3D printing powder proportion optimization method and system based on artificial intelligence

Abstract

The invention relates to the technical field of intelligent optimization, and discloses a 3D printing powder proportion optimization method and system based on artificial intelligence. The method comprises the following steps: collecting historical data corresponding to a powder material based on a printing process of a 3D printed finished product, and performing multivariate analysis on the historical data to generate a plurality of powder compositions; the material simulation model is used for carrying out property simulation on any powder composition, outputting material property data and extracting a powder ratio meeting the printing requirement; the finished product prediction model performs finished product prediction on any powder ratio, outputs a finished product prediction result, performs matching optimization on the finished product prediction result and ideal finished product data, and screens out an optimal powder ratio and a corresponding optimal printing process; and performance testing is conducted on the actual printing finished product, if the actual printing finished product does not meet the requirement, the step is executed repeatedly, and a new optimal powder proportion and a new optimal printing technology are screened out again. According to the method, the precision of 3D printing powder proportion optimization is improved.

Description

Technical Field

[0001] The present application relates to the field of intelligent optimization technology, and in particular to a 3D printing powder ratio optimization method and system based on artificial intelligence. Background Art

[0002] In the field of 3D printing, the ratio of powder materials plays a crucial role in the quality, performance, and efficiency of the printed product. Traditional powder ratio methods rely primarily on empirical formulas and extensive experimental verification. This approach is not only time-consuming and labor-intensive, but also difficult to adapt to complex and changing printing requirements and material properties. With the rapid development of artificial intelligence technology, its application in materials science has gradually attracted attention. Utilizing artificial intelligence algorithms, particularly machine learning and deep learning techniques, it is possible to analyze and mine large amounts of experimental data, thereby achieving precise optimization of 3D printing powder ratios. However, the current application of artificial intelligence technology in 3D printing powder ratio optimization is relatively limited and lacks systematicity and efficiency. Therefore, the development of an artificial intelligence-based 3D printing powder ratio optimization method and system is of great practical significance for improving 3D printing efficiency, reducing costs, and enhancing product quality.

[0003] Similar prior art includes a Chinese patent application with publication number CN119294270A, which discloses a parameter optimization method for a powder laser 3D printing process. The method includes: determining the solution space of a particle swarm algorithm, performing finite element analysis on each process parameter, determining the objective function of the particle swarm algorithm, obtaining simulated particles and measured particles within the solution space, and using the measured particles to iteratively train a pre-built support vector machine. The credibility of the simulated particles at any training session is calculated. When the product of the credibility and the normalized simulated fitness value is greater than a preset value, and the measured fitness value of the simulated particles is greater than a conversion threshold, the simulated particles are used as measured particles in subsequent training processes, the training process is repeated, and the target process parameters are obtained after the training is completed. This invention can improve the accuracy of the determined target process parameters while reducing the amount of calculation and time cost. There is also a Chinese patent application with publication number CN119720647A, which discloses a machine learning-based 3D printing parameter optimization method and system. This relates to the field of 3D printing technology and includes: obtaining a data set; preprocessing the data set; constructing a machine learning-based training model based on the data set; training the training model using the data set, and using the trained training model as a 3D printing parameter optimization model; and optimizing the 3D printing parameters to be optimized using the 3D printing parameter optimization model. This invention constructs and trains a machine learning-based training model to obtain a 3D printing parameter optimization model, which is then used to optimize the 3D printing parameters to be optimized. This solves the problem in the prior art of requiring a large amount of computing resources and time to determine 3D printing parameters, and improves the calculation effect and accuracy of 3D printing parameters.

[0004] The shortcomings of the existing technologies are mainly reflected in the fact that they all focus on the optimization of process parameters, do not introduce the optimization of the ratio of powder materials, and do not consider that the characteristics of different materials will affect the performance of printed products under the same process parameters. In actual situations, it is necessary to combine the printing process parameters with the printing powder ratio to improve the performance of the printed products. Summary of the Invention

[0005] The present application provides an artificial intelligence-based 3D printing powder ratio optimization method and system for improving the accuracy of artificial intelligence-based 3D printing powder ratio optimization.

[0006] In a first aspect, the present application provides an artificial intelligence-based 3D printing powder ratio optimization method, the artificial intelligence-based 3D printing powder ratio optimization method comprising: collecting historical data corresponding to powder materials based on a printing process of a 3D printed finished product, the historical data including material data and finished product performance data, and performing multivariate analysis on the historical data based on the printing process and the finished product performance data to generate a plurality of powder compositions; Constructing a material simulation model, wherein the material simulation model simulates the properties of any of the powder compositions, outputs material property data, and extracts a powder ratio that meets printing requirements from all of the powder compositions based on the material property data; Constructing a finished product prediction model, wherein the finished product prediction model predicts a finished product for any of the powder ratios based on the printing process, outputs a finished product prediction result, matches and optimizes the finished product prediction result with ideal finished product data, and screens out an optimal powder ratio and a corresponding optimal printing process; The optimal powder ratio is used to perform 3D printing using the optimal printing process, and the actual printed product is subjected to a performance test. If the actual printed product does not meet the requirements, this step is repeated to re-screen a new optimal powder ratio and a new optimal printing process.

[0007] In combination with the first aspect, the generating of multiple powder compositions comprises: Normalizing the historical data and then performing factor analysis to obtain explanatory factors between any variables in the historical data, obtaining the impact of the explanatory factors on the finished product performance data based on interaction analysis, and extracting multiple key variables from all variables based on the impact values; A proportion range of any material in the material data is obtained based on the key variable, and a plurality of the powder compositions are generated by random sampling based on the proportion range.

[0008] In combination with the first aspect, the constructing of the material simulation model includes: Setting the characteristic quantity of the material data as an explanatory variable, setting the characteristic quantity of the finished product performance data as a target variable, and collecting and summarizing the explanatory variables and the target variables to generate a first data set; Setting the number of nodes in the input layer based on the number of the explanatory variables, setting the number of nodes in the output layer based on the type of the target variable, building a multilayer perceptron model using the input layer, the hidden layer, and the output layer, training the first data set using the multilayer perceptron model, and outputting a first simulated value based on the target variable; Adjusting model parameters of a multilayer perceptron model based on a comparison result between the finished product performance data and the first simulation value to generate a first model; Using a theoretical calculation method to obtain a simulated variable of the explanatory variable, collecting and summarizing the explanatory variable, the simulated variable and the target variable to generate a second data set; setting the number of nodes in a new input layer based on the number of the explanatory variables and the simulation variables, building a new multilayer perceptron model using the new input layer, hidden layer, and output layer, training the second data set using the new multilayer perceptron model, and outputting a second simulation value based on the target variable; Adjusting the model parameters of the new multilayer perceptron model based on the comparison result between the finished product performance data and the second simulation value to generate a second model; The first model and the second model are combined to form the material simulation model.

[0009] In conjunction with the first aspect, the outputting of material property data includes: extracting a first characteristic value of the powder composition, preprocessing the first characteristic value and inputting it into the first model, and outputting a first estimation result; Performing cluster analysis on the first estimation result and all the first simulation values, extracting a cluster containing the first estimation result, and setting the explanatory variable corresponding to the cluster as a reference variable; A first simulation variable corresponding to the first characteristic quantity is obtained using a theoretical calculation method, the first characteristic quantity, the first simulation variable and the reference variable are combined and input into the second model, a second estimation result is output, and the second estimation result is set as the material property data.

[0010] In combination with the first aspect, the step of constructing a finished product prediction model includes: Acquiring printing parameters of different powder materials in the printing process, screening printing process parameters from all the printing parameters based on the correlation between the finished product performance data and the printing parameters, and setting the remaining printing parameters as auxiliary parameters; generating a mathematical formula between the printing process parameters and the finished product performance data based on a regression analysis method, calculating a first predicted value of the target performance based on the mathematical formula, and calculating a difference between the first predicted value and the finished product performance data based on a type of the target performance; Constructing a supplementary prediction model based on a neural network, setting the auxiliary parameter and the difference as input variables, setting the finished product performance data as output variables, and setting the combination of the input variables and the output variables as a training data set; The supplementary prediction model is trained using the training data set based on the input variables, outputs a prediction correction value based on the output variables, and the trained prediction model is set as the finished product prediction model.

[0011] In combination with the first aspect, the printing parameters and the powder ratio are input into the finished product prediction model, a second prediction value is output based on the target performance, the second prediction value is added to the prediction correction value, and set as the finished product prediction result.

[0012] In combination with the first aspect, the matching and optimization of the finished product prediction result with the ideal finished product data includes: Comparing the finished product prediction results of the same category with the ideal finished product data, calculating an evaluation value, and combining and storing the evaluation value with the corresponding powder ratio and the finished product prediction results; Extracting low-dimensional features associated with the evaluation value from the finished product prediction result based on principal component analysis, and combining the low-dimensional features with the powder ratio and the evaluation value for storage; generating transformation parameters based on the powder ratio and the low-dimensional feature quantity, screening the transformation parameters corresponding to the minimum evaluation value using a gradient descent algorithm, and setting them as optimal transformation parameters, converting the optimal transformation parameters into new input parameters based on an inverse transformation, generating a new powder ratio based on the new input parameters, and combining the new powder ratio with the optimal transformation parameters and the minimum evaluation value for storage; Repeat this step based on the stored records to complete the matching optimization.

[0013] Combined with the first aspect, the optimal powder ratio and the corresponding optimal printing process are screened out, including: Setting the new powder ratio generated after matching optimization as the optimal powder ratio; The finished product prediction result corresponding to the optimal powder ratio is extracted from the storage record, the printing parameters corresponding to the finished product prediction result are output based on the finished product prediction model, and the printing parameters are set as the optimal printing process.

[0014] In combination with the first aspect, generating the transformation parameters based on the powder ratio and the low-dimensional feature quantity includes: A parameter prediction model is constructed based on polynomial regression, and the parameter prediction model is trained based on the feature quantities of all the powder ratios and all the low-dimensional feature quantities. The feature quantity of any of the powder ratios is input into the trained parameter prediction model, and the predicted feature quantity is output, and the predicted feature quantity is set as the transformation parameter.

[0015] In a second aspect, the present application provides an artificial intelligence-based 3D printing powder ratio optimization system, the artificial intelligence-based 3D printing powder ratio optimization system comprising: a collection module for collecting historical data corresponding to powder materials according to the printing process of the 3D printed product, the historical data including material data and finished product performance data, performing multivariate analysis on the historical data based on the printing process and the finished product performance data to generate multiple powder compositions; a simulation module for constructing a material simulation model, wherein the material simulation model simulates the properties of any of the powder compositions, outputs material property data, and extracts a powder ratio that meets printing requirements from all of the powder compositions based on the material property data; A prediction module is used to build a finished product prediction model, wherein the finished product prediction model predicts a finished product for any of the powder ratios based on the printing process, outputs a finished product prediction result, matches and optimizes the finished product prediction result with ideal finished product data, and screens out an optimal powder ratio and a corresponding optimal printing process; The printing module uses the optimal powder ratio to perform 3D printing using the optimal printing process, and performs performance testing on the actual printed product. If the actual printed product does not meet the requirements, this step is repeated to re-screen a new optimal powder ratio and a new optimal printing process.

[0016] In the technical solution provided by the present application, first, by collecting historical data on the printing process and finished product performance of 3D printed products, multivariate analysis is performed to identify the key factors affecting the performance of the finished product, and multiple potential powder compositions can be generated, avoiding the limitations of relying on experience or trial and error in traditional methods. Then, a material simulation model is constructed, the powder composition is simulated, material property data is output, and based on these data, a powder ratio that meets the printing requirements is screened out. The powder ratio that meets the printing requirements can be accurately screened out, thereby improving the accuracy of the ratio optimization. Next, a finished product prediction model is constructed, and the finished product performance is predicted for the powder ratio based on the printing process. By matching and optimizing with the ideal finished product data, the optimal powder ratio and the optimal printing process are screened out. Through multi-model collaboration and correction mechanism, accurate prediction of finished product performance is achieved, providing a reliable basis for optimization, and improving the intelligent and automated level of optimization. Finally, through iterative optimization and feedback mechanism, continuous improvement of ratio and process is achieved, ensuring that the performance of the finished product continues to improve, and adaptively adjusting the ratio and process parameters according to the actual printing results improves the adaptability and robustness of the system.

[0017] This application also achieves fine-tuning of powder ratio and printing parameters through parameter transformation, further improving the performance of the finished product. The powder ratio, printing parameters, finished product performance data and intermediate results of the optimization process are stored to form a knowledge base to provide a reference for subsequent optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0019] Figure 1 This is a schematic diagram of an embodiment of a method for optimizing 3D printing powder ratio based on artificial intelligence in an embodiment of the present application; Figure 2 This is a schematic diagram of an embodiment of the multivariate analysis and powder composition generation process in the embodiments of the present application; Figure 3 This is a schematic diagram of an embodiment of the finished product prediction model construction and optimization process in the embodiment of this application; Figure 4 This is a schematic diagram of an embodiment of an artificial intelligence-based 3D printing powder ratio optimization system in an embodiment of the present application. DETAILED DESCRIPTION

[0020] The embodiments of the present application provide a method and system for optimizing 3D printing powder ratios based on artificial intelligence. The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of this application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units that are not explicitly listed or inherent to these processes, methods, products or devices.

[0021] For ease of understanding, the specific process of the embodiment of the present application is described below. Figure 1 In the embodiments of the present application, an embodiment of the 3D printing powder ratio optimization method based on artificial intelligence includes: Step S101: Collect historical data corresponding to powder materials based on the printing process of the 3D printed product, the historical data including material data and finished product performance data, perform multivariate analysis on the historical data based on the printing process and finished product performance data, and generate multiple powder compositions.

[0022] It is understood that the execution subject of this application can be an artificial intelligence-based 3D printing powder ratio optimization device, or a terminal or server, which is not limited here. The embodiment of this application is described by taking the server as the execution subject as an example.

[0023] Specifically, a 3D-printed finished product refers to an object manufactured using 3D printing technology, and the printing process refers to the process data used when using 3D printing technology, such as printing temperature, printing speed, layer thickness, and laser power. Powder materials refer to the powder material data used in different 3D-printed finished products, including but not limited to various metal powder materials such as aluminum alloy powder and titanium alloy powder. Historical data refers to a data set generated by aggregating historically collected 3D printing-related data. Material data includes but is not limited to parameters such as the ratio, particle size, purity, and physical properties of various powder materials. Finished product performance data refers to the performance data of 3D-printed finished products, including but not limited to mechanical properties (such as tensile strength, hardness, and toughness), physical properties (such as density and porosity), and chemical properties (such as corrosion resistance and heat resistance). Multivariate analysis can identify key factors and interrelationships that influence finished product performance data, allowing the various powders to be recombined to generate multiple powder compositions. These compositions are then preliminarily screened to eliminate those that clearly do not conform to physical and chemical principles or actual production conditions.

[0024] Step S102: constructing a material simulation model, which simulates the properties of any powder composition, outputs material property data, and extracts a powder ratio that meets printing requirements from all powder compositions based on the material property data.

[0025] Specifically, a material simulation model uses historical data to simulate the performance of each powder composition after printing a finished product, regardless of the printing process. Material property data refers to the performance data of the simulated powder composition after printing, such as physical and chemical properties such as density, hardness, melting point, and thermal expansion coefficient. Printing requirements refer to the expected requirements for the printed product, and the powder ratio of the powder composition that may meet these requirements is extracted.

[0026] Step S103: Construct a finished product prediction model. The finished product prediction model predicts the finished product for any powder ratio based on the printing process, outputs the finished product prediction result, matches and optimizes the finished product prediction result with the ideal finished product data, and screens out the optimal powder ratio and the corresponding optimal printing process.

[0027] Specifically, the finished product prediction model predicts the finished product produced by printing with different powder ratios using different printing processes. The finished product prediction results represent the performance data of the finished product generated by printing with different powder ratios using different printing processes. Ideal finished product data refers to the performance data of the finished product corresponding to the printing requirements. Based on the ideal finished product data, the printing process can be adjusted, and the finished product prediction results can be adjusted and optimized to ensure that the finished product prediction results match the ideal finished product data. This allows the powder ratio and printing process that best meet the printing requirements to be selected, i.e., the optimal powder ratio and printing process.

[0028] Step S104: Optimal powder ratio. Use the optimal printing process for 3D printing, and perform performance testing on the actual printed product. If the actual printed product does not meet the requirements, repeat this step to re-screen a new optimal powder ratio and a new optimal printing process.

[0029] Specifically, the actual printed product refers to the real-time finished product after 3D printing using the optimal powder ratio and the optimal printing process. Performance testing refers to the performance testing of the actual printed product, including but not limited to mechanical, physical, and chemical properties. If the actual printed product does not meet the requirements, it means that the performance test results of the actual printed product are different from the ideal product data, and the parameters of the optimal powder ratio and optimal printing process need to be adjusted. Through the above steps, a new optimal powder ratio and a new optimal printing process can be re-screened until the actual printed product meets the requirements.

[0030] In one embodiment, a plurality of powder compositions are generated, comprising: (1) After normalizing the historical data, factor analysis is performed to obtain the explanatory factors between any variables in the historical data. Based on the interaction analysis, the impact value of the explanatory factors on the finished product performance data is obtained. Based on the impact value, multiple key variables are extracted from all variables.

[0031] (2) Obtaining the proportion range of any material in the material data based on the key variables, and randomly sampling based on the proportion range to generate multiple powder compositions.

[0032] Specifically, Figure 2 Generate a flow chart for multivariate analysis and powder composition. Normalization is the process of converting historical data of different dimensions to the same scale. Factor analysis is the process of identifying common factors between potential variables and explaining the correlation between variables. Explanatory factors are common factors between variables. Regression analysis is performed using interaction terms as a method of interaction analysis to calculate the impact of explanatory factors on the finished product performance data, that is, output impact values. For example, in the historical data, printing speed has a significant impact on tensile strength, corresponding to a large impact value. Variables corresponding to explanatory factors with impact values greater than a preset threshold are set as key variables.

[0033] Ratio range refers to determining the optimal ratio of different materials based on key variable and interaction analysis using optimization algorithms (such as linear programming and genetic algorithms). Multiple powder compositions are generated using random sampling or grid sampling to ensure that each powder composition is within the ratio range.

[0034] In a specific embodiment, constructing a material simulation model includes: (1) Set the characteristic quantity of material data as explanatory variables, set the characteristic quantity of finished product performance data as target variables, and collect and summarize the explanatory variables and target variables to generate the first data set.

[0035] (2) The number of nodes in the input layer is set based on the number of explanatory variables, and the number of nodes in the output layer is set based on the type of the target variable. A multilayer perceptron model is constructed by the input layer, hidden layer, and output layer. The multilayer perceptron model is used to train the first data set, and a first simulation value is output based on the target variable.

[0036] (3) By comparing the performance data of the finished product with the first simulation value, the model parameters of the multilayer perceptron model are adjusted to generate the first model.

[0037] (4) Use theoretical calculation methods to obtain simulated variables of explanatory variables, collect and summarize explanatory variables, simulated variables and target variables to generate the second data set.

[0038] (5) The number of nodes in the new input layer is set based on the number of explanatory variables and simulation variables, a new multilayer perceptron model is constructed by the new input layer, hidden layer, and output layer, the second data set is trained using the new multilayer perceptron model, and a second simulation value is output based on the target variable.

[0039] (6) By comparing the performance data of the finished product with the second simulation value, the model parameters of the new multilayer perceptron model are adjusted to generate a second model.

[0040] (7) The first model and the second model are combined to form a material simulation model.

[0041] Specifically, the characteristic quantities of material data include, but are not limited to, material composition ratios, material physical and chemical properties, and other related parameters. The characteristic quantities of finished product performance data include, but are not limited to, mechanical properties (such as tensile strength, hardness, and toughness), physical properties (such as density and porosity), and chemical properties (such as corrosion resistance and heat resistance).

[0042] The hidden layer in a multilayer perceptron (MLP) model contains multiple neurons. It uses explanatory variables as input, and the output layer outputs a predicted value based on the type of the target variable, known as the first simulated value. During the training process of the multilayer perceptron model, a loss function and optimization algorithm are set. The forward and backward propagation processes are repeated until the model converges or a predetermined number of training rounds are reached, and the model parameters are adjusted. The deviation between the finished product performance data corresponding to the same target variable type in the test set and the first simulated value is compared, and the comparison result is output. The model parameters are then adjusted to generate the first model. The primary function of the first model is to use historical data to preliminarily simulate and predict the first simulated values corresponding to various material data, thereby preliminarily simulating material properties.

[0043] Theoretical calculation refers to the use of quantum chemical calculations, molecular dynamics simulations and other methods to calculate the material properties corresponding to the explanatory variables as simulation variables.

[0044] The simulated variables can be used to correct and verify the first simulated values output by the first model, while also supplementing the first dataset to generate a second dataset. Retraining the second model with the second dataset improves its generalization capabilities, adapting it to diverse material data and enhancing the accuracy and reliability of material property simulations. Therefore, the first and second models are combined to form a material simulation model.

[0045] In one embodiment, outputting material property data includes: (1) Extracting a first characteristic value of the powder composition, preprocessing the first characteristic value and inputting it into a first model, and outputting a first estimation result.

[0046] (2) Perform cluster analysis on the first estimation results and all the first simulation values, extract the clusters containing the first estimation results, and set the explanatory variables corresponding to the clusters as reference variables.

[0047] (3) Using a theoretical calculation method to obtain a first simulation variable corresponding to the first characteristic quantity, combining the first characteristic quantity, the first simulation variable, and the reference variable and inputting them into a second model, outputting a second estimation result, and setting the second estimation result as the material property data.

[0048] Specifically, the first feature quantity refers to a feature quantity corresponding to the powder composition extracted based on the type of the feature quantity of the material data, and the first estimation result refers to a first simulation value generated by the first model based on the simulation of the first feature quantity.

[0049] The first estimation result and all first simulation values are aggregated and clustered. K-means clustering can be used to divide the clusters into multiple clusters, and clusters containing the first estimation result are further extracted. The clusters can be used to find a reference variable similar to the first characteristic value among all explanatory variables, indicating that the powder material corresponding to the reference variable is associated with the powder composition.

[0050] The first simulation variable is used to simulate the material properties of the powder composition. Combining the first characteristic quantity, the first simulation variable, and the reference variable can improve the accuracy of the characteristic quantity prediction, and the second model then outputs the corresponding second estimation result. For example, if the input characteristic quantities are simulated tensile strength (480 MPa), simulated hardness (190 HV), Fe: 60%, Ni: 20%, Cr: 20%, particle size: 15-53 μm, laser power: 200 W, and scanning speed: 1000 mm / s, the corresponding second estimation results are tensile strength (510 MPa), hardness (210 HV), etc.

[0051] In a specific embodiment, building a finished product prediction model includes: (1) Obtain the printing parameters of different powder materials in the printing process, and based on the correlation between the performance data of the finished product and the printing parameters, select the printing process parameters from all the printing parameters based on the correlation, and set the remaining printing parameters as auxiliary parameters.

[0052] (2) Generate a mathematical formula between the printing process parameters and the finished product performance data based on the regression analysis method, calculate a first predicted value of the target performance based on the mathematical formula, and calculate the difference between the first predicted value and the finished product performance data based on the type of the target performance.

[0053] (3) Construct a supplementary prediction model based on a neural network, set the auxiliary parameters and difference values as input variables, set the finished product performance data as output variables, and set the combination of input variables and output variables as a training data set.

[0054] (4) The supplementary prediction model is trained using the training data set based on the input variables, the prediction correction value is output based on the output variables, and the trained prediction model is set as the finished prediction model.

[0055] Specifically, Figure 3 A flowchart for building and optimizing a finished product prediction model. Building a finished product prediction model requires integrating the printing process for analysis and prediction. Printing parameters refer to the parameter data involved in the printing process, such as printing temperature, printing speed, layer thickness, and laser power. Correlation is used to select printing parameters that significantly impact the performance of the finished product, namely, printing process parameters. The remaining printing parameters are used as auxiliary features, namely, auxiliary parameters.

[0056] A mathematical formula is a calculation formula corresponding to a multivariate regression equation generated through regression analysis. It describes the mathematical relationship between printing process parameters and finished product performance data. Target performance refers to the performance type corresponding to any finished product performance data. The mathematical formula can be used to make a preliminary estimate of any target performance and output a first predicted value. The difference is used to describe the deviation of the regression prediction using the mathematical formula.

[0057] The supplementary prediction model is used to learn the relationship between auxiliary parameters and the difference, introducing a correction mechanism to improve the accuracy of the final prediction. Auxiliary parameters are screened printing parameters with low correlation. By constructing a training dataset based on input and output variables, it is possible to accurately predict the performance of the finished product.

[0058] The supplementary prediction model is trained based on the training data set to generate a finished product prediction model, which can accurately predict the finished product performance data in combination with the criticality of the printing process.

[0059] In one embodiment, the printing parameters and the powder ratio are input into the finished product prediction model, a second prediction value is output based on the target performance, the second prediction value is added to the prediction correction value, and the result is set as the finished product prediction result.

[0060] Specifically, since there are multiple printing parameters and powder ratios, inputting any powder ratio and printing parameter into the finished product prediction model will output a second prediction value corresponding to the target performance. Because the finished product prediction model incorporates a correction mechanism, the second prediction value is added to the prediction correction value to generate the finished product prediction result.

[0061] In a specific embodiment, matching and optimizing the finished product prediction result with the ideal finished product data includes: (1) Compare the finished product prediction results of the same category with the ideal finished product data, calculate the evaluation value, and combine the evaluation value with the corresponding powder ratio and finished product prediction results for storage.

[0062] (2) Based on principal component analysis, low-dimensional features associated with the evaluation value are extracted from the finished product prediction results, and the low-dimensional features are combined with the powder ratio and the evaluation value and stored.

[0063] (3) Generate transformation parameters based on powder ratio and low-dimensional feature quantity, use gradient descent algorithm to screen out the transformation parameters corresponding to the minimum evaluation value, and set them as the optimal transformation parameters. Convert the optimal transformation parameters into new input parameters based on inverse transformation, generate a new powder ratio based on the new input parameters, and store the new powder ratio in combination with the optimal transformation parameters and the minimum evaluation value.

[0064] (4) Repeat this step based on the stored records to complete the matching optimization.

[0065] Specifically, the evaluation value represents the comparison result between the finished product prediction result and the ideal finished product data. The weight coefficient can be set according to the category of the finished product prediction result. The difference between the finished product prediction results of all categories and the ideal finished product data can be summarized according to the weight coefficient to calculate the evaluation value.

[0066] Principal component analysis is suitable for linear dimensionality reduction and can extract the dimension with the largest variance as the feature quantity. Since the finished product prediction results are diverse, the feature quantities of the finished product prediction results are high-dimensional. It is necessary to extract low-dimensional features that are more directly related to the evaluation value from the features of the high-dimensional finished product performance data. Correlation can be expressed using the Pearson correlation coefficient.

[0067] The transformation parameters can transform the high-dimensional input parameter space into a low-dimensional transformation parameter space, thereby reducing the complexity of the optimization problem. The gradient descent algorithm can be used as an optimization algorithm to find the transformation parameters that minimize or maximize the evaluation value. In this application, the smaller the evaluation value, the closer the finished product prediction result is to the ideal finished product data. Therefore, the optimal transformation parameters can be screened out. Since the transformation parameters are obtained through principal component analysis transformation, the inverse transformation can be used to convert the optimal transformation parameters into new input parameters. The category of the new input parameter corresponds to the category of the powder ratio. Therefore, a new powder ratio can be generated.

[0068] By storing all the above optimization records, it is possible to achieve the process of matching and optimizing the finished product prediction results with the ideal finished product data.

[0069] In a specific embodiment, screening out the optimal powder ratio and the corresponding optimal printing process includes: (1) The new powder ratio generated after matching optimization is set as the optimal powder ratio.

[0070] (2) Extract the finished product prediction result corresponding to the optimal powder ratio from the stored records, output the printing parameters corresponding to the finished product prediction result based on the finished product prediction model, and set the printing parameters to the optimal printing process.

[0071] Specifically, new powder ratios are generated from the matching optimization process, which are converted from the optimal transformation parameters corresponding to the minimum evaluation value screened by the gradient descent algorithm.

[0072] In the finished product prediction model, the reverse mapping function of the finished product prediction model is used to input the optimal powder ratio and target performance indicators, and the corresponding printing parameters, that is, the optimal printing process, are output.

[0073] In a specific embodiment, generating transformation parameters based on the powder ratio and the low-dimensional feature quantity includes: A parameter prediction model is constructed based on polynomial regression, and the parameter prediction model is trained based on the feature quantities of all powder ratios and all low-dimensional feature quantities. The feature quantity of any powder ratio is input into the trained parameter prediction model, and the predicted feature quantity is output, which is set as the transformation parameter.

[0074] Specifically, polynomial regression can capture nonlinear relationships between variables, is applicable to complex data relationships, and is used to build parameter prediction models. The characteristic quantities corresponding to the powder ratio and the low-dimensional characteristic quantities are set as input variables, and the predicted characteristic quantities are set as output variables. The output variables can be transformed characteristic quantities or other indicators that need to be predicted. The appropriate polynomial order is selected to balance the complexity and generalization ability of the model. The characteristic quantities of all powder ratios and all low-dimensional characteristic quantities are used as the training set to train the parameter prediction model. The model performs prediction calculations based on the input data and outputs the predicted characteristic quantities.

[0075] The above describes the 3D printing powder ratio optimization method based on artificial intelligence in the embodiment of the present application. The following describes the 3D printing powder ratio optimization system based on artificial intelligence in the embodiment of the present application. Figure 4 In the embodiments of the present application, an embodiment of the 3D printing powder ratio optimization system based on artificial intelligence includes: The collection module 201 is used to collect historical data corresponding to the powder material according to the printing process of the 3D printed product, the historical data including material data and finished product performance data, and perform multivariate analysis on the historical data based on the printing process and finished product performance data to generate multiple powder compositions.

[0076] The simulation module 202 is used to construct a material simulation model. The material simulation model simulates the properties of any powder composition, outputs material property data, and extracts a powder ratio that meets the printing requirements from all powder compositions based on the material property data.

[0077] The prediction module 203 is used to build a finished product prediction model. The finished product prediction model predicts the finished product for any powder ratio based on the printing process, outputs the finished product prediction result, matches and optimizes the finished product prediction result with the ideal finished product data, and screens out the optimal powder ratio and the corresponding optimal printing process.

[0078] Printing module 204 uses the optimal powder ratio to perform 3D printing using the optimal printing process, and performs performance testing on the actual printed product. If the actual printed product does not meet the requirements, this step is repeated to re-screen a new optimal powder ratio and a new optimal printing process.

[0079] Through the collaborative efforts of these components, the system first collects historical data on the printing process and performance of 3D-printed products, performs multivariate analysis, and identifies key factors influencing product performance. This allows the generation of multiple potential powder compositions, avoiding the limitations of traditional methods that rely on experience or trial-and-error. A material simulation model is then constructed to simulate the properties of the powder composition, outputting material property data. Based on this data, a powder ratio that meets the printing requirements is selected. This allows for precise selection of the powder ratio that meets the printing requirements, improving the accuracy of the ratio optimization. Next, a finished product prediction model is constructed to predict the finished product performance of the powder ratio based on the printing process. By optimizing the matching with the ideal finished product data, the optimal powder ratio and printing process are selected. Through multi-model collaboration and a correction mechanism, accurate prediction of finished product performance is achieved, providing a reliable basis for optimization and enhancing the intelligent and automated level of optimization. Finally, through iterative optimization and feedback mechanisms, continuous improvement of the ratio and process is achieved, ensuring continuous improvement in finished product performance. The ratio and process parameters are adaptively adjusted based on actual printing results, improving the adaptability and robustness of the system.

[0080] This application also achieves fine-tuning of powder ratio and printing parameters through parameter transformation, further improving the performance of the finished product. The powder ratio, printing parameters, finished product performance data and intermediate results of the optimization process are stored to form a knowledge base to provide a reference for subsequent optimization.

[0081] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described systems, systems and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0082] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, and other media that can store program code.

[0083] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A 3D printing powder ratio optimization method based on artificial intelligence, characterized in that: The artificial intelligence-based 3D printing powder ratio optimization method includes: collecting historical data corresponding to powder materials based on a printing process of a 3D printed finished product, the historical data including material data and finished product performance data, and performing multivariate analysis on the historical data based on the printing process and the finished product performance data to generate a plurality of powder compositions; Constructing a material simulation model, wherein the material simulation model simulates the properties of any of the powder compositions, outputs material property data, and extracts a powder ratio that meets printing requirements from all of the powder compositions based on the material property data; Constructing a finished product prediction model, wherein the finished product prediction model predicts a finished product for any of the powder ratios based on the printing process, outputs a finished product prediction result, matches and optimizes the finished product prediction result with ideal finished product data, and screens out an optimal powder ratio and a corresponding optimal printing process; The optimal powder ratio is used to perform 3D printing using the optimal printing process, and the actual printed product is subjected to a performance test. If the actual printed product does not meet the requirements, this step is repeated to re-screen a new optimal powder ratio and a new optimal printing process.

2. The artificial intelligence-based 3D printing powder ratio optimization method according to claim 1, characterized in that: The method of generating a plurality of powder compositions comprises: Normalizing the historical data and then performing factor analysis to obtain explanatory factors between any variables in the historical data, obtaining the impact of the explanatory factors on the finished product performance data based on interaction analysis, and extracting multiple key variables from all variables based on the impact values; A proportion range of any material in the material data is obtained based on the key variable, and a plurality of the powder compositions are generated by random sampling based on the proportion range.

3. The artificial intelligence-based 3D printing powder ratio optimization method according to claim 1, characterized in that: The constructing of the material simulation model comprises: Setting the characteristic quantity of the material data as an explanatory variable, setting the characteristic quantity of the finished product performance data as a target variable, and collecting and summarizing the explanatory variables and the target variables to generate a first data set; Setting the number of nodes in the input layer based on the number of the explanatory variables, setting the number of nodes in the output layer based on the type of the target variable, building a multilayer perceptron model using the input layer, the hidden layer, and the output layer, training the first data set using the multilayer perceptron model, and outputting a first simulated value based on the target variable; Adjusting model parameters of a multilayer perceptron model based on a comparison result between the finished product performance data and the first simulation value to generate a first model; Using a theoretical calculation method to obtain a simulated variable of the explanatory variable, collecting and summarizing the explanatory variable, the simulated variable and the target variable to generate a second data set; setting the number of nodes in a new input layer based on the number of the explanatory variables and the simulation variables, building a new multilayer perceptron model using the new input layer, hidden layer, and output layer, training the second data set using the new multilayer perceptron model, and outputting a second simulation value based on the target variable; Adjusting the model parameters of the new multilayer perceptron model based on the comparison result between the finished product performance data and the second simulation value to generate a second model; The first model and the second model are combined to form the material simulation model.

4. The artificial intelligence-based 3D printing powder ratio optimization method according to claim 3, characterized in that: The output material property data includes: extracting a first characteristic value of the powder composition, preprocessing the first characteristic value and inputting it into the first model, and outputting a first estimation result; Performing cluster analysis on the first estimation result and all the first simulation values, extracting a cluster containing the first estimation result, and setting the explanatory variable corresponding to the cluster as a reference variable; A first simulation variable corresponding to the first characteristic quantity is obtained using a theoretical calculation method, the first characteristic quantity, the first simulation variable and the reference variable are combined and input into the second model, a second estimation result is output, and the second estimation result is set as the material property data.

5. The artificial intelligence-based 3D printing powder ratio optimization method according to claim 1, characterized in that: The step of constructing a finished product prediction model includes: Acquiring printing parameters of different powder materials in the printing process, screening printing process parameters from all the printing parameters based on the correlation between the finished product performance data and the printing parameters, and setting the remaining printing parameters as auxiliary parameters; generating a mathematical formula between the printing process parameters and the finished product performance data based on a regression analysis method, calculating a first predicted value of the target performance based on the mathematical formula, and calculating a difference between the first predicted value and the finished product performance data based on a type of the target performance; Constructing a supplementary prediction model based on a neural network, setting the auxiliary parameter and the difference as input variables, setting the finished product performance data as output variables, and setting the combination of the input variables and the output variables as a training data set; The supplementary prediction model is trained using the training data set based on the input variables, outputs a prediction correction value based on the output variables, and the trained prediction model is set as the finished product prediction model.

6. The artificial intelligence-based 3D printing powder ratio optimization method according to claim 5, characterized in that: The printing parameters and the powder ratio are input into the finished product prediction model, a second prediction value is output based on the target performance, the second prediction value is added to the prediction correction value, and the result is set as the finished product prediction result.

7. The artificial intelligence-based 3D printing powder ratio optimization method according to claim 1, characterized in that: The matching and optimization of the finished product prediction result with the ideal finished product data includes: Comparing the finished product prediction results of the same category with the ideal finished product data, calculating an evaluation value, and combining and storing the evaluation value with the corresponding powder ratio and the finished product prediction results; Extracting low-dimensional features associated with the evaluation value from the finished product prediction result based on principal component analysis, and combining the low-dimensional features with the powder ratio and the evaluation value for storage; generating transformation parameters based on the powder ratio and the low-dimensional feature quantity, screening the transformation parameters corresponding to the minimum evaluation value using a gradient descent algorithm, and setting them as optimal transformation parameters, converting the optimal transformation parameters into new input parameters based on an inverse transformation, generating a new powder ratio based on the new input parameters, and combining the new powder ratio with the optimal transformation parameters and the minimum evaluation value for storage; Repeat this step based on the stored records to complete the matching optimization.

8. The artificial intelligence-based 3D printing powder ratio optimization method according to claim 7, characterized in that: Screen out the best powder ratio and the corresponding best printing process, including: Setting the new powder ratio generated after matching optimization as the optimal powder ratio; The finished product prediction result corresponding to the optimal powder ratio is extracted from the storage record, the printing parameters corresponding to the finished product prediction result are output based on the finished product prediction model, and the printing parameters are set as the optimal printing process.

9. The artificial intelligence-based 3D printing powder ratio optimization method according to claim 7, characterized in that: The generating of transformation parameters based on the powder ratio and the low-dimensional feature quantity includes: A parameter prediction model is constructed based on polynomial regression, and the parameter prediction model is trained based on the feature quantities of all the powder ratios and all the low-dimensional feature quantities. The feature quantity of any of the powder ratios is input into the trained parameter prediction model, and the predicted feature quantity is output, and the predicted feature quantity is set as the transformation parameter.

10. A 3D printing powder ratio optimization system based on artificial intelligence, characterized in that: The artificial intelligence-based 3D printing powder ratio optimization system includes: a collection module for collecting historical data corresponding to powder materials according to the printing process of the 3D printed product, the historical data including material data and finished product performance data, performing multivariate analysis on the historical data based on the printing process and the finished product performance data to generate multiple powder compositions; a simulation module for constructing a material simulation model, wherein the material simulation model simulates the properties of any of the powder compositions, outputs material property data, and extracts a powder ratio that meets printing requirements from all of the powder compositions based on the material property data; A prediction module is used to build a finished product prediction model, wherein the finished product prediction model predicts a finished product for any of the powder ratios based on the printing process, outputs a finished product prediction result, matches and optimizes the finished product prediction result with ideal finished product data, and screens out an optimal powder ratio and a corresponding optimal printing process; The printing module uses the optimal powder ratio to perform 3D printing using the optimal printing process, and performs performance testing on the actual printed product. If the actual printed product does not meet the requirements, this step is repeated to re-screen a new optimal powder ratio and a new optimal printing process.

Images