A method for longitudinal performance enhancement of lightweight alloys by solid phase additive manufacturing and related devices

Abstract

The application discloses a longitudinal performance strengthening method for light alloy solid-phase additive manufacturing and a related device, and the method comprises the following steps: obtaining a material to be subjected to additive manufacturing for performing solid-phase additive manufacturing, and inputting the material to be subjected to additive manufacturing into a longitudinal strengthening system; calling a process recommendation model corresponding to the material to be subjected to additive manufacturing in the longitudinal strengthening system to perform process recommendation processing, and obtaining recommended process parameters; taking the recommended process parameters as initial setting values of the longitudinal strengthening system, performing solid-phase additive manufacturing processing according to the initial setting values based on adaptive pressure feedback depth control, and forming a solid-phase additive manufacturing deposition layer; and performing strengthening post-processing on the solid-phase additive manufacturing deposition layer by using a friction stir processing method based on preset processing parameters. In the embodiment of the application, axial pressure is used as an indirect criterion for depth state, layer combination defects caused by excessive depth or excessive shallowness are prevented, and the layer combination performance is further improved through strengthening post-processing.

Description

Technical Field

[0001] This invention relates to the field of additive manufacturing technology, and in particular to a method and related apparatus for enhancing the longitudinal properties of lightweight alloy solid-phase additive manufacturing. Background Technology

[0002] With the increasing demand for lightweight structural components in the aerospace, rail transportation, and automotive industries, the forming and manufacturing of complex lightweight alloy components such as aluminum and magnesium alloys has become a research hotspot. Traditional fusion additive manufacturing technologies, such as selective laser melting and arc additive manufacturing, are prone to problems such as porosity, oxide inclusions, and grain coarsening when processing materials like aluminum and magnesium due to their high heat input, which limits their application in the field of high-performance component fabrication.

[0003] The emergence of solid-phase additive manufacturing technology has provided a new solution to the aforementioned bottlenecks. This technology is based on the principle of friction stir welding. Through the mechanical stirring action of the stirring head, the material undergoes intense plastic deformation and heat accumulation, achieving interlayer solid-phase deposition. It features no welding defects, fine microstructure, and excellent mechanical properties.

[0004] However, solid-state additive manufacturing still faces several technical challenges in controlling longitudinal performance. First, the metallurgical quality of the interlayer bonding interface is difficult to control stably. Due to the complex thermal field distribution during additive manufacturing, if the surface temperature of the previous layer is too low or the upsetting force is insufficient when the next layer is deposited, the interlayer bonding is easily weak, resulting in anisotropy. If the heat input is too large, it may cause grain coarsening or excessive interface reaction. Second, the mapping relationship between process parameters and process state exhibits highly nonlinear characteristics. The coupling effect of multiple parameters such as rotation speed, travel speed, feed rate, and downpressure makes it difficult to achieve global optimization using traditional trial and error methods. Third, the depth control of the additive manufacturing process has a disturbance sensitivity problem. Factors such as heat accumulation effect and changes in fixture rigidity can cause the actual penetration depth to deviate from the set value, thereby affecting the interlayer bonding quality. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art. This invention provides a method and related apparatus for enhancing the longitudinal performance of lightweight alloy solid-phase additive manufacturing. By using axial pressure as an indirect criterion for the depth of penetration, it prevents interlayer bonding defects caused by excessive or shallow penetration. The interlayer bonding performance is further improved through post-treatment.

[0006] To address the aforementioned technical problems, this invention provides a method for enhancing the longitudinal properties of lightweight alloy solid-state additive manufacturing, applied to a longitudinal strengthening system. The longitudinal strengthening system includes an additive manufacturing base plate, a friction stir additive manufacturing actuator, a spindle control unit, and a data processing and control unit. The method includes: Obtain the material to be added in solid-state additive manufacturing, and input the material to be added in the longitudinal strengthening system. In the longitudinal strengthening system, call the process recommendation model corresponding to the material to be added in the longitudinal strengthening system to perform process recommendation processing and obtain the recommended process parameters of the material to be added in the longitudinal strengthening system. The recommended process parameters are used as the initial settings of the longitudinal strengthening system. The longitudinal strengthening system, based on pressure feedback adaptive depth control, performs solid-state additive manufacturing of the material to be additively manufactured according to the initial settings to form a solid-state additive manufacturing deposition layer of the material to be additively manufactured. The solid additive manufacturing deposition layer is strengthened by using a stirring friction processing method based on preset processing parameters.

[0007] Optionally, the process recommendation model is a training dataset constructed using historical additive manufacturing data of the corresponding additive manufacturing material, and the training dataset is used to train the preset model. The preset model includes a feature extraction layer and a parameter optimization layer, wherein the feature extraction layer uses a convolutional neural network or a stacked autoencoder network, and the parameter optimization layer uses a genetic algorithm or a Bayesian algorithm for optimization. The training dataset constructed using historical additive manufacturing data of the corresponding additive manufacturing materials includes: Obtain each set of historical process parameters in the corresponding additive manufacturing material and the longitudinal mechanical property parameters corresponding to each set of historical process parameter data; Establish a mapping relationship between each set of historical process parameters and the corresponding longitudinal mechanical performance parameters to form the training dataset.

[0008] Optionally, the step of calling the process recommendation model corresponding to the material to be additively manufactured within the longitudinal strengthening system to perform process recommendation processing and obtain recommended process parameters for the material to be additively manufactured includes: Within the longitudinal reinforcement system, a process recommendation model corresponding to the material to be additively manufactured is invoked, and the feature extraction layer within the process recommendation model is used for output processing to obtain output pressure, output temperature data, and output raw process parameters. Based on the output pressure, output temperature data, and output original process parameters, process optimization definition processing is performed in the parameter optimization layer. Based on the process optimization definition results, a coded genetic algorithm or Bayesian algorithm is used for optimization to form the recommended process parameters for the material to be additively manufactured.

[0009] Optionally, the process optimization definition processing based on output pressure, output temperature data, and original output process parameters within the parameter optimization layer includes: Based on the output pressure, output temperature data, and original process parameters, the interlaminar tensile shear strength and the thickness of the intermetallic compound layer at the interface are obtained within the parameter optimization layer, and the variables are defined as follows:

[0010] in, The rotational speed of the agitator head in the friction stir additive actuator is [900, 1500] r / min; The travel speed of the agitator head in the friction stir additive actuator is [20, 200] mm / min; The downward pressure of the agitator head in the friction stir additive actuator ranges from [0.1, 1] mm. The constraints are: Peak axial pressure ; Obtained from pressure threshold window calibration; peak temperature , The solidus temperature of the alloy; The process optimization problem is transformed into a single-objective optimization problem using a weighted method, as follows: ; in, and These represent the maximum tensile shear strength and the minimum intermetallic compound layer thickness in the historical dataset; where... , As weight, , .

[0011] Optionally, the optimization based on the process optimization definition results using a coded genetic algorithm includes: After encoding and initializing the process optimization definition results, a fitness function is constructed based on the encoding and initialization results; After completing the fitness function construction, the selection operation, crossover operation, mutation operation, and optimization termination condition setting are performed in sequence. Based on the fitness function, the recommended process parameters generated by the encoded genetic algorithm after optimization are output.

[0012] Optionally, the optimization using a Bayesian algorithm based on the process optimization definition results includes: A Gaussian process is used as a surrogate model for the single-objective optimization function in the process optimization definition result, and Latin hypercube sampling is used to generate several initial observation points to train the surrogate model. During training, an improvement-oriented approach is adopted to determine the next sampling point in each iteration by maximizing the acquisition function. The acquisition function is optimized by using the L-BFGS-B algorithm combined with the improvement of maximizing the expected value through 5 random restarts. The proxy model is iteratively updated based on the optimized acquisition function, and an iterative update termination condition is set. When the termination condition is met, recommended process parameters are generated.

[0013] Optionally, the longitudinal strengthening system, based on pressure feedback and adaptive depth control, performs solid-state additive manufacturing of the material to be additively manufactured according to the initial set value, forming a solid-state additive manufacturing deposition layer of the material to be additively manufactured, including: The axial pressure of the main shaft of the longitudinal reinforcement system is used as an indirect characterization parameter of the compression depth state to set a pressure threshold window, while the axial pressure of the main shaft is detected in real time. When the axial pressure of the spindle is lower than the lower limit of the pressure threshold window, it is determined that the interlayer contact is poor or the plasticization is insufficient. The spindle control unit drives the additive stirring head in the friction stir actuator to increase the downward pressure by a first preset amount to perform solid-phase additive manufacturing of the material to be added, thereby forming a solid-phase additive manufacturing deposition layer of the material to be added. When the axial pressure of the main spindle is higher than the upper limit of the pressure threshold window, it is determined that the forging force is too large. The main spindle control unit drives the additive stirring head in the friction stir additive actuator to rise upward by a second preset amount to perform solid-phase additive manufacturing process on the material to be additively manufactured, forming a solid-phase additive manufacturing deposition layer on the material to be additively manufactured.

[0014] In addition, this invention also provides a longitudinal performance enhancement device for solid-state additive manufacturing of lightweight alloys, applied to a longitudinal performance enhancement system. The longitudinal performance enhancement system includes an additive manufacturing base plate, a friction stir additive manufacturing actuator, a spindle control unit, and a data processing and control unit. The method includes: Process recommendation module: used to obtain the material to be added in solid-state additive manufacturing, input the material to be added in the longitudinal strengthening system, call the process recommendation model corresponding to the material to be added in the longitudinal strengthening system to perform process recommendation processing, and obtain the recommended process parameters of the material to be added; Additive manufacturing module: used to use the recommended process parameters as the initial set values ​​of the longitudinal reinforcement system. The longitudinal reinforcement system, based on pressure feedback adaptive depth control, performs solid-phase additive manufacturing processing on the material to be additively manufactured according to the initial set values ​​to form a solid-phase additive manufacturing deposition layer on the material to be additively manufactured. Strengthening processing module: used to strengthen the solid additive manufacturing deposited layer by means of friction stirring processing based on preset processing parameters.

[0015] In addition, embodiments of the present invention also provide an electronic device, including a processor and a memory, wherein the processor runs a computer program or code stored in the memory to implement the method for enhancing the longitudinal properties of lightweight alloy solid-phase additive manufacturing as described in any of the above embodiments.

[0016] In addition, embodiments of the present invention also provide a computer-readable storage medium for storing a computer program or code, which, when executed by a processor, implements the method for enhancing the longitudinal properties of lightweight alloy solid-phase additive manufacturing as described above.

[0017] In this embodiment of the invention, adaptive optimization of process parameters is achieved, overcoming the limitations of traditional trial-and-error methods in heterogeneous material additive manufacturing. By using axial pressure as an indirect criterion for the pressing depth, the effects of thermal deformation and fixture rigidity fluctuations can be effectively compensated, preventing interlayer bonding defects caused by excessive or shallow insertion. Immediately after solid-phase additive manufacturing, friction stir processing is applied to strengthen the structure. The secondary stirring action refines the microstructure and introduces longitudinal upsetting force, further improving the interlayer bonding performance and making the longitudinal mechanical properties of the component approach the level of homogeneous forgings. Attached Figure Description

[0018] 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 only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a schematic diagram of the structural composition of the longitudinal reinforcement system in an embodiment of the present invention; Figure 2 This is a schematic flowchart of the method for enhancing the longitudinal properties of lightweight alloy solid-phase additive manufacturing in an embodiment of the present invention. Figure 3 This is a schematic diagram of the pressure depth adaptive adjustment curve based on pressure feedback in an embodiment of the present invention; Figure 4 This is a schematic diagram of the structural composition of the lightweight alloy solid-phase additive manufacturing longitudinal performance enhancement device in an embodiment of the present invention; Figure 5 This is a schematic diagram of the structural composition of the electronic device in an embodiment of the present invention. Detailed Implementation

[0020] 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. 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.

[0021] Example 1, please refer to Figure 1 and Figure 2 , Figure 1 This is a schematic diagram of the structural composition of the longitudinal reinforcement system in an embodiment of the present invention; Figure 2 This is a schematic flowchart of the method for enhancing the longitudinal properties of lightweight alloy solid-phase additive manufacturing in an embodiment of the present invention.

[0022] like Figure 1 As shown, the longitudinal reinforcement system includes an additive manufacturing base plate 1, a friction stir additive manufacturing actuator, a spindle control unit 2, and a data processing and control unit 3. The friction stir additive manufacturing actuator includes an additive stirring head 7, a material handling box 6, a feeding channel 5, and a feeding bin 4. The additive stirring head 7 shares an electric spindle system with the friction stir machining stirring head, and can be freely switched between the friction stir machining and additive manufacturing processes. The feeding bin 4 can feed different types of raw materials, including but not limited to granules, powders, filaments, and rods. The end of the electric spindle integrates a strain gauge force sensor, a photoelectric encoder, and an infrared thermal imager, and shares a spindle with the additive manufacturing actuator and the friction stir machining.

[0023] like Figure 2 As shown, a method for enhancing the longitudinal properties of lightweight alloy solid-phase additive manufacturing is applied to a longitudinal strengthening system. The longitudinal strengthening system includes an additive manufacturing base plate, a friction stir additive manufacturing actuator, a spindle control unit, and a data processing and control unit. The method includes: S201: Obtain the material to be added in solid-state additive manufacturing and input the material to be added in the longitudinal strengthening system. In the longitudinal strengthening system, call the process recommendation model corresponding to the material to be added in the system to perform process recommendation processing and obtain the recommended process parameters of the material to be added in the system. In the specific implementation of this invention, the process recommendation model is a training dataset constructed using historical additive manufacturing data of the corresponding additive manufacturing material, and the training dataset is used to train a preset model. The preset model includes a feature extraction layer and a parameter optimization layer, wherein the feature extraction layer uses a convolutional neural network or a stacked autoencoder network, and the parameter optimization layer uses a genetic algorithm or a Bayesian algorithm for optimization. The training dataset constructed using historical additive manufacturing data of the corresponding additive manufacturing material includes: obtaining each set of historical process parameters in the corresponding additive manufacturing material and the longitudinal mechanical property parameters corresponding to each set of historical process parameter data; establishing a mapping relationship between each set of historical process parameters and the longitudinal mechanical property parameters corresponding to each set of historical process parameter data to form the training dataset.

[0024] Furthermore, the step of calling the process recommendation model corresponding to the material to be additively manufactured within the longitudinal strengthening system to perform process recommendation processing and obtain recommended process parameters for the material to be additively manufactured includes: calling the process recommendation model corresponding to the material to be additively manufactured within the longitudinal strengthening system, and using the feature extraction layer within the process recommendation model for output processing to obtain output pressure, output temperature data, and output original process parameters; performing process optimization definition processing within the parameter optimization layer based on the output pressure, output temperature data, and output original process parameters, and using a coding genetic algorithm or Bayesian algorithm to optimize according to the process optimization definition results to form recommended process parameters for the material to be additively manufactured.

[0025] Furthermore, the process optimization definition process based on output pressure, output temperature data, and original output process parameters within the parameter optimization layer includes: obtaining the interlayer tensile shear strength and the thickness of the intermetallic compound layer based on the output pressure, output temperature data, and original output process parameters within the parameter optimization layer, and defining the variables as follows:

[0026] in, The rotational speed of the agitator head in the friction stir additive actuator is [900, 1500] r / min; The travel speed of the agitator head in the friction stir additive actuator is [20, 200] mm / min; The downward pressure of the agitator head in the friction stir additive actuator ranges from [0.1, 1] mm. The constraint condition is: peak axial pressure ; Obtained from pressure threshold window calibration; peak temperature , Let be the solidus temperature of the alloy; the process optimization is transformed into a single-objective optimization problem using a weighted method, as follows: ; in, and These represent the maximum tensile shear strength and the minimum intermetallic compound layer thickness in the historical dataset; where... , As weight, , .

[0027] Furthermore, the optimization based on the process optimization definition result using a coded genetic algorithm includes: encoding and initializing the process optimization definition result, and then constructing a fitness function based on the encoding and initialization result; after completing the construction of the fitness function, performing selection, crossover, mutation operations, and setting optimization termination conditions in sequence; and outputting recommended process parameters formed by the coded genetic algorithm optimization based on the fitness function.

[0028] Furthermore, the optimization based on the process optimization definition using a Bayesian algorithm includes: using a Gaussian process as a surrogate model for the single-objective optimization function in the process optimization definition, and using Latin hypercube sampling to generate several initial observation points to train the surrogate model; during the training process, using the expected improvement method, in each iteration, the next sampling point is determined by maximizing the acquisition function; the acquisition function is optimized using the L-BFGS-B algorithm combined with 5 random restarts to maximize the expected improvement; the surrogate model is iteratively updated based on the optimized acquisition function, and an iterative update termination condition is set, and recommended process parameters are formed when the termination condition is met.

[0029] Specifically, the first step is to construct a pre-defined model, which includes a feature extraction layer and a parameter optimization layer. The feature extraction layer uses a convolutional neural network or a stacked autoencoder network, while the parameter optimization layer uses a genetic algorithm or a Bayesian algorithm for optimization. The feature extraction layer is used to perform dimensionality reduction and key feature extraction on historical process parameters and original process signals (such as pressure-time series and temperature field features). The parameter optimization layer is used to optimize the parameters using a genetic algorithm or Bayesian algorithm, with the optimization objectives being to maximize the interlayer tensile shear load and minimize the thickness of the intermetallic compound layer at the interface, and to output the optimal combination of process parameters.

[0030] The next step is to build a training dataset, train the preset model using the training dataset, and then form a process recommendation model after training is completed.

[0031] When constructing the training dataset, it is necessary to collect historical additive manufacturing data, specifically including different historical process parameters (rotation speed 900-1500 r / min, travel speed 50-200 mm / min, compression 0.1-0.5 mm) for different additive manufacturing materials, as well as their corresponding process signals (axial pressure, torque, temperature) and longitudinal mechanical property parameters (interlaminar tensile shear strength, interface microstructure), etc. Then, by establishing the mapping relationship between each set of historical process parameters, the corresponding machine process signals, and the corresponding longitudinal mechanical property parameters, a training dataset is formed. When the training data is input into a preset model for training, its features are used... The extraction layer extracts features from pressure-time series and temperature field images, and then inputs the extracted features into the parameter optimization layer. A genetic algorithm or Bayesian algorithm is then used for multi-objective optimization to establish a process recommendation model from process parameters to interlaminar tensile-shear strength. This allows the operator to input the material specifications (AA6061) to be manufactured into the longitudinal strengthening system at the start of additive manufacturing. The longitudinal strengthening system then calls the corresponding process recommendation model for the material to be manufactured to obtain recommended process parameters, such as: rotational speed 1200 r / min, travel speed 120 mm / min, and preset compression amount 0.3 mm.

[0032] Within the vertical reinforcement system, a process recommendation model corresponding to the material to be additively manufactured is invoked. The feature extraction layer within the process recommendation model is used for output processing to obtain output pressure, output temperature data, and original output process parameters. Finally, the output pressure, output temperature data, and original output process parameters are used in the parameter optimization layer for process optimization definition processing. Based on the process optimization definition results, a genetic algorithm or Bayesian algorithm is used for optimization to form recommended process parameters for the material to be additively manufactured.

[0033] The process optimization definition process involves inputting the pressure and temperature data output from the feature extraction layer along with the original process parameters into the parameter optimization layer to obtain two target predicted values: interlayer tensile shear strength. (Unit: MPa, maximum), thickness of intermetallic compound layer at the interface (Unit: μm, Minimize), Decision variables are: ;in, The rotational speed of the agitator head in the friction stir additive actuator ranges from [900, 1500] r / min. The travel speed of the agitator head in the friction stir additive actuator is [20, 200] mm / min; The downward pressure of the agitator head in the friction stir additive actuator ranges from [0.1, 1] mm.

[0034] The constraint condition is: peak axial pressure ; Obtained from pressure threshold window calibration; peak temperature , is the solidus temperature of the alloy.

[0035] The process optimization problem is transformed into a single-objective optimization problem using a weighted method, as follows: ; in, and These represent the maximum tensile shear strength and the minimum intermetallic compound layer thickness in the historical dataset; where... , As weight, , .

[0036] Then, based on the process optimization definition, a coding genetic algorithm is used for optimization. Specifically, when the process recommendation model exhibits non-convex or multi-peak patterns, a real-number coding genetic algorithm is used for global optimization. Encoding and initialization are performed using real-number encoding, with each individual directly represented as... Initial population: N generated by Latin hypercube sampling (LHS) pop = 50 individuals; LHS can cover the search space more evenly than random sampling, avoiding initial population clustering; Elite retention: retained in each generation N elite = 2 optimal individuals, directly replicated to the next generation.

[0037] Fitness function, the fitness function is directly adopted For individuals that do not meet the constraints (e.g., predicted peak pressure exceeds the pressure window), a penalty term is applied: ; Among them, the penalty coefficient λ 1= λ 2 = 10; This penalty function significantly reduces the fitness of violating individuals, ensuring that feasible solutions are selected first during the search process.

[0038] The selection operation uses tournament selection with a tournament size of k=3. The specific steps are as follows: randomly select 3 individuals from the population and choose the one with the highest fitness as the parent; repeat this process until the parent pool is full. If the tournament size is 2, the selection pressure is insufficient and the convergence is slow; if it is 4, it is prone to premature convergence, and 3 is the optimal size.

[0039] The interleaving operation uses analog binary interleaving (SBX), which is suitable for real number encoding; it involves two parent numbers. , Produce offspring , The formula is as follows: ; ; ; in Distribution index ;SBX can generate offspring that are interpolated within the parent interval, maintaining the reasonableness of the variable range; Experiments have verified that this method is optimal, allowing for both thorough exploration and preservation of population diversity.

[0040] The mutation operation employs polynomial mutation; for each dimension of the offspring individual... With probability = 0.1 to perform mutation: ; Where δ is generated by a multinomial distribution: ; Distribution index The variable length is approximately 1% of the search range; if it is too large, it will destroy the superior genes, and if it is too small, it will be unable to escape the local optimum.

[0041] Termination conditions: The process terminates when any of the following conditions are met: the number of iterations reaches 50; or the absolute value of the change in optimal fitness over 10 consecutive generations is less than 0.001.

[0042] Output the optimal individual. This is the optimal combination of process parameters.

[0043] Based on the process optimization definition, Bayesian algorithm is used for optimization. When the process recommendation model has high calling cost or limited experimental resources, Bayesian optimization is used as an alternative, which can converge with fewer evaluations.

[0044] The surrogate model uses a Gaussian process (GP) as the objective function. The proxy model. GP is composed of the mean function. m ( and covariance function Confirmed. This invention uses the Matern 5 / 2 kernel function: ; in , For signal variance, For length scales; the Matern 5 / 2 core, compared to the RBF core, can adapt to rougher response surfaces, making it more suitable for the potentially non-smooth relationship between solid-state additive manufacturing process parameters and performance. Hyperparameters , It is obtained by maximizing the marginal log-likelihood estimate.

[0045] The initial design uses Latin hypercube sampling to generate... Initial observation points Use these points to train the initial GP model.

[0046] The acquisition function, Expected Improvement (EI), is used in each iteration to determine the next sampling point by maximizing the acquisition function. In this embodiment, Expected Improvement (EI) is used, and its expression is: ; in, and These are the mean and standard deviation of the GP predictions, respectively. This is the current best observation value; To explore and develop balance coefficients; , These are the cumulative distribution function and probability density function of the standard normal distribution, respectively.

[0047] To encourage exploration of regions with high forecast uncertainty; To determine through sensitivity analysis, if Full development can easily lead to local optima; if Overexploration will double the convergent algebra.

[0048] The sampling function is optimized using the L-BFGS-B algorithm combined with 5 random restarts to maximize the efficiency (EI). The specific steps are: 1. Randomly generate 5 initial points in the search space; 2. Run L-BFGS-B (an extended version of the L-BFGS (Limited-memory Broyden–Fletcher–Goldfarb–Shanno) algorithm, specifically designed for handling optimization problems with boundary constraints; where "B" stands for Boundary, allowing upper and lower bounds on the optimization variables) for each initial point to perform local optimization (gradients are obtained through automatic differentiation); 3. Select the point with the largest EI from the 5 local optima as the next sampling point. .

[0049] If the number of random restarts is less than 3, the global optimum is easily missed; if it is more than 7, there is computational redundancy; 5 is the optimal number.

[0050] Update the surrogate model and evaluate it on the true objective function (i.e., the invocation prediction model). Add the new sample Retrain the GP (update the hyperparameters).

[0051] Termination conditions: Termination will occur if any of the following conditions are met: The total number of assessments (including the initial point) reaches [a certain threshold]. The improvement in the optimal value after 15 consecutive iterations is less than 0.5%.

[0052] Output the result, output the result The largest (posterior mean) (Or select the evaluated point with the highest EI), which is the optimal combination of process parameters; the genetic algorithm and Bayesian optimization are not arbitrarily selected, but automatically switched according to the actual working conditions: when there is sufficient offline training time, GA is used to perform 50 generations of global search to ensure that the global optimum is found; when each model call takes a long time or needs to be combined with a small number of actual experiments for verification, BO is used to converge efficiently within 70 evaluations.

[0053] S202: The recommended process parameters are used as the initial settings of the longitudinal strengthening system. The longitudinal strengthening system performs solid-phase additive manufacturing of the material to be additively manufactured according to the initial settings based on the adaptive depth control of pressure feedback, thereby forming a solid-phase additive manufacturing deposition layer of the material to be additively manufactured. In a specific implementation of this invention, the longitudinal strengthening system, based on pressure feedback and adaptive pressure depth control, performs solid-phase additive manufacturing of the material to be added according to the initial set value, forming a solid-phase additive manufacturing deposition layer of the material to be added. This includes: setting a pressure threshold window using the axial pressure of the main shaft of the longitudinal strengthening system as an indirect characterization parameter of the pressure depth state, while simultaneously detecting the axial pressure of the main shaft in real time; when the axial pressure of the main shaft is lower than the lower limit of the pressure threshold window, it is determined that there is poor interlayer contact or insufficient plasticization, and the main shaft control unit drives the additive stirring head in the friction stir actuator to increase the downward pressure by a first preset amount to perform the solid-phase additive manufacturing of the material to be added, forming a solid-phase additive manufacturing deposition layer of the material to be added; when the axial pressure of the main shaft is higher than the upper limit of the pressure threshold window, it is determined that the upsetting force is too large, and the main shaft control unit drives the additive stirring head in the friction stir actuator to rise by a second preset amount to perform the solid-phase additive manufacturing of the material to be added, forming a solid-phase additive manufacturing deposition layer of the material to be added.

[0054] Specifically, after obtaining the recommended process parameters, these parameters are used as the initial settings for the longitudinal strengthening system. The longitudinal strengthening system then performs solid-state additive manufacturing using an adaptive pressure depth control method based on pressure feedback. That is, during the additive deposition process, the spindle axial pressure is used as an indirect characterization parameter of the pressure depth state, and a pressure threshold window is set. When the spindle axial pressure is lower than the lower limit of the pressure threshold window, it is determined that there is poor interlayer contact or insufficient plasticization. Then, the spindle control unit drives the additive stirring head in the friction stir actuator to increase the downward pressure by a first preset amount to perform solid-state additive manufacturing of the material to be added, forming a solid-state additive manufacturing deposition layer of the material to be added. When the spindle axial pressure is higher than the upper limit of the pressure threshold window, it is determined that the upsetting force is too large. The spindle control unit drives the additive stirring head in the friction stir actuator to rise by a second preset amount to perform solid-state additive manufacturing of the material to be added, forming a solid-state additive manufacturing deposition layer of the material to be added.

[0055] refer to Figure 3 As shown, during the first layer deposition in the additive manufacturing process, the additive stirring head rotates and presses down, while the wire feeding mechanism feeds it in simultaneously; the force sensor collects the axial pressure F(t) in real time; the longitudinal reinforcement system has a built-in pressure threshold window calibrated based on historical excellent deposition data; for AA6061 material, the calibration window is [2.8kN, 3.4kN]; when the stirring head moves to a certain position, the pressure monitoring value is 2.6kN, which is lower than the lower limit of the window; it is determined that the interlayer contact is insufficient, and a correction signal is immediately output to drive the stirring head to slightly increase the downward pressure; when the downward pressure increases to 0.35mm, the axial pressure rises back to 3.1kN, falling within the threshold window; the current depth is locked, and this pressure state is maintained to complete the deposition of this layer.

[0056] S203: Based on preset processing parameters, the solid additive manufacturing deposition layer is strengthened by friction stirring processing.

[0057] In the specific implementation of this invention, after the deposition of a single or multi-layer additive manufacturing process is completed, a stirring friction processing step is performed along the surface of the deposited layer; through secondary stirring, the surface structure is refined, the interface oxides are broken, and additional longitudinal upsetting force is introduced to further strengthen the interlayer bonding; the processing parameters (rotation speed, travel speed, and downsetting amount) and the additive deposition stage parameters can be set independently to meet the needs of surface structure refinement.

[0058] That is, after the Nth layer is deposited, the system will automatically switch to the friction stir processing enhancement mode according to the settings, and perform a wire-free friction stir processing along the surface of the deposited layer. The processing parameters are set as follows: rotation speed 1000 r / min, travel speed 200 mm / min, and downward pressure 0.1 mm. Through the secondary stirring action, the surface structure undergoes dynamic recrystallization, the grains are further refined, and at the same time, the longitudinal upsetting force acts on the interlayer interface, which strengthens the bonding effect.

[0059] In this embodiment of the invention, adaptive optimization of process parameters is achieved, overcoming the limitations of traditional trial-and-error methods in heterogeneous material additive manufacturing. By using axial pressure as an indirect criterion for the pressing depth, the effects of thermal deformation and fixture rigidity fluctuations can be effectively compensated, preventing interlayer bonding defects caused by excessive or shallow insertion. Immediately after solid-phase additive manufacturing, friction stir processing is applied to strengthen the structure. The secondary stirring action refines the microstructure and introduces longitudinal upsetting force, further improving the interlayer bonding performance and making the longitudinal mechanical properties of the component approach the level of homogeneous forgings.

[0060] Example 2, please refer to Figure 4 , Figure 4 This is a schematic diagram of the structural composition of the lightweight alloy solid-phase additive manufacturing longitudinal performance enhancement device in an embodiment of the present invention.

[0061] like Figure 4 As shown, a longitudinal performance enhancement device for solid-state additive manufacturing of lightweight alloys is applied to a longitudinal performance enhancement system. The longitudinal performance enhancement system includes an additive manufacturing base plate, a friction stir additive manufacturing actuator, a spindle control unit, and a data processing and control unit. The method includes: Process recommendation module 401: used to obtain the material to be added in solid-state additive manufacturing, input the material to be added in the longitudinal strengthening system, call the process recommendation model corresponding to the material to be added in the longitudinal strengthening system to perform process recommendation processing, and obtain the recommended process parameters of the material to be added. In the specific implementation of this invention, the process recommendation model is a training dataset constructed using historical additive manufacturing data of the corresponding additive manufacturing material, and the training dataset is used to train a preset model. The preset model includes a feature extraction layer and a parameter optimization layer, wherein the feature extraction layer uses a convolutional neural network or a stacked autoencoder network, and the parameter optimization layer uses a genetic algorithm or a Bayesian algorithm for optimization. The training dataset constructed using historical additive manufacturing data of the corresponding additive manufacturing material includes: obtaining each set of historical process parameters in the corresponding additive manufacturing material and the longitudinal mechanical property parameters corresponding to each set of historical process parameter data; establishing a mapping relationship between each set of historical process parameters and the longitudinal mechanical property parameters corresponding to each set of historical process parameter data to form the training dataset.

[0062] Furthermore, the step of calling the process recommendation model corresponding to the material to be additively manufactured within the longitudinal strengthening system to perform process recommendation processing and obtain recommended process parameters for the material to be additively manufactured includes: calling the process recommendation model corresponding to the material to be additively manufactured within the longitudinal strengthening system, and using the feature extraction layer within the process recommendation model for output processing to obtain output pressure, output temperature data, and output original process parameters; performing process optimization definition processing within the parameter optimization layer based on the output pressure, output temperature data, and output original process parameters, and using a coding genetic algorithm or Bayesian algorithm to optimize according to the process optimization definition results to form recommended process parameters for the material to be additively manufactured.

[0063] Furthermore, the process optimization definition process based on output pressure, output temperature data, and original output process parameters within the parameter optimization layer includes: obtaining the interlayer tensile shear strength and the thickness of the intermetallic compound layer based on the output pressure, output temperature data, and original output process parameters within the parameter optimization layer, and defining the variables as follows:

[0064] in, The rotational speed of the agitator head in the friction stir additive actuator is [900, 1500] r / min; The travel speed of the agitator head in the friction stir additive actuator is [20, 200] mm / min; The downward pressure of the agitator head in the friction stir additive actuator ranges from [0.1, 1] mm. The constraint condition is: peak axial pressure ; Obtained from pressure threshold window calibration; peak temperature , Let be the solidus temperature of the alloy; the process optimization is transformed into a single-objective optimization problem using a weighted method, as follows: ; in, and These represent the maximum tensile shear strength and the minimum intermetallic compound layer thickness in the historical dataset; where... , As weight, , .

[0065] Furthermore, the optimization based on the process optimization definition result using a coded genetic algorithm includes: encoding and initializing the process optimization definition result, and then constructing a fitness function based on the encoding and initialization result; after completing the construction of the fitness function, performing selection, crossover, mutation operations, and setting optimization termination conditions in sequence; and outputting recommended process parameters formed by the coded genetic algorithm optimization based on the fitness function.

[0066] Furthermore, the optimization based on the process optimization definition using a Bayesian algorithm includes: using a Gaussian process as a surrogate model for the single-objective optimization function in the process optimization definition, and using Latin hypercube sampling to generate several initial observation points to train the surrogate model; during the training process, using the expected improvement method, in each iteration, the next sampling point is determined by maximizing the acquisition function; the acquisition function is optimized using the L-BFGS-B algorithm combined with 5 random restarts to maximize the expected improvement; the surrogate model is iteratively updated based on the optimized acquisition function, and an iterative update termination condition is set, and recommended process parameters are formed when the termination condition is met.

[0067] Specifically, the first step is to construct a pre-defined model, which includes a feature extraction layer and a parameter optimization layer. The feature extraction layer uses a convolutional neural network or a stacked autoencoder network, while the parameter optimization layer uses a genetic algorithm or a Bayesian algorithm for optimization. The feature extraction layer is used to perform dimensionality reduction and key feature extraction on historical process parameters and original process signals (such as pressure-time series and temperature field features). The parameter optimization layer is used to optimize the parameters using a genetic algorithm or Bayesian algorithm, with the optimization objectives being to maximize the interlayer tensile shear load and minimize the thickness of the intermetallic compound layer at the interface, and to output the optimal combination of process parameters.

[0068] The next step is to build a training dataset, train the preset model using the training dataset, and then form a process recommendation model after training is completed.

[0069] When constructing the training dataset, it is necessary to collect historical additive manufacturing data, specifically including different historical process parameters (rotation speed 900-1500 r / min, travel speed 50-200 mm / min, compression 0.1-0.5 mm) for different additive manufacturing materials, as well as their corresponding process signals (axial pressure, torque, temperature) and longitudinal mechanical property parameters (interlaminar tensile shear strength, interface microstructure), etc. Then, by establishing the mapping relationship between each set of historical process parameters, the corresponding machine process signals, and the corresponding longitudinal mechanical property parameters, a training dataset is formed. When the training data is input into a preset model for training, its features are used... The extraction layer extracts features from pressure-time series and temperature field images, and then inputs the extracted features into the parameter optimization layer. A genetic algorithm or Bayesian algorithm is then used for multi-objective optimization to establish a process recommendation model from process parameters to interlaminar tensile-shear strength. This allows the operator to input the material specifications (AA6061) to be manufactured into the longitudinal strengthening system at the start of additive manufacturing. The longitudinal strengthening system then calls the corresponding process recommendation model for the material to be manufactured to obtain recommended process parameters, such as: rotational speed 1200 r / min, travel speed 120 mm / min, and preset compression amount 0.3 mm.

[0070] Within the vertical reinforcement system, a process recommendation model corresponding to the material to be additively manufactured is invoked. The feature extraction layer within the process recommendation model is used for output processing to obtain output pressure, output temperature data, and original output process parameters. Finally, the output pressure, output temperature data, and original output process parameters are used in the parameter optimization layer for process optimization definition processing. Based on the process optimization definition results, a genetic algorithm or Bayesian algorithm is used for optimization to form recommended process parameters for the material to be additively manufactured.

[0071] The process optimization definition process involves inputting the pressure and temperature data output from the feature extraction layer along with the original process parameters into the parameter optimization layer to obtain two target predicted values: interlayer tensile shear strength. (Unit: MPa, maximum), thickness of intermetallic compound layer at the interface (Unit: μm, Minimize), Decision variables are: ;in, The rotational speed of the agitator head in the friction stir additive actuator ranges from [900, 1500] r / min. The travel speed of the agitator head in the friction stir additive actuator is [20, 200] mm / min; The downward pressure of the agitator head in the friction stir additive actuator ranges from [0.1, 1] mm.

[0072] The constraint condition is: peak axial pressure ; Obtained from pressure threshold window calibration; peak temperature , is the solidus temperature of the alloy.

[0073] The process optimization problem is transformed into a single-objective optimization problem using a weighted method, as follows: ; in, and These represent the maximum tensile shear strength and the minimum intermetallic compound layer thickness in the historical dataset; where... , As weight, , .

[0074] Then, based on the process optimization definition, a coding genetic algorithm is used for optimization. Specifically, when the process recommendation model exhibits non-convex or multi-peak patterns, a real-number coding genetic algorithm is used for global optimization. Encoding and initialization are performed using real-number encoding, with each individual directly represented as... Initial population: N generated by Latin hypercube sampling (LHS) pop = 50 individuals; LHS can cover the search space more evenly than random sampling, avoiding initial population clustering; Elite retention: retained in each generation N elite = 2 optimal individuals, directly replicated to the next generation.

[0075] Fitness function, the fitness function is directly adopted For individuals that do not meet the constraints (e.g., predicted peak pressure exceeds the pressure window), a penalty term is applied: ; Among them, the penalty coefficient λ 1= λ 2 = 10; This penalty function significantly reduces the fitness of violating individuals, ensuring that feasible solutions are selected first during the search process.

[0076] The selection operation uses tournament selection with a tournament size of k=3. The specific steps are as follows: randomly select 3 individuals from the population and choose the one with the highest fitness as the parent; repeat this process until the parent pool is full. If the tournament size is 2, the selection pressure is insufficient and the convergence is slow; if it is 4, it is prone to premature convergence, and 3 is the optimal size.

[0077] The interleaving operation uses analog binary interleaving (SBX), which is suitable for real number encoding; it involves two parent numbers. , Produce offspring , The formula is as follows: ; ; ; in Distribution index ;SBX can generate offspring that are interpolated within the parent interval, maintaining the reasonableness of the variable range; Experiments have verified that this method is optimal, allowing for both thorough exploration and preservation of population diversity.

[0078] The mutation operation employs polynomial mutation; for each dimension of the offspring individual... With probability = 0.1 to perform mutation: ; Where δ is generated by a multinomial distribution: ; Distribution index The variable length is approximately 1% of the search range; if it is too large, it will destroy the superior genes, and if it is too small, it will be unable to escape the local optimum.

[0079] Termination conditions: The process terminates when any of the following conditions are met: the number of iterations reaches 50; or the absolute value of the change in optimal fitness over 10 consecutive generations is less than 0.001.

[0080] Output the optimal individual. This is the optimal combination of process parameters.

[0081] Based on the process optimization definition, Bayesian algorithm is used for optimization. When the process recommendation model has high calling cost or limited experimental resources, Bayesian optimization is used as an alternative, which can converge with fewer evaluations.

[0082] The surrogate model uses a Gaussian process (GP) as the objective function. The proxy model. GP is composed of the mean function. m ( and covariance function Confirmed. This invention uses the Matern 5 / 2 kernel function: ; in , For signal variance, For length scales; the Matern 5 / 2 core, compared to the RBF core, can adapt to rougher response surfaces, making it more suitable for the potentially non-smooth relationship between solid-state additive manufacturing process parameters and performance. Hyperparameters , It is obtained by maximizing the marginal log-likelihood estimate.

[0083] The initial design uses Latin hypercube sampling to generate... Initial observation points Use these points to train the initial GP model.

[0084] The acquisition function, Expected Improvement (EI), is used in each iteration to determine the next sampling point by maximizing the acquisition function. In this embodiment, Expected Improvement (EI) is used, and its expression is: ; in, and These are the mean and standard deviation of the GP predictions, respectively. This is the current best observation value; To explore and develop balance coefficients; , These are the cumulative distribution function and probability density function of the standard normal distribution, respectively.

[0085] To encourage exploration of regions with high forecast uncertainty; To determine through sensitivity analysis, if Full development can easily lead to local optima; if Overexploration will double the convergent algebra.

[0086] The sampling function is optimized using the L-BFGS-B algorithm combined with 5 random restarts to maximize the efficiency (EI). The specific steps are: 1. Randomly generate 5 initial points in the search space; 2. Run L-BFGS-B (an extended version of the L-BFGS (Limited-memory Broyden–Fletcher–Goldfarb–Shanno) algorithm, specifically designed for handling optimization problems with boundary constraints; where "B" stands for Boundary, allowing upper and lower bounds on the optimization variables) for each initial point to perform local optimization (gradients are obtained through automatic differentiation); 3. Select the point with the largest EI from the 5 local optima as the next sampling point. .

[0087] If the number of random restarts is less than 3, the global optimum is easily missed; if it is more than 7, there is computational redundancy; 5 is the optimal number.

[0088] Update the surrogate model and evaluate it on the true objective function (i.e., the invocation prediction model). Add the new sample Retrain the GP (update the hyperparameters).

[0089] Termination conditions: Termination will occur if any of the following conditions are met: The total number of assessments (including the initial point) reaches [a certain threshold]. The improvement in the optimal value after 15 consecutive iterations is less than 0.5%.

[0090] Output the result, output the result The largest (posterior mean) (Or select the evaluated point with the highest EI), which is the optimal combination of process parameters; the genetic algorithm and Bayesian optimization are not arbitrarily selected, but automatically switched according to the actual working conditions: when there is sufficient offline training time, GA is used to perform 50 generations of global search to ensure that the global optimum is found; when each model call takes a long time or needs to be combined with a small number of actual experiments for verification, BO is used to converge efficiently within 70 evaluations.

[0091] Additive manufacturing module 402: used to use the recommended process parameters as the initial set values ​​of the longitudinal reinforcement system, the longitudinal reinforcement system based on pressure feedback adaptive depth control to perform solid-phase additive manufacturing processing of the material to be additively manufactured according to the initial set values, to form a solid-phase additive manufacturing deposition layer of the material to be additively manufactured; In a specific implementation of this invention, the longitudinal strengthening system, based on pressure feedback and adaptive pressure depth control, performs solid-phase additive manufacturing of the material to be added according to the initial set value, forming a solid-phase additive manufacturing deposition layer of the material to be added. This includes: setting a pressure threshold window using the axial pressure of the main shaft of the longitudinal strengthening system as an indirect characterization parameter of the pressure depth state, while simultaneously detecting the axial pressure of the main shaft in real time; when the axial pressure of the main shaft is lower than the lower limit of the pressure threshold window, it is determined that there is poor interlayer contact or insufficient plasticization, and the main shaft control unit drives the additive stirring head in the friction stir actuator to increase the downward pressure by a first preset amount to perform the solid-phase additive manufacturing of the material to be added, forming a solid-phase additive manufacturing deposition layer of the material to be added; when the axial pressure of the main shaft is higher than the upper limit of the pressure threshold window, it is determined that the upsetting force is too large, and the main shaft control unit drives the additive stirring head in the friction stir actuator to rise by a second preset amount to perform the solid-phase additive manufacturing of the material to be added, forming a solid-phase additive manufacturing deposition layer of the material to be added.

[0092] Specifically, after obtaining the recommended process parameters, these parameters are used as the initial settings for the longitudinal strengthening system. The longitudinal strengthening system then performs solid-state additive manufacturing using an adaptive pressure depth control method based on pressure feedback. That is, during the additive deposition process, the spindle axial pressure is used as an indirect characterization parameter of the pressure depth state, and a pressure threshold window is set. When the spindle axial pressure is lower than the lower limit of the pressure threshold window, it is determined that there is poor interlayer contact or insufficient plasticization. Then, the spindle control unit drives the additive stirring head in the friction stir actuator to increase the downward pressure by a first preset amount to perform solid-state additive manufacturing of the material to be added, forming a solid-state additive manufacturing deposition layer of the material to be added. When the spindle axial pressure is higher than the upper limit of the pressure threshold window, it is determined that the upsetting force is too large. The spindle control unit drives the additive stirring head in the friction stir actuator to rise by a second preset amount to perform solid-state additive manufacturing of the material to be added, forming a solid-state additive manufacturing deposition layer of the material to be added.

[0093] refer to Figure 3 As shown, during the first layer deposition in the additive manufacturing process, the additive stirring head rotates and presses down, while the wire feeding mechanism feeds it in simultaneously; the force sensor collects the axial pressure F(t) in real time; the longitudinal reinforcement system has a built-in pressure threshold window calibrated based on historical excellent deposition data; for AA6061 material, the calibration window is [2.8kN, 3.4kN]; when the stirring head moves to a certain position, the pressure monitoring value is 2.6kN, which is lower than the lower limit of the window; it is determined that the interlayer contact is insufficient, and a correction signal is immediately output to drive the stirring head to slightly increase the downward pressure; when the downward pressure increases to 0.35mm, the axial pressure rises back to 3.1kN, falling within the threshold window; the current depth is locked, and this pressure state is maintained to complete the deposition of this layer.

[0094] Strengthening processing module 403: used to strengthen the solid additive manufacturing deposited layer by means of friction stirring processing based on preset processing parameters.

[0095] In the specific implementation of this invention, after the deposition of a single or multi-layer additive manufacturing process is completed, a stirring friction processing step is performed along the surface of the deposited layer; through secondary stirring, the surface structure is refined, the interface oxides are broken, and additional longitudinal upsetting force is introduced to further strengthen the interlayer bonding; the processing parameters (rotation speed, travel speed, and downsetting amount) and the additive deposition stage parameters can be set independently to meet the needs of surface structure refinement.

[0096] That is, after the Nth layer is deposited, the system will automatically switch to the friction stir processing enhancement mode according to the settings, and perform a wire-free friction stir processing along the surface of the deposited layer. The processing parameters are set as follows: rotation speed 1000 r / min, travel speed 200 mm / min, and downward pressure 0.1 mm. Through the secondary stirring action, the surface structure undergoes dynamic recrystallization, the grains are further refined, and at the same time, the longitudinal upsetting force acts on the interlayer interface, which strengthens the bonding effect.

[0097] In this embodiment of the invention, adaptive optimization of process parameters is achieved, overcoming the limitations of traditional trial-and-error methods in heterogeneous material additive manufacturing. By using axial pressure as an indirect criterion for the pressing depth, the effects of thermal deformation and fixture rigidity fluctuations can be effectively compensated, preventing interlayer bonding defects caused by excessive or shallow insertion. Immediately after solid-phase additive manufacturing, friction stir processing is applied to strengthen the structure. The secondary stirring action refines the microstructure and introduces longitudinal upsetting force, further improving the interlayer bonding performance and making the longitudinal mechanical properties of the component approach the level of homogeneous forgings.

[0098] This invention provides a computer-readable storage medium storing a computer program. When executed by a processor, the program implements the longitudinal performance enhancement method for solid-state additive manufacturing of lightweight alloys according to any of the above embodiments. The computer-readable storage medium includes, but is not limited to, any type of disk (including floppy disks, hard disks, optical disks, CD-ROMs, and magneto-optical disks), ROM (Read-Only Memory), RAM (Random Access Memory), EPROM (Erasable Programmable Read-Only Memory), EEPROM (Electrically Erasable Programmable Read-Only Memory), flash memory, magnetic cards, or optical cards. In other words, the storage device includes any medium that stores or transmits information in a readable form by a device (e.g., a computer, a mobile phone), and can be a read-only memory, a disk, or an optical disk, etc.

[0099] This invention also provides a computer application running on a computer, which is used to execute the longitudinal performance enhancement method for lightweight alloy solid-phase additive manufacturing in any of the above embodiments.

[0100] also, Figure 5 This is a schematic diagram of the structural composition of the electronic device in an embodiment of the present invention.

[0101] This invention also provides an electronic device, such as... Figure 5As shown. The electronic device includes a processor 502, a memory 503, an input unit 504, and a display unit 505, among other devices. Those skilled in the art will understand that... Figure 5 The structural components of the illustrated electronic device do not constitute a limitation on all devices and may include more or fewer components than illustrated, or combine certain components. Memory 503 can be used to store application program 501 and various functional modules. Processor 502 runs application program 501 stored in memory 503, thereby performing various functional applications and data processing of the device. Memory can be internal memory or external memory, or both. Internal memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), flash memory, or random access memory. External memory may include hard disks, floppy disks, ZIP disks, USB flash drives, magnetic tapes, etc. The memory disclosed in this invention includes, but is not limited to, these types of memory. The memory disclosed in this invention is only an example and not a limitation.

[0102] Input unit 504 is used to receive signal input and user-input keywords. Input unit 504 may include a touch panel and other input devices. The touch panel can collect user touch operations on or near it (such as operations performed by the user using a finger, stylus, or any suitable object or accessory on or near the touch panel) and drive the corresponding connection device according to a pre-set program; other input devices may include, but are not limited to, one or more of physical keyboards, function keys (such as play control buttons, power buttons, etc.), trackballs, mice, joysticks, etc. Display unit 505 can be used to display user-input information or information provided to the user, as well as various menus of the terminal device. Display unit 505 may be in the form of a liquid crystal display, organic light-emitting diode, etc. Processor 502 is the control center of the terminal device, connecting various parts of the entire device through various interfaces and lines, performing various functions and processing data by running or executing software programs and / or modules stored in memory 503, and calling data stored in memory.

[0103] As one embodiment, the electronic device includes: one or more processors 502, a memory 503, and one or more application programs 501, wherein the one or more application programs 501 are stored in the memory 503 and configured to be executed by the one or more processors 502, and the one or more application programs 501 are configured to perform the longitudinal performance enhancement method for lightweight alloy solid-phase additive manufacturing corresponding to any of the embodiments described above.

[0104] In this embodiment of the invention, adaptive optimization of process parameters is achieved, overcoming the limitations of traditional trial-and-error methods in heterogeneous material additive manufacturing. By using axial pressure as an indirect criterion for the pressing depth, the effects of thermal deformation and fixture rigidity fluctuations can be effectively compensated, preventing interlayer bonding defects caused by excessive or shallow insertion. Immediately after solid-phase additive manufacturing, friction stir processing is applied to strengthen the structure. The secondary stirring action refines the microstructure and introduces longitudinal upsetting force, further improving the interlayer bonding performance and making the longitudinal mechanical properties of the component approach the level of homogeneous forgings.

[0105] Furthermore, the above provides a detailed description of the method and related apparatus for enhancing the longitudinal properties of lightweight alloy solid-phase additive manufacturing. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only intended to help understand the method and core ideas of the present invention. At the same time, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for enhancing the longitudinal properties of lightweight alloy solid-phase additive manufacturing, characterized in that, The method is applied to a longitudinal strengthening system, which includes an additive manufacturing base plate, a friction stir additive manufacturing actuator, a spindle control unit, and a data processing and control unit; the longitudinal strengthening system includes: Obtain the material to be added in solid-state additive manufacturing, and input the material to be added in the longitudinal strengthening system. In the longitudinal strengthening system, call the process recommendation model corresponding to the material to be added in the longitudinal strengthening system to perform process recommendation processing and obtain the recommended process parameters of the material to be added in the longitudinal strengthening system. The recommended process parameters are used as the initial settings of the longitudinal strengthening system. The longitudinal strengthening system, based on pressure feedback adaptive depth control, performs solid-state additive manufacturing of the material to be additively manufactured according to the initial settings to form a solid-state additive manufacturing deposition layer of the material to be additively manufactured. The solid additive manufacturing deposition layer is strengthened by using a stirring friction processing method based on preset processing parameters.

2. The method for enhancing the longitudinal properties of lightweight alloy solid-phase additive manufacturing according to claim 1, characterized in that, The process recommendation model is a training dataset constructed using historical additive manufacturing data of the corresponding additive manufacturing materials, and the training dataset is used to train the preset model. The preset model includes a feature extraction layer and a parameter optimization layer, wherein the feature extraction layer uses a convolutional neural network or a stacked autoencoder network, and the parameter optimization layer uses a genetic algorithm or a Bayesian algorithm for optimization. The training dataset constructed using historical additive manufacturing data of the corresponding additive manufacturing materials includes: Obtain each set of historical process parameters in the corresponding additive manufacturing material and the longitudinal mechanical property parameters corresponding to each set of historical process parameter data; Establish a mapping relationship between each set of historical process parameters and the corresponding longitudinal mechanical performance parameters to form the training dataset.

3. The method for enhancing the longitudinal properties of lightweight alloy solid-phase additive manufacturing according to claim 1, characterized in that, The process of calling the process recommendation model corresponding to the material to be additively manufactured within the longitudinal strengthening system to perform process recommendation processing and obtain the recommended process parameters for the material to be additively manufactured includes: Within the longitudinal reinforcement system, a process recommendation model corresponding to the material to be additively manufactured is invoked, and the feature extraction layer within the process recommendation model is used for output processing to obtain output pressure, output temperature data, and output raw process parameters. Based on the output pressure, output temperature data, and output original process parameters, process optimization definition processing is performed in the parameter optimization layer. Based on the process optimization definition results, a coded genetic algorithm or Bayesian algorithm is used for optimization to form the recommended process parameters for the material to be additively manufactured.

4. The method for enhancing the longitudinal properties of lightweight alloy solid-phase additive manufacturing according to claim 3, characterized in that, The process optimization definition process based on output pressure, output temperature data, and original output process parameters within the parameter optimization layer includes: Based on the output pressure, output temperature data, and original process parameters, the interlaminar tensile shear strength and the thickness of the intermetallic compound layer at the interface are obtained within the parameter optimization layer, and the variables are defined as follows: in, The rotational speed of the agitator head in the friction stir additive actuator ranges from [900, 1500] r / min. The travel speed of the agitator head in the friction stir additive actuator is [20, 200] mm / min; The downward pressure of the agitator head in the friction stir additive actuator ranges from [0.1, 1] mm. The constraints are: Peak axial pressure ; Obtained from pressure threshold window calibration; peak temperature , The solidus temperature of the alloy; The process optimization problem is transformed into a single-objective optimization problem using a weighted method, as follows: ； in, and These represent the maximum tensile shear strength and the minimum intermetallic compound layer thickness in the historical dataset; where... , As weight, , .

5. The method for enhancing the longitudinal properties of lightweight alloy solid-phase additive manufacturing according to claim 3, characterized in that, The optimization based on the process optimization definition results using a coded genetic algorithm includes: After encoding and initializing the process optimization definition results, a fitness function is constructed based on the encoding and initialization results; After completing the fitness function construction, the selection operation, crossover operation, mutation operation, and optimization termination condition setting are performed in sequence. Based on the fitness function, the recommended process parameters generated by the encoded genetic algorithm after optimization are output.

6. The method for enhancing the longitudinal properties of lightweight alloy solid-phase additive manufacturing according to claim 3, characterized in that, The optimization based on the process optimization definition results using the Bayesian algorithm includes: A Gaussian process is used as a surrogate model for the single-objective optimization function in the process optimization definition result, and Latin hypercube sampling is used to generate several initial observation points to train the surrogate model. During training, an improvement-oriented approach is adopted to determine the next sampling point in each iteration by maximizing the acquisition function. The acquisition function is optimized by using the L-BFGS-B algorithm combined with the improvement of maximizing the expected value through 5 random restarts. The proxy model is iteratively updated based on the optimized acquisition function, and an iterative update termination condition is set. When the termination condition is met, recommended process parameters are generated.

7. The method for enhancing the longitudinal properties of lightweight alloy solid-phase additive manufacturing according to claim 1, characterized in that, The longitudinal strengthening system, based on pressure feedback and adaptive depth control, performs solid-state additive manufacturing of the material to be additively manufactured according to the initial set value, forming a solid-state additive manufacturing deposition layer of the material to be additively manufactured, including: The axial pressure of the main shaft of the longitudinal reinforcement system is used as an indirect characterization parameter of the compression depth state to set a pressure threshold window, while the axial pressure of the main shaft is detected in real time. When the axial pressure of the spindle is lower than the lower limit of the pressure threshold window, it is determined that the interlayer contact is poor or the plasticization is insufficient. The spindle control unit drives the additive stirring head in the friction stir actuator to increase the downward pressure by a first preset amount to perform solid-phase additive manufacturing of the material to be added, thereby forming a solid-phase additive manufacturing deposition layer of the material to be added. When the axial pressure of the main spindle is higher than the upper limit of the pressure threshold window, it is determined that the forging force is too large. The main spindle control unit drives the additive stirring head in the friction stir additive actuator to rise upward by a second preset amount to perform solid-phase additive manufacturing process on the material to be additively manufactured, forming a solid-phase additive manufacturing deposition layer on the material to be additively manufactured.

8. A device for enhancing the longitudinal properties of lightweight alloy solid-phase additive manufacturing, characterized in that, The method is applied to a longitudinal strengthening system, which includes an additive manufacturing base plate, a friction stir additive manufacturing actuator, a spindle control unit, and a data processing and control unit; the longitudinal strengthening system includes: Process recommendation module: used to obtain the material to be added in solid-state additive manufacturing, input the material to be added in the longitudinal strengthening system, call the process recommendation model corresponding to the material to be added in the longitudinal strengthening system to perform process recommendation processing, and obtain the recommended process parameters of the material to be added; Additive manufacturing module: used to use the recommended process parameters as the initial set values ​​of the longitudinal reinforcement system. The longitudinal reinforcement system, based on pressure feedback adaptive depth control, performs solid-phase additive manufacturing processing on the material to be additively manufactured according to the initial set values ​​to form a solid-phase additive manufacturing deposition layer on the material to be additively manufactured. Strengthening processing module: used to strengthen the solid additive manufacturing deposited layer by means of friction stirring processing based on preset processing parameters.

9. An electronic device comprising a processor and a memory, characterized in that, The processor runs a computer program or code stored in the memory to implement the method for enhancing the longitudinal properties of lightweight alloy solid-phase additive manufacturing as described in any one of claims 1 to 7.

10. A computer-readable storage medium for storing computer programs or code, characterized in that, When the computer program or code is executed by a processor, the method for enhancing the longitudinal properties of lightweight alloy solid-phase additive manufacturing as described in any one of claims 1 to 7 is implemented.

Images