Machine learning-oriented aluminum alloy tensile strength feature extension method

Abstract

The application provides an aluminum alloy tensile strength feature expansion method for machine learning, which comprises the following steps: collecting tensile strength data, selecting multiple tensile strength data as feature values for machine learning to predict tensile strength; performing one-hot encoding processing on the selected multiple tensile strength data; performing preprocessing on the multiple tensile strength data after the one-hot encoding processing; establishing a machine learning model, training the established machine learning model by using a data set, and predicting the tensile strength by using the trained machine learning model; analyzing the feature importance, finding out the feature variable that has the greatest influence on the prediction result of the machine learning model; and performing feature expansion to obtain an expanded feature with greater importance. The application can expand the feature value with the greatest influence from multiple tensile strength feature variables.

Description

Technical Field

[0001] This invention relates to the fields of aluminum alloy material properties and artificial intelligence technology, and more specifically, to a method for extending the tensile strength characteristics of aluminum alloys for machine learning. Background Technology

[0002] Tensile strength is one of the most important properties of aluminum alloys, indicating the actual load-bearing capacity of tough metallic materials. Currently, most researchers lack a quantitative description of the relationship between tensile strength and its influencing factors, leading to a lack of systematic theoretical guidance in experimental research, a degree of uncertainty in experiments, and increased experimental costs and time. To reduce the "trial and error" costs of experiments, combining machine learning with materials research is becoming increasingly important.

[0003] Machine learning is a multidisciplinary field that aims to enable computers to learn autonomously by leveraging large amounts of existing data and accumulated experience through data processing and model building. Since features have varying impacts on target attributes, the performance of a machine learning model largely depends on feature selection; therefore, choosing appropriate features is crucial for model building. Features represent factors that influence performance, but in machine learning, more features are not necessarily better; rather, more meaningful features are preferable. Features that are more meaningful to performance increase the interpretability of the machine learning model and greatly help improve the model's predictive accuracy. Some features, while impacting performance, have a relatively small impact compared to others and do not play a significant role in machine learning model training. Therefore, adding meaningful features is very important in machine learning. A common feature expansion method in machine learning is to use the PolynomialFeatures function for polynomial expansion. While this method can increase the dimensionality of features, some less important features may also participate in the expansion, potentially resulting in unimportant expanded features. Summary of the Invention

[0004] Based on this, and to address the problems existing in the aforementioned background technology, the present invention provides a machine learning-oriented method for extending the tensile strength feature of aluminum alloys, the specific technical solution of which is as follows:

[0005] A machine learning-oriented method for extending the tensile strength feature of aluminum alloys includes the following steps:

[0006] Collect tensile strength data and select multiple tensile strength data as feature values ​​for machine learning to predict tensile strength;

[0007] Perform unique thermal encoding on the selected tensile strength data;

[0008] Preprocessing is performed on multiple tensile strength data after unique thermal encoding;

[0009] Establish a machine learning model, train the established machine learning model using the dataset, and use the trained machine learning model to predict tensile strength.

[0010] Analyze the relationships between feature importance to identify the feature variables that have the greatest impact on the prediction results of the machine learning model;

[0011] Feature expansion is performed to obtain expanded features that are of greater importance.

[0012] Among the features extended using the machine learning-oriented aluminum alloy tensile strength feature extension method, we found that one extended feature exhibits exceptional importance in the tensile strength model, far exceeding any of the original features. This is extremely helpful for materials science research. Feature extension not only avoids losing crucial feature information and increasing computation time, but also improves the accuracy of model predictions, enhances the interpretability of machine learning models, and facilitates researchers in further unraveling the "black box" of the relationship between materials and machine learning.

[0013] Furthermore, multiple tensile strength data include the contents of Si, Fe, Cu, Mn, Mg, Zn, Pb, Cd, and Sn metal elements, with the tensile strength value used as the target variable.

[0014] Furthermore, the OneHotEncoder function is used to perform hot encoding on the selected tensile strength data, encoding the recasting method as 10 and the die-casting method as 01 in the forming method and adding them to the feature combination.

[0015] Furthermore, z-score normalization was used to preprocess the multiple tensile strength data after one-hot encoding.

[0016] Furthermore, the dataset was trained using the RF machine learning model in scikit-learn, and the importance of features was analyzed using the features_importance function in the algorithm.

[0017] Furthermore, the EVAL function is used to expand the features, and the expanded features are added together with the original features to train the machine learning model. The features_importance function under the algorithm is used to analyze the importance of the features.

[0018] Furthermore, a computer-readable storage medium stores a computer program that, when executed by a processor, implements the machine learning-oriented method for extending the tensile strength characteristics of aluminum alloys. Attached Figure Description

[0019] The invention will be further understood from the following description taken in conjunction with the accompanying drawings. The components in the drawings are not necessarily drawn to scale, but rather the emphasis is on illustrating the principles of the embodiments. In different views, the same reference numerals designate corresponding parts.

[0020] Figure 1 This is a flowchart illustrating a method for extending the tensile strength feature of aluminum alloys for machine learning, according to one embodiment of the present invention.

[0021] Figure 2 This is a comparison chart showing the analysis of feature importance using the features_importance function in the algorithm after training the dataset with the RandomForest (RF) machine learning model in scikit-learn during model training in one embodiment of the present invention. The ranking is performed using the Gini coefficient-based method in RF.

[0022] Figure 3 This is a feature importance comparison chart of all features after feature expansion in one embodiment of the present invention, which is sorted using the Gini coefficient-based method in RF. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to its embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and do not limit the scope of protection of the invention.

[0024] It should be noted that when an element is referred to as being "fixed to" another element, it can be directly attached to the other element or there may be an intervening element. When an element is referred to as being "connected to" another element, it can be directly connected to the other element or there may be an intervening element. The terms "vertical," "horizontal," "left," "right," and similar expressions used herein are for illustrative purposes only and do not represent the only possible implementation.

[0025] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0026] In this invention, "first" and "second" do not represent a specific quantity or order, but are merely used to distinguish names.

[0027] like Figure 1 As shown, an embodiment of the present invention provides a method for extending the tensile strength feature of aluminum alloys based on machine learning, which includes the following steps:

[0028] S1. Collect tensile strength data and select multiple tensile strength data as feature values ​​for machine learning to predict tensile strength.

[0029] Specifically, the tensile strength data comes from experimental data from Guangzhou Zhiyuan New Material Technology Co., Ltd., an aluminum alloy casting company. There are a total of 285 sets of tensile strength data. Nine features, including the content of metal elements such as Si, Fe, Cu, Mn, Mg, Zn, Pb, Cd and Sn, were selected as feature values ​​for machine learning to predict tensile strength. The tensile strength value is used as the target variable, i.e. the predicted value.

[0030] S2 performs unique thermal encoding on multiple selected tensile strength data.

[0031] Because the aluminum alloys in this dataset differ in their forming states, being divided into two forming methods: recasting and die casting, the forming method, as a categorical feature, is not continuous and cannot be directly used in machine learning model training.

[0032] Therefore, we use one-hot encoding for processing. One-hot encoding, also known as single-bit encoding, uses an N-bit state register to encode N states. Each state has its own independent register bit, and at any given time, only one bit is valid. We use the `OneHotEncoder` function from `sklearn.preprocessing` to perform one-hot encoding, encoding the recasting mode as 10 and the die-casting mode as 01, and add them to the feature combination.

[0033] S3 preprocesses multiple tensile strength data after unique thermal encoding.

[0034] Specifically, z-score normalization is used to preprocess multiple tensile strength data after one-hot encoding.

[0035] In machine learning algorithms, it's essential to ensure that features are not highly correlated. z-score standardization maps the original feature data to a distribution with a mean of 0 and a standard deviation of 1. The formula is: .

[0036] in, Representing data characteristics, Represents the mean of the feature data. It represents the standard deviation.

[0037] S4. Establish a machine learning model, train the established machine learning model using the dataset, and use the trained machine learning model to predict tensile strength.

[0038] S5 analyzes the relationships between feature importance to identify the feature variables that have the greatest impact on the prediction results of the machine learning model.

[0039] The dataset was trained using the RF (Random Forest) machine learning model from scikit-learn. The features_importance function under the algorithm was used to analyze the importance of features. The feature variable with the greatest impact on the model's prediction results was identified as Si, followed by Zn, and Sn with the least impact.

[0040] S6. Perform feature expansion to obtain extended features that are of greater importance.

[0041] Based on the obtained feature importance and the intrinsic relationship of material properties, we selected six alloying element features with the highest feature importance, namely Si, Fe, Cu, Mn, Mg, and Zn, for feature extension.

[0042] We added new features by adding existing features together, using the `EVAL` function in Python to expand the features, such as "Si+Fe", and so on. Ultimately, 15 features were added to the feature combination. Then, we trained the dataset using the Random Forest (RF) machine learning model from scikit-learn. Analyzing the importance of features using the `features_importance` function revealed that the combination of Si and Zn had the highest and most prominent importance, far exceeding the importance of any single feature. Therefore, we can conclude that the Si + Zn content has a significant impact on tensile strength.

[0043] Figure 2 This image shows a comparison of feature importance after training the dataset using the RandomForest (RF) machine learning model from scikit-learn. The algorithm's `features_importance` function was used to analyze feature importance, ranking them using the Gini coefficient-based method. A higher Gini coefficient indicates a more important feature. It can be seen that the feature variable with the greatest impact on the model's prediction results is Si, followed by Zn, with Sn having the least impact.

[0044] Figure 3The chart shows a comparison of the importance of features after feature expansion, analyzed using the features_importance function. It can be seen that the feature variable with the greatest impact on the model prediction results is Si_Zn, which is the sum of the contents of Si and Zn. Its feature importance is much greater than that of other features.

[0045] In other words, the machine learning-oriented aluminum alloy tensile strength feature extension method can be used to extend the feature value with the greatest influence from multiple tensile strength feature variables.

[0046] Among the features extended using the machine learning-oriented aluminum alloy tensile strength feature extension method, we found that one extended feature exhibits exceptional importance in the tensile strength model, far exceeding any of the original features. This is extremely helpful for materials science research. Feature extension not only avoids losing crucial feature information and increasing computation time, but also improves the accuracy of model predictions, enhances the interpretability of machine learning models, and facilitates researchers in further unraveling the "black box" of the relationship between materials and machine learning.

[0047] In one embodiment, the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the machine learning-oriented method for extending the tensile strength characteristics of aluminum alloys.

[0048] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0049] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A method for extending the tensile strength feature of aluminum alloys based on machine learning, characterized in that, The machine learning-oriented method for extending the tensile strength feature of aluminum alloys includes the following steps: Collect tensile strength data and select multiple tensile strength data as feature values ​​for machine learning to predict tensile strength; Perform unique thermal encoding on the selected tensile strength data; Preprocessing is performed on multiple tensile strength data after unique thermal encoding; Establish a machine learning model, train the established machine learning model using the dataset, and use the trained machine learning model to predict tensile strength. Analyze the relationships between feature importance to identify the feature variables that have the greatest impact on the prediction results of the machine learning model; Feature expansion is performed to obtain expanded features that are of greater importance; Specifically, the EVAL function is used to extend the features of the top six alloying elements in terms of feature importance, and new features are added by adding features together.

2. The method for extending the tensile strength feature of aluminum alloys based on machine learning as described in claim 1, characterized in that, Multiple tensile strength data include the contents of Si, Fe, Cu, Mn, Mg, Zn, Pb, Cd, and Sn metal elements, with the tensile strength value used as the target variable.

3. The method for extending the tensile strength feature of aluminum alloys based on machine learning as described in claim 2, characterized in that, The OneHotEncoder function is used to perform one-hot encoding on the selected tensile strength data. The recasting method is encoded as 10 and the die-casting method is encoded as 01 and added to the feature combination.

4. The method for extending the tensile strength feature of aluminum alloys based on machine learning as described in claim 3, characterized in that, z-score normalization was used to preprocess multiple tensile strength data after one-hot encoding.

5. The method for extending the tensile strength feature of aluminum alloys based on machine learning as described in claim 4, characterized in that, The dataset was trained using the RF machine learning model in scikit-learn, and the importance of features was analyzed using the features_importance function in the algorithm.

6. The method for extending the tensile strength feature of aluminum alloys for machine learning as described in claim 5, characterized in that, The extended features are added together with the original features to train the machine learning model, and the importance of features is analyzed using the features_importance function in the algorithm.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the machine learning-oriented method for extending the tensile strength characteristics of aluminum alloys as described in any one of claims 1-6.

Images