CN114580271B - Method for realizing solid-liquid phase temperature prediction of multi-element noble metal alloy solder - Google Patents
Method for realizing solid-liquid phase temperature prediction of multi-element noble metal alloy solder Download PDFInfo
- Publication number
- CN114580271B CN114580271B CN202210139496.5A CN202210139496A CN114580271B CN 114580271 B CN114580271 B CN 114580271B CN 202210139496 A CN202210139496 A CN 202210139496A CN 114580271 B CN114580271 B CN 114580271B
- Authority
- CN
- China
- Prior art keywords
- phase temperature
- liquid phase
- machine learning
- solid
- noble metal
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 239000007791 liquid phase Substances 0.000 title claims abstract description 58
- 229910045601 alloy Inorganic materials 0.000 title claims abstract description 57
- 239000000956 alloy Substances 0.000 title claims abstract description 57
- 229910000510 noble metal Inorganic materials 0.000 title claims abstract description 34
- 238000000034 method Methods 0.000 title claims abstract description 33
- 229910000679 solder Inorganic materials 0.000 title claims abstract description 32
- 238000010801 machine learning Methods 0.000 claims abstract description 40
- 239000000126 substance Substances 0.000 claims abstract description 32
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 23
- 238000012216 screening Methods 0.000 claims abstract description 16
- 239000007790 solid phase Substances 0.000 claims abstract description 16
- 238000002360 preparation method Methods 0.000 claims abstract description 9
- 230000002068 genetic effect Effects 0.000 claims abstract description 6
- 230000000694 effects Effects 0.000 claims description 7
- 238000012360 testing method Methods 0.000 claims description 7
- 238000007637 random forest analysis Methods 0.000 claims description 6
- 101001095088 Homo sapiens Melanoma antigen preferentially expressed in tumors Proteins 0.000 claims description 5
- 102100037020 Melanoma antigen preferentially expressed in tumors Human genes 0.000 claims description 5
- 230000015572 biosynthetic process Effects 0.000 claims description 5
- 238000009529 body temperature measurement Methods 0.000 claims description 5
- 238000001914 filtration Methods 0.000 claims description 5
- 238000012417 linear regression Methods 0.000 claims description 5
- 238000003786 synthesis reaction Methods 0.000 claims description 5
- 230000008018 melting Effects 0.000 claims description 4
- 238000002844 melting Methods 0.000 claims description 4
- 238000012549 training Methods 0.000 claims description 4
- 238000004458 analytical method Methods 0.000 claims description 3
- 238000013528 artificial neural network Methods 0.000 claims description 3
- 238000000889 atomisation Methods 0.000 claims description 3
- 238000003066 decision tree Methods 0.000 claims description 3
- 230000008020 evaporation Effects 0.000 claims description 3
- 238000001704 evaporation Methods 0.000 claims description 3
- 230000007786 learning performance Effects 0.000 claims description 3
- 238000012706 support-vector machine Methods 0.000 claims description 3
- 238000011156 evaluation Methods 0.000 claims 1
- 238000005219 brazing Methods 0.000 abstract description 11
- 239000000945 filler Substances 0.000 abstract description 9
- 229910052751 metal Inorganic materials 0.000 abstract description 8
- 239000002184 metal Substances 0.000 abstract description 8
- 238000005266 casting Methods 0.000 description 9
- KDLHZDBZIXYQEI-UHFFFAOYSA-N Palladium Chemical compound [Pd] KDLHZDBZIXYQEI-UHFFFAOYSA-N 0.000 description 5
- 239000000463 material Substances 0.000 description 5
- 229910052759 nickel Inorganic materials 0.000 description 4
- BASFCYQUMIYNBI-UHFFFAOYSA-N platinum Chemical compound [Pt] BASFCYQUMIYNBI-UHFFFAOYSA-N 0.000 description 4
- 229910052709 silver Inorganic materials 0.000 description 4
- 239000000654 additive Substances 0.000 description 3
- 230000000996 additive effect Effects 0.000 description 3
- 238000010276 construction Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 3
- 239000011159 matrix material Substances 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 229910018651 Mn—Ni Inorganic materials 0.000 description 2
- 238000013461 design Methods 0.000 description 2
- 229910001325 element alloy Inorganic materials 0.000 description 2
- PCHJSUWPFVWCPO-UHFFFAOYSA-N gold Chemical compound [Au] PCHJSUWPFVWCPO-UHFFFAOYSA-N 0.000 description 2
- 239000010931 gold Substances 0.000 description 2
- 229910052737 gold Inorganic materials 0.000 description 2
- 229910052763 palladium Inorganic materials 0.000 description 2
- 229910052697 platinum Inorganic materials 0.000 description 2
- 229910052684 Cerium Inorganic materials 0.000 description 1
- 229910017518 Cu Zn Inorganic materials 0.000 description 1
- 238000012952 Resampling Methods 0.000 description 1
- BQCADISMDOOEFD-UHFFFAOYSA-N Silver Chemical compound [Ag] BQCADISMDOOEFD-UHFFFAOYSA-N 0.000 description 1
- 229910007610 Zn—Sn Inorganic materials 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- FFBHFFJDDLITSX-UHFFFAOYSA-N benzyl N-[2-hydroxy-4-(3-oxomorpholin-4-yl)phenyl]carbamate Chemical compound OC1=C(NC(=O)OCC2=CC=CC=C2)C=CC(=C1)N1CCOCC1=O FFBHFFJDDLITSX-UHFFFAOYSA-N 0.000 description 1
- 229910052802 copper Inorganic materials 0.000 description 1
- 238000002790 cross-validation Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 229910001092 metal group alloy Inorganic materials 0.000 description 1
- 238000004377 microelectronic Methods 0.000 description 1
- 238000004806 packaging method and process Methods 0.000 description 1
- 238000010587 phase diagram Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 239000004332 silver Substances 0.000 description 1
- 229910052718 tin Inorganic materials 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Business, Economics & Management (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Physics & Mathematics (AREA)
- Biophysics (AREA)
- Evolutionary Computation (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Economics (AREA)
- Human Resources & Organizations (AREA)
- Strategic Management (AREA)
- Evolutionary Biology (AREA)
- Software Systems (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Computational Linguistics (AREA)
- Development Economics (AREA)
- Computing Systems (AREA)
- Game Theory and Decision Science (AREA)
- Genetics & Genomics (AREA)
- Physiology (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Investigating Or Analyzing Materials Using Thermal Means (AREA)
Abstract
The invention relates to a method for predicting solid-liquid phase temperature of a multi-element noble metal alloy solder, which comprises the following steps: searching the chemical formula, the preparation process and the solid phase temperature and the liquid phase temperature values of the noble metal alloy brazing filler metal from the literature to be used as a data set sample; firstly, constructing a physical and chemical parameter set, and then constructing a feature set according to the collected alloy chemical formula to replace direct input of the chemical formula; the feature set is primarily screened through correlation screening, and then a genetic algorithm is adopted to screen the feature set after the primary screening and different machine learning algorithms, so as to find the optimal combination of key features and the machine learning algorithms; establishing a machine learning model based on the screening result to predict performance; in addition, an active learning method is adopted to iteratively improve the established machine learning model. The invention can realize the prediction of the solid-liquid phase temperature of the multi-element noble metal alloy solder.
Description
Technical Field
The invention relates to the field of noble metal alloy brazing filler metal, in particular to a method for realizing solid-liquid phase temperature prediction of a multi-element noble metal alloy brazing filler metal.
Technical Field
Noble metal solders are important in the fields of electronics industry, microelectronic packaging, vacuum multi-stage brazing, high temperature technology, ornament manufacturing industry, aerospace and the like. There are thousands of noble metal solders in the solder series of various countries. The alloy is mainly composed of silver-based brazing filler metal, gold-based brazing filler metal, palladium-based brazing filler metal and partial platinum-based brazing filler metal according to alloy components. The silver solder has the most wide application and is mainly used for medium and low temperature brazing. The solders with certain special properties at low and high temperatures are most gold-based, palladium-based and partially platinum-based solders. At present, the components of the noble metal solder alloy mainly adopt binary and ternary components, and the research on the multi-component (such as quaternary, penta and even more) noble metal solder alloy is less. More importantly, the solid-liquid phase temperature (solid phase temperature and liquid phase temperature value) of binary and ternary noble metal alloy brazing filler metal can be obtained from a traditional phase diagram, the solid-liquid phase temperature of non-noble metal multi-element alloy can be predicted approximately through melting point calculation software, and the solid-liquid phase temperature prediction of related multi-element noble metal alloy cannot be realized due to the lack of basic data of noble metal thermodynamics, so that guidance is lacked in designing the noble metal brazing filler metal alloy with ideal solid-liquid phase temperature. Therefore, there is an urgent need to propose a method for realizing the solid-liquid phase temperature prediction of the multi-element noble metal alloy solder.
With its ability to automatically address complex tasks, machine learning is being used as a completely new approach to help discover material relatedness, to learn about the nature of materials and to speed up the discovery of materials. During machine learning, the material characteristics and models determine the accuracy of machine learning predictions, and therefore, for a given problem, an optimal combination of characteristics sets and models should be determined. The number of the machine learning features and the number of the models are large, and the performance of the machine learning model is actually estimated through test errors by adopting resampling methods such as cross-validation, bootstrap and the like. This exhaustive search of all possibilities ensures global optimization, but creates a significant challenge for computing resources. Therefore, a reasonable framework for selecting the most suitable combination of feature sets and machine learning models needs to be proposed to optimize such problems, and such problems have been a problem in the field of predicting material performance by machine technology learning.
Disclosure of Invention
The invention aims to overcome the defects that the solid-liquid phase temperature of the noble metal alloy solder is high in experimental test cost and low in experimental efficiency, and the conventional thermodynamic data is lack to cause difficulty in predicting quaternary or even other multi-element alloys, and provides a simple, convenient, quick, low-cost and labor-saving method for realizing solid-liquid phase temperature prediction of the noble metal alloy solder based on a machine learning technology. In addition, the invention provides a reasonable framework for selecting the most suitable combination of the feature group and the machine learning model based on the genetic algorithm.
The aim of the invention is achieved by the following technical scheme:
A method for realizing solid-liquid phase temperature prediction of a multi-element noble metal alloy solder comprises the following steps:
1) Searching chemical formula, preparation process, solid phase temperature and liquid phase temperature values of the noble metal solder from the channels of literature, public databases, self databases and the like, and taking the chemical formula, the preparation process, the solid phase temperature and the liquid phase temperature values as data set samples;
2) Randomly dividing a data set into a training set (80-90%) and a testing set (10-20%), wherein a chemical formula and a preparation process are used as inputs, and the solid phase temperature and the liquid phase temperature are prediction target values of a machine learning model;
3) Constructing a physicochemical parameter set, wherein the physicochemical parameters comprise a plurality of (such as k) basic physicochemical parameters of elements including element serial number, relative atomic mass, lattice constant, melting point, atomization enthalpy, evaporation enthalpy, vacancy enthalpy change and the like; in addition, a feature set for evaluating the influence degree of each parameter on the target amount is constructed according to the chemical proportion of the collected alloy chemical formula, and the direct input of the chemical formula is replaced. The construction process of the target feature set for evaluating the influence degree of each physical and chemical parameter is as follows: the k mean factor f mi features for each alloy were calculated using equation (1) and the k variance factor f vi features for each alloy were calculated using equation (2). Taking f mi and f vi as inputs of a machine learning performance prediction model;
fmi＝∑(fi×ci)/∑ci (1)
fvi＝∑[(fi-fmi)2×ci]/∑ci (2)
In the above formula, f i represents the physicochemical parameter of the ith element, i represents the element number of the alloy, where i=1, 2, …, n, n represents the number of the elements of the alloy, and c i represents the content of the ith element.
4) And (3) primarily screening the features by linear correlation filtering, and classifying the features with strong linear correlation in the feature set into the same group by taking the Pearson correlation coefficient larger than 0.9 as strong linear correlation. After grouping, the features within each group are strongly linear, and the features between each group are not strongly linear. In the process of linear correlation screening, a random forest algorithm is selected as a machine learning model, and in addition, the effects of comprehensive analysis modeling of parameters such as a linear regression correlation coefficient R, an average absolute percentage error MAPE, a root mean square error RMSE and the like are adopted.
5) And then, adopting a genetic algorithm to screen the initially screened feature group and different machine learning algorithms (support vector machines, neural networks, naive Bayes, random forests, decision trees and the like), and searching the optimal combination of key features and the machine learning algorithms. The modeling effect is comprehensively analyzed by adopting parameters such as a linear regression correlation coefficient R, an average absolute percentage error MAPE, a root mean square error RMSE and the like;
6) Based on the optimal combination of the screened characteristics and the machine learning algorithm, the characteristic quantity related to the target component is taken as input, and the solid-liquid phase temperature prediction of the unknown alloy is carried out. Through model parameter optimization, the degree of model overfitting is low, and the model prediction error is less than 10%;
7) And (3) adopting an active learning method, selecting a plurality of components with the largest uncertainty of a predicted result, performing experimental synthesis and solid-liquid phase temperature measurement, expanding the components to an initial data set, iteratively improving a machine learning model, and obtaining a predicted value of the solid-liquid phase temperature based on a final improved model.
Compared with the prior art, the invention has the following beneficial effects:
1. the method has higher accuracy on the prediction of the solid-liquid phase temperature of the multi-element noble metal alloy solder, and can be popularized to the prediction of the solid-liquid phase temperature of other non-noble metal alloy solders, and has universality;
2. The method can effectively save the experimental cost of solid-liquid phase temperature experimental measurement of the noble metal alloy solder, and finally, the noble metal alloy solder with ideal solid-liquid phase temperature is designed under the guidance of high efficiency and low cost.
Drawings
FIG. 1 is a basic flow chart of the design method of the present invention.
Detailed Description
The invention will now be described in detail with reference to the drawings and a preferred embodiment.
Example 1: solid-liquid phase temperature prediction of Ag-Cu-Zn-Mn-Ni series alloy.
The method comprises the following specific steps:
1) The design concept and flow of the embodiment are shown in fig. 1, and the chemical formula, the preparation process, the solid phase temperature and the liquid phase temperature value of the noble metal solder are searched from the channels of literature, public databases, self databases and the like, and are taken as data set samples. The chemical formula, solid phase temperature and liquid phase temperature experimental values of the partial noble metal alloy solder are shown in table 1:
Experimental values of chemical formula, solid phase temperature and liquid phase temperature of the partial noble metal alloy solder collected in table 1
| No | Ag | Cu | Zn | Mn | Ni | Si | B | P | Au | Pd | Sn | Solid phase temperature | Liquid phase temperature | Preparation process |
| 1 | 45 | 33.5 | 20 | 0 | 1.5 | 0 | 0 | 0 | 0 | 0 | 0 | 666 | 746 | Casting |
| 2 | 45 | 33 | 20 | 0 | 2 | 0 | 0 | 0 | 0 | 0 | 0 | 667 | 747 | Casting |
| 3 | 45 | 32 | 20 | 0 | 3 | 0 | 0 | 0 | 0 | 0 | 0 | 672 | 753 | Casting |
| 4 | 49.25 | 29.55 | 19.7 | 0 | 1.5 | 0 | 0 | 0 | 0 | 0 | 0 | 680.1 | 750.1 | Casting |
| 5 | 0 | 93.5 | 0 | 0 | 0 | 0 | 0 | 6.5 | 0 | 0 | 0 | 690 | 745 | Casting |
| 6 | 0 | 0 | 0 | 0 | 92 | 5 | 3 | 0 | 0 | 0 | 0 | 980 | 1040 | Casting |
| 7 | 0 | 4.5 | 0 | 22.5 | 66 | 7 | 0 | 0 | 0 | 0 | 0 | 980 | 1010 | Casting |
| 8 | 34 | 36.5 | 27 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 2.5 | 630 | 730 | Casting |
| 9 | 0 | 0 | 0 | 0 | 36 | 0 | 0 | 0 | 30 | 34 | 0 | 1135 | 1169 | Casting |
| … | … | … | … | … | … | … | … | … | … | … | … | … | … | … |
2) Randomly dividing a data set into a training set (80-90%) and a testing set (10-20%), wherein chemical formulas are used as input, and the solid phase temperature and the liquid phase temperature are prediction target values of a machine learning model;
3) The basic physicochemical parameters of the element concerned are extracted as a physical and chemical parameter set for construction and screening, the basic physicochemical parameters comprise a plurality of (such as k) basic physicochemical parameters of the element including element serial number, relative atomic mass, lattice constant, melting point, atomization enthalpy, evaporation enthalpy, vacancy enthalpy change and the like, and part of the basic physicochemical parameters are shown in table 2. In addition, a feature set for evaluating the influence degree of each parameter on the target amount is constructed according to the chemical proportion of the collected alloy chemical formula, and the direct input of the chemical formula is replaced. The construction process of the target feature set for evaluating the influence degree of each physical and chemical parameter is as follows: the k mean factor f mi features for each alloy were calculated using equation (1) and the k variance factor f vi features for each alloy were calculated using equation (2). Taking f mi and f vi as inputs of a machine learning performance prediction model;
fmi＝∑(fi×ci)/∑ci (1)
fvi＝∑[(fi-fmi)2×ci]/∑ci (2)
In the above formula, f i represents the physicochemical parameter of the ith element, i represents the element number of the alloy, where i=1, 2, …, n, n represents the number of the elements of the alloy, and c i represents the content of the ith element.
TABLE 2 basic physicochemical parameters
For example 1, specific usage of the above formula (1) and formula (2) is as follows: each time a physicochemical parameter is brought into the formula (1) and the formula (2), the mean factor f mi characteristic and the variance factor f vi characteristic of a corresponding physicochemical parameter are respectively obtained, and after K physicochemical parameters are sequentially brought into the formula, 2K characteristic quantity sets are constructed for subsequent characteristic screening. Taking the first data 45Ag-33.5Cu-20Zn-1.5Ni in Table 1 and the corresponding 1 st physicochemical parameter element number in Table 2 as an example, the mean factor f mi characteristic and the variance factor f vi characteristic of other physicochemical parameters for the first data 45Ag-33.5Cu-20Zn-1.5Ni are described by analogy:
For equation (1), n represents the number of components of the alloy, where the first data in table 1 is 45Ag-33.5Cu-20Zn-1.5Ni is a 4-component alloy, so where n=4, i represents the element number of the alloy, i=1, 2,3,4, respectively corresponding to the elements Ag, cu, zn and Ni in sequence, c i represents the content of the i-th element, and for 45Ag-33.5Cu-20Zn-1.5Ni, respectively corresponding to the contents of Ag, cu, ni and Ce 4 elements are 45, 33.5, 20,1.5; in addition, f i represents the physicochemical parameter of the ith element, and for the 1 st physicochemical parameter element number of 45Ag-33.5Cu-20Zn-1.5Ni, the element number values corresponding to the respective elements Ag, cu, zn and Ni 4 are (as can be found from the values in Table 2): 47,29,30 and 28.
Therefore, the mean factor characteristic of the 1 st physical and chemical parameter element sequence number is obtained by the formula (1) fmi＝(47*45)/(45+33.5+20+1.5)+(29*33.5)/(45+33.5+20+1.5)+(30*20)/(45+33.5+20+1.5)+(28*1.5)/(45+33.5+20+1.5)＝37.285;
The variance factor characteristic of the 1 st physical and chemical parameter element sequence number is obtained by the formula (2) fvi＝[(47-37.285)2*45]/(45+33.5+20+1.5)+[(29-37.285)2*33.5]/(45+33.5+20+1.5)+[(30-37.285)2*20]/(45+33.5+20+1.5)+[(28-37.285)2*1.5]/(45+33.5+20+1.5)
Similarly, the mean factor f mi characteristic and the variance factor f vi characteristic of other physicochemical parameters of the first data can be obtained sequentially. Similarly, 2k number of feature sets of other respective data can be obtained.
4) And (3) primarily screening the features by linear correlation filtering, and classifying the features with strong linear correlation in the feature set into the same group by taking the Pearson correlation coefficient larger than 0.9 as strong linear correlation. After grouping, the features within each group are strongly linear, and the features between each group are not strongly linear. In the process of linear correlation screening, a random forest algorithm is selected as a machine learning model, and in addition, the effects of comprehensive analysis modeling of parameters such as a linear regression correlation coefficient R, an average absolute percentage error MAPE, a root mean square error RMSE and the like are adopted.
5) And then, adopting a genetic algorithm to screen the initially screened feature group and different machine learning algorithms (support vector machines, neural networks, naive Bayes, random forests, decision trees and the like), and searching the optimal combination of key features and the machine learning algorithms. The modeling effect is comprehensively analyzed by adopting parameters such as a linear regression correlation coefficient R, an average absolute percentage error MAPE, a root mean square error RMSE and the like;
6) Based on the optimal combination of the screened characteristics and the machine learning algorithm, the test set data is low in degree of fitting through model parameter optimization, and the model prediction error is less than 10%;
7) Since the data collected is from the same process, the effect of process conditions is not considered here. Based on the established model, taking Ag as a matrix element, and Cu, zn, mn and Ni 4 elements as additive elements, and selecting the search range (mass fraction) of each element as Cu: 10-50%, zn: 0-10%, mn: 0-10%, ni: 0-10%, and each element establishes a component space at a spacing of 2% (mass fraction) for solid-liquid phase temperature prediction.
8) And adopting an active learning method, selecting a plurality of components with the largest uncertainty of the prediction result, performing experimental synthesis and solid-liquid phase temperature measurement, expanding the components to an initial data set, and iteratively improving a machine learning model. Based on the final improved model, solid-liquid phase temperature prediction is carried out on the Ag-16Cu-23Zn-7.5Mn-4.5Ni alloy, and the predicted value is as follows: solid phase temperature: 661 ℃, liquid phase temperature: 682 ℃. Experiments prove that the errors of the predicted value and the experimental value of the solid-liquid phase temperature are less than 7 percent. Therefore, the method has better effect in the aspect of realizing the solid-liquid phase temperature prediction of the noble metal alloy solder.
Example 2: solid-liquid phase temperature prediction of Ag-Cu-Zn-Sn series alloy.
This example is essentially identical to steps 1) -6) of example 1 above, with the exception of steps 7) and 8):
Ag is used as a matrix element, cu, zn and Sn 3 elements are used as additive elements, and the search range (mass fraction) of each element is selected as Cu: 10-50%, zn: 0-10%, sn: 0-10%, and each element establishes a component space at a spacing of 1% (mass fraction) for solid-liquid phase temperature prediction.
And adopting an active learning method, selecting a plurality of components with the largest uncertainty of the prediction result, performing experimental synthesis and solid-liquid phase temperature measurement, expanding the components to an initial data set, and iteratively improving a machine learning model. Based on the final improved model, solid-liquid phase temperature prediction is carried out on the Ag-27.5Cu-25Zn-2.5Sn alloy, and the predicted value is as follows: solid phase temperature: 620 ℃, liquid phase temperature: 654 ℃. Experiments prove that the errors of the predicted value and the experimental value of the solid-liquid phase temperature are less than 7 percent.
Example 3: and predicting solid-liquid phase temperature of Pd-Cu-Mn-Ni series alloy.
This example is essentially identical to steps 1) -6) of example 1 above, with the exception of steps 7) and 8):
Pd is used as a matrix element, cu, mn and Ni 3 elements are used as additive elements, and the search range (mass fraction) of each element is selected as Cu: 10-50%, mn: 0-20%, ni: 0-20%, and each element establishes a component space at a spacing of 2% (mass fraction) for solid-liquid phase temperature prediction.
And adopting an active learning method, selecting a plurality of components with the largest uncertainty of the prediction result, performing experimental synthesis and solid-liquid phase temperature measurement, expanding the components to an initial data set, and iteratively improving a machine learning model. Based on the final improved model, solid-liquid phase temperature prediction is carried out on Pd-55Cu-10Mn-15Ni alloy, and the predicted value is as follows: solid phase temperature: 1136 ℃, liquid phase temperature: 1158 ℃. Experiments prove that the errors of the predicted value and the experimental value of the solid-liquid phase temperature are less than 7 percent.
In summary, the above embodiment is a method for implementing solid-liquid phase temperature prediction of a multi-element noble metal alloy solder, firstly, searching chemical formula, preparation process, solid phase temperature and liquid phase temperature values of the noble metal alloy solder from literature as a data set sample; firstly, constructing a physical and chemical parameter set, and then constructing a feature set according to the collected alloy chemical formula to replace direct input of the chemical formula; the feature set is primarily screened through correlation screening, and then a genetic algorithm is adopted to screen the feature set after the primary screening and different machine learning algorithms, so as to find the optimal combination of key features and the machine learning algorithms; establishing a machine learning model based on the screening result to predict performance; in addition, an active learning method is adopted to iteratively improve the established machine learning model. The invention can realize the prediction of the solid-liquid phase temperature of the quaternary to polynary noble metal alloy solder.
The above embodiments are only examples of the present invention, and the above embodiments may be changed without departing from the scope of the present invention, and are not to be considered as illustrative, and not limiting the scope of the present invention.
Claims (6)
1. The method for realizing solid-liquid phase temperature prediction of the multi-element noble metal alloy solder is characterized by comprising the following steps of:
1) Searching the chemical formula, the preparation process, the solid phase temperature and the liquid phase temperature value of the multi-element noble metal alloy solder, and taking the values as data set samples;
2) Randomly dividing a data set into a training set and a testing set, wherein a chemical formula and a preparation process are used as inputs, and a solid phase temperature and a liquid phase temperature are used as prediction target values of a machine learning model;
3) Constructing a physical and chemical parameter set, and constructing a feature set for evaluating the influence degree of each parameter on the target quantity according to the chemical proportion of the chemical formula of the collected alloy solder to replace the direct input of the chemical formula;
4) Primary screening of the features by linear correlation filtering, wherein the features with strong linear correlation in the feature set are classified into the same group by taking the Pearson correlation coefficient larger than 0.9 as strong linear correlation;
5) Screening the initially screened feature group and different machine learning algorithms by adopting a genetic algorithm, and searching an optimal combination of key features and the machine learning algorithm;
6) Based on the screened key characteristics and the optimal combination of the machine learning algorithm, taking the characteristic quantity related to the target component as input, and carrying out solid-liquid phase temperature prediction of the unknown alloy solder;
7) Adopting an active learning method, selecting a plurality of components with the largest uncertainty of a predicted result, performing experimental synthesis and solid-liquid phase temperature measurement, expanding the components to an initial data set, iteratively improving a machine learning model, and obtaining the predicted value of the solid-liquid phase temperature based on the final improved model
Calculating the characteristic quantity of a mean value factor f mi of each basic physicochemical parameter of each alloy by using the formula (1), and calculating the characteristic quantity of a variance factor f vi of each basic physicochemical parameter of each alloy by using the formula (2); taking f mi and f vi as inputs of a machine learning performance prediction model;
fmi ＝ ∑ ( fi × ci )/ ∑ ci (1)
fvi ＝ ∑ [( fi - f mi ) 2 × ci ]/ ∑ ci (2)
Wherein f i represents the physicochemical parameter of the i-th element (i=1, 2, … n), n represents the number of components of the alloy, and c i represents the content of the i-th element.
2. The method of predicting according to claim 1, wherein the basic physicochemical parameters in the step 2) include 7 physicochemical parameters including an element number, a relative atomic mass, a lattice constant, a melting point, an atomization enthalpy, an evaporation enthalpy, and an enthalpy change of a vacancy.
3. The method of claim 1, wherein in step 4), a random forest algorithm is selected as the machine learning model during the primary filtering of the features by linear correlation filtering.
4. The method of predicting according to claim 1, wherein the machine learning algorithm in step 5) includes a support vector machine, a neural network, naive bayes, random forests, and decision trees.
5. The method according to claim 1, wherein the effects of the linear regression correlation coefficient R, the mean absolute percentage error MAPE and the root mean square error RMSE parameter analysis modeling are respectively used in the evaluation of the alloy feature screening, the model screening and the final prediction model in the steps 4), 5) and 6).
6. The method of claim 1 to 5, wherein in step 3), the training set is 80-90% and the test set is 10-20%.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210139496.5A CN114580271B (en) | 2022-02-16 | 2022-02-16 | Method for realizing solid-liquid phase temperature prediction of multi-element noble metal alloy solder |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210139496.5A CN114580271B (en) | 2022-02-16 | 2022-02-16 | Method for realizing solid-liquid phase temperature prediction of multi-element noble metal alloy solder |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114580271A CN114580271A (en) | 2022-06-03 |
| CN114580271B true CN114580271B (en) | 2024-07-02 |
Family
ID=81770694
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210139496.5A Active CN114580271B (en) | 2022-02-16 | 2022-02-16 | Method for realizing solid-liquid phase temperature prediction of multi-element noble metal alloy solder |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114580271B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115964787B (en) * | 2022-12-26 | 2024-06-14 | 北京建筑大学 | Phase redistribution-based method for extracting and characterizing initial geometric defects of lasso-type spinal rod |
| CN116580791B (en) * | 2023-04-28 | 2024-06-18 | 贵研铂业股份有限公司 | Method for simultaneously optimizing wettability and braze joint strength of alloy solder |
| CN116720058A (en) * | 2023-04-28 | 2023-09-08 | 贵研铂业股份有限公司 | Method for realizing key feature combination screening of machine learning candidate features |
| CN118430701A (en) * | 2024-07-03 | 2024-08-02 | 吉林大学 | Selective laser melting NiTi alloy phase transition temperature prediction method based on machine learning |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107122502A (en) * | 2016-02-24 | 2017-09-01 | 中南大学 | A kind of method of optimized alloy extrusion process |
| AU2020101874A4 (en) * | 2020-08-18 | 2020-09-24 | Hefei General Machinery Research Institute Co., Ltd. | A method for predicting high-temperature mechanical properties of heat-resistant alloys based on deep learning |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112132182B (en) * | 2020-08-20 | 2024-03-22 | 上海大学 | Method for rapidly predicting resistivity of ternary gold alloy based on machine learning |
-
2022
- 2022-02-16 CN CN202210139496.5A patent/CN114580271B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107122502A (en) * | 2016-02-24 | 2017-09-01 | 中南大学 | A kind of method of optimized alloy extrusion process |
| AU2020101874A4 (en) * | 2020-08-18 | 2020-09-24 | Hefei General Machinery Research Institute Co., Ltd. | A method for predicting high-temperature mechanical properties of heat-resistant alloys based on deep learning |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114580271A (en) | 2022-06-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114580271B (en) | Method for realizing solid-liquid phase temperature prediction of multi-element noble metal alloy solder | |
| Yang et al. | A machine learning-based alloy design system to facilitate the rational design of high entropy alloys with enhanced hardness | |
| Risum et al. | Using deep learning to evaluate peaks in chromatographic data | |
| CN108228716B (en) | SMOTE _ Bagging integrated sewage treatment fault diagnosis method based on weighted extreme learning machine | |
| Johannesson et al. | Combined electronic structure and evolutionary search approach to materials design | |
| Stumpf et al. | Sampling properties of random graphs: the degree distribution | |
| CN112382352A (en) | Method for quickly evaluating structural characteristics of metal organic framework material based on machine learning | |
| Frary et al. | Connectivity and percolation behaviour of grain boundary networks in three dimensions | |
| Helmus et al. | Phylogenetic diversity–area curves | |
| Palla et al. | Statistical mechanics of topological phase transitions in networks | |
| CN116720058A (en) | Method for realizing key feature combination screening of machine learning candidate features | |
| Balis et al. | Towards an operational database for real-time environmental monitoring and early warning systems | |
| Tewari et al. | Machine learning approaches to identify and design low thermal conductivity oxides for thermoelectric applications | |
| Pollet et al. | Optimal monte carlo updating | |
| CN114580272A (en) | Design method for simultaneously optimizing conductivity and hardness of multi-element electric contact alloy | |
| Tetsassi Feugmo et al. | Neural evolution structure generation: High entropy alloys | |
| Serra et al. | Soft computing techniques applied to combinatorial catalysis: A new approach for the discovery and optimization of catalytic materials | |
| CN115795225B (en) | Screening method and device for near infrared spectrum correction set | |
| CN117275599A (en) | High-temperature alloy component and process design method | |
| Juhász et al. | Partially asymmetric zero-range process with quenched disorder | |
| Diaz-Ortiz et al. | Cluster expansions in multicomponent systems: precise expansions from noisy databases | |
| CN115274023A (en) | Method and device for predicting optimal performance of alloy and electronic equipment | |
| Yuan et al. | Temporal Web Service QoS Prediction via Kalman Filter-Incorporated Latent Factor Analysis | |
| CN112132182A (en) | Method for rapidly predicting resistivity of ternary gold alloy based on machine learning | |
| CN114564884B (en) | Design method for simultaneously optimizing multiple electric contact performances of electric contact material |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant |