CN114373523B - Glass hardness prediction method based on squirrel optimization algorithm and machine learning algorithm - Google Patents

Glass hardness prediction method based on squirrel optimization algorithm and machine learning algorithm Download PDF

Info

Publication number
CN114373523B
CN114373523B CN202210281841.9A CN202210281841A CN114373523B CN 114373523 B CN114373523 B CN 114373523B CN 202210281841 A CN202210281841 A CN 202210281841A CN 114373523 B CN114373523 B CN 114373523B
Authority
CN
China
Prior art keywords
squirrel
glass
tree
algorithm
parameters
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
Application number
CN202210281841.9A
Other languages
Chinese (zh)
Other versions
CN114373523A (en
Inventor
杨勇
翟华
韩江
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Hefei University of Technology
Original Assignee
Hefei University of Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Hefei University of Technology filed Critical Hefei University of Technology
Priority to CN202210281841.9A priority Critical patent/CN114373523B/en
Publication of CN114373523A publication Critical patent/CN114373523A/en
Application granted granted Critical
Publication of CN114373523B publication Critical patent/CN114373523B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
    • G16C60/00—Computational materials science, i.e. ICT specially adapted for investigating the physical or chemical properties of materials or phenomena associated with their design, synthesis, processing, characterisation or utilisation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G06N20/20—Ensemble learning
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/004—Artificial life, i.e. computing arrangements simulating life
    • G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
    • G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
    • G16C20/30—Prediction of properties of chemical compounds, compositions or mixtures
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
    • G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
    • G16C20/70—Machine learning, data mining or chemometrics
    • 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)
  • Theoretical Computer Science (AREA)
  • Computing Systems (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Software Systems (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Artificial Intelligence (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Physics & Mathematics (AREA)
  • Medical Informatics (AREA)
  • General Health & Medical Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • Crystallography & Structural Chemistry (AREA)
  • Health & Medical Sciences (AREA)
  • Chemical & Material Sciences (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Computational Linguistics (AREA)
  • Biophysics (AREA)
  • Databases & Information Systems (AREA)
  • Biomedical Technology (AREA)
  • Molecular Biology (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses a glass hardness prediction method based on a squirrel optimization algorithm and a machine learning algorithm, and relates to the field of electric digital data processing; the method comprises the following steps: constructing a hardness database of glass materials with different components; constructing a feature descriptor; providing a training set sample, and initializing parameters of a squirrel search algorithm; setting optimized parameters of a squirrel search algorithm, optimizing parameters of a Catboost algorithm by using the optimized parameters to obtain optimal parameters of the Catboost algorithm, and establishing a glass hardness prediction model; test sample data is input, and glass hardness is predicted. The method constructs a unique descriptor, uses the squirrel search optimization algorithm for optimizing the parameters of the Catboost algorithm, has a simple structure, improves the convergence speed and precision, can obviously improve the performance of the Catboost algorithm by the optimized optimal Catboost algorithm parameters, and has practical significance for improving the accuracy of predicting the glass hardness.

Description

Glass hardness prediction method based on squirrel optimization algorithm and machine learning algorithm
Technical Field
The invention relates to the field of electric digital data processing, in particular to a glass hardness prediction method based on a squirrel optimization algorithm and a machine learning algorithm.
Background
Glass is an unbalanced, amorphous material that spontaneously relaxes to a supercooled liquid state. Unlike crystals, glasses do not need to meet strict stoichiometric rules and can be considered as continuous solutions of chemical elements. A large number of elements may become components constituting the glass material. Variation of 80 chemical elements in an amount of 1 mol% can result
Figure 965089DEST_PATH_IMAGE001
A possible glass component. However, the only reported amounts of inorganic glasses are
Figure 342981DEST_PATH_IMAGE002
On the other hand, this means that there is a great room for exploring new glass-forming ingredients having specific properties.
Hardness, as one of the most important mechanical properties, reflects the behaviour of the material against permanent deformation. The hardness of a glass material depends not only on the chemical composition but also on its network structure, and in general, a network former oxide (e.g., a network former oxide) that constructs a glass network structure
Figure 678147DEST_PATH_IMAGE003
，
Figure 141489DEST_PATH_IMAGE004
Etc.) the higher the hardness value, and the more network-modifying body oxides (such as alkali metal and alkaline earth metal oxides, etc.) having a mesh breaking effect, the lower the hardness value. The main factors affecting hardness are: the thermal history, pressure history and external factors of the glass such as temperature, humidity at the time of measurement, and the magnitude of applied load. The hardness prediction in the existing glass material design method is mainly based on a weighted sum formula of components, and the prediction method has the defect of low prediction accuracy.
Disclosure of Invention
The invention provides a glass hardness prediction method based on a squirrel search optimization algorithm and a machine learning algorithm, aiming at solving the problem that the glass hardness cannot be accurately and quickly predicted in the prior art.
The technical scheme adopted by the invention is as follows:
the glass hardness prediction method based on the squirrel optimization algorithm and the machine learning algorithm is characterized by comprising the following steps of:
step 1: acquiring hardness data of oxide glass with different components, and constructing a glass hardness database, wherein the glass hardness database comprises glass components mapped one by one and hardness corresponding to the glass components;
step 2: based on chemical characteristics, in terms of element molar content
Figure 282621DEST_PATH_IMAGE005
Coulomb force between atoms
Figure 515019DEST_PATH_IMAGE006
Short-range interaction relation based on force field potential
Figure 21087DEST_PATH_IMAGE007
A descriptor as an input parameter;
and step 3: taking the descriptor constructed in the step 2 as the input of the model, taking the hardness database constructed in the step 1 as the output of the model, constructing a training set and a testing set, and establishing a Catboost model;
and 4, step 4: introducing a squirrel search optimization algorithm, and optimizing the parameters of the selected Catboost model;
and 5: establishing a Catboost model with optimal performance based on the optimized parameters;
step 6: and (3) predicting the glass hardness of the glass component to be predicted by utilizing an optimal Catboost model.
Further, the glass hardness prediction method based on the squirrel optimization algorithm and the machine learning algorithm is characterized in that the step 2 comprises the following steps:
step 2-1: in terms of the molar content of each component constituting the glass
Figure 34042DEST_PATH_IMAGE005
Is a set of descriptors;
step 2-2: construction of descriptors by coulomb forces between different atoms of the components of the constituent glasses
Figure 916547DEST_PATH_IMAGE006
：
Figure 3452DEST_PATH_IMAGE008
Wherein the content of the first and second substances,
Figure 680421DEST_PATH_IMAGE009
and
Figure 665826DEST_PATH_IMAGE010
is an ion
Figure 352022DEST_PATH_IMAGE011
And
Figure 293433DEST_PATH_IMAGE012
the effective ionic charge of (a) is,
Figure 203620DEST_PATH_IMAGE013
and
Figure 128851DEST_PATH_IMAGE014
are respectively a constituent element
Figure 87580DEST_PATH_IMAGE011
And
Figure 414656DEST_PATH_IMAGE012
effective ionic charge
Figure 495744DEST_PATH_IMAGE009
And
Figure 642692DEST_PATH_IMAGE010
the mole fraction of (a);
step 2-3: by short-range interaction of different interatomic force field potentials of various components of glass
Figure 405112DEST_PATH_IMAGE007
Construction descriptor
Figure 586694DEST_PATH_IMAGE007
：
Figure 838684DEST_PATH_IMAGE015
Wherein, the first and the second end of the pipe are connected with each other,Sis a collection of constituent elements, and is,
Figure 472928DEST_PATH_IMAGE016
is the mole fraction of the effective ionic charge of the constituent elements,
Figure 39038DEST_PATH_IMAGE017
for the Buckingham potential parameter,p= -4, -3, -2, -1, 0, 1, 2, 3, 4。
further, the glass hardness prediction method based on the squirrel optimization algorithm and the machine learning algorithm is characterized in that the step 4 comprises the following steps:
step 4-1: selecting parameters needing to be optimized by a Catboost algorithm, wherein the position coordinates of the squirrel are the parameters needing to be optimized, and defining the population scale, the dimensionality, the maximum iteration times, the sliding distance parameters and the predator existence probability of the squirrel search optimization algorithm;
step 4-2: the position of the population is initialized to
Figure 622597DEST_PATH_IMAGE018
Squirrel alone generated random positions, second
Figure 248751DEST_PATH_IMAGE011
The position of only squirrel can be determined by a vector; the positions of all squirrels were randomly initialized within the bounds, as follows:
Figure 104711DEST_PATH_IMAGE019
wherein the content of the first and second substances,
Figure 474513DEST_PATH_IMAGE020
represents the first
Figure 427425DEST_PATH_IMAGE011
Only squirrel (a Chinese character of pine)
Figure 958901DEST_PATH_IMAGE012
The value of the dimension(s) is,
Figure 567737DEST_PATH_IMAGE021
、
Figure 272388DEST_PATH_IMAGE022
respectively the upper and lower boundaries of the variable,randis [0, 1 ]]A random number in between;
step 4-3: calculating the model accuracy of the parameters represented by the position of each squirrel according to the position of each squirrel in the step 4-2;
step 4-4: arranging their positions in ascending order according to the accuracy of flying squirrels;
and 4-5: according to the sequence of the step 4-4, sequentially distributing the flying squirrels to a hickory nut, an oak tree and a common tree, wherein the hickory nut represents the global optimal solution position, and the oak tree represents the local optimal solution position;
and 4-6: renewing squirrel position, as follows:
(1) the squirrel on the oak moves towards the pecan tree,
Figure 283069DEST_PATH_IMAGE023
wherein the content of the first and second substances,
Figure 985446DEST_PATH_IMAGE024
is the random sliding distance of the sliding block,
Figure 81578DEST_PATH_IMAGE025
is [0, 1 ]]A random number within the range of the random number,
Figure 589919DEST_PATH_IMAGE026
is the position of the hickory tree, t represents the current iteration; sliding constant
Figure 720686DEST_PATH_IMAGE027
Realizes the balance between the global search and the local search, and through a large amount of analysis and demonstration,
Figure 593964DEST_PATH_IMAGE027
the value of (d) is set to 1.9;
(2) squirrels on ordinary trees move toward the oak,
Figure 459283DEST_PATH_IMAGE028
wherein the content of the first and second substances,
Figure 974578DEST_PATH_IMAGE029
is [0, 1 ]]Random within range;
(3) the squirrel on the common tree moves toward the hickory tree,
Figure 694273DEST_PATH_IMAGE030
wherein the content of the first and second substances,
Figure 4031DEST_PATH_IMAGE031
is [0, 1 ]]A random number within a range;
and 4-7: calculating the model accuracy of the parameters represented by the position of each squirrel according to the updated position of each squirrel in the steps 4-6, arranging the positions in an ascending order, and redistributing the flying squirrel to the pecan tree, the oak tree and the common tree in sequence;
and 4-8: judging whether the seasonal variation condition is satisfied, if the updating of the position of the squirrel on the common tree is satisfied, the original position is not satisfied, and the following formula is shown:
(1) calculating seasonal constants
Figure 605914DEST_PATH_IMAGE032
：
Figure 659321DEST_PATH_IMAGE033
(2) Calculating seasonal variation conditions
Figure 233521DEST_PATH_IMAGE034
：
Figure 714181DEST_PATH_IMAGE035
Wherein t and
Figure 68939DEST_PATH_IMAGE036
current and maximum iteration values, respectively;
(3) randomly changing the position of squirrels on the common tree if seasonal conditions are met:
Figure 394878DEST_PATH_IMAGE037
and 4-9: recalculating the accuracy of each squirrel position, arranging the positions in an ascending order, and distributing the flying squirrels to the pecan trees, the oak trees and the common trees;
step 4-10: repeating the steps 4-6 to 4-9, and finishing the optimization process when the iteration condition or the maximum iteration times is met;
and 4-11: and outputting the position and the accuracy of the squirrel on the hickory nut tree, wherein the position of the squirrel on the hickory nut tree is a Catboost parameter value.
The invention has the beneficial effects that:
(1) the invention establishes a descriptor based on chemical characteristics, considers the interrelation of ion acting force and short-range acting force among different atoms, better accords with the rule of actual glass, and has more accurate prediction result;
(2) the invention uses the squirrel search optimization algorithm to optimize the parameter optimization of the Catboost algorithm, has simple structure, improves the convergence speed and precision, can obviously improve the performance of the Catboost algorithm by the optimized optimal Catboost algorithm parameter, and has practical significance for improving the accuracy of predicting the glass hardness.
Drawings
FIG. 1 is a flow chart of an embodiment of the method of the present invention;
FIG. 2 is a flow diagram of descriptor construction according to the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments.
Example 1, as shown in fig. 1 and 2:
step 1: the hardness data of oxide glass with different components are collected to construct a glass hardness database, and 400 groups of hardness databases are specifically collected
Figure 354744DEST_PATH_IMAGE038
The hardness data of the glass is used for constructing a glass hardness database;
step 2: based on chemical characteristics, in terms of element molar content
Figure 803043DEST_PATH_IMAGE005
Interatomic coulomb force
Figure 582780DEST_PATH_IMAGE006
Short-range interaction relation based on force field potential
Figure 712410DEST_PATH_IMAGE007
A descriptor as an input parameter; the method comprises the following specific steps:
step 2-1: in 400 groups
Figure 526782DEST_PATH_IMAGE038
Molar content
Figure 893785DEST_PATH_IMAGE039
、
Figure 160819DEST_PATH_IMAGE040
And
Figure 94140DEST_PATH_IMAGE041
three groups of descriptors;
step 2-2: constructing descriptors by coulomb force among Si, Na, Ca and O atoms
Figure 763018DEST_PATH_IMAGE006
：
Figure 553120DEST_PATH_IMAGE008
Wherein the content of the first and second substances,
Figure 41870DEST_PATH_IMAGE009
and
Figure 44461DEST_PATH_IMAGE010
is an ion
Figure 364584DEST_PATH_IMAGE011
And
Figure 528849DEST_PATH_IMAGE012
the effective ionic charge of (a) is,
Figure 504895DEST_PATH_IMAGE013
and
Figure 45598DEST_PATH_IMAGE014
are respectively a constituent element
Figure 220227DEST_PATH_IMAGE011
And
Figure 555394DEST_PATH_IMAGE012
effective ionic charge
Figure 18736DEST_PATH_IMAGE009
And
Figure 910600DEST_PATH_IMAGE010
the mole fraction of (c);
step 2-3: short-range interaction relation of force field potential among Si, Na, Ca and O atoms
Figure 142998DEST_PATH_IMAGE007
Construction descriptor
Figure 649066DEST_PATH_IMAGE007
：
Figure 599704DEST_PATH_IMAGE015
Wherein the content of the first and second substances,Sis a collection of constituent elements, and is,
Figure 278947DEST_PATH_IMAGE016
is the mole fraction of the effective ionic charge of the constituent elements,
Figure 631431DEST_PATH_IMAGE017
for the Buckingham potential parameter,p= -4, -3, -2, -1, 0, 1, 2, 3, 4；
and step 3: taking the descriptor constructed in the step 2 as the input of the model, taking the hardness database constructed in the step 1 as the output of the model, constructing a training set by 320 groups of data, constructing a test set by 80 groups of data, and establishing a Catboost model;
step 4-1: selecting the number of parameter iterations, learning rate and depth of the Catboost algorithm to be optimized, establishing the position of the squirrel, and defining the population scale, dimensionality and maximum iteration number of the squirrel search optimization algorithm;
step 4-2: initializing population position, generating random position for 50 squirrels, the second
Figure 308400DEST_PATH_IMAGE011
The position of the squirrel alone can be determined by a vector. The positions of all squirrels were randomly initialized within the bounds, as follows:
Figure 543073DEST_PATH_IMAGE019
wherein the content of the first and second substances,
Figure 229269DEST_PATH_IMAGE020
represents the first
Figure 170680DEST_PATH_IMAGE011
Only squirrel (a Chinese character of pine)
Figure 284130DEST_PATH_IMAGE012
The value of the dimension(s) is,
Figure 6098DEST_PATH_IMAGE021
、
Figure 230406DEST_PATH_IMAGE022
respectively the upper and lower boundaries of the variable,randis [0, 1 ]]A random number in between;
step 4-3: calculating the model accuracy of the parameters represented by the position of each squirrel according to the position of each squirrel in the step 4-2;
step 4-4: arranging their positions in ascending order according to the accuracy of flying squirrels;
and 4-5: according to the sequence of the step 4-4, sequentially distributing the flying squirrels to the pecan trees, the acorn trees and the common trees;
and 4-6: renewing squirrel position, as follows:
(1) the squirrel on the oak moves towards the pecan tree,
Figure 291903DEST_PATH_IMAGE023
wherein the content of the first and second substances,
Figure 310674DEST_PATH_IMAGE024
is the random sliding distance of the sliding block,
Figure 270671DEST_PATH_IMAGE025
is [0, 1 ]]A random number within the range of the random number,
Figure 33091DEST_PATH_IMAGE026
is the position of the hickory tree and t represents the current iteration. Sliding constant
Figure 214673DEST_PATH_IMAGE027
Realizes the balance between the global search and the local search, and through a large amount of analysis and demonstration,
Figure 466663DEST_PATH_IMAGE027
the value of (d) is set to 1.9;
(2) squirrels on ordinary trees move toward the oak,
Figure 100907DEST_PATH_IMAGE028
wherein the content of the first and second substances,
Figure 401438DEST_PATH_IMAGE029
is [0, 1 ]]Random numbers within a range.
(3) The squirrel on the common tree moves toward the hickory tree,
Figure 703107DEST_PATH_IMAGE030
wherein
Figure 125998DEST_PATH_IMAGE031
Is [0, 1 ]]Random numbers within a range.
And 4-7: calculating the model accuracy of the parameters represented by the position of each squirrel according to the updated position of each squirrel in the steps 4-6, arranging the positions in an ascending order, and redistributing the flying squirrel to the pecan tree, the oak tree and the common tree in sequence;
and 4-8: judging whether the seasonal variation condition is satisfied, if the updating of the position of the squirrel on the common tree is satisfied, the original position is not satisfied, and the following formula is shown:
(1) calculating seasonal constants
Figure 247537DEST_PATH_IMAGE032
：
Figure 351760DEST_PATH_IMAGE033
(2) Calculating seasonal variation conditions
Figure 304672DEST_PATH_IMAGE034
：
Figure 101727DEST_PATH_IMAGE035
Wherein t and
Figure 444984DEST_PATH_IMAGE036
current and maximum iteration values, respectively;
(3) randomly changing the position of squirrels on the common tree if seasonal conditions are met:
Figure 352897DEST_PATH_IMAGE037
and 4-9: recalculating the accuracy of each squirrel position, arranging the positions in an ascending order, and distributing the flying squirrels to the pecan trees, the oak trees and the common trees;
step 4-10: repeating the steps 4-6 to 4-9, and finishing the optimization process when the iteration condition or the maximum iteration times is met;
and 4-11: and outputting the position (Catboost parameter value) and accuracy of the squirrel on the hickory nut.
And 5: establishing a Catboost model with optimal performance based on the optimized parameters;
step 6: and (3) predicting the glass hardness of the glass component to be predicted by utilizing an optimal Catboost model.
Example 2, as shown in fig. 1 and 2:
step 1: the hardness data of oxide glass with different components are collected to construct a glass hardness database, and 800 groups of hardness databases are specifically collected
Figure 911048DEST_PATH_IMAGE038
The hardness data of the glass is used for constructing a glass hardness database;
step 2: based on chemical characteristics, in terms of element molar content respectively
Figure 879004DEST_PATH_IMAGE005
Inter-atomic coulombsActing force
Figure 709557DEST_PATH_IMAGE006
Short-range interaction relation based on force field potential
Figure 217899DEST_PATH_IMAGE007
A descriptor as an input parameter; the method comprises the following specific steps:
step 2-1: in 800 groups
Figure 83086DEST_PATH_IMAGE038
Molar content
Figure 956365DEST_PATH_IMAGE039
、
Figure 539793DEST_PATH_IMAGE040
And
Figure 851825DEST_PATH_IMAGE041
three groups of descriptors;
step 2-2: with Si, Na、Ca. Construction descriptor of coulomb force between O atoms
Figure 571520DEST_PATH_IMAGE006
：
Figure 615699DEST_PATH_IMAGE008
Wherein the content of the first and second substances,
Figure 483161DEST_PATH_IMAGE009
and
Figure 536567DEST_PATH_IMAGE010
is an ion
Figure 110768DEST_PATH_IMAGE011
And
Figure 591428DEST_PATH_IMAGE012
effective ion ofThe charge is applied to the surface of the substrate,
Figure 705708DEST_PATH_IMAGE013
and
Figure 297226DEST_PATH_IMAGE014
are respectively a constituent element
Figure 725933DEST_PATH_IMAGE011
And
Figure 643074DEST_PATH_IMAGE012
effective ionic charge
Figure 219549DEST_PATH_IMAGE009
And
Figure 349179DEST_PATH_IMAGE010
the mole fraction of (c);
step 2-3: with Si, Na、Ca. Short-range interaction relationship of force field potential between O atoms
Figure 163551DEST_PATH_IMAGE007
Construction descriptor
Figure 782751DEST_PATH_IMAGE007
：
Figure 49784DEST_PATH_IMAGE015
Wherein the content of the first and second substances,Sis a collection of constituent elements, and is,
Figure 983105DEST_PATH_IMAGE016
is the mole fraction of the effective ionic charge of the constituent elements,
Figure 651984DEST_PATH_IMAGE017
for the Buckingham potential parameter,p= -4, -3, -2, -1, 0, 1, 2, 3, 4；
and 3, step 3: taking the descriptor constructed in the step 2 as the input of the model, taking the hardness database constructed in the step 1 as the output of the model, constructing a training set by using 640 groups of data, constructing a test set by using 160 groups of data, and constructing a Catboost model;
step 4-1: selecting the number of parameter iterations, learning rate and depth of the Catboost algorithm to be optimized, establishing the position of the squirrel, and defining the population scale, dimensionality and maximum iteration number of the squirrel search optimization algorithm;
step 4-2: initializing population position, generating random position for 50 squirrels, the secondiThe position of only squirrels can be determined by a vector. The positions of all squirrels were randomly initialized within the bounds, as follows:
Figure 442085DEST_PATH_IMAGE019
wherein the content of the first and second substances,
Figure 196415DEST_PATH_IMAGE020
represents the first
Figure 667847DEST_PATH_IMAGE011
Only squirrel (a Chinese character of pine)
Figure 738703DEST_PATH_IMAGE012
The value of the dimension(s) is,
Figure 168547DEST_PATH_IMAGE021
、
Figure 144593DEST_PATH_IMAGE022
respectively the upper and lower boundaries of the variable,randis [0, 1 ]]A random number in between;
step 4-3: calculating the model accuracy of the parameters represented by the position of each squirrel according to the position of each squirrel in the step 4-2;
step 4-4: arranging their positions in ascending order according to the accuracy of flying squirrels;
and 4-5: according to the sequence of the step 4-4, sequentially distributing the flying squirrels to the pecan trees, the acorn trees and the common trees;
and 4-6: renewing squirrel position, as follows:
(1) the squirrel on the oak moves towards the pecan tree,
Figure 419717DEST_PATH_IMAGE023
wherein the content of the first and second substances,
Figure 859925DEST_PATH_IMAGE024
is the random sliding distance of the sliding block,
Figure 195092DEST_PATH_IMAGE025
is [0, 1 ]]A random number within the range of the random number,
Figure 658434DEST_PATH_IMAGE026
is the position of the hickory tree,trepresenting the current iteration. Sliding constant
Figure 737249DEST_PATH_IMAGE027
Realizes the balance between the global search and the local search, and through a large amount of analysis and demonstration,
Figure 31964DEST_PATH_IMAGE027
the value of (d) is set to 1.9;
(2) squirrels on ordinary trees move toward the oak,
Figure 538031DEST_PATH_IMAGE028
wherein the content of the first and second substances,
Figure 488670DEST_PATH_IMAGE029
is [0, 1 ]]Random numbers within a range.
(3) The squirrel on the general tree moves toward the pecan tree,
Figure 167913DEST_PATH_IMAGE030
wherein
Figure 520397DEST_PATH_IMAGE031
Is [0, 1 ]]Random numbers within a range.
And 4-7: calculating the model accuracy of the parameters represented by the position of each squirrel according to the updated position of each squirrel in the steps 4-6, arranging the positions in an ascending order, and redistributing the flying squirrel to the pecan tree, the oak tree and the common tree in sequence;
and 4-8: judging whether the seasonal variation condition is satisfied, if the updating of the position of the squirrel on the common tree is satisfied, the original position is not satisfied, and the following formula is shown:
(1) calculating seasonal constants
Figure 197366DEST_PATH_IMAGE032
:
Figure 635300DEST_PATH_IMAGE033
(2) Calculating seasonal variation conditions
Figure 603388DEST_PATH_IMAGE034
：
Figure 75957DEST_PATH_IMAGE035
Wherein the content of the first and second substances,tand
Figure 923828DEST_PATH_IMAGE036
current and maximum iteration values, respectively.
(3) Randomly changing the position of squirrels on the common tree if seasonal conditions are met:
Figure 645796DEST_PATH_IMAGE037
and 4-9: recalculating the accuracy of each squirrel position, arranging the positions in an ascending order, and distributing the flying squirrels to the pecan trees, the oak trees and the common trees;
step 4-10: repeating the steps 4-6 to 4-9, and finishing the optimization process when the iteration condition or the maximum iteration times is met;
and 4-11: and outputting the position (Catboost parameter value) and accuracy of the squirrel on the hickory nut.
And 5: establishing a Catboost model with optimal performance based on the optimized parameters;
step 6: and (3) predicting the glass hardness of the glass component to be predicted by utilizing an optimal Catboost model.
Example model properties are given in the following table:
table one: examples 1, 2 model Performance tables
Figure 870104DEST_PATH_IMAGE042
The above description is only for the preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art should be considered to be within the technical scope of the present invention, and the technical solutions and the inventive concepts thereof according to the present invention should be equivalent or changed within the scope of the present invention.

Claims (2)

1. The glass hardness prediction method based on the squirrel optimization algorithm and the machine learning algorithm is characterized by comprising the following steps of:
step 1: acquiring hardness data of oxide glass with different components, and constructing a glass hardness database, wherein the glass hardness database comprises glass components mapped one by one and hardness corresponding to the glass components;
step 2: based on chemical characteristics, in terms of element molar content
Figure RE-DEST_PATH_IMAGE001
Coulomb force between atoms
Figure RE-DEST_PATH_IMAGE002
Short-range interaction relation based on force field potential
Figure RE-DEST_PATH_IMAGE003
A descriptor as an input parameter;
and step 3: taking the descriptor constructed in the step 2 as the input of the model, taking the hardness database constructed in the step 1 as the output of the model, constructing a training set and a test set, and establishing a Catboost model;
and 4, step 4: introducing a squirrel search optimization algorithm to optimize the parameters of the selected Catboost model;
and 5: establishing a Catboost model with optimal performance based on the optimized parameters;
and 6: predicting the glass hardness of the glass component by utilizing an optimal Catboost model aiming at the glass component to be predicted;
the step 2 comprises the following steps:
step 2-1: in terms of the molar content of each component constituting the glass
Figure 568274DEST_PATH_IMAGE001
Is a set of descriptors;
step 2-2: construction of descriptors by coulomb forces between different atoms of the components of the constituent glasses
Figure 649362DEST_PATH_IMAGE002
：
Figure RE-DEST_PATH_IMAGE004
Wherein the content of the first and second substances,
Figure RE-DEST_PATH_IMAGE005
and
Figure RE-DEST_PATH_IMAGE006
is an ion
Figure RE-DEST_PATH_IMAGE007
And
Figure RE-DEST_PATH_IMAGE008
the effective ionic charge of (a) is,
Figure RE-DEST_PATH_IMAGE009
and
Figure RE-DEST_PATH_IMAGE010
are respectively a constituent element
Figure 874938DEST_PATH_IMAGE007
And
Figure 637358DEST_PATH_IMAGE008
effective ionic charge
Figure 818941DEST_PATH_IMAGE005
And
Figure 805351DEST_PATH_IMAGE006
the mole fraction of (c);
step 2-3: by short-range interaction of different interatomic force field potentials of various components of glass
Figure 439595DEST_PATH_IMAGE003
Construction descriptor
Figure 5705DEST_PATH_IMAGE003
：
Figure RE-DEST_PATH_IMAGE011
Wherein S is a collection of constituent elements,
Figure RE-DEST_PATH_IMAGE012
is the mole fraction of the effective ionic charge of the constituent elements,
Figure RE-DEST_PATH_IMAGE013
p = -4, -3, -2, -1, 0, 1, 2, 3, 4 for Buckingham potential parameters.
2. The squirrel optimization algorithm and machine learning algorithm-based glass hardness prediction method according to claim 1, wherein the step 4 comprises the steps of:
step 4-1: selecting parameters needing to be optimized by a Catboost algorithm, wherein the position coordinates of the squirrel are the parameters needing to be optimized, and defining the population scale, the dimensionality, the maximum iteration times, the sliding distance parameters and the predator existence probability of the squirrel search optimization algorithm;
step 4-2: the position of the population is initialized to
Figure RE-DEST_PATH_IMAGE014
Random positions were generated by squirrels alone, second
Figure 383073DEST_PATH_IMAGE007
The position of only squirrel can be determined by a vector; the positions of all squirrels were randomly initialized within the bounds, as follows:
Figure RE-DEST_PATH_IMAGE015
wherein, the first and the second end of the pipe are connected with each other,
Figure RE-DEST_PATH_IMAGE016
represents the first
Figure 274805DEST_PATH_IMAGE007
Only squirrel (a Chinese character of pine)
Figure 661924DEST_PATH_IMAGE008
The value of the dimension(s) is,
Figure RE-DEST_PATH_IMAGE017
、
Figure RE-DEST_PATH_IMAGE018
respectively the upper and lower boundaries of the variable, and rand is [0, 1 ]]A random number in between;
step 4-3: calculating the model accuracy of the parameters represented by the position of each squirrel according to the position of each squirrel in the step 4-2;
step 4-4: arranging their positions in ascending order according to the accuracy of flying squirrels;
and 4-5: according to the sequence of the step 4-4, sequentially distributing the flying squirrels to a hickory nut, an oak tree and a common tree, wherein the hickory nut represents the global optimal solution position, and the oak tree represents the local optimal solution position;
and 4-6: updating squirrel position as follows:
(1) the squirrel on the oak moves towards the pecan tree,
Figure RE-DEST_PATH_IMAGE019
wherein the content of the first and second substances,
Figure RE-DEST_PATH_IMAGE020
is the random sliding distance of the sliding block,
Figure RE-DEST_PATH_IMAGE021
is [0, 1 ]]A random number within the range of the random number,
Figure RE-DEST_PATH_IMAGE022
is the position of the hickory tree, t represents the current iteration; sliding constant
Figure RE-DEST_PATH_IMAGE023
Realizes the balance between the global search and the local search, and through a large amount of analysis and demonstration,
Figure 828463DEST_PATH_IMAGE023
the value of (d) is set to 1.9;
(2) squirrels on ordinary trees move toward the oak,
Figure RE-DEST_PATH_IMAGE024
wherein the content of the first and second substances,
Figure RE-DEST_PATH_IMAGE025
is [0, 1 ]]A random number within a range;
(3) the squirrel on the common tree moves toward the hickory tree,
Figure RE-DEST_PATH_IMAGE026
wherein the content of the first and second substances,
Figure RE-DEST_PATH_IMAGE027
is [0, 1 ]]A random number within a range;
and 4-7: calculating the model accuracy of the parameters represented by the position of each squirrel according to the updated position of each squirrel in the steps 4-6, arranging the positions in an ascending order, and redistributing the flying squirrel to the pecan tree, the oak tree and the common tree in sequence;
and 4-8: judging whether the seasonal variation condition is satisfied, if the updating of the position of the squirrel on the common tree is satisfied, the original position is not satisfied, and the following formula is shown:
(1) calculating seasonal constants
Figure RE-DEST_PATH_IMAGE028
：
Figure RE-DEST_PATH_IMAGE029
(2) Calculating seasonal variation conditions
Figure RE-DEST_PATH_IMAGE030
：
Figure RE-DEST_PATH_IMAGE031
Wherein the content of the first and second substances,t and
Figure RE-DEST_PATH_IMAGE032
current and maximum iteration values, respectively;
(3) randomly changing the position of squirrels on the common tree if seasonal conditions are met:
Figure RE-DEST_PATH_IMAGE033
and 4-9: recalculating the accuracy of each squirrel position, arranging the positions in an ascending order, and distributing the flying squirrels to the pecan trees, the oak trees and the common trees;
step 4-10: repeating the steps 4-6 to 4-9, and finishing the optimization process when the iteration condition or the maximum iteration times is met;
and 4-11: and outputting the position and the accuracy of the squirrel on the hickory nut tree, wherein the position of the squirrel on the hickory nut tree is a Catboost parameter value.
CN202210281841.9A 2022-03-22 2022-03-22 Glass hardness prediction method based on squirrel optimization algorithm and machine learning algorithm Active CN114373523B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202210281841.9A CN114373523B (en) 2022-03-22 2022-03-22 Glass hardness prediction method based on squirrel optimization algorithm and machine learning algorithm

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202210281841.9A CN114373523B (en) 2022-03-22 2022-03-22 Glass hardness prediction method based on squirrel optimization algorithm and machine learning algorithm

Publications (2)

Publication Number Publication Date
CN114373523A CN114373523A (en) 2022-04-19
CN114373523B true CN114373523B (en) 2022-06-03

Family

ID=81146137

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202210281841.9A Active CN114373523B (en) 2022-03-22 2022-03-22 Glass hardness prediction method based on squirrel optimization algorithm and machine learning algorithm

Country Status (1)

Country Link
CN (1) CN114373523B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN117497087B (en) * 2023-12-20 2024-04-26 浙江大学 Oxide glass performance prediction method based on interpretable high-dimensional spatial prediction model

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104062422A (en) * 2014-06-10 2014-09-24 华东理工大学 Predicating method for transformation temperature and hardness of glass
CN106444625A (en) * 2016-09-18 2017-02-22 合肥工业大学 Cutter head servo control method and cutter head servo control device for glass cutting machine
CN110683749A (en) * 2019-10-31 2020-01-14 江苏瑞特钢化玻璃制品有限公司 Production process of high-strength toughened glass
CN110728401A (en) * 2019-10-10 2020-01-24 郑州轻工业学院 Short-term power load prediction method of neural network based on squirrel and weed hybrid algorithm

Family Cites Families (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR102523472B1 (en) * 2016-08-01 2023-04-18 삼성전자주식회사 Method and apparatus for searching new material
CN110850477A (en) * 2019-11-07 2020-02-28 中建四局第一建筑工程有限公司 Resistivity karst cave identification method based on squirrel search algorithm

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104062422A (en) * 2014-06-10 2014-09-24 华东理工大学 Predicating method for transformation temperature and hardness of glass
CN106444625A (en) * 2016-09-18 2017-02-22 合肥工业大学 Cutter head servo control method and cutter head servo control device for glass cutting machine
CN110728401A (en) * 2019-10-10 2020-01-24 郑州轻工业学院 Short-term power load prediction method of neural network based on squirrel and weed hybrid algorithm
CN110683749A (en) * 2019-10-31 2020-01-14 江苏瑞特钢化玻璃制品有限公司 Production process of high-strength toughened glass

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
Explainable machine learning algorithms for predicting galss transition temperatures;Edesio Alcobaca,et al.;《Acta Materialia》;20200415;92-100 *
触控屏用盖板玻璃的硬度分析;刘亚茹 等;《玻璃搪瓷与眼镜》;20211231;7-13 *

Also Published As

Publication number Publication date
CN114373523A (en) 2022-04-19

Similar Documents

Publication Publication Date Title
Shazeer et al. Swivel: Improving embeddings by noticing what's missing
Swenson Phylogenetic imputation of plant functional trait databases
Ghosh et al. Evolutionary algorithms for multi-criteria optimization: A survey
Örkcü et al. Estimating the parameters of 3-p Weibull distribution using particle swarm optimization: A comprehensive experimental comparison
CN114373523B (en) Glass hardness prediction method based on squirrel optimization algorithm and machine learning algorithm
CN110390561B (en) User-financial product selection tendency high-speed prediction method and device based on momentum acceleration random gradient decline
CN111931067A (en) Interest point recommendation method, device, equipment and medium
CN108875233B (en) Structure reliability method based on dynamic weight response surface under mixed uncertainty
CN106529185B (en) A kind of combination forecasting method and system of ancient building displacement
CN113407185B (en) Compiler optimization option recommendation method based on Bayesian optimization
JP7144676B2 (en) Information processing device, quality-related expression generation method, and quality-related expression generation program
CN111753215A (en) Multi-objective recommendation optimization method and readable medium
CN108764546A (en) A kind of paper impact factor prediction technique based on academic big data
CN115147192A (en) Recommendation method and recommendation system based on double-view-angle deviation correction
Vanclay Compatible deterministic and stochastic predictions by probabilistic modeling of individual trees
CN114255879A (en) Preventive medical facility network design software considering congestion effect
CN109960762B (en) Weather and time perception personalized POI sequence recommendation method
Kalweit et al. Composite q-learning: Multi-scale q-function decomposition and separable optimization
CN110060086A (en) A kind of on-line prediction method based on User reliability in Web cloud service
Dufays Infinite-state Markov-switching for dynamic volatility and correlation models
CN114912331A (en) Cabin reinforcing rib optimization method, device, equipment and medium
Wang et al. A new algorithm to estimate monotone nonparametric link functions and a comparison with parametric approach
Keller-Schmidt et al. Anomalous scaling in an age-dependent branching model
CN117497087B (en) Oxide glass performance prediction method based on interpretable high-dimensional spatial prediction model
WO2024139684A1 (en) Machine-learning-based method and system for predicting stress layer depth after glass tempering

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
GR01 Patent grant