Aluminum alloy extrusion prediction optimization method based on machine learning, and related apparatus
By using machine learning-based methods to generate simulation data from finite element models and train aluminum alloy extrusion models, the problem of time-consuming and costly traditional finite element analysis is solved. This enables rapid and accurate prediction of aluminum alloy extrusion parameters, reducing the cost and time of mold trial runs.
Patent Information
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- GUANGDONG INST OF NEW MATERIALS
- Filing Date
- 2024-12-06
- Publication Date
- 2026-06-04
AI Technical Summary
Traditional finite element analysis methods involve large computational loads and long computation times during aluminum alloy extrusion, making them inefficient for handling the massive mold demands of large enterprises. This results in high mold trial costs and long processing times.
A machine learning-based approach is adopted, using finite element models to generate simulation data, training an aluminum alloy extrusion model, and quickly predicting the maximum extrusion stress and exit temperature. The prediction process is then optimized using machine learning algorithms such as random forest models.
It enables rapid and accurate prediction of aluminum alloy extrusion parameters, reduces testing time and costs, supports rapid adjustments to production plans, and improves production efficiency.
Smart Images
Figure CN2024137284_04062026_PF_FP_ABST
Abstract
Description
A Machine Learning-Based Prediction and Optimization Method for Aluminum Alloy Extrusion and Related Devices
[0001] Cross-reference to related applications
[0002] This disclosure claims priority to Chinese Patent Application No. 2024117107851, filed on November 27, 2024, entitled "Method and Apparatus for Predicting and Optimizing Aluminum Alloy Extrusion Based on Machine Learning". Technical Field
[0003] This disclosure relates to the field of plastic processing of materials, and more specifically, to a machine learning-based method for predicting and optimizing aluminum alloy extrusion and related apparatus. Background Technology
[0004] Hot extrusion molding is mainly used to manufacture profiles, pipes, bars, and various ultra-wide, thin-walled profiles and small, high-precision profiles. Due to the high flexibility of hot extrusion molding and the diverse types and high precision of the products it produces, it has become a key area of industrial manufacturing. With the continuous development of manufacturing technology, hot-extruded products have been widely used in aerospace, transportation, and other fields.
[0005] Aluminum extrusion technology is an important branch of hot extrusion forming technology in the field of aluminum alloy processing. During aluminum extrusion, extrusion stress and exit temperature are key factors determining whether defects occur in the product. Therefore, effectively predicting the maximum extrusion stress and exit temperature and determining the optimal parameters is crucial. The extrusion stress and exit temperature of aluminum alloy extrusion are usually obtained through extrusion tests, but these tests require extensive preparation and consume a significant amount of time.
[0006] For example, before conducting an extrusion test, preheating and heat preservation of the extruder and die blank are required. During the test, the die needs to be adjusted and repaired multiple times, which not only prolongs the entire test but also increases the workload. Furthermore, since extrusion tests consume a large amount of aluminum, they also incur high material costs.
[0007] Therefore, the finite element method (FEM) is employed in related technologies. Since FEM can systematically solve boundary value problems, compared to aluminum alloy extrusion tests, it allows for the simultaneous simulation of multiple models without the need for preheating during testing, significantly reducing testing time and costs. This provides an important engineering foundation for achieving efficient aluminum extrusion production. This method is particularly important for predicting the maximum extrusion stress and exit temperature during aluminum alloy extrusion and reducing time costs.
[0008] However, in practice, it has been found that large enterprises need to use more than 2,000 different molds every month, of which more than 50% are new molds. The cost of a single mold trial ranges from RMB 1 million to RMB 10 million, and the annual mold trial cost is as high as RMB 40 billion. Faced with such a huge demand for molds, traditional finite element analysis methods are inadequate due to the large amount of computation and long calculation time, resulting in a significant time consumption even when simulations are performed using finite element tools. Summary of the Invention
[0009] The purpose of this disclosure includes, for example, providing a machine learning-based method and apparatus for predicting and optimizing aluminum alloy extrusion, capable of generating a large amount of simulation data using a finite element model to train an aluminum alloy extrusion model. This aluminum alloy extrusion model can quickly predict the maximum extrusion stress and exit temperature of the aluminum alloy based on input design process parameters, thereby providing a guiding basis for actual extrusion production and facilitating rapid adjustments to production plans.
[0010] The embodiments of this disclosure can be implemented as follows:
[0011] Embodiments of this disclosure provide a machine learning-based prediction and optimization method for aluminum alloy extrusion, the method comprising:
[0012] A simulation model of aluminum alloy extrusion process is obtained, wherein the simulation model can simulate the maximum extrusion stress and exit temperature of aluminum alloy using the input simulation process parameters;
[0013] Multiple simulation data points are generated using the simulation model;
[0014] The model is trained using multiple sets of simulation data to obtain a trained aluminum alloy extrusion model, wherein the aluminum alloy extrusion model is configured to predict the maximum extrusion stress and exit temperature of the aluminum alloy based on the input design process parameters.
[0015] Optionally, the simulation model is a finite element model that has been verified by measured data.
[0016] Optionally, generating multiple simulation data sets through the simulation model includes:
[0017] Receive multiple sets of simulation process parameters set by the user for the finite element model;
[0018] The finite element model was used to simulate and calculate the multiple sets of parameters to obtain multiple sets of maximum extrusion stress and outlet temperature.
[0019] Multiple sets of simulation data were obtained based on the multiple sets of simulation process parameters, the multiple sets of maximum extrusion stress and the exit temperature.
[0020] Optionally, the model is trained using multiple sets of simulation data to obtain a trained aluminum alloy extrusion model, including:
[0021] Multiple training models are trained using multiple sets of simulation data to obtain multiple candidate models.
[0022] The aluminum alloy extrusion model is selected from the multiple candidate models based on its best prediction performance.
[0023] Optionally, the plurality of models to be trained includes K-nearest neighbor algorithm model, ridge regression model, lasso regression, decision tree algorithm model, random forest model, XGBoost model, support vector machine regression-linear kernel function model, support vector machine regression-Gaussian kernel function model, and AdaBoost model.
[0024] Optionally, the aluminum alloy extrusion model is the model obtained by training the random forest model.
[0025] Optionally, each piece of simulation data includes the mold type and extrusion parameters;
[0026] Multiple training models are trained using the aforementioned simulation data to obtain multiple candidate models, including:
[0027] The process parameters of each simulation data are preprocessed to obtain preprocessed data for each simulation data. The preprocessing method for the mold category is thermal unique encoding, and the preprocessing method for the extrusion parameters is standardization.
[0028] Multiple sets of preprocessed data are input into the multiple models to be trained for training, resulting in multiple candidate models.
[0029] Optionally, the extrusion parameters are preprocessed using Z-Score normalization.
[0030] Optionally, the mold category is one of profiles, bars, tubes, and plates.
[0031] Optionally, the extrusion parameters include extrusion speed, extrusion temperature, extrusion cylinder temperature, and coefficient of friction.
[0032] Optionally, the friction coefficient in the extrusion parameters is obtained by mapping the extrusion cylinder temperature in the extrusion parameters.
[0033] Optionally, the relationship between the friction coefficient and the extrusion cylinder temperature in the extrusion parameters is as follows:
[0034] In the formula, μ(T) represents the friction coefficient when the extrusion cylinder temperature is T, T0 represents room temperature, μ(T0) represents the friction coefficient at room temperature, and a, b, and n represent preset coefficients, respectively.
[0035] Embodiments of this disclosure also provide a machine learning-based aluminum alloy extrusion prediction and optimization device, the device comprising:
[0036] The data generation module is configured to acquire a simulation model of the aluminum alloy extrusion process, wherein the simulation model can simulate the maximum extrusion stress and exit temperature of the aluminum alloy using the input simulation process parameters.
[0037] The data generation module is also configured to generate multiple simulation data through the simulation model;
[0038] The model training module is configured to train the model using multiple sets of simulation data to obtain a trained aluminum alloy extrusion model. The aluminum alloy extrusion model is configured to predict the maximum extrusion stress and exit temperature of the aluminum alloy based on the input design process parameters.
[0039] Optionally, the simulation model is a finite element model that has been verified by measured data.
[0040] Optionally, the data generation module is further configured to:
[0041] Receive multiple sets of simulation process parameters set by the user for the finite element model;
[0042] The finite element model was used to simulate and calculate the multiple sets of parameters to obtain multiple sets of maximum extrusion stress and outlet temperature.
[0043] Multiple sets of simulation data were obtained based on the multiple sets of simulation process parameters, the multiple sets of maximum extrusion stress and the exit temperature.
[0044] Optionally, the model training module is further configured to:
[0045] Multiple training models are trained using multiple sets of simulation data to obtain multiple candidate models.
[0046] The aluminum alloy extrusion model is selected from the multiple candidate models based on its best prediction performance.
[0047] Optionally, the plurality of models to be trained includes K-nearest neighbor algorithm model, ridge regression model, lasso regression, decision tree algorithm model, random forest model, XGBoost model, support vector machine regression-linear kernel function model, support vector machine regression-Gaussian kernel function model, and AdaBoost model.
[0048] Optionally, the aluminum alloy extrusion model is the model obtained by training the random forest model.
[0049] Optionally, each piece of simulation data includes a mold type and extrusion parameters; the model training module is further configured to:
[0050] The process parameters of each simulation data are preprocessed to obtain preprocessed data for each simulation data. The preprocessing method for the mold category is thermal unique encoding, and the preprocessing method for the extrusion parameters is standardization.
[0051] Multiple sets of preprocessed data are input into the multiple models to be trained for training, resulting in multiple candidate models.
[0052] Optionally, the extrusion parameters are preprocessed using Z-Score normalization.
[0053] Optionally, the mold category is one of profiles, bars, tubes, and plates.
[0054] Optionally, the extrusion parameters include extrusion speed, extrusion temperature, extrusion cylinder temperature, and coefficient of friction.
[0055] Optionally, the friction coefficient in the extrusion parameters is obtained by mapping the extrusion cylinder temperature in the extrusion parameters.
[0056] Optionally, the relationship between the friction coefficient and the extrusion cylinder temperature in the extrusion parameters is as follows:
[0057] In the formula, μ(T) represents the friction coefficient when the extrusion cylinder temperature is T, T0 represents room temperature, μ(T0) represents the friction coefficient at room temperature, and a, b, and n represent preset coefficients, respectively.
[0058] Embodiments of this disclosure also provide a storage medium storing a computer program that, when executed by a processor, implements the machine learning-based aluminum alloy extrusion prediction and optimization method.
[0059] Embodiments of this disclosure also provide an electronic device, which includes a processor and a memory. The memory stores a computer program, and when the computer program is executed by the processor, it implements the machine learning-based aluminum alloy extrusion prediction and optimization method.
[0060] In combination with the above technical solutions, the beneficial effects brought about by the embodiments of this disclosure include, for example:
[0061] An electronic device acquires a simulation model of the aluminum alloy extrusion process. This simulation model can simulate the maximum extrusion stress and exit temperature of the aluminum alloy using input process parameters. Multiple simulation data points are generated from the model, and these data points are used to train the model, resulting in a trained aluminum alloy extrusion model. In this way, a large amount of simulation data is generated using the finite element model to train the aluminum alloy extrusion model. This aluminum alloy extrusion model can quickly predict the maximum extrusion stress and exit temperature of the aluminum alloy based on input design process parameters, thus providing a guiding basis for actual extrusion production and facilitating rapid adjustments to production plans. Attached Figure Description
[0062] To more clearly illustrate the technical solutions of the embodiments of this disclosure, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of this disclosure and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.
[0063] Figure 1 is a flowchart illustrating the machine learning-based aluminum alloy extrusion prediction and optimization method provided in this embodiment of the present disclosure.
[0064] Figure 2 is a comparison chart of the test results of the compressive stress provided in the embodiments of this disclosure;
[0065] Figure 3 is a test comparison chart of the outlet temperature provided in the embodiments of this disclosure;
[0066] Figure 4 is a schematic diagram of the structure of the machine learning-based aluminum alloy extrusion prediction and optimization method provided in the embodiments of this disclosure;
[0067] Figure 5 is a schematic diagram of the structure of the electronic device provided in an embodiment of this disclosure.
[0068] Icons: 11-Data generation module; 12-Model training module; 21-Memory; 22-Processor; 23-Communication unit; 24-System bus. Detailed Implementation
[0069] To make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, and not all embodiments. The components of the embodiments of this disclosure described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.
[0070] Therefore, the following detailed description of the embodiments of this disclosure provided in the accompanying drawings is not intended to limit the scope of the claimed disclosure, but merely to illustrate selected embodiments of the disclosure. All other embodiments obtained by those skilled in the art based on the embodiments of this disclosure without inventive effort are within the scope of protection of this disclosure.
[0071] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.
[0072] In the description of this disclosure, it should be noted that the terms "first," "second," "third," etc., are used only for distinguishing descriptions and should not be construed as indicating or implying relative importance. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.
[0073] Based on the above statement, as introduced in the background section, although finite element models can simulate multiple models simultaneously without conducting experiments and without the preheating steps required in experimental processes, thus significantly reducing experimental time and costs, traditional finite element analysis methods are inadequate for the massive mold demands of large enterprises due to their large computational load and long computation time. Even simulations using finite element tools still require a significant amount of time.
[0074] Based on the discovery of the aforementioned technical problems, the inventors, through creative labor, proposed the following technical solutions to solve or improve these problems. It should be noted that the defects in the solutions of the aforementioned related technologies are all results derived by the inventors after practical experience and careful research. Therefore, the discovery process of the aforementioned problems and the solutions proposed in the embodiments of this disclosure below should be considered contributions made by the inventors to this disclosure during the inventive process, and should not be construed as technical content known to those skilled in the art.
[0075] In view of the above problems, embodiments of this disclosure provide a machine learning-based prediction and optimization method for aluminum alloy extrusion. In this method, an electronic device acquires a simulation model of the aluminum alloy extrusion process. This simulation model can simulate the maximum extrusion stress and exit temperature of the aluminum alloy using input simulation process parameters. Multiple simulation data points are generated from the simulation model. These multiple simulation data points are then used to train the model, resulting in a trained aluminum alloy extrusion model. Thus, a large amount of simulation data is generated using a finite element model to train the aluminum alloy extrusion model. This aluminum alloy extrusion model can quickly predict the maximum extrusion stress and exit temperature of the aluminum alloy based on input design process parameters, thereby providing a guiding basis for actual extrusion production and facilitating rapid adjustments to production plans.
[0076] It should be noted that the electronic device implementing the above-described machine learning-based aluminum alloy extrusion prediction and optimization method can be, but is not limited to, mobile terminals, tablet computers, laptop computers, desktop computers, servers, etc., as long as they can provide sufficient computing power. The server can be a single server or a group of servers. The server group can be centralized or distributed (e.g., the servers can be a distributed system). In some embodiments, the server can be local or remote relative to the user terminal. In some embodiments, the server can be implemented on a cloud platform; by way of example only, the cloud platform can include private cloud, public cloud, hybrid cloud, community cloud, distributed cloud, inter-cloud, multi-cloud, etc., or any combination thereof. In some embodiments, the server can be implemented on an electronic device having one or more components.
[0077] To make the solutions provided by the embodiments of this disclosure clearer, a server is used as the electronic device implementing the method below, and the various steps of the method are described in detail with reference to FIG1. However, it should be understood that the operations in the flowchart may not be implemented in sequence, and steps without logical contextual relationships may be reversed in order or implemented simultaneously. Furthermore, those skilled in the art, guided by the content of this disclosure, may add one or more other operations to the flowchart, or remove one or more operations from the flowchart. As shown in FIG1, the method includes:
[0078] S1, obtain the simulation model of aluminum alloy extrusion process.
[0079] The simulation model can simulate the maximum extrusion stress and exit temperature of the aluminum alloy using the input simulation process parameters. In the embodiments of this disclosure, the simulation model can be a finite element model verified by measured data.
[0080] To validate the constructed finite element model, as an optional implementation, an actual aluminum alloy extrusion test can be conducted first to collect data such as the maximum extrusion stress and exit temperature during the extrusion process. This data can then be used to verify the accuracy of the subsequent finite element model. Next, the user can set the billet size according to actual requirements and create a three-dimensional geometric model of the extrusion cylinder, billet, and extrusion rod. This three-dimensional geometric model is then input into finite element simulation software (e.g., Deform-3D) running on the server. The server then receives the material processing parameters, material parameters, and simulation control parameters set by the user according to the actual experimental requirements. These parameters specifically include extrusion speed, extrusion temperature, die temperature, friction coefficient, thermal conductivity coefficient, and the mesh size of the three-dimensional geometric model. The server simulates the user-input parameters using the finite element simulation software, generates simulation data, and compares the simulation data with the experimental results for verification. If the deviation between the two is outside the error range, the server continues to receive adjustments from the user to the finite element model until both results are within the error range, thus ensuring the accuracy of the finite element model.
[0081] In the embodiments of this disclosure, a finite element model is constructed using 7N01 aluminum alloy as the research object, wherein the material parameters of 7N01 are shown in the table below:
[0082] In addition, the material properties of 7N01 aluminum alloy are shown in the table below:
[0083] Of course, the methods provided in the embodiments of this disclosure are not limited to 7N01 aluminum alloy, but are also applicable to other types of aluminum alloys.
[0084] Based on the above description of the simulation model in step S1, and referring to Figure 1, the method further includes:
[0085] S2 generates multiple simulation data points through a simulation model.
[0086] It should be noted that the final prediction performance of the neural network model is directly related to the quality and quantity of the training samples. However, during the research process, it was found that relevant data on the hot extrusion forming process of aluminum alloys is extremely scarce. Therefore, in order to obtain a sufficient number of training samples, embodiments of this disclosure generate a large number of samples through a simulation model. In embodiments of this disclosure, the server can receive multiple sets of simulation process parameters set by the user for the finite element model; perform simulation calculations on the multiple sets of parameters through the finite element model to obtain multiple sets of maximum extrusion stress and exit temperature; and obtain multiple sets of simulation data based on the multiple sets of simulation process parameters and the multiple sets of maximum extrusion stress and exit temperature.
[0087] For example, the server can receive different parameters set by the user according to actual experimental requirements, simulate and record the generated extrusion stress and outlet temperature, forming multiple simulation data sets to constitute a dataset, which is then used for subsequent machine learning training. As an optional implementation, each simulation data set includes a die type and extrusion parameters. The die type in each simulation data set is one of profiles, bars, tubes, or plates. The extrusion parameters include extrusion speed, extrusion temperature, die temperature, extrusion barrel temperature, and coefficient of friction.
[0088] Extrusion speed refers to the speed at which material passes through the die during the extrusion process. Extrusion speed has a significant impact on product quality; therefore, an appropriate extrusion speed can ensure uniform material flow, reduce defects, and improve product precision. Specifically, if the extrusion speed is too fast, it may lead to uneven material flow, resulting in internal cracks or surface defects; while if the extrusion speed is too slow, it will affect production efficiency and may result in a rough product surface.
[0089] Extrusion temperature refers to the temperature reached by the material during the extrusion process. Extrusion temperature is one of the key parameters in the aluminum alloy extrusion process, as it affects the material's plasticity, fluidity, and final mechanical properties.
[0090] Die temperature refers to the operating temperature of the die during the extrusion process. Die temperature has a significant impact on the surface quality, dimensional accuracy, and die life of the product. Appropriate die temperature can reduce friction between the material and the die, improve the surface finish of the product, and extend the die's lifespan.
[0091] Extrusion cylinder temperature refers to the temperature of the extrusion cylinder (the container in which the material is extruded). The extrusion cylinder temperature needs to match the material's extrusion temperature to ensure good flowability and uniform deformation during extrusion. Therefore, controlling the extrusion cylinder temperature is crucial for maintaining consistent product quality.
[0092] The coefficient of friction refers to the degree of friction between the material and the surfaces of the die and extrusion cylinder during the extrusion process. The magnitude of the coefficient of friction affects the magnitude of the extrusion pressure and the material flow behavior. An excessively high coefficient of friction will lead to increased extrusion pressure, increased energy consumption, and may cause a decrease in product surface quality; while an excessively low coefficient of friction may lead to unstable material flow, affecting the accuracy of product shape and dimensions.
[0093] It is worth noting that, during the research, it was found that, unlike other parameters in the extrusion parameters which can be arbitrarily set by the user, the friction coefficient is not a constant about the material, but rather a parameter that varies with temperature. Further research revealed that the friction coefficient is affected by the extrusion cylinder temperature; therefore, the friction coefficient in the extrusion parameters is obtained by mapping it to the extrusion cylinder temperature. In this regard, the server can obtain a gain coefficient for the friction coefficient at room temperature based on the ratio between the extrusion cylinder temperature and a preset room temperature; applying this gain coefficient to the friction coefficient at room temperature yields the friction coefficient. To obtain the mapping relationship between the two, the embodiments of this disclosure utilize a large amount of measured data for fitting, resulting in the following relationship between the friction coefficient in the extrusion parameters and the extrusion cylinder temperature:
[0094] In the formula, μ(T) represents the friction coefficient when the extrusion cylinder temperature is T, T0 represents room temperature, μ(T0) represents the friction coefficient at room temperature, and a, b, n represent preset coefficients, which are all obtained through fitting.
[0095] Based on the above description of the multiple simulation data in step S2, and referring to Figure 1, the method further includes:
[0096] S3 uses multiple simulation data to train the model and obtain the trained aluminum alloy extrusion model.
[0097] The aluminum alloy extrusion model is configured to predict the maximum extrusion stress and exit temperature of the aluminum alloy based on the input design process parameters. Given the large number of machine learning models available, but not all models being suitable for the application scenarios of the embodiments of this disclosure, the embodiments of this disclosure provide the following optional implementations of step S3:
[0098] S3-1 uses multiple simulation data to train multiple models to obtain multiple candidate models.
[0099] During the research, it was learned that Scikit-learn, as an open-source machine learning library, is built upon open-source projects such as NumPy, SciPy, and matplotlib, providing a series of simple and effective data mining and data analysis tools. Therefore, Scikit-learn has extensive machine learning and data mining capabilities, including but not limited to classification, regression, clustering, data preprocessing, model selection, and evaluation.
[0100] Therefore, the aforementioned multiple models to be trained can include the multiple models to be trained described in Scikit-learn, including the K-nearest neighbor algorithm model, ridge regression model, lasso regression model, decision tree algorithm model, random forest model, XGBoost model, support vector machine regression-linear kernel function model, support vector machine regression-Gaussian kernel function model, and AdaBoost model. The following sections will provide a detailed explanation of each of these algorithms.
[0101] The K-Nearest Neighbors (KNN) algorithm is a basic classification and regression method. Given a training dataset, for a new input instance, the KNN algorithm finds the K nearest neighbors in the training set and then uses the output values of these K neighbors to predict the output value of the new instance through voting (classification problems) or averaging (regression problems).
[0102] Ridge regression is a regression method used for analyzing collinear data. It reduces the risk of overfitting by introducing an L2 regularization term. This is achieved by adding a term to the loss function that is proportional to the sum of the squares of the regression coefficients.
[0103] The LASSO Regression model is a regression model that uses L1 regularization to compress the estimation. Its feature is that it can perform feature selection because some regression coefficients can be reduced to zero, thus making the model more concise.
[0104] Decision Tree is a common machine learning algorithm that uses a tree-like structure for decision-making. Each internal node represents a test on an attribute, each branch represents a test output, and each leaf node represents a classification or regression result.
[0105] The Random Forest model is a classifier consisting of multiple decision trees, where the output class is determined by the mode of the classes output by the individual trees. When training each tree, the Random Forest randomly selects a subset of features to increase the model's generalization ability.
[0106] XGBoost is an optimized distributed gradient boosting library designed for high efficiency, flexibility, and portability. It is a machine learning framework based on Gradient Boosting Decision Trees (GBDT), offering exceptional efficiency and scalability.
[0107] Support Vector Machine Regression - Linear Kernel Function Model: This algorithmic model uses a linear kernel function in Support Vector Machines for classification and regression analysis. It separates data into different categories by finding a hyperplane that maximizes the margin.
[0108] Support Vector Machine Regression with Gaussian Kernel Function: This algorithm, which uses the Gaussian Radial Basis Function (RBF) as the kernel function, can handle linearly inseparable data. The RBF kernel function maps the input space to a higher-dimensional space, allowing the finding of a separating hyperplane.
[0109] The AdaBoost model is a boosting algorithm that iteratively trains a classifier, adjusting the weights of data points in each iteration so that data points misclassified by the previous classifier receive more attention in the next classifier. Finally, these classifiers are merged using a weighted majority vote to form the final strong classifier.
[0110] Based on the above embodiments introducing multiple models to be trained, as an optional implementation method, the server can preprocess the process parameters of each simulation data to obtain preprocessed data for each simulation data. The preprocessing method for mold category is thermal unique encoding, and the preprocessing method for extrusion parameters is standardization. Multiple preprocessed data are input into multiple models to be trained for training to obtain multiple candidate models trained.
[0111] Regarding the above implementation method, it should be understood that mold categories such as profiles, bars, pipes, and plates are not continuous numerical features and cannot be directly used in machine learning model training. This implementation uses one-hot encoding for them. Specifically, the server can use the OneHotEncoder function in sklearn.preprocessing to perform one-hot encoding, encoding profiles as 1000, plates as 0100, pipes as 0010, and bars as 0001, and adding them to the feature combination.
[0112] Furthermore, it should be understood that in machine learning algorithms, features should not have a high degree of correlation. If the magnitudes of the features differ significantly, the algorithm will struggle to converge or require a very long training time. Therefore, in the embodiments of this disclosure, Z-Score normalization can be used to preprocess the squeezing parameters to obtain standardized data.
[0113] Z-Score standardization, also known as standardization or standard score standardization, scales features by adjusting their mean and standard deviation, ensuring that each feature has a mean of 0 and a standard deviation of 1. The formula used for Z-Score standardization for each parameter in the squeezing parameters is as follows:
[0114] In the formula, x z-score σ represents the extrusion parameters after standard deviation, where x represents the original extrusion parameters, μ represents the mean of the characteristic data, and σ represents the standard deviation of the extrusion parameters.
[0115] In this way, multiple preprocessed data are obtained using the above implementation method, and then input into different training models to obtain multiple candidate models.
[0116] S3-2: Select the model with the best prediction performance from multiple candidate models as the aluminum alloy extrusion model.
[0117] To continue using the above nine models as examples, we will refer to them as KNN, RIDGE, LASSO, DT, RF, XGBOOST, SVM_LINEAR, SVM_RBF, and ADA, respectively. In this example, we will use the coefficient of determination R0. 2 As an evaluation metric for generalization performance, R 2 A value closer to 1 indicates a better fit. The nine models were tested using a test set, and the results are shown in Figures 2 and 3. Figure 2 shows the test set R for each model after 10 training cycles under extrusion stress. 2 Figure 3 shows the comparison of mean results for the test set R of each model after 10 training runs for the outlet temperature. 2 The mean results are compared in the graph. After comparison, it's clear that the Random Forest (RF) model has a lower R-value in the two comparison graphs. 2 The values are 0.88 and 0.9 respectively. Therefore, this model performs the best among the nine models, that is, the aluminum alloy extrusion model is the model trained by the random forest model.
[0118] Finally, embodiments of this disclosure also use the trained aluminum alloy extrusion model to predict random parameters, conduct experiments on the prediction results, and compare and analyze the error and optimization effect between the prediction results of the aluminum alloy extrusion model and the experimental data. Some parameters in the experimental process are shown in the table below:
[0119] During the experiment, the billet undergoes rapid deformation after extrusion, generating a large amount of heat and causing a sharp rise in the exit temperature. As the extrusion reaches a steady state, the billet's exit temperature gradually stabilizes. The average exit temperature obtained from the experiment is approximately 513.5 degrees Celsius. Under the same parameters, the aluminum alloy extrusion model predicts an exit temperature of 507.5 degrees Celsius, with an error of 1.1%, thus meeting the requirements.
[0120] Based on the same inventive concept as the machine learning-based aluminum alloy extrusion prediction and optimization method provided in the embodiments of this disclosure, the embodiments of this disclosure also provide a machine learning-based aluminum alloy extrusion prediction and optimization apparatus. This apparatus includes at least one software functional module that can be stored in memory or embedded in an electronic device. The processor in the electronic device is configured to execute the executable module stored in memory. For example, the software functional module and computer program included in the apparatus. Referring to Figure 4, functionally, the apparatus may include:
[0121] The data generation module 11 is configured to acquire a simulation model of the aluminum alloy extrusion process, wherein the simulation model can use the input simulation process parameters to simulate the maximum extrusion stress and exit temperature of the aluminum alloy.
[0122] The data generation module 11 is also configured to generate multiple simulation data through the simulation model;
[0123] Model training module 12 is configured to train the model using multiple simulation data to obtain a trained aluminum alloy extrusion model. The aluminum alloy extrusion model is configured to predict the maximum extrusion stress and exit temperature of the aluminum alloy based on the input design process parameters.
[0124] In this embodiment, the data generation module 11 is configured to implement steps S1 and S2 in Figure 1, and the model training module 12 is configured to implement step S3 in Figure 1. Therefore, for a detailed description of each module, please refer to the specific implementation method of the corresponding step. Of course, since it shares the same inventive concept as the machine learning-based aluminum alloy extrusion prediction and optimization method, this machine learning-based aluminum alloy extrusion prediction and optimization device can also implement other steps or sub-steps of the method through the above modules, which will not be elaborated upon in this embodiment.
[0125] In addition, the functional modules in the various embodiments of this disclosure can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.
[0126] It should also be understood that if the above embodiments are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this disclosure, in essence, or the part that contributes to the related technology, or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this disclosure.
[0127] Therefore, embodiments of this disclosure also provide a storage medium, which is a computer-readable storage medium. This storage medium stores a computer program, which, when executed by a processor, implements the machine learning-based aluminum alloy extrusion prediction and optimization method provided in embodiments of this disclosure. The storage medium can be any medium capable of storing program code, such as a USB flash drive, portable hard drive, read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.
[0128] This disclosure provides an electronic device for implementing a machine learning-based aluminum alloy extrusion prediction and optimization method. As shown in FIG5, the electronic device may include a processor 22 and a memory 21. The memory 21 stores a computer program, and the processor implements the machine learning-based aluminum alloy extrusion prediction and optimization method provided in this disclosure by reading and executing the computer program corresponding to the above embodiments in the memory 21.
[0129] Referring again to Figure 5, the electronic device may also include a communication unit 23. The memory 21, processor 22, and communication unit 23 are electrically connected to each other directly or indirectly via system bus 24 to realize data transmission or interaction.
[0130] The memory 21 can be an information recording device based on any electronic, magnetic, optical, or other physical principles, configured to record execution instructions, data, etc. In some embodiments, the memory 21 can be, but is not limited to, volatile memory, non-volatile memory, memory drive, etc.
[0131] In some embodiments, the volatile memory may be random access memory (RAM); in some embodiments, the non-volatile memory may be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), flash memory, etc.; in some embodiments, the storage drive may be a disk drive, solid-state drive, any type of storage disk (such as optical disc, DVD, etc.), or similar storage media, or a combination thereof.
[0132] The communication unit 23 is configured to send and receive data via a network. In some embodiments, the network may include a wired network, a wireless network, a fiber optic network, a telecommunications network, an intranet, the Internet, a local area network (LAN), a wide area network (WAN), a wireless local area network (WLAN), a metropolitan area network (MAN), a public switched telephone network (PSTN), a Bluetooth network, a ZigBee network, or a near field communication (NFC) network, or any combination thereof. In some embodiments, the network may include one or more network access points. For example, the network may include wired or wireless network access points, such as base stations and / or network switching nodes, through which one or more components of the service request processing system can connect to the network to exchange data and / or information.
[0133] The processor 22 may be an integrated circuit chip with signal processing capabilities, and may include one or more processing cores (e.g., a single-core processor or a multi-core processor). By way of example only, the processor described above may include a Central Processing Unit (CPU), an Application Specific Integrated Circuit (ASIC), an Application Specific Instruction-set Processor (ASIP), a Graphics Processing Unit (GPU), a Physics Processing Unit (PPU), a Digital Signal Processor (DSP), a Field Programmable Gate Array (FPGA), a Programmable Logic Device (PLD), a controller, a microcontroller unit, a Reduced Instruction Set Computing (RISC) computer, or a microprocessor, or any combination thereof.
[0134] It is understood that the structure shown in Figure 5 is for illustrative purposes only. The electronic device 100 may also have more or fewer components than shown in Figure 5, or may have a different configuration than shown in Figure 5. The components shown in Figure 5 may be implemented using hardware, software, or a combination thereof.
[0135] It should be understood that the apparatus and methods disclosed in the above embodiments can also be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the flowcharts and block diagrams in the accompanying drawings show the architecture, functionality, and operation of possible implementations of apparatus, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions that implement the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.
[0136] The above descriptions are merely various embodiments of this disclosure, but the scope of protection of this disclosure is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this disclosure should be included within the scope of protection of this disclosure. Therefore, the scope of protection of this disclosure should be determined by the scope of the claims. Industrial applicability
[0137] In summary, this disclosure provides a machine learning-based method and apparatus for predicting and optimizing aluminum alloy extrusion. It can generate a large amount of simulation data using a finite element model to train an aluminum alloy extrusion model. This aluminum alloy extrusion model can quickly predict the maximum extrusion stress and exit temperature of the aluminum alloy based on the input design process parameters. Therefore, the aluminum alloy extrusion model can provide a guiding basis for actual extrusion production, facilitating rapid adjustments to production plans.
Claims
1. A machine learning based aluminum alloy extrusion prediction optimization method, characterized by, The method includes: A simulation model of aluminum alloy extrusion process is obtained, wherein the simulation model can simulate the maximum extrusion stress and exit temperature of aluminum alloy using the input simulation process parameters; Multiple simulation data points are generated using the simulation model; The model is trained using multiple sets of simulation data to obtain a trained aluminum alloy extrusion model, wherein the aluminum alloy extrusion model is configured to predict the maximum extrusion stress and exit temperature of the aluminum alloy based on the input design process parameters.
2. The machine learning based aluminum alloy extrusion prediction optimization method of claim 1, wherein, The simulation model is a finite element model that has been verified by measured data.
3. The machine learning based aluminum alloy extrusion prediction optimization method of claim 2, wherein, The generation of multiple simulation data through the simulation model includes: Receive multiple sets of simulation process parameters set by the user for the finite element model; The finite element model was used to simulate and calculate the multiple sets of parameters to obtain multiple sets of maximum extrusion stress and outlet temperature. Multiple sets of simulation data were obtained based on the multiple sets of simulation process parameters, the multiple sets of maximum extrusion stress and the exit temperature.
4. The machine learning based aluminum alloy extrusion prediction optimization method of claim 1, wherein, The model was trained using multiple sets of simulation data to obtain a trained aluminum alloy extrusion model, including: Multiple training models are trained using multiple sets of simulation data to obtain multiple candidate models. The aluminum alloy extrusion model is selected from the multiple candidate models based on its best prediction performance.
5. The machine learning based aluminum alloy extrusion prediction optimization method of claim 4, wherein, The multiple models to be trained include the K-nearest neighbor algorithm model, ridge regression model, lasso regression model, decision tree algorithm model, random forest model, XGBoost model, support vector machine regression-linear kernel model, support vector machine regression-Gaussian kernel model, and AdaBoost model.
6. The machine learning based aluminum alloy extrusion prediction optimization method of claim 5, wherein, The aluminum alloy extrusion model is a model obtained by training the random forest model.
7. The machine learning based aluminum alloy extrusion prediction optimization method of claim 4, wherein, Each piece of simulation data includes the mold type and extrusion parameters; Multiple training models are trained using the aforementioned simulation data to obtain multiple candidate models, including: The process parameters of each simulation data are preprocessed to obtain preprocessed data for each simulation data. The preprocessing method for the mold category is thermal unique encoding, and the preprocessing method for the extrusion parameters is standardization. Multiple sets of preprocessed data are input into the multiple models to be trained for training, resulting in multiple candidate models.
8. The machine learning based aluminum alloy extrusion prediction optimization method of claim 7, wherein, The extrusion parameters were preprocessed using Z-Score normalization.
9. The machine learning based aluminum alloy extrusion prediction optimization method of claim 7, wherein, The mold category is one of the following: profile, bar, tube, or plate.
10. The machine learning based aluminum alloy extrusion prediction optimization method of claim 7, wherein, The extrusion parameters include extrusion speed, extrusion temperature, extrusion cylinder temperature, and coefficient of friction.
11. The machine learning based aluminum alloy extrusion prediction optimization method of claim 10, wherein, The friction coefficient in the extrusion parameters is obtained by mapping the extrusion cylinder temperature in the extrusion parameters.
12. The machine learning-based aluminum alloy extrusion prediction and optimization method according to claim 11, the method further includes: The gain coefficient of friction coefficient at room temperature is obtained based on the ratio between the extrusion cylinder temperature and the preset room temperature. The friction coefficient is obtained by applying the gain coefficient to the friction coefficient at room temperature.
13. The machine learning based aluminum alloy extrusion prediction optimization method of claim 12, wherein, The relationship between the friction coefficient and the barrel temperature in the extrusion parameters is: In the formula, μ(T) represents the friction coefficient when the extrusion cylinder temperature is T, T0 represents room temperature, μ(T0) represents the friction coefficient at room temperature, and a, b, and n represent preset coefficients, respectively.
14. A machine learning based aluminum alloy extrusion prediction optimization apparatus, characterized by, The device includes: The data generation module is configured to obtain a simulation model of an aluminum alloy extrusion process, wherein the simulation model is capable of simulating a maximum extrusion stress and an outlet temperature of the aluminum alloy by using input simulation process parameters; The data generation module is further configured to generate a plurality of simulation data by using the simulation model; The model training module is configured to perform model training by using the plurality of simulation data, and obtain a trained aluminum alloy extrusion model, wherein the aluminum alloy extrusion model is configured to predict the maximum extrusion stress and the outlet temperature of the aluminum alloy according to input design process parameters.
15. A storage medium, characterized by The storage medium stores a computer program, and the computer program is executed by the processor to implement the machine learning-based aluminum alloy extrusion prediction and optimization method in any one of claims 1-13.
16. An electronic device, comprising: The electronic device includes a processor and a memory, and the memory stores a computer program, and the computer program is executed by the processor to implement the machine learning-based aluminum alloy extrusion prediction and optimization method in any one of claims 1-13.