A method and system for optimizing electrolytic refining high-purity indium product qualification rate
By establishing a high-purity indium product quality prediction model based on statistical machine learning, the problem of unstable product qualification rate caused by the instability of crude indium raw materials was solved, realizing the stable and rapid optimization process of high-purity indium products, and improving production efficiency and product quality.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- YUNNAN TIN IND TIN MATERIAL CO LTD
- Filing Date
- 2022-11-14
- Publication Date
- 2026-07-21
Smart Images
Figure CN115831263B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of quality control and optimization of high-purity metal products, and in particular to a method and system for optimizing the yield of electrolytically refined high-purity indium products. Background Technology
[0002] High-purity indium refers to elemental indium with a purity of 99.999% or higher. It is soft, malleable, and ductile, making it an indispensable material in semiconductors, high-purity alloys, and thin-film solar cells. Currently, the main methods for preparing high-purity indium include electrolytic refining, vacuum distillation, and zone melting. Among these, electrolytic refining is a commonly used method to purify crude indium to a purity of 99.999% or higher. However, the instability of impurities in the crude indium raw material leads to instability in the yield of electrolytically refined products. Ensuring the stability of electrolytically refined products with crude indium raw materials of different compositions is a crucial issue. Summary of the Invention
[0003] The purpose of this invention is to provide a method and system for optimizing the yield of electrolytically refined high-purity indium products. This method utilizes statistical machine learning to optimize the yield of electrolytically refined high-purity indium products and ensure the stability of the electrolytically refined products.
[0004] To achieve the above objectives, the present invention provides the following solution:
[0005] A method for optimizing the yield of high-purity indium products refined by electrolysis includes:
[0006] Within a preset time period, the proportion of raw material components, electrolytic cell process parameters, and corresponding high-purity indium product quality results of the electrolytic refining process are collected; the high-purity indium product quality results are either qualified or unqualified.
[0007] Using the proportion of raw material components and the process parameters of the electrolytic cell as independent variables, and the test results of high-purity indium product quality as dependent variables, a statistical machine learning method was used to establish a high-purity indium product quality prediction model.
[0008] The raw material composition ratio and electrolytic cell process parameters to be predicted for high-purity indium electrolytic refining are input into the high-purity indium product quality prediction model, and the predicted high-purity indium product quality results are output.
[0009] When the predicted high-purity indium product quality result is unqualified, optimize the raw material composition ratio and / or electrolytic cell process parameters until the predicted high-purity indium product quality result is qualified.
[0010] Optionally, the raw material components include lead, copper, sulfur, and tin;
[0011] The proportions of the raw material components include the mass concentrations of lead, copper, sulfur, and tin.
[0012] The electrolytic cell process parameters include indium ion concentration, temperature, pH value, and electrolytic cell voltage.
[0013] Optionally, the step of establishing a high-purity indium product quality prediction model using statistical machine learning methods, with the proportion of raw material components and electrolytic cell process parameters as independent variables and the test results of high-purity indium product quality as the dependent variable, specifically includes:
[0014] A dataset was established with the proportion of raw material components and the process parameters of the electrolytic cell as independent variables, and the test results of the high-purity indium product quality as the dependent variable.
[0015] Multiple dimensionality reduction methods were used to reduce the dimensionality of the dataset, resulting in various dimensionality-reduced datasets.
[0016] Multiple machine learning classification algorithms are trained using each dimensionality-reduced dataset to obtain multiple prediction models corresponding to each dimensionality-reduced dataset.
[0017] Five-fold cross-validation was used to calculate the area under the ROC curve for each prediction model, and the prediction model with the largest ROC curve area was taken as the prediction model for the quality of high-purity indium products.
[0018] Optionally, various dimensionality reduction methods include principal component analysis, Fisher projection, and t-SNE dimensionality reduction.
[0019] The various machine learning classification algorithms mentioned include decision trees, support vector machines, logistic regression, and random forests.
[0020] A system for optimizing the yield of high-purity indium products obtained through electrolytic refining includes:
[0021] The sample data collection module is used to collect the proportion of raw material components, electrolytic cell process parameters, and corresponding high-purity indium product quality results during the electrolytic refining of high-purity indium within a preset time period; the high-purity indium product quality results are either qualified or unqualified.
[0022] The model building module is used to establish a high-purity indium product quality prediction model using statistical machine learning methods, with the proportion of raw material components and electrolytic cell process parameters as independent variables and the test results of high-purity indium product quality as dependent variables.
[0023] The prediction module is used to input the raw material composition ratio of the electrolytically refined high-purity indium and the electrolytic cell process parameters into the high-purity indium product quality prediction model, and output the predicted high-purity indium product quality results.
[0024] An optimization module is used to optimize the raw material composition ratio and / or electrolytic cell process parameters when the predicted high-purity indium product quality result is unqualified, until the predicted high-purity indium product quality result is qualified.
[0025] Optionally, the raw material components include lead, copper, sulfur, and tin;
[0026] The proportions of the raw material components include the mass concentrations of lead, copper, sulfur, and tin.
[0027] The electrolytic cell process parameters include indium ion concentration, temperature, pH value, and electrolytic cell voltage.
[0028] Optionally, the model building module specifically includes:
[0029] The dataset creation unit is used to create a dataset with the proportion of raw material components and electrolytic cell process parameters as independent variables and the test results of high-purity indium product quality as dependent variables.
[0030] The dimensionality reduction unit is used to reduce the dimensionality of the dataset using multiple dimensionality reduction methods to obtain multiple dimensionality-reduced datasets.
[0031] The training unit is used to train multiple machine learning classification algorithms using each dimensionality-reduced dataset to obtain multiple prediction models corresponding to each dimensionality-reduced dataset.
[0032] The optimization unit is used to calculate the area under the ROC curve for each prediction model using five-fold cross-validation, and the prediction model with the largest ROC curve area is used as the prediction model for high-purity indium product quality.
[0033] Optionally, various dimensionality reduction methods include principal component analysis, Fisher projection, and t-SNE dimensionality reduction.
[0034] The various machine learning classification algorithms mentioned include decision trees, support vector machines, logistic regression, and random forests.
[0035] An electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method described above for optimizing the yield of high-purity indium products obtained through electrolytic refining.
[0036] A computer-readable storage medium having a computer program stored thereon, which, when executed, implements the method described above for optimizing the yield of high-purity indium products obtained through electrolytic refining.
[0037] According to specific embodiments provided by the present invention, the present invention discloses the following technical effects:
[0038] This invention discloses a method and system for optimizing the yield rate of electrolytically refined high-purity indium products. It employs statistical machine learning to establish a high-purity indium product quality prediction model, which predicts whether the high-purity indium product is qualified. If the product is unqualified, the method optimizes the raw material composition ratio and / or electrolytic cell process parameters. This invention utilizes statistical machine learning to optimize the yield rate of electrolytically refined high-purity indium products, ensuring the stability of the electrolytically refined products. Attached Figure Description
[0039] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.
[0040] Figure 1 A flowchart of a method for optimizing the yield of high-purity indium products through electrolytic refining, provided in an embodiment of the present invention;
[0041] Figure 2 A schematic diagram illustrating the principle of the method for optimizing the yield of high-purity indium products through electrolytic refining, as provided in an embodiment of the present invention.
[0042] Figure 3 The diagram shows the dimensionality reduction results of principal component analysis provided in this embodiment of the invention. Detailed Implementation
[0043] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0044] The purpose of this invention is to provide a method and system for optimizing the yield of electrolytically refined high-purity indium products. This method utilizes statistical machine learning to optimize the yield of electrolytically refined high-purity indium products and ensure the stability of the electrolytically refined products.
[0045] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.
[0046] Statistical machine learning methods are data-driven statistical approaches that can fit statistical patterns to data without understanding the underlying logic, thus avoiding complex logical reasoning and mechanism verification. This invention employs three dimensionality reduction algorithms—Principal Component Analysis (PCA), Fisher projection, and t-SNE (t-distributed stochastic neighbor embedding)—as well as four machine learning classification algorithms: decision tree, support vector machine, logistic regression, and random forest.
[0047] To address the different compositions of raw materials and electrolytic cell process parameters, this invention, without altering existing equipment, interfering with current production, or requiring repeated small-scale and pilot-scale tests, uses statistical machine learning techniques to predict the qualification rate of electrolytically refined high-purity indium products. It also provides a method for optimizing the qualification rate of electrolytically refined high-purity indium products.
[0048] This invention provides a method for optimizing the yield of high-purity indium products obtained through electrolytic refining, such as... Figure 1 and Figure 2 As shown, the method includes the following steps:
[0049] Step S1: Collect the proportion of raw material components, electrolytic cell process parameters, and corresponding high-purity indium product quality results during the electrolytic refining of high-purity indium within a preset time period; the high-purity indium product quality results are either qualified or unqualified.
[0050] Over a period of time, the composition of the raw materials for the electrolysis equipment was determined to obtain the elemental composition percentage of non-indium impurities in the raw materials, and the corresponding electrolysis process parameters of the equipment, including the In electrolyte, were recorded. 3+ Parameters include indium ion concentration, temperature, pH value, and electrolytic cell voltage. The raw material composition includes the mass concentrations of lead, copper, sulfur, and tin.
[0051] The purity of high-purity indium products produced by the proportion of raw material components and the process parameters of the electrolytic cell is tested to determine whether the electrolytically refined products are qualified, and the quality results of the tested high-purity indium products are obtained.
[0052] Step S2: Using the proportion of raw material components and the process parameters of the electrolytic cell as independent variables, and the test results of high-purity indium product quality as the dependent variable, a statistical machine learning method is used to establish a high-purity indium product quality prediction model.
[0053] Using raw material composition and process parameters as independent variables, and product qualification as the dependent variable, a machine learning classification model was established. The dataset underwent two steps: dimensionality reduction and modeling. For dimensionality reduction, Principal Component Analysis (PCA), Fisher projection, or t-SNE were used. The dimensionality-reduced data was then input into four machine learning classification algorithms: decision tree, support vector machine, logistic regression, and random forest. Finally, the model with the highest area under the ROC curve (receiver operating characteristic curve) in 5-fold cross-validation on the training set was selected. The results of the principal component analysis dimensionality reduction are shown below. Figure 3 As shown, the horizontal axis represents the first principal component, and the vertical axis represents the second principal component.
[0054] Step S3: Input the raw material composition ratio and electrolytic cell process parameters of the high-purity indium to be predicted into the high-purity indium product quality prediction model, and output the predicted high-purity indium product quality results.
[0055] Step S4: When the predicted high-purity indium product quality result is unqualified, optimize the raw material composition ratio and / or electrolytic cell process parameters until the predicted high-purity indium product quality result is qualified.
[0056] Using a high-purity indium product quality prediction model, the system can predict whether the produced product is qualified based on the current production parameters and the composition of the raw materials. If it is qualified, production can proceed directly; if it is not qualified, the system can optimize the product qualification rate by adjusting the electrolytic cell process parameters or changing to another type of raw material.
[0057] This invention is based on reliable actual production data and modeling methods. The optimization process does not require additional experiments and does not affect normal industrial production. It helps to improve the product qualification rate while improving the level of digital production in enterprises.
[0058] Example 1: Data from 50 electrolytic refining processes are shown in Table 1.
[0059] Table 1. Raw material composition and operating parameters for products produced by electrolytic refining process
[0060]
[0061]
[0062]
[0063] Principal component analysis (PCA) was used to reduce the dimensionality of the dataset. The PCA dimensionality reduction results are shown in Table 2. Subsequently, decision tree, random forest, logistic regression, and support vector machine algorithms were used to build models. Five-fold cross-validation was used to evaluate the performance of the models, and the results are shown in Table 2. The model with the highest AUC (a combination of PCA and support vector machine) was selected to predict whether the product is qualified under certain raw material and process parameters for a specific component. The prediction accuracy of the model was 96%.
[0064] Table 2. Five-fold cross-validation of different machine learning algorithms
[0065]
[0066] Example 2: The composition of 10 new raw materials and current process parameters were determined and input into the established machine learning classification model to predict the products. The prediction results are shown in Table 3. All model predictions were accurate.
[0067] Table 3. Prediction Results of Products from Electrolytic Refining Process
[0068]
[0069] The present invention has the following advantages:
[0070] I. The method of optimizing the pass rate of high-purity indium products based on statistical machine learning is simple and quick: just input the raw material composition and the current electrolytic cell process parameters, and the prediction result can be obtained in seconds.
[0071] Second, the optimization process does not require additional testing and does not affect normal industrial production.
[0072] Third, only the equipment parameters are adjusted, without changing the equipment: This invention does not require any modification to the equipment during implementation, requiring little investment and yielding quick results.
[0073] This invention also provides a system for optimizing the yield of high-purity indium products obtained through electrolytic refining, comprising:
[0074] The sample data collection module is used to collect the proportion of raw material components, electrolytic cell process parameters, and corresponding high-purity indium product quality results during the electrolytic refining of high-purity indium within a preset time period; the high-purity indium product quality results are either qualified or unqualified.
[0075] The model building module is used to establish a high-purity indium product quality prediction model using statistical machine learning methods, with the proportion of raw material components and electrolytic cell process parameters as independent variables and the test results of high-purity indium product quality as dependent variables.
[0076] The prediction module is used to input the raw material composition ratio of the electrolytically refined high-purity indium and the electrolytic cell process parameters into the high-purity indium product quality prediction model, and output the predicted high-purity indium product quality results.
[0077] An optimization module is used to optimize the raw material composition ratio and / or electrolytic cell process parameters when the predicted high-purity indium product quality result is unqualified, until the predicted high-purity indium product quality result is qualified.
[0078] The raw material components include lead, copper, sulfur, and tin;
[0079] The proportions of the raw material components include the mass concentrations of lead, copper, sulfur, and tin.
[0080] The electrolytic cell process parameters include In 3+ Concentration, temperature, pH value, and electrolytic cell voltage.
[0081] The model building module specifically includes:
[0082] The dataset creation unit is used to create a dataset with the proportion of raw material components and electrolytic cell process parameters as independent variables and the test results of high-purity indium product quality as dependent variables.
[0083] The dimensionality reduction unit is used to reduce the dimensionality of the dataset using multiple dimensionality reduction methods to obtain multiple dimensionality-reduced datasets.
[0084] The training unit is used to train multiple machine learning classification algorithms using each dimensionality-reduced dataset to obtain multiple prediction models corresponding to each dimensionality-reduced dataset.
[0085] The optimization unit is used to calculate the area under the ROC curve for each prediction model using five-fold cross-validation, and the prediction model with the largest ROC curve area is used as the prediction model for high-purity indium product quality.
[0086] The various dimensionality reduction methods mentioned include principal component analysis, Fisher projection, and t-SNE dimensionality reduction.
[0087] The various machine learning classification algorithms mentioned include decision trees, support vector machines, logistic regression, and random forests.
[0088] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method described above for optimizing the yield of high-purity indium products obtained through electrolytic refining.
[0089] Furthermore, when the computer program in the aforementioned memory is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a 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, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory, random access memory, magnetic disks, or optical disks.
[0090] Furthermore, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed, implements the method described above for optimizing the yield of high-purity indium products obtained through electrolytic refining.
[0091] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple; relevant parts can be referred to the method section.
[0092] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.
Claims
1. A method for optimizing the yield of high-purity indium products obtained through electrolytic refining, characterized in that, include: Collect the proportion of raw material components, electrolytic cell process parameters, and corresponding high-purity indium product quality results during the electrolytic refining of high-purity indium within a preset time period. The quality result of the high-purity indium product is qualified or unqualified; the raw material composition ratio refers to the elemental composition ratio of non-indium impurity elements in the raw material; Using the proportion of raw material composition and electrolytic cell process parameters as independent variables, and the test results of high-purity indium product quality as dependent variables, a statistical machine learning method is used to establish a high-purity indium product quality prediction model; specifically, this includes: establishing a dataset using the proportion of raw material composition and electrolytic cell process parameters as independent variables, and the test results of high-purity indium product quality as dependent variables; Multiple dimensionality reduction methods were used to reduce the dimensionality of the dataset, resulting in various dimensionality-reduced datasets. Multiple machine learning classification algorithms were trained using each dimensionality-reduced dataset to obtain multiple prediction models corresponding to each dimensionality-reduced dataset. Five-fold cross-validation was used to calculate the area under the ROC curve for each prediction model, and the prediction model with the largest area under the ROC curve was used as the high-purity indium product quality prediction model. The raw material composition ratio and electrolytic cell process parameters to be predicted for high-purity indium electrolytic refining are input into the high-purity indium product quality prediction model, and the predicted high-purity indium product quality results are output. When the predicted high-purity indium product quality result is unqualified, optimize the raw material composition ratio and / or electrolytic cell process parameters until the predicted high-purity indium product quality result is qualified.
2. The method for optimizing the yield of high-purity indium products by electrolytic refining according to claim 1, characterized in that, The raw material components include lead, copper, sulfur, and tin; The proportions of the raw material components include the mass concentrations of lead, copper, sulfur, and tin. The electrolytic cell process parameters include indium ion concentration, temperature, pH value, and electrolytic cell voltage.
3. The method for optimizing the yield of high-purity indium products by electrolytic refining according to claim 1, characterized in that, The various dimensionality reduction methods mentioned include principal component analysis, Fisher projection, and t-SNE dimensionality reduction. The various machine learning classification algorithms mentioned include decision trees, support vector machines, logistic regression, and random forests.
4. A system for optimizing the yield of high-purity indium products obtained through electrolytic refining, characterized in that, The system is used to implement the method for optimizing the yield of high-purity indium products by electrolytic refining as described in any one of claims 1-3, the system comprising: The sample data collection module is used to collect the proportion of raw material components, electrolytic cell process parameters, and corresponding high-purity indium product quality results during the electrolytic refining of high-purity indium within a preset time period; the high-purity indium product quality results are either qualified or unqualified. The model building module is used to establish a high-purity indium product quality prediction model using statistical machine learning methods, with the proportion of raw material components and electrolytic cell process parameters as independent variables and the test results of high-purity indium product quality as dependent variables. The prediction module is used to input the raw material composition ratio of the electrolytically refined high-purity indium and the electrolytic cell process parameters into the high-purity indium product quality prediction model, and output the predicted high-purity indium product quality results. An optimization module is used to optimize the raw material composition ratio and / or electrolytic cell process parameters when the predicted high-purity indium product quality result is unqualified, until the predicted high-purity indium product quality result is qualified.
5. The system for optimizing the yield of high-purity indium products through electrolytic refining according to claim 4, characterized in that, The raw material components include lead, copper, sulfur, and tin; The proportions of the raw material components include the mass concentrations of lead, copper, sulfur, and tin. The electrolytic cell process parameters include indium ion concentration, temperature, pH value, and electrolytic cell voltage.
6. The system for optimizing the yield of high-purity indium products through electrolytic refining according to claim 4, characterized in that, The model building module specifically includes: The dataset creation unit is used to create a dataset with the proportion of raw material components and electrolytic cell process parameters as independent variables and the test results of high-purity indium product quality as dependent variables. The dimensionality reduction unit is used to reduce the dimensionality of the dataset using multiple dimensionality reduction methods to obtain multiple dimensionality-reduced datasets. The training unit is used to train multiple machine learning classification algorithms using each dimensionality-reduced dataset to obtain multiple prediction models corresponding to each dimensionality-reduced dataset. The optimization unit is used to calculate the area under the ROC curve for each prediction model using five-fold cross-validation, and the prediction model with the largest ROC curve area is used as the prediction model for high-purity indium product quality.
7. The system for optimizing the yield of high-purity indium products through electrolytic refining according to claim 6, characterized in that, The various dimensionality reduction methods mentioned include principal component analysis, Fisher projection, and t-SNE dimensionality reduction. The various machine learning classification algorithms mentioned include decision trees, support vector machines, logistic regression, and random forests.
8. An electronic device, characterized in that, The method includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement a method for optimizing the yield of high-purity indium products as described in any one of claims 1 to 3.
9. A computer-readable storage medium, characterized in that, It stores a computer program that, when executed, implements a method for optimizing the yield of electrolytically refined high-purity indium products as described in any one of claims 1 to 3.