;

[0157] The performance data, mechanical requirement data, and lightweight requirement data are used as test data, and the test data are input into the trained structural evaluation model, and an evaluation label is output. The corresponding evaluation result is obtained according to the evaluation label, and whether the generated gradient lattice structure meets the standard is evaluated according to the evaluation result; the evaluation label is a digital label corresponding to the evaluation result, and the evaluation results include meeting the standard and not meeting the standard. Different evaluation results have different corresponding digital labels. For example, the digital label is set to 0.1 for meeting the standard, and the digital label is set to 0.2 for not meeting the standard.

[0158] The specific training process of the structure evaluation model includes:

[0159] A group of training samples are collected in advance, and each group of training samples includes test data and judgment results corresponding to the test data. A is an integer greater than 1. The judgment results corresponding to the test data are collected by technical personnel in the historical gradient lattice structure generation process. Different test data of group A are collected. According to actual experience, technical personnel in this field will compare the performance data in different test data of group A with the corresponding mechanical requirement data and lightweight requirement data in turn to determine whether the corresponding gradient lattice structure meets the standards, and set corresponding judgment results for different test data of group A in turn.

[0160] The training samples are divided into training set and test set according to the ratio of 7:3.

[0161] The training set is used to train the structural evaluation model. The structural evaluation model uses the evaluation label corresponding to each set of test data as output and the actual evaluation label corresponding to each set of test data as the prediction target. The actual evaluation label is the evaluation label corresponding to the test data collected in advance. The mean absolute percentage error MAPE is used to evaluate the model accuracy of the prediction results. When the calculated MAPE is less than the preset MAPE, the structural evaluation model training is completed, and a structural evaluation model that predicts the evaluation label based on the test data is generated. The calculation formula of MAPE is: ;in, is the actual evaluation label corresponding to the d-th group of test data, is the predicted evaluation label corresponding to the dth group of test data, d is the group number of the feature vector corresponding to the test data, A represents the number of predicted evaluation labels, ; Generate a structural evaluation model that predicts evaluation labels based on test data; wherein the structural evaluation model is a deep belief network model, and the preset MAPE is pre-set by technicians in this field according to the accuracy required by the structural evaluation model.

[0162] The structure optimization module is used to optimize the gradient lattice structure if the gradient lattice structure does not meet the standards.

[0163] The steps for optimizing the gradient lattice structure include:

[0164] Step a: Use finite element software to discretize the gradient lattice structure into finite element units, obtain the volume of each unit, and set the initial density value for each unit. is 1; , which represents the distribution state of the material; the finite element software automatically calculates the volume of each unit and stores it, so it can be read directly.

[0165] Step b: Define the objective function and volume constraints.

[0166] The expression of the objective function is: ; In the formula, For softness, is the stiffness matrix, which changes with the density value. is the displacement vector, is the transpose of the displacement vector; the displacement vector and the stiffness matrix are obtained by analyzing the gradient lattice structure using the finite element method. The finite element method is an existing technology and will not be described in detail here; the flexibility is the deformation ability of the gradient lattice structure under the action of external force. Specifically, the flexibility can be defined as the relationship between the load and the corresponding displacement of the gradient lattice structure. It is usually used to evaluate the stiffness characteristics of the gradient lattice structure. The smaller the flexibility, the greater the stiffness of the gradient lattice structure.

[0167] The expression of volume constraint is: ; In the formula, is the volume of the gradient lattice structure, For the The volume of a unit, The maximum allowable volume of the gradient lattice structure is set according to the actual shape requirements (such as lightweight goals or load-bearing capacity) to ensure that the design meets both functional requirements and resource constraints. , M is the number of units in the gradient lattice structure; wherein, the volume of the gradient lattice structure is also obtained by analyzing the gradient lattice structure using the finite element method, and the maximum allowable volume of the gradient lattice structure is obtained through the shape requirement data.

[0168] It should be noted that in the volume constraint expression, dividing by The purpose is to standardize the volume constraint, that is, to convert the volume constraint into a relative standard quantity, so as to facilitate the control and management of material distribution during the optimization process of the gradient lattice structure.

[0169] The volume constraint formula limits the material usage of the gradient lattice structure by normalizing the weighted volume sum: The left side of the formula represents the volume of all units With its density (or weight) The sum of the products of , ensuring that this ratio does not exceed the maximum permissible volume The core logic is to control the distribution or total amount of materials within the structure to avoid excessive consumption of resources. At the same time, through normalization processing, the constraints are made independent of the structure size and only focus on the reasonable distribution of volume.

[0170] Step c: Calculate the sensitivity of the current gradient lattice structure. The sensitivity is reflected in the effect of the change in density value on the overall structural compliance in each unit.

[0171] Methods for calculating the sensitivity of the current gradient lattice structure include:

[0172] The finite element method is used to calculate the displacement vector and stiffness matrix of the current gradient lattice structure, and the displacement vector and stiffness matrix are substituted into the objective function to calculate the corresponding compliance, and the sensitivity is calculated based on the compliance.

[0173] The expression of sensitivity is: ; In the formula, is the sensitivity, is the penalty factor, usually 2 or 3. For the The negative sign in the formula indicates that areas with high compliance (i.e., units with softer structures or larger deformations) will significantly reduce the overall sensitivity, thereby guiding the design variables to adjust in the direction of reducing compliance (enhancing stiffness) during the optimization process, which meets the needs of lightweight and toughened structures.

[0174] when =2, It will amplify the sensitivity of low-density or intermediate-density areas, push the design variables toward 0 or 1, thereby eliminating fuzzy boundaries and accelerating convergence to a clear structure. Control nonlinear effects: By adjusting The value of (usually 2 or 3) balances the sensitivity to different density areas and avoids the optimization process falling into local extremes.

[0175] Compliance reflects the ability of a unit to deform when subjected to force, with high compliance corresponding to low stiffness. By incorporating it into the sensitivity calculation, the formula directly relates the structural performance (stiffness) to the design variables, ensuring that the optimization goal (such as minimizing compliance) is achieved.

[0176] Step d: Update the density value of each cell , and perform smoothing operations to ensure that the density value changes continuously in the neighborhood and avoid numerical instability problems; smoothing operations are existing technologies and will not be described in detail here.

[0177] Update the density value of each cell The methods include:

[0178] ;

[0179] In the formula, For the During the optimization process The density value of each unit, is the gradient coefficient, For the The sensitivity of each unit, .

[0180] The steps for obtaining the gradient coefficient include:

[0181] Step d1: Preset the initial coefficient search interval , this embodiment preferably , ; Set the split ratio , this embodiment preferably ;

[0182] Step d2: According to the segmentation ratio , the initial coefficient search interval Create two new points and , , ;

[0183] Step d3: According to and Update the density value of each unit, obtain the displacement vector and stiffness matrix of the gradient lattice structure after the unit density value is updated, and calculate the corresponding compliance respectively; The flexibility calculated after updating the cell density value is marked as the first flexibility. The flexibility calculated after the unit density value is updated is marked as the second flexibility;

[0184] Step d4: If the first flexibility is greater than or equal to the second flexibility, the minimum value of the updated coefficient search interval is the point corresponding to the second flexibility, and the maximum value is the maximum value of the coefficient search interval before updating; if the first flexibility is less than the second flexibility, the minimum value of the updated coefficient search interval is the minimum value of the coefficient search interval before updating, and the maximum value is the point corresponding to the first flexibility; illustratively, the initial coefficient search interval Passing point and After updating, if the first flexibility is greater than or equal to the second flexibility, the update coefficient search interval is , if the first flexibility is less than the second flexibility, the update coefficient search interval is ;

[0185] Step d5: preset a width threshold, which is preset by those skilled in the art according to actual conditions, and obtain the interval width by subtracting the minimum value from the maximum value in the updated coefficient search interval; compare the interval width with the width threshold; if the interval width is greater than or equal to the width threshold, the updated coefficient search area is divided according to the segmentation ratio. , re-dividing two new points and , and return to step d3; if the interval width is less than the width threshold, the average of the maximum and minimum values ​​in the updated coefficient search area is used as the gradient coefficient.

[0186] Step e: After calculating the density value update, the volume of each unit in the current gradient lattice structure is determined to determine whether the volume constraint is satisfied. If not, the density value of each unit is updated. Adjust and then enter step f. If satisfied, directly enter step f. It should be understood that when the density value of each unit in the gradient lattice structure is After the change, the gradient lattice structure will also change, so the volume of each unit in the current gradient lattice structure will also change.

[0187] Methods for determining whether the volume constraint is satisfied include:

[0188] Substitute the volume of each unit in the current gradient lattice structure after the density value is updated into the expression of the volume constraint. If the expression of the volume constraint is established, the volume constraint is satisfied, that is, ; If the volume constraint expression does not hold, the volume constraint is not satisfied, that is .

[0189] The density value for each cell Ways to make adjustments include:

[0190] Calculate the scaling factor , scaling factor The expression is: ; The density value of each unit Multiply by the scaling factor , get the adjusted density value of each unit .

[0191] Step f: Calculate the sensitivity of the current gradient lattice structure and determine whether the optimization is completed. If not, return to step d; if so, obtain the current gradient lattice structure as the optimized gradient lattice structure.

[0192] Methods for determining whether optimization is completed include:

[0193] A change threshold is preset, and the change threshold is preset by technical personnel in this field according to actual conditions; the sensitivity change calculated in the current optimization process is subtracted from the sensitivity change calculated in the previous optimization process to obtain a change difference; the change difference is compared with the change threshold; if the change difference is less than the change threshold, the optimization ends, and if the change difference is greater than or equal to the change threshold, the optimization continues.

[0194] It should be noted that the purpose of optimizing the gradient lattice structure by adopting the above steps a to f is:

[0195] 1. Improve structural performance: Through the optimization process, the flexibility is reduced and the structural stiffness is increased, so that the gradient lattice structure has better bearing capacity and stability under external forces;

[0196] 2. Meet design requirements: Ensure that the optimized structure can meet specific mechanical requirements, material requirements and lightweight goals, and ensure the reliability and efficiency of the gradient lattice structure in practical applications.

[0197] 3. Achieve efficient use of materials: By controlling density distribution and volume constraints, the optimization process can achieve effective use of materials, reduce unnecessary material waste, and thus reduce manufacturing costs;

[0198] 4. Enhance the adaptability of the structure: Through sensitivity analysis, the optimization process can adjust the material distribution according to different loads and environmental conditions, making the structure more adaptable.

[0199] 5. Improve production efficiency: By optimizing the design, the complexity of subsequent processing and manufacturing can be reduced, thereby improving overall production efficiency and reducing costs.

[0200] This embodiment uses an improved evolutionary algorithm to automatically generate a gradient lattice structure based on the collected design requirement data, realizes the automatic conversion of the design requirement data into the gradient lattice structure, and solves the problem that the existing method relies on manual modeling; at the same time, multiple design indicators are considered to realize multi-objective optimization of the gradient lattice structure, which can improve the performance of the gradient lattice structure from multiple aspects and effectively generate an excellent gradient lattice structure that meets complex design requirements; in addition, if the generated gradient lattice structure does not meet the standards, the gradient lattice structure can also be automatically optimized, which can shorten the design cycle and improve product performance, thereby realizing more efficient gradient lattice structure design in the field of laser additive manufacturing.

[0201] Example 2

[0202] See also Figure 3 As shown, the part not described in detail in this embodiment is described in Example 1, and a method for generating and optimizing a gradient lattice structure for laser additive manufacturing is provided, the method comprising:

[0203] Collect design requirement data;

[0204] According to the design requirement data, the improved evolutionary algorithm is used to generate the gradient lattice structure;

[0205] Evaluate whether the generated gradient lattice structure meets the standards;

[0206] If the gradient lattice structure does not meet the standard, the gradient lattice structure is discretized into multiple finite element units, and the density of each unit is initially set to the maximum value; the goal is to minimize the structural flexibility, and the volume of the unit is set not to exceed the maximum allowable volume; the unit density value is iteratively adjusted through analytical sensitivity, and the distribution is optimized by smoothing processing, and the volume constraint is repeatedly checked; if not, the density value is readjusted until the result converges, and the optimized gradient lattice structure is determined.

[0207] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

[0208] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for generating and optimizing a gradient lattice structure for laser additive manufacturing, characterized in that: include: Collect design requirement data; According to the design requirement data, the improved evolutionary algorithm is used to generate the gradient lattice structure; Evaluate whether the generated gradient lattice structure meets the standards; If the gradient lattice structure does not meet the standard, the gradient lattice structure is discretized into multiple finite element units, and the density of each unit is initially set to the maximum value; the goal is to minimize the structural flexibility, and the volume of the unit is set not to exceed the maximum allowable volume; the unit density value is iteratively adjusted through analytical sensitivity, and the distribution is optimized by smoothing processing, and the volume constraint is repeatedly checked; if not, the density value is readjusted until the result converges, and the optimized gradient lattice structure is determined.

2. The method for generating and optimizing a gradient lattice structure for laser additive manufacturing according to claim 1, characterized in that: The steps of generating a gradient lattice structure using the improved clonal selection algorithm include: Step 1: Collect m structure samples, set different numerical labels for different structure samples, and mark them as structure labels, where m is an integer greater than 1; Step 2: Randomly generate an initial antibody group, where the antibodies in the initial antibody group correspond one to one with the structural tags; Step 3: Determine the fitness function; Step 4: Calculate the fitness of each antibody in the initial antibody group, and perform antibody screening to generate a memory antibody group; Step 5: Calculate the corresponding clone size for each antibody in the memory antibody group, and clone it to generate a clone antibody group; Step 6: Calculate the selection probability of each antibody in the cloned antibody group, and determine whether to select the antibody based on the selection probability to generate a selected antibody group; Step 7: Perform cloud adaptive mutation on each antibody in the selected antibody group using a cloud adaptive mutation operator to generate a mutant antibody group; Step 8: Using an interpolation method to generate new antibodies for the antibodies in the variant antibody group to generate an interpolated antibody group; Step 9: Recombining the antibodies in the variant antibody group to generate a recombinant antibody group; Step 10: merging the interpolated antibody group and the recombinant antibody group to generate a merged antibody group; Step 11: Determine whether the iteration is finished; if so, calculate the fitness of each antibody in the combined antibody group, and obtain the structural sample corresponding to the structural label of the antibody with the largest fitness; if not, perform antibody screening on the combined antibody group, regenerate the memory antibody group, and return to step 5.

3. The method for generating and optimizing a gradient lattice structure for laser additive manufacturing according to claim 2, characterized in that: In the step 1, the structure sample is a gradient lattice structure sample; In step 2, the initial antibody population , is the qth antibody, that is, the size of the initial antibody population is q; the range of antibodies is m structural samples; the number of iterations t of the initial antibody population is 0; In step 4, the method for antibody screening is: screening out The antibody with the highest fitness, ; In step 5, the clone size is calculated by combining the fitness of the antibody with the affinity of the antibody.

4. The method for generating and optimizing a gradient lattice structure for laser additive manufacturing according to claim 3, characterized in that: The design requirement data includes additive manufacturing objects, mechanical requirement data, material requirement data, shape requirement data and lightweight requirement data; the additive manufacturing objects include product types, functional requirements and usage environments; the mechanical requirement data include stress distribution, load conditions and stiffness and flexibility requirements; the material requirement data include material types and material distribution; the shape requirement data include geometric dimensions, geometric shapes and local density gradients; the lightweight requirement data include density distribution and weight distribution; In step 3, the expression of the fitness function is: ; In the formula, For fitness, is the demand satisfaction; the method for obtaining the demand satisfaction is: different digital labels are set for the data that are not numerical values ​​in the design demand data, and marked as demand labels, wherein the digital labels of different data are different; the data that are not numerical values ​​in the design demand data are replaced with corresponding demand labels, and the replaced design demand data are marked as replacement data; the replacement data and the structural label corresponding to the antibody are marked as analysis data, and the analysis data is input into the trained demand prediction model to predict the corresponding demand satisfaction.

5. The method for generating and optimizing a gradient lattice structure for laser additive manufacturing according to claim 4, characterized in that: In step 6, the method for generating the selected antibody group comprises: Calculate the fitness of each antibody in the cloned antibody group, and add them up in sequence to obtain the total antibody fitness, divide the fitness of each antibody by the total antibody fitness, and obtain the selection probability of each antibody; sort the selection probability of each antibody from large to small, and replace each selection probability in order according to the positive order with the sum of each selection probability and all selection probabilities before the corresponding selection probability; obtain all selection probabilities after replacement and mark them as replacement probabilities; the antibody corresponding to the replacement probability is consistent with the antibody corresponding to the selection probability before replacement; take every two adjacent replacement probabilities as an analysis set according to the positive order; mark the replacement probability with a large value in each analysis set as the first probability, and mark the replacement probability with a small value as the second probability; construct the probability screening range corresponding to each analysis set according to the first probability and the second probability of each analysis set, wherein the first probability is the maximum value of the probability screening range, the second probability is the minimum value of the probability screening range, and the minimum value is an open interval, and the maximum value is a closed interval; each analysis set corresponds to the antibody corresponding to the second probability; wherein the minimum value of the probability screening range corresponding to the first antibody is 0, and the maximum value is the corresponding selection probability; randomly generate indivual A random number between The probability screening range of the random numbers corresponds to the antibodies to generate a selection antibody group; In step 8, the method for generating the interpolated antibody group includes: Calculate the fitness of each antibody in the variant antibody group, and sort them from large to small to generate a sorting table; select half of the antibodies in the variant antibody group in positive order according to the sorting table, and mark them as screening antibodies; use the interpolation method to generate new antibodies for every two adjacent screening antibodies; calculate the fitness corresponding to each new antibody, and add it to the sorting table, retain the antibodies corresponding to the first y fitness in the sorting table, and generate an interpolated antibody group, where y is the number of antibodies in the variant antibody group; New antibodies are obtained by weighted fusion of two adjacent screened antibodies with random weights.

6. The method for generating and optimizing a gradient lattice structure for laser additive manufacturing according to claim 5, characterized in that: In step 9, the method for antibody recombination comprises: Randomly select Y antibodies from the mutant antibody group as parent antibodies for antibody recombination to generate progeny antibodies , until the number of unselected antibodies in the mutant antibody group is less than Y, the antibody recombination is completed; Progeny Antibodies Generated by the weighted average of Y randomly selected parent antibodies; In step 11, the method for determining whether the iteration is completed includes: Preset an iteration threshold; compare the number of iterations t with the iteration threshold; if the number of iterations t is greater than or equal to the iteration threshold, the iteration ends; if the number of iterations t is less than the iteration threshold, mark the maximum fitness in the combined antibody group as the single maximum fitness; subtract the single maximum fitness in the previous iteration from the single maximum fitness in this iteration and take the absolute value as the fitness change; Preset variable threshold and quantity threshold; compare the fitness change with the variable threshold. If the fitness change is greater than or equal to the variable threshold, no smoothing instruction is generated; if the fitness change is less than the variable threshold, a smoothing instruction is generated; count the number of smoothing instructions generated and compare it with the quantity threshold; if the number of smoothing instructions is less than the quantity threshold, the iteration continues and the number of iterations is set to , and select the one with the largest fitness among the combined antibody groups antibodies form a new memory antibody group; if the number of smooth instructions is greater than or equal to the threshold, the iteration ends.

7. The method for generating and optimizing a gradient lattice structure for laser additive manufacturing according to claim 1, characterized in that: The method for evaluating whether the generated gradient lattice structure meets the standards includes: Finite element analysis is performed on the generated gradient lattice structure using finite element software to obtain corresponding performance data; the performance data includes mechanical data and lightweight data, wherein the data in the mechanical data is consistent with the data in the mechanical requirement data, and the data in the lightweight data is consistent with the data in the lightweight requirement data; the performance data, mechanical requirement data, and lightweight requirement data are used as test data, the test data are input into a trained structural evaluation model, and an evaluation label is output, and the corresponding evaluation result is obtained according to the evaluation label, and whether the generated gradient lattice structure meets the standard is evaluated according to the evaluation result; the evaluation label is a digital label corresponding to the evaluation result, and the evaluation results include meeting the standard and not meeting the standard, and different evaluation results correspond to different digital labels.

8. The method for generating and optimizing a gradient lattice structure for laser additive manufacturing according to claim 7, characterized in that: The steps for optimizing the gradient lattice structure include: Step a: Discretize the gradient lattice structure into finite element units, obtain the volume of each unit, and set the initial density value for each unit is 1; ; Step b: define the objective function and volume constraint; Step c: calculate the sensitivity of the current gradient lattice structure; Step d: update the density value of each unit , and perform smoothing operation; Step e: After calculating the density value update, the volume of each unit in the current gradient lattice structure is determined to determine whether the volume constraint is met. If not, the density value of each unit is Make adjustments and then go to step f. If satisfied, go directly to step f. Step f: Calculate the sensitivity of the current gradient lattice structure and determine whether the optimization is completed. If not, return to step d; if yes, obtain the current gradient lattice structure as the optimized gradient lattice structure; In step b, the expression of the objective function is: ; In the formula, For softness, is the stiffness matrix, is the displacement vector, is the transpose of the displacement vector; The expression of volume constraint is: ; In the formula, is the volume of the gradient lattice structure, For the The volume of a unit, is the maximum permissible volume of the gradient lattice structure, , M is the number of units in the gradient lattice structure.

9. The method for generating and optimizing a gradient lattice structure for laser additive manufacturing according to claim 8, characterized in that: In the step c, the method for calculating the sensitivity of the current gradient lattice structure includes: using the finite element method to calculate the displacement vector and stiffness matrix of the current gradient lattice structure, substituting them into the objective function, calculating the corresponding compliance, and calculating the sensitivity according to the compliance; The expression of sensitivity is: ; In the formula, is the sensitivity, is the penalty factor, For the The flexibility of each unit; In step d, the density value of each unit is updated The method is: ; In the formula, For the During the optimization process The density value of each unit, is the gradient coefficient, For the The sensitivity of each unit, ; In the step e, the method for determining whether the volume constraint is satisfied is: substituting the volume of each unit in the current gradient lattice structure after the density value is updated into the expression of the volume constraint, if the expression of the volume constraint is established, the volume constraint is satisfied; if the expression of the volume constraint is not established, the volume constraint is not satisfied; The density value for each cell The adjustment is done by calculating the scaling factor , scaling factor The expression is: ; The density value of each unit Multiply by the scaling factor , get the adjusted density value of each unit ; In step f, the method for determining whether the optimization is finished is: presetting a change threshold; subtracting the sensitivity change calculated in the previous optimization process from the sensitivity change calculated in the current optimization process to obtain the change difference; if the change difference is less than the change threshold, the optimization is finished; if the change difference is greater than or equal to the change threshold, the optimization continues.

10. The method for generating and optimizing a gradient lattice structure for laser additive manufacturing according to claim 9, characterized in that: The steps for obtaining the gradient coefficient include: Step d1: Preset the initial coefficient search interval ; Set the split ratio ; Step d2: According to the segmentation ratio , the initial coefficient search interval Create two new points and , , ; Step d3: According to and Update the density value of each unit, obtain the displacement vector and stiffness matrix of the gradient lattice structure after the unit density value is updated, and calculate the corresponding compliance respectively; The flexibility calculated after updating the cell density value is marked as the first flexibility. The flexibility calculated after the unit density value is updated is marked as the second flexibility; Step d4: If the first flexibility is greater than or equal to the second flexibility, the minimum value of the updated coefficient search interval is the point corresponding to the second flexibility, and the maximum value is the maximum value of the coefficient search interval before updating; if the first flexibility is less than the second flexibility, the minimum value of the updated coefficient search interval is the minimum value of the coefficient search interval before updating, and the maximum value is the point corresponding to the first flexibility; Step d5: preset a width threshold, subtract the minimum value from the maximum value in the updated coefficient search interval to obtain the interval width; compare the interval width with the width threshold; if the interval width is greater than or equal to the width threshold, the updated coefficient search area is divided according to the segmentation ratio , re-dividing two new points and , and return to step d3; if the interval width is less than the width threshold, the average of the maximum and minimum values ​​in the updated coefficient search area is used as the gradient coefficient.

11. A device for generating and optimizing a gradient lattice structure for laser additive manufacturing, implementing the method for generating and optimizing a gradient lattice structure for laser additive manufacturing according to any one of claims 1 to 10, characterized in that: include: Data collection module, used to collect design requirement data; The structure generation module uses an improved evolutionary algorithm to generate a gradient lattice structure based on the design requirement data; The structure evaluation module is used to evaluate whether the generated gradient lattice structure meets the standards; The structure optimization module is used to optimize the gradient lattice structure if the gradient lattice structure does not meet the standards.

Images