3D printing parameter optimization method and system based on machine learning

Abstract

The invention discloses a 3D printing parameter optimization method and system based on machine learning, and relates to the technical field of 3D printing, and the method comprises the steps: obtaining a data set; preprocessing the data set; constructing a training model based on machine learning according to the data set; training a training model by using the data set, and taking the trained training model as a 3D printing parameter optimization model; and optimizing the 3D printing parameters to be optimized by using the 3D printing parameter optimization model. According to the method, the 3D printing parameter optimization model is obtained by constructing and training the training model based on machine learning, and the 3D printing parameters to be optimized are optimized by using the 3D printing parameter optimization model, so that the problem that determination of the 3D printing parameters in the prior art needs a large amount of computing resources and time is solved, and the determination efficiency of the 3D printing parameters is improved. And the calculation effect and accuracy of the 3D printing parameters are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of finite element analysis, and in particular to a 3D printing parameter optimization method and system based on machine learning. Background Art

[0002] In the 3D printing process, the design of 3D printing parameters has an important impact on the entity after molding. The determination of 3D printing parameters in the prior art usually relies on empirical methods or trial and error methods to select parameters, which not only consumes a lot of computing resources and time, but also makes it difficult to ensure the accuracy of the final result. Therefore, a method that can reduce computing resources and time and improve accuracy is needed to optimize 3D printing parameters. Summary of the invention

[0003] In order to solve the technical problems existing in the background technology, the present invention proposes a 3D printing parameter optimization method and system based on machine learning.

[0004] In a first aspect, the present invention proposes a 3D printing parameter optimization method based on machine learning, wherein the model to be 3D printed is a three-dimensional finite element model of a target molded part;

[0005] The method comprises:

[0006] Acquire a data set; wherein the data set includes 3D printing parameters and their corresponding output results;

[0007] Preprocess the dataset;

[0008] Build a machine learning-based training model based on the data set;

[0009] The training model is trained using the data set, and the trained model is used as a 3D printing parameter optimization model;

[0010] The 3D printing parameter optimization model is used to optimize the 3D printing parameters to be optimized.

[0011] Preferably, the 3D printing parameters include: laser power, scanning speed, layer thickness and substrate preheating temperature; and the output results are stress and deformation distribution results.

[0012] Preferably, obtaining a data set specifically includes:

[0013] Get 3D printing parameters;

[0014] Perform thermal analysis simulation on the three-dimensional finite element model according to the 3D printing parameters to obtain thermal analysis results;

[0015] According to the thermal analysis results, the stress-deformation analysis is performed on the three-dimensional finite element model to obtain the stress-deformation distribution results;

[0016] The 3D printing parameters and the corresponding stress-deformation distribution results are combined to form a data set.

[0017] Preferably, a training model based on machine learning is constructed according to the data set, specifically including:

[0018] Choose a machine learning algorithm based on the size of the dataset;

[0019] Build a machine learning-based training model based on the selected machine learning algorithm.

[0020] Preferably, a machine learning algorithm is selected according to the size of the data set, including:

[0021] When the size of the data set is less than or equal to a first preset size, selecting a neural network algorithm;

[0022] When the size of the data set is larger than the first preset size, a more efficient random forest or gradient boosting tree algorithm is selected.

[0023] Preferably, the training model is trained using the data set, and the trained training model is used as a 3D printing parameter optimization model, specifically including:

[0024] Divide the dataset into training set, test set and validation set;

[0025] The training model is trained, tested and verified in sequence using the training set, test set and verification set to obtain a verified training model;

[0026] The verified training model is used as the 3D printing parameter optimization model.

[0027] Preferably, if there is no training model that has passed the test or verification, the training set, test set and verification set are repeatedly used to train, test and verify the training model in sequence until there is a training model that has passed the verification.

[0028] Preferably, the 3D printing parameters and their corresponding output results are preprocessed, specifically including:

[0029] Eliminate outliers on 3D printing parameters and their corresponding output results;

[0030] The 3D printing parameters and their corresponding output results after outlier removal are normalized.

[0031] In a second aspect, the present invention further proposes a 3D printing parameter optimization system based on machine learning, wherein the model to be 3D printed is a three-dimensional finite element model of a target molded part, including:

[0032] An acquisition module, used to acquire a data set; wherein the data set includes 3D printing parameters and their corresponding output results;

[0033] Preprocessing module, used to preprocess the data set;

[0034] The model building module is used to build a machine learning-based training model based on the data set;

[0035] A model training module is used to train the training model using the data set, and use the trained training model as a 3D printing parameter optimization model;

[0036] The parameter optimization module is used to optimize the 3D printing parameters to be optimized by using the 3D printing parameter optimization model.

[0037] Preferably, the 3D printing parameters include: laser power, scanning speed, layer thickness and substrate preheating temperature; and the output results are stress and deformation distribution results.

[0038] Preferably, the model building module includes:

[0039] The algorithm selection submodule is used to select a machine learning algorithm based on the size of the data set;

[0040] The construction submodule is used to build a machine learning-based training model according to the selected machine learning algorithm.

[0041] Preferably, the preprocessing includes outlier removal and normalization.

[0042] Preferably, the model training module includes:

[0043] The partitioning module is used to divide the data set into training set, test set and validation set;

[0044] The training submodule is used to train the training model using the training set to obtain a trained training model;

[0045] The test submodule is used to test the trained model using the test set to obtain a training model that passes the test;

[0046] The verification submodule is used to verify the model that has passed the test and obtain a verified training model;

[0047] Preferably, it also includes a judgment submodule, which is used to judge whether there is a training model that has passed the verification;

[0048] If not, the training model is retrained, tested and verified in sequence through the training submodule, the testing submodule and the verification submodule until there is a training model that passes the verification.

[0049] In the present invention, the proposed 3D printing parameter optimization method and system based on machine learning obtains a 3D printing parameter optimization model by constructing and training a training model based on machine learning, and optimizes the 3D printing parameters to be optimized using the 3D printing parameter optimization model, thereby solving the problem in the prior art that the determination of 3D printing parameters requires a large amount of computing resources and time, and improving the calculation effect and accuracy of 3D printing parameters. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 A schematic flow chart of a 3D printing parameter optimization method based on machine learning in one embodiment of the present invention. DETAILED DESCRIPTION

[0051] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments of the present invention can be combined with each other. The present invention will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0052] Reference Figure 1 The present invention proposes a 3D printing parameter optimization method based on machine learning, comprising:

[0053] The model to be 3D printed is a three-dimensional finite element model of the target molded part;

[0054] The optimization method comprises:

[0055] Acquire a data set; wherein the data set includes 3D printing parameters and their corresponding output results;

[0056] Preprocess the dataset;

[0057] Build a machine learning-based training model based on the data set;

[0058] The training model is trained using the data set, and the trained model is used as a 3D printing parameter optimization model;

[0059] The 3D printing parameter optimization model is used to optimize the 3D printing parameters to be optimized.

[0060] The present invention obtains a 3D printing parameter optimization model by constructing and training a training model based on machine learning. The 3D printing parameter optimization model is used to optimize the 3D printing parameters to be optimized, thereby solving the problem in the prior art that the determination of 3D printing parameters requires a large amount of computing resources and time, and improving the calculation effect and accuracy of 3D printing parameters.

[0061] In this embodiment, the 3D printing parameters include: laser power, scanning speed, layer thickness, substrate preheating temperature, and the output results are stress and deformation distribution results.

[0062] In this embodiment, obtaining a data set specifically includes:

[0063] Get 3D printing parameters;

[0064] Perform thermal analysis simulation on the three-dimensional finite element model according to the 3D printing parameters to obtain thermal analysis results;

[0065] According to the thermal analysis results, the stress-deformation analysis is performed on the three-dimensional finite element model to obtain the stress-deformation distribution results;

[0066] The 3D printing parameters and the corresponding stress-deformation distribution results are combined to form a data set.

[0067] In this embodiment, a training model based on machine learning is constructed according to the data set, specifically including:

[0068] Choose a machine learning algorithm based on the size of the dataset;

[0069] Build a machine learning-based training model based on the selected machine learning algorithm.

[0070] Among them, when the scale of the data set is less than or equal to the first preset scale, the neural network algorithm is selected as the machine learning algorithm of this embodiment; when the scale of the data set is greater than the first preset scale, a more efficient random forest or gradient boosting tree algorithm is selected as the machine learning algorithm of this embodiment.

[0071] In order to ensure the accuracy of the 3D printing parameter optimization model, in a further implementation method, the training model is trained using the data set, and the trained training model is used as the 3D printing parameter optimization model, specifically including:

[0072] Divide the dataset into training set, test set and validation set;

[0073] The training model is trained, tested and verified in sequence using the training set, test set and verification set to obtain a verified training model;

[0074] The verified training model is used as the 3D printing parameter optimization model.

[0075] In a further implementation, if there is no training model that has passed the test or verification, the training set, test set and verification set are repeatedly used to train, test and verify the training model in sequence until there is a training model that has passed the verification, so as to further ensure the accuracy of the 3D printing parameter optimization model.

[0076] In order to ensure the quality of data, in this embodiment, the 3D printing parameters and their corresponding output results are preprocessed, specifically including:

[0077] Eliminate outliers on 3D printing parameters and their corresponding output results;

[0078] The 3D printing parameters and their corresponding output results after outlier removal are normalized.

[0079] In a further embodiment, the 3D printing parameters and their corresponding output results are subjected to outlier elimination processing, specifically including:

[0080] The Z-score statistical method is used to eliminate outliers in the 3D printing parameters and their corresponding output results.

[0081] Among them, the Z-score statistical method is used to eliminate outliers in the 3D printing parameters and their corresponding output results, including:

[0082] The mean μ and standard deviation σ of each 3D printing parameter in the data set are calculated;

[0083] Each 3D printing parameter is calculated using a first preset formula; wherein the first preset formula is Where X is the value of a 3D printing parameter;

[0084] When the absolute value of Z corresponding to each 3D printing parameter is greater than 3, the 3D printing parameter is determined to be an abnormal value, and the abnormal value and its corresponding output result are eliminated.

[0085] In a further embodiment, the 3D printing parameters after the outlier removal process and the corresponding output results are normalized, specifically including:

[0086] Using a second preset formula to normalize the 3D printing parameters after outlier elimination and their corresponding output results;

[0087] Among them, the second preset formula is Where X' is the normalized value of X.

[0088] The present invention also proposes a 3D printing parameter optimization system based on machine learning, wherein the model to be 3D printed is a three-dimensional finite element model of a target molded part, including:

[0089] An acquisition module, used to acquire a data set; wherein the data set includes 3D printing parameters and their corresponding output results;

[0090] Preprocessing module, used to preprocess the data set;

[0091] The model building module is used to build a machine learning-based training model based on the data set;

[0092] A model training module is used to train the training model using the data set, and use the trained training model as a 3D printing parameter optimization model;

[0093] The parameter optimization module is used to optimize the 3D printing parameters to be optimized by using the 3D printing parameter optimization model.

[0094] In this embodiment, the 3D printing parameters include: laser power, scanning speed, layer thickness and substrate preheating temperature; the output results are stress and deformation distribution results.

[0095] In this embodiment, the model building module includes:

[0096] The algorithm selection submodule is used to select a machine learning algorithm based on the size of the data set;

[0097] The construction submodule is used to build a machine learning-based training model according to the selected machine learning algorithm.

[0098] Among them, when the scale of the data set is less than or equal to the first preset scale, the neural network algorithm is selected as the machine learning algorithm of this embodiment; when the scale of the data set is greater than the first preset scale, a more efficient random forest or gradient boosting tree algorithm is selected as the machine learning algorithm of this embodiment.

[0099] The process of acquiring the data set in this embodiment includes:

[0100] Get 3D printing parameters;

[0101] Perform thermal analysis simulation on the three-dimensional finite element model according to the 3D printing parameters to obtain thermal analysis results;

[0102] According to the thermal analysis results, stress-deformation analysis is performed on the three-dimensional finite element model to obtain the stress-deformation distribution results; the 3D printing parameters and the corresponding stress-deformation distribution results are combined to form a data set.

[0103] In this embodiment, the model training module includes:

[0104] The partitioning module is used to divide the data set into training set, test set and validation set;

[0105] The training submodule is used to train the training model using the training set to obtain a trained training model;

[0106] The test submodule is used to test the trained model using the test set to obtain a training model that passes the test;

[0107] The verification submodule is used to verify the model that has passed the test and obtain a verified training model;

[0108] The verified training model is used as the 3D printing parameter optimization model.

[0109] In a further embodiment, a judgment submodule is further included, and the judgment submodule is used to judge whether there is a training model that has passed the verification;

[0110] If not, the training model is retrained, tested and verified in sequence through the training submodule, the testing submodule and the verification submodule until there is a training model that passes the verification.

[0111] In order to ensure the quality of data, in this embodiment, the 3D printing parameters and their corresponding output results are preprocessed, specifically including:

[0112] Eliminate outliers on 3D printing parameters and their corresponding output results;

[0113] The 3D printing parameters and their corresponding output results after outlier removal are normalized.

[0114] In a further embodiment, the 3D printing parameters and their corresponding output results are subjected to outlier elimination processing, specifically including:

[0115] The Z-score statistical method is used to eliminate outliers in the 3D printing parameters and their corresponding output results.

[0116] Among them, the Z-score statistical method is used to eliminate outliers in the 3D printing parameters and their corresponding output results, including:

[0117] Calculate the mean μ and standard deviation σ of each parameter in the data set;

[0118] Each 3D printing parameter is calculated using a first preset formula; wherein the first preset formula is In the formula, X is the value of a parameter of the data set;

[0119] When the absolute value of Z corresponding to each 3D printing parameter is greater than 3, the 3D printing parameter is determined to be an abnormal value, and the abnormal value and its corresponding output result are eliminated.

[0120] In a further embodiment, the 3D printing parameters after the outlier removal process and the corresponding output results are normalized, specifically including:

[0121] Using a second preset formula to normalize the 3D printing parameters after outlier elimination and their corresponding output results;

[0122] Among them, the second preset formula is Where X' is the normalized value of X.

[0123] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A 3D printing parameter optimization method based on machine learning, wherein the model to be 3D printed is a three-dimensional finite element model of a target molded part; characterized in that: include: Acquire a data set; wherein the data set includes 3D printing parameters and their corresponding output results; Preprocess the dataset; Build a machine learning-based training model based on the data set; The training model is trained using the data set, and the trained model is used as a 3D printing parameter optimization model; The 3D printing parameter optimization model is used to optimize the 3D printing parameters to be optimized.

2. The 3D printing parameter optimization method based on machine learning according to claim 1, characterized in that: 3D printing parameters include: laser power, scanning speed, layer thickness and substrate preheating temperature; the output results are stress and deformation distribution results; Preferably, obtaining a data set specifically includes: Get 3D printing parameters; Perform thermal analysis simulation on the three-dimensional finite element model according to the 3D printing parameters to obtain thermal analysis results; According to the thermal analysis results, the stress-deformation analysis is performed on the three-dimensional finite element model to obtain the stress-deformation distribution results; The 3D printing parameters and the corresponding stress-deformation distribution results are combined to form a data set.

3. The 3D printing parameter optimization method based on machine learning according to claim 1, characterized in that: According to the data set, a training model based on machine learning is constructed, including: Choose a machine learning algorithm based on the size of the dataset; Build a machine learning-based training model based on the selected machine learning algorithm; Preferably, a machine learning algorithm is selected according to the size of the data set, including: When the size of the data set is less than or equal to a first preset size, selecting a neural network algorithm; When the size of the data set is larger than the first preset size, a more efficient random forest or gradient boosting tree algorithm is selected.

4. The 3D printing parameter optimization method based on machine learning according to claim 1, characterized in that: The training model is trained using the data set, and the trained model is used as a 3D printing parameter optimization model, including: Divide the dataset into training set, test set and validation set; The training model is trained, tested and verified in sequence using the training set, test set and verification set to obtain a verified training model; The verified training model is used as the 3D printing parameter optimization model; Preferably, if there is no training model that has passed the test or verification, the training set, test set and verification set are repeatedly used to train, test and verify the training model in sequence until there is a training model that has passed the verification.

5. The 3D printing parameter optimization method based on machine learning according to claim 1, characterized in that: Preprocessing the 3D printing parameters and their corresponding output results, including: Eliminate outliers on 3D printing parameters and their corresponding output results; The 3D printing parameters and their corresponding output results after outlier removal are normalized.

6. A 3D printing parameter optimization system based on machine learning, wherein the model to be 3D printed is a three-dimensional finite element model of a target molded part, characterized in that: include: An acquisition module, used to acquire a data set; wherein the data set includes 3D printing parameters and their corresponding output results; Preprocessing module, used to preprocess the data set; The model building module is used to build a machine learning-based training model based on the data set; A model training module is used to train the training model using the data set, and use the trained training model as a 3D printing parameter optimization model; The parameter optimization module is used to optimize the 3D printing parameters to be optimized by using the 3D printing parameter optimization model.

7. The 3D printing parameter optimization system based on machine learning according to claim 6, characterized in that: 3D printing parameters include: laser power, scanning speed, layer thickness and substrate preheating temperature; the output results are stress and deformation distribution results.

8. The 3D printing parameter optimization system based on machine learning according to claim 6, characterized in that: Model building modules, including: The algorithm selection submodule is used to select a machine learning algorithm based on the size of the data set; The construction submodule is used to build a machine learning-based training model according to the selected machine learning algorithm.

9. The 3D printing parameter optimization system based on machine learning according to claim 6, characterized in that: Preprocessing includes outlier removal and normalization.

10. The 3D printing parameter optimization system based on machine learning according to claim 6, characterized in that: The model training module includes: The partitioning module is used to divide the data set into training set, test set and validation set; The training submodule is used to train the training model using the training set to obtain a trained training model; The test submodule is used to test the trained model using the test set to obtain a training model that passes the test; The verification submodule is used to verify the model that has passed the test and obtain a verified training model; Preferably, it also includes a judgment submodule, which is used to judge whether there is a training model that has passed the verification; If not, the training model is retrained, tested and verified in sequence through the training submodule, the testing submodule and the verification submodule until there is a training model that passes the verification.

Images