CN115310623B - Intelligent decision-making method, system and terminal for sapphire chemical mechanical polishing processing technology - Google Patents
Intelligent decision-making method, system and terminal for sapphire chemical mechanical polishing processing technology Download PDFInfo
- Publication number
- CN115310623B CN115310623B CN202210816729.0A CN202210816729A CN115310623B CN 115310623 B CN115310623 B CN 115310623B CN 202210816729 A CN202210816729 A CN 202210816729A CN 115310623 B CN115310623 B CN 115310623B
- Authority
- CN
- China
- Prior art keywords
- chemical mechanical
- mechanical polishing
- decision
- instance
- library
- 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
Images
Classifications
-
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/906—Clustering; Classification
-
- 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
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- 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
Abstract
The invention belongs to the technical field of sapphire chemical mechanical polishing processing control, and discloses an intelligent decision-making method, an intelligent decision-making system and an intelligent decision-making terminal for a sapphire chemical mechanical polishing processing process, wherein a framework of the intelligent decision-making system for the sapphire chemical mechanical polishing process is established based on a 7R model; establishing a perfect chemical mechanical polishing database based on the database structure; carrying out example optimization by combining three decision theory, analytic hierarchy process and comprehensive confidence factors of global similarity, confidence and activity calculated by nearest neighbor algorithm; establishing a prediction model combining heterogeneous integrated learning of a decision tree and a deep neural network and particle swarm optimization to obtain an optimal polishing process scheme output; a sapphire chemical mechanical polishing processing technology decision system is developed based on Qt 4.8.7 and SQLite 3 and is applied to a domestic numerical control polishing machine tool. The invention provides an industrial application paradigm for the intelligent decision-making of the sapphire chemical mechanical polishing processing technology, and realizes the intelligentization of the sapphire chemical mechanical polishing.
Description
Technical Field
The invention belongs to the technical field of sapphire chemical mechanical polishing processing control, and particularly relates to an intelligent decision-making method, system and terminal for a sapphire chemical mechanical polishing processing technology.
Background
Sapphire has important application in both national defense and military fields and civil consumption fields at present. In recent years, the semiconductor illumination has remarkable effects on energy conservation and emission reduction, and the development of the semiconductor illumination has positive effects on ecological civilization promotion. The sapphire substrate is used as a key material of the semiconductor illumination high-end chip of the LED and the Micro LED, and the processing quality of the sapphire substrate can directly have great influence on the service life and the service performance of the LED chip. However, sapphire belongs to a typical hard and brittle class of difficult-to-process ceramic class materials, resulting in significant challenges in its processing.
The chemical mechanical polishing is used as the current ultra-precise polishing technology of planar substrate materials, and is an important guarantee for realizing the efficient and high-precision processing of sapphire substrates. The chemical mechanical polishing is an ultra-precise polishing processing technology adopting free abrasive particles, and polishing pressure, polishing rotation speed, polishing liquid (pH, dispersing agent, oxidant, catalyst, active agent and the like), abrasive particles (types, concentration, particle size, shape and the like), chemical reaction rate and the like all play roles and influence on the sapphire chemical mechanical polishing, and the free abrasive particles enable the processing track to be random and indefinite. In addition, the cost of chemical mechanical polishing is more than 80% in the sapphire substrate processing chain, but the polished result is not predictive, which results in too high process test cost. Therefore, this places demands on the choice of sapphire chemical mechanical polishing process. The optimal process selection in the existing polishing process can only be completed on the basis of one-by-one experiment of workers, the cost is too high, the efficiency is too low, and the huge requirements on sapphire substrates in the military and civil fields cannot be met.
In conclusion, the realization of the intelligent decision in the chemical mechanical polishing process of the sapphire substrate material is an important method for reducing the processing cost of the sapphire substrate and improving the production efficiency of the sapphire substrate, and the method lays a foundation for the high-end tamping of the domestic polishing machine tool.
Through the above analysis, the problems and defects existing in the prior art are as follows:
(1) The existing sapphire chemical mechanical polishing is an ultra-precise polishing processing technology adopting free abrasive particles, and polishing pressure, polishing rotating speed, polishing liquid, abrasive particles, chemical reaction rate and the like all have an effect and influence on the sapphire chemical mechanical polishing, and the free abrasive particles enable the processing track to be random.
(2) The existing sapphire chemical mechanical polishing process has the cost accounting for more than 80% of the sapphire substrate processing chain, but the polished result is not predictive, which results in too high process test cost.
(3) The optimal process selection in the existing polishing process can only be completed on the basis of one-by-one experiment of workers, the cost is too high, the efficiency is too low, and the huge requirements on sapphire substrates in the military and civil fields cannot be met.
Disclosure of Invention
Aiming at the problems existing in the prior art, the invention provides an intelligent decision method, an intelligent decision system and an intelligent decision terminal for a sapphire chemical mechanical polishing processing technology.
The invention is realized in such a way that the intelligent decision method for the sapphire chemical mechanical polishing processing technology comprises the following steps:
the system comprises a chemical mechanical polishing intelligent process decision system framework, a chemical mechanical polishing processing basic database, a chemical mechanical polishing processing knowledge base, a chemical mechanical polishing processing problem definition, chemical mechanical polishing processing example optimization, chemical mechanical polishing processing scheme prediction and chemical mechanical polishing processing application.
Further, the chemical mechanical polishing processing basic database comprises a polishing machine library, a polishing head library, a polishing pad library, a polishing liquid library, a material library and a finishing library;
the chemical mechanical polishing processing knowledge base comprises a chemical mechanical polishing processing instance base, a model base, an algorithm base and a rule base; the example library comprises process problem description, process problem solution, process problem application efficiency and evaluation, and the model library, algorithm library and rule library comprise corresponding information required by the intelligent decision process of the sapphire chemical mechanical polishing process;
the chemical mechanical polishing processing applications include chemical mechanical polishing process example preferences and chemical mechanical polishing process recipe reasoning.
Further, the chemical mechanical polishing processing problem definition is to modularly input basic information required for a subsequent intelligent process decision process by corresponding to a process problem description part in an example library, including sapphire size, thickness, material removal rate, surface roughness, subsurface damage and flatness; and after inputting the basic information, generating a process problem definition file, splicing decision and application of residual information after evaluation in an instance recovery stage, and forming a new process instance to be stored in a chemical mechanical polishing processing knowledge base.
Further, the chemical mechanical polishing process example is preferably realized based on three decision theory, analytic hierarchy process and comprehensive confidence factors of global similarity, confidence and activity calculated by nearest neighbor algorithm.
The chemical mechanical polishing process scheme reasoning is realized based on a prediction model combining heterogeneous integrated learning of a decision tree and a deep neural network and particle swarm optimization.
The chemical mechanical polishing processing example is preferably carried out, and if the processing effect of the example which does not meet the threshold setting or is obtained by the optimization cannot meet the processing requirement in the pre-experiment, the chemical mechanical polishing processing scheme is predicted. In the optimization process of the chemical mechanical polishing processing example, a method of combining subjective and objective weights is adopted, wherein three decision theories are selected to calculate objective weight values of all attributes, a chromatographic analysis method is selected to calculate subjective weight values of all the attributes, and a linear weighting principle is adopted to obtain comprehensive characteristic attribute weight values; calculating the local similarity value of the attribute between the new process problem and each instance by adopting a nearest neighbor algorithm, then calculating the overall similarity value of the new process problem and each instance by combining the weight of each characteristic attribute, arranging the values in descending order, and outputting the instance number meeting the set threshold value through instance preference; when the instance meeting the threshold cannot be searched or the searched instance cannot meet the processing requirement, a process scheme prediction module is started, a process optimization algorithm based on heterogeneous integrated learning of a decision tree and a deep neural network algorithm is adopted to optimize a combined weight value of machine learning and deep learning by adopting a particle swarm algorithm, an optimal learning prediction result is obtained, and a final processing scheme is formed by combining instance optimization.
Further, the intelligent decision method for the sapphire chemical mechanical polishing processing technology comprises the following steps:
step one, building a sapphire chemical mechanical polishing intelligent process decision system frame based on a 7R model;
step two, a perfect chemical mechanical polishing database is established based on the database structure;
thirdly, carrying out example optimization by combining three decision theories, an analytic hierarchy process and comprehensive confidence factors of global similarity, confidence and activity calculated by a nearest neighbor algorithm;
establishing a prediction model combining heterogeneous integrated learning of a decision tree and a deep neural network and optimization of a particle swarm algorithm to obtain an optimal polishing process scheme output;
and fifthly, developing a sapphire chemical mechanical polishing processing technology decision system based on Qt 4.8.7 and SQLite 3 and applying the system to a domestic numerical control polishing machine tool.
Further, the 7R model in step one includes instance expression, instance reuse, instance retrieval, instance reasoning, instance modification, instance reclamation, and instance reclassification.
Another object of the present invention is to provide an intelligent decision system for a sapphire chemical mechanical polishing process, which applies the intelligent decision method for a sapphire chemical mechanical polishing process, the intelligent decision system for a sapphire chemical mechanical polishing process comprising: the system comprises a user layer, a system algorithm layer, a data management layer, a data operation layer and a data storage layer, wherein the user layer, the system algorithm layer, the data operation layer and the data storage layer are used for inputting process problem information of workpiece types and material types and importing new process problems through texts; the system performs a comprehensive confidence factor calculation and the user will either receive recommended instances within a threshold from the existing instance database or acquire a machining process recipe using heterogeneous integrated learning.
The user layer comprises a computer end, a processing machine tool terminal, a man-machine interaction interface and a user end;
the system algorithm layer comprises a grinding process example optimization algorithm and a grinding process scheme reasoning algorithm; the grinding process example optimization algorithm comprises a rough set, a three-branch decision, a CRITIC method and an AHP method, and the grinding process scheme reasoning algorithm comprises a neural network and a support vector machine;
the data management layer comprises basic data management, experience data management, decision data management and data security management; the basic data management comprises a machine tool library, a material library, a polishing solution library, a polishing pad library and a finishing library, the experience data management comprises an instance library, an algorithm library, a model library and a rule library, the decision data management comprises processing technology information, technology instance reasoning, technology intelligent reasoning and technology scheme output, and the data security management comprises user management, authority management, data backup and security log;
the data operation layer comprises data access/file operation;
the data storage layer comprises a basic database, a process knowledge base and a user database.
It is another object of the present invention to provide a computer device comprising a memory and a processor, the memory storing a computer program which, when executed by the processor, causes the processor to perform the steps of the smart decision method for sapphire chemical mechanical polishing process.
Another object of the present invention is to provide a computer readable storage medium storing a computer program, which when executed by a processor, causes the processor to perform the steps of the intelligent decision method for a sapphire chemical mechanical polishing process.
The invention further aims at providing an information data processing terminal which is used for realizing the intelligent decision system for the sapphire chemical mechanical polishing processing technology.
In combination with the above technical solution and the technical problems to be solved, please analyze the following aspects to provide the following advantages and positive effects:
first, aiming at the technical problems in the prior art and the difficulty in solving the problems, the technical problems solved by the technical proposal of the invention are analyzed in detail and deeply by tightly combining the technical proposal to be protected, the results and data in the research and development process, and the like, and some technical effects brought after the problems are solved have creative technical effects. The specific description is as follows:
the invention provides an intelligent decision-making method for a sapphire chemical mechanical polishing processing technology. Aiming at the related problems of high test cost, low test efficiency, insufficient production capacity, poor intelligent degree of the processing process and the like of the existing sapphire chemical mechanical polishing processing technology, the invention provides an intelligent decision method of the sapphire chemical mechanical polishing processing technology by analyzing the problems of sapphire chemical mechanical polishing processing characteristic extraction, decision optimization key technology, knowledge expression, reuse and the like contained in the intelligent decision of the sapphire chemical mechanical polishing, and develops a decision system responding to a domestic polishing machine tool based on a software platform and a database platform.
The invention provides an intelligent decision-making method of a sapphire chemical mechanical polishing process based on deep learning and machine learning algorithms and comprising process instance optimization and process scheme prediction, and a decision-making system for developing and responding to a domestic polishing machine tool based on information interaction of a software platform and a database platform of Qt 4.8.7 and SQLite 3 by analyzing the problems of sapphire chemical mechanical polishing processing characteristics extraction, decision-making optimization key technology, knowledge expression and reuse and the like.
The invention specifically discloses an intelligent decision-making method of a sapphire chemical mechanical polishing process, which specifically comprises a chemical mechanical polishing intelligent process decision-making system framework, a chemical mechanical polishing process basic database, a chemical mechanical polishing process knowledge base, chemical mechanical polishing process problem definition, chemical mechanical polishing process instance optimization, chemical mechanical polishing process scheme prediction and chemical mechanical polishing process application; establishing a sapphire chemical mechanical polishing intelligent process decision system frame based on a 7R model; establishing a perfect chemical mechanical polishing database based on the database structure; carrying out example optimization by combining three decision theory, analytic hierarchy process and comprehensive confidence factors of global similarity, confidence and activity calculated by nearest neighbor algorithm; establishing a prediction model combining heterogeneous integrated learning of a decision tree and a deep neural network and particle swarm optimization to obtain an optimal polishing process scheme output; a sapphire chemical mechanical polishing processing technology decision system is developed based on Qt 4.8.7 and SQLite 3 and is applied to a domestic numerical control polishing machine tool.
Secondly, the technical scheme is regarded as a whole or from the perspective of products, and the technical scheme to be protected has the following technical effects and advantages:
the intelligent decision-making method for the sapphire chemical mechanical polishing processing technology solves a series of problems of complex processing working conditions, high difficulty in process parameter decision-making optimization, high test cost, low processing efficiency and the like in the existing sapphire chemical mechanical polishing processing, can obviously reduce the process decision-making time of the sapphire wafer chemical mechanical polishing processing, effectively improves the processing efficiency, and fully exerts the chemical mechanical polishing processing efficiency.
The invention provides an industrial application model for intelligent decision-making of the sapphire chemical mechanical polishing processing technology, solves the problems that the existing optimal technological selection in the polishing processing process is too high in cost and too low in efficiency and cannot meet the huge requirements on the sapphire substrate in the military and civil fields only based on the completion of experiments one by one of workers, reduces the processing cost of the sapphire substrate, improves the production efficiency of the sapphire substrate, realizes the intelligent progress of the sapphire chemical mechanical polishing, and lays a foundation for the domestic polishing machine tool to be highly-advanced. In addition, the invention can be popularized to ultra-precise polishing of other planar substrate materials.
Thirdly, as inventive supplementary evidence of the claims of the present invention, the following important aspects are also presented:
(1) The expected benefits and commercial values after the technical scheme of the invention is converted are as follows:
the invention can be applied to the domestic polishing machine tool, improves the intelligent level of the domestic polishing machine tool, enhances the commercial competitiveness of the domestic polishing machine tool, can gradually accumulate the original industrial data in the polishing field, upgrades the tamping foundation for the later production line, and improves the core commercial competitiveness of the domestic polishing machine tool.
(2) The technical scheme of the invention fills the technical blank in the domestic and foreign industries:
at present, the polishing process at home and abroad is in a test mode and basically depends on the experience of workers, and no intelligent process decision for realizing polishing process by the method disclosed by the patent is reported.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings that are needed in the embodiments of the present invention will be briefly described below, and it is obvious that the drawings described below are only some embodiments of the present invention, and other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of an intelligent decision-making method for a sapphire chemical mechanical polishing process provided by an embodiment of the present invention;
FIG. 2 is a schematic diagram of a "7R" model provided by an embodiment of the present invention;
FIG. 3 is a flow chart of intelligent decision making for the sapphire chemical mechanical polishing process provided by the embodiment of the invention;
FIG. 4 is a schematic diagram of a base database for sapphire CMP processing provided by an embodiment of the present invention;
FIG. 5 is a schematic diagram of a knowledge base of a sapphire chemical mechanical polishing process according to an embodiment of the present invention;
FIG. 6 is a flowchart of a sapphire chemical mechanical polishing process scheme prediction algorithm based on heterogeneous ensemble learning provided by an embodiment of the present invention;
fig. 7 is a diagram of a software framework for intelligent decision making for a sapphire chemical mechanical polishing process according to an embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the following examples in order to make the objects, technical solutions and advantages of the present invention more apparent. It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
Aiming at the problems in the prior art, the invention provides an intelligent decision method, an intelligent decision system and an intelligent decision terminal for a sapphire chemical mechanical polishing processing technology, and the invention is described in detail below with reference to the accompanying drawings.
In order to fully understand how the invention may be embodied by those skilled in the art, this section is an illustrative embodiment in which the claims are presented for purposes of illustration.
The intelligent decision-making method for the sapphire chemical mechanical polishing process provided by the embodiment of the invention specifically comprises a chemical mechanical polishing intelligent process decision-making system framework, a chemical mechanical polishing process basic database, a chemical mechanical polishing process knowledge base, chemical mechanical polishing process problem definition, chemical mechanical polishing process example optimization, chemical mechanical polishing process scheme prediction and chemical mechanical polishing process application.
As shown in fig. 1, the intelligent decision method for the sapphire chemical mechanical polishing processing technology provided by the embodiment of the invention comprises the following steps:
s101, establishing a sapphire chemical mechanical polishing intelligent process decision system frame based on a 7R model;
s102, establishing a perfect chemical mechanical polishing database based on a database structure;
s103, carrying out example optimization by combining three decision theories, analytic hierarchy process and comprehensive confidence factors of global similarity, confidence and activity calculated by a nearest neighbor algorithm;
s104, establishing a prediction model combining heterogeneous integrated learning of a decision tree and a deep neural network and particle swarm optimization to obtain an optimal polishing process scheme output;
s105, developing a sapphire chemical mechanical polishing processing technology decision system based on Qt 4.8.7 and SQLite 3 and applying the system to a domestic numerical control polishing machine tool.
The 7R model provided by the embodiment of the invention comprises instance expression (rendering), instance Reuse (Reuse), instance retrieval (retrieval), instance Reasoning (retrieval), instance modification (Revise), instance recovery (retrieval) and instance Reclassification (Reclassification).
The framework of the intelligent process decision system for the chemical mechanical polishing provided by the embodiment of the invention is a framework which is built based on a 7R model and comprises optimization and process scheme prediction of a sapphire chemical mechanical polishing processing process example, a process scheme for reference can be decided based on an input new polishing process problem for workers to select and realize processing and testing, and a test result can be recovered to the system for next use after evaluation.
The chemical mechanical polishing processing basic database provided by the embodiment of the invention is a collection containing various libraries related to the chemical mechanical polishing processing of sapphire, and mainly comprises a polishing machine library, a polishing head library, a polishing pad library, a polishing liquid library, a material library and a finishing library. The chemical mechanical polishing processing basic database is established for providing data support for the follow-up example optimization and process scheme prediction, the established design logic follows an E-R model (namely Entity-Relationship Model) method to model the related data information of the chemical mechanical polishing processing, and each Entity library is ensured to have corresponding attributes and corresponding attribute values.
The chemical mechanical polishing knowledge base provided by the embodiment of the invention mainly stores knowledge information of sapphire chemical mechanical polishing, and mainly comprises a chemical mechanical polishing instance base, a model base, an algorithm base and a rule base. The example library mainly comprises three parts of process problem description, process problem solution, process problem application efficiency and evaluation information. The model library, algorithm library and rule library mainly contain corresponding information required by the intelligent decision process of the sapphire chemical mechanical polishing process.
The chemical mechanical polishing processing problem definition provided by the embodiment of the invention is a first step of intelligent decision-making of the sapphire chemical mechanical polishing processing process, and the modular input is carried out through the process problem description part corresponding to the example library, namely the basic information required by the subsequent intelligent process decision-making process, such as sapphire size, thickness, material removal rate, surface roughness, subsurface damage, flatness and the like. After basic information is input, a process problem definition file is generated, and the process problem definition file can be called in multiple links of the following process intelligent decision whole process. And in the example recycling stage, the decision and the residual information after the evaluation are spliced to form a new process example which is stored in a chemical mechanical polishing processing knowledge base.
The chemical mechanical polishing processing example provided by the embodiment of the invention is preferable, and the chemical mechanical polishing processing scheme prediction belongs to the core step of intelligent decision of the sapphire chemical mechanical polishing processing technology. Based on the requirements defined by the chemical mechanical polishing processing problem, firstly, carrying out optimization of the chemical mechanical polishing processing example, and if the example meeting the threshold setting is not available or the processing effect of the optimized example in the pre-experiment cannot meet the processing requirement, carrying out prediction of the chemical mechanical polishing processing scheme. In the optimization process of the chemical mechanical polishing processing example, in order to reflect the weights of all the attributes more accurately, a method of combining subjective and objective weights is adopted, wherein three decision theory (3 WD) is selected to calculate the objective weight value of each attribute, chromatography (APH) is selected to calculate the subjective weight value of each attribute, and a linear weighting principle is adopted to obtain the comprehensive characteristic attribute weight value. Then, calculating the local similarity value of the attribute between the new process problem and each instance by adopting a nearest neighbor algorithm, calculating the overall similarity value of the new process problem and each instance by combining the weight of each characteristic attribute, arranging the overall similarity values according to descending order, and outputting the instance number meeting the set threshold value through instance preference. When the instance meeting the threshold cannot be searched or the searched instance cannot meet the processing requirement, a process scheme prediction module is started, a process optimization algorithm based on heterogeneous integrated learning of a Decision Tree (DT) and a Deep Neural Network (DNN) algorithm is adopted, a Particle Swarm Optimization (PSO) is adopted to optimize a machine learning and deep learning combination weight value, so that an optimal learning prediction result is obtained, and a final processing scheme is formed by combining instance optimization.
The chemical mechanical polishing processing application provided by the embodiment of the invention is based on Qt 4.8.7 and SQLite 3, and develops a program interactive interface of an intelligent decision system of a sapphire chemical mechanical polishing process, and the program interactive interface comprises a user layer, a system algorithm layer, a data management layer, a data operation layer and a data storage layer. The interface can realize the input of the process problem information of workpiece types, material types and the like, and can also directly import new process problems through texts. The system then performs a comprehensive confidence factor calculation, and the user will either receive recommended instances within a threshold from the existing instance database or acquire a process recipe using heterogeneous integrated learning.
The intelligent decision system for the sapphire chemical mechanical polishing processing technology provided by the embodiment of the invention comprises the following components: the system comprises a user layer, a system algorithm layer, a data management layer, a data operation layer and a data storage layer, wherein the user layer, the system algorithm layer, the data operation layer and the data storage layer are used for inputting process problem information of workpiece types and material types and importing new process problems through texts; the system performs a comprehensive confidence factor calculation and the user will either receive recommended instances within a threshold from the existing instance database or acquire a machining process recipe using heterogeneous integrated learning.
The user layer comprises a computer end, a machine tool terminal, a man-machine interaction interface and a user end;
the system algorithm layer comprises a grinding process example optimization algorithm and a grinding process scheme reasoning algorithm; the grinding process example optimization algorithm comprises a rough set, a three-branch decision, a CRITIC method and an AHP method, and the grinding process scheme reasoning algorithm comprises a neural network and a support vector machine;
the data management layer comprises basic data management, experience data management, decision data management and data security management; the basic data management comprises a machine tool library, a material library, a polishing solution library, a polishing pad library and a finishing library, the experience data management comprises an instance library, an algorithm library, a model library and a rule library, the decision data management comprises processing technology information, technology instance reasoning, technology intelligent reasoning and technology scheme output, and the data security management comprises user management, authority management, data backup and security log;
the data operation layer comprises data access/file operation;
the data storage layer comprises a basic database, a process knowledge base and a user database.
The technical scheme of the invention is further described below with reference to specific embodiments.
Example 1:
as shown in fig. 2 and 3, the embodiment of the invention provides an intelligent decision method for a sapphire chemical mechanical polishing processing technology based on a 7R model, which mainly comprises the following steps:
step 1, the expression of the sapphire chemical mechanical polishing processing process example is realized, and the imperfect example is subjected to rule reasoning supplement by combining a rule reasoning method, so that the stored example is ensured to have operability. The stored instance becomes an instance library for subsequent operations.
And 2, reusing the example library obtained in the step 1, so as to make intelligent process decisions on the new sapphire chemical mechanical polishing process problem input by a user.
And 3, performing instance retrieval on the new process problems, firstly performing attribute reduction on the instances based on a method combining three decision theories with a hierarchical analysis method, then performing local similarity calculation on different types of information based on a nearest neighbor algorithm, and then combining all characteristic attribute weights, calculating the overall similarity value of the new process problems and each instance, arranging the overall similarity value according to descending order, and outputting instance numbers meeting a set threshold through instance preference.
And 4, when the instance meeting the threshold cannot be searched out or the searched instance cannot meet the processing requirement, starting a process scheme prediction module, optimizing a process optimization algorithm based on heterogeneous integrated learning of a Decision Tree (DT) and a Deep Neural Network (DNN) algorithm, optimizing a machine learning and deep learning combination weight value by adopting a Particle Swarm Optimization (PSO), thereby obtaining an optimal learning prediction result, and optimizing a final processing scheme by combining the instance.
And 5, modifying the processing scheme originally stored in the database based on the optimal scheme obtained in the step 4 to form a new example.
And 6, recovering the new instance in the step 5 and storing the new instance in an instance library.
And 7, disassembling the instance information recovered in the step 6, and reclassifying each item of information into each database.
Example 2: a method for calculating the weight value of the comprehensive characteristic attribute of the comprehensive three decision theory and the chromatographic analysis method.
The weights of different characteristic attributes in the decision process are different, which has important influence on the decision process, and obtaining the weight value of the characteristic attribute which is more in line with the reality obviously improves the accuracy of the decision process.
Before the three decisions are classified, the samples need to be divided into domains. According to the definition of the rough set, x and a can be divided into three relations, x e a,
x ε BND (A), where x ε BND (A) represents the boundary that element x belongs to concept ADomain. In the coarse decision set, X is a subset of the corpus U, and the state set is expressed as
X and->
Representing belonging to X and not belonging to X. The set of actions corresponding to state X is Γ= { P, B, N }, where P, B, N represents 3 kinds of determination actions, i.e., X e POS (X), X e BND (X), X e NEG (X), respectively. The loss function of the three decisions is determined by the losses brought about by the individual actions. As shown in Table 1, wherein lambda PP 、λ BP 、λ NP Representing the loss of action P, B, N when X belongs to X, lambda PN 、λ BN 、λ NN Indicating that x is->
The loss due to action P, B, N is taken. />
Table 1 three decision loss functions
According to a minimum risk decision rule:
(P) when P r (X|[x]) When the alpha is not less than the alpha, X is E POS (X);
(B) When beta is less than P r (X|[x]) When < alpha, X is E BND (X);
(N) when P r (X|[x]) When beta is less than or equal to beta, X is less than NEG (X);
wherein, the liquid crystal display device comprises a liquid crystal display device,
and beta is more than or equal to 0 and less than alpha is more than or equal to 1.
Based on the above classification, all the characteristic attributes in the sapphire chemical mechanical polishing process are classified into three, and finally all the characteristic attributes are included in a positive domain interval POS (X) and a negative domain interval NEG (X) through a flow as shown in fig. 4, wherein the characteristic attribute of the negative domain interval is a less or no-influence attribute in the decision process, and is not considered as a priority attribute in the case reasoning process.
And carrying out discrete processing on the multidimensional continuous attributes by adopting a rough set discretization algorithm based on forest optimization, and determining objective weights of the characteristic attributes based on an improved three-branch decision theory weight construction method.
In a given one of the information systems is= (U, a, V, f),
and->
R is any->
And an equivalence relation on U, the attribute certainty of attribute a is:
where ζ (a) is the certainty of attribute a.
Let the domain be U, n attributes, m reduction, F (x, F) k ) Representing all reduction f in U k The number of attributes in the list is the reduced number of x,
representing the number of combinations of x from n elements, the degree of reduction of attribute a is:
where ρ (a) is the degree of reduction of attribute a.
The comprehensive objective weight of the attribute a can be obtained after normalization processing by combining the attribute certainty factor with the reduction factor:
in the method, in the process of the invention,
the ith characteristic attribute weight coefficient is obtained by using three decision theory; λ is a defined weight parameter, typically taking a value of 0.5./>
Subjective weight calculation is carried out based on an analytic hierarchy process, expert experience knowledge is considered in the weight calculation process, and the weighting result is relatively real and reliable. The method mainly comprises the following steps:
(1) Constructing a sapphire chemical mechanical polishing processing characteristic attribute comparison judgment matrix = [ T ] ij ] n×n . T in matrix ij The value of (1) is the importance of attribute i to attribute j;
(2) Consultation was made to an expert in the field of machine tool spindle numerical control sapphire chemical mechanical polishing, the invention was defined using a 1-9 ratio scale, as shown in table 2.
Table 2 compares ratio scale definition in a judgment matrix
(3) After normalizing each column of the judgment matrix T, the weight corresponding to each characteristic attribute can be obtained, and the calculation method is shown as follows:
in the method, in the process of the invention,
and (5) obtaining the ith characteristic attribute weight coefficient by using an AHP method.
Three decision theory are used for calculating objective weights, AHP is used for calculating subjective weights, and in order to enable the characteristic attribute weight values of the sapphire chemical mechanical polishing process to be more reasonable, the weight coefficients calculated by the two methods are calculated to be comprehensive weights through a linear weighting principle. Then for a certain characteristic attribute a i The comprehensive weight is as follows:
in the method, in the process of the invention,
the comprehensive weight coefficient of the characteristic attribute i; />
The weight value of the characteristic attribute i is obtained by using a subjective weighting method AHP method; />
The weight value of the characteristic attribute i is obtained by using three decision theories of an objective weighting method; ζ is a weighting factor, typically 0.5.
According to the characteristic attribute described by the sapphire chemical mechanical polishing process problem, the example characteristic attribute is classified according to different types, and the methods for calculating the similarity of the characteristic attributes of different types are different. The present invention classifies the characteristic properties into the following three types: numerical type attribute, fuzzy logic type attribute, irrelevant type attribute. The three types of local similarity calculation formulas are shown as follows:
wherein sim (x k ,y k ) Local similarity of the new process problem x and the characteristic attribute k in the old example y; x is x k The value of the characteristic attribute k in the new process problem x is taken; y is k The value of the characteristic attribute k in the old instance y is taken; MAX (k) is the maximum value of the feature attribute k; MIN (k) is the minimum value of the characteristic attribute k; m is the maximum difference of the values of the fuzzy logic type k.
Example 3: technological scheme optimization and example evaluation based on heterogeneous integrated learning.
When an instance meeting the threshold cannot be retrieved or the retrieved instance cannot meet the processing requirement, the process optimization module is enabled. In order to enhance the semantic interpretation of the process optimization process and reflect the nonlinear relation of the parameters of the sapphire chemical mechanical polishing process, the embodiment of the invention provides a process optimization algorithm based on heterogeneous integrated learning of a Decision Tree (DT) and a Deep Neural Network (DNN) algorithm, and a Particle Swarm Optimization (PSO) is adopted to optimize a machine learning and deep learning combination weight value, so that an optimal learning prediction result is obtained. The main algorithm flow is shown in fig. 6.
In the PSO algorithm, a test function is set to be a linear combination of a predicted result value of a decision tree algorithm and a predicted result value of a deep neural network, and the following formula is shown:
f(V DT ,V DNN )＝ω 1 ·V DT +ω 2 ·V DNN -V TRUE
wherein f (V) DT ,V DNN ) Is a test function; v (V) DT 、V DNN Representing a decision tree algorithm prediction result and a deep neural network algorithm prediction result respectively; v (V) TRUE Representing the actual value of the test set sample; omega 1 、ω 2 The weight vectors occupied by the decision tree algorithm prediction result and the deep neural network algorithm prediction result are represented respectively, and a group of corresponding weight values exist for each output variable.
To ensure that the data is suitable for practical use, the output of the system needs to be evaluated. The similarity derived from the data indicates how well the new process problem matches the instance in the instance library. It has some objectivity but is susceptible to noise data. The confidence is obtained through judgment of people and has certain subjectivity. In order to improve accuracy and anti-interference performance of example matching, a comprehensive evaluation method combining similarity, confidence and activity is adopted. Based on the similarity, confidence and activity, a comprehensive confidence factor E (C i ) The method comprises the following steps:
wherein ε (C) i ) Confidence for the i-th instance; a (C) i ) The activity level of the i-th example;
kappa, mu represent similarity and confidence pair integrated confidence factor E (C i ) The degree of influence of (2) and has +.>
Example 4:
the embodiment of the invention develops a sapphire chemical mechanical polishing process decision system program interaction interface based on Qt 4.8.7 and SQLite 3, and the interface comprises a user layer, a system algorithm layer, a data management layer, a data operation layer and a data storage layer, as shown in figure 7. The interface can realize the direct input of the technical problem information, and can also directly import new technical problems through texts. The system then performs a comprehensive confidence factor calculation, and the user will either receive recommended instances within a threshold from the existing instance database or acquire a process recipe using heterogeneous integrated learning. And detecting the surface quality and the precision of the processed workpiece through processing application, evaluating the new instance meeting the requirements, and then recovering the instance and storing the instance into a database for later use. The data and examples required by the whole flow process are sourced from the databases in fig. 4 and 5.
It should be noted that the embodiments of the present invention can be realized in hardware, software, or a combination of software and hardware. The hardware portion may be implemented using dedicated logic; the software portions may be stored in a memory and executed by a suitable instruction execution system, such as a microprocessor or special purpose design hardware. Those of ordinary skill in the art will appreciate that the apparatus and methods described above may be implemented using computer executable instructions and/or embodied in processor control code, such as provided on a carrier medium such as a magnetic disk, CD or DVD-ROM, a programmable memory such as read only memory (firmware), or a data carrier such as an optical or electronic signal carrier. The device of the present invention and its modules may be implemented by hardware circuitry, such as very large scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, etc., or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., as well as software executed by various types of processors, or by a combination of the above hardware circuitry and software, such as firmware.
The foregoing is merely illustrative of specific embodiments of the present invention, and the scope of the invention is not limited thereto, but any modifications, equivalents, improvements and alternatives falling within the spirit and principles of the present invention will be apparent to those skilled in the art within the scope of the present invention.
Claims (6)
1. The intelligent decision-making method for the sapphire chemical mechanical polishing processing technology is characterized by comprising the following steps of:
a chemical mechanical polishing intelligent process decision system framework, a chemical mechanical polishing processing basic database, a chemical mechanical polishing processing knowledge base, a chemical mechanical polishing processing problem definition, a chemical mechanical polishing processing example preference, a chemical mechanical polishing processing scheme prediction and a chemical mechanical polishing processing application;
the chemical mechanical polishing processing basic database comprises a polishing machine library, a polishing head library, a polishing pad library, a polishing liquid library, a material library and a finishing library;
the chemical mechanical polishing processing knowledge base comprises a chemical mechanical polishing processing instance base, a model base, an algorithm base and a rule base; the example library comprises process problem description, process problem solution, process problem application efficiency and evaluation, and the model library, algorithm library and rule library comprise corresponding information required by the intelligent decision process of the sapphire chemical mechanical polishing process;
the chemical mechanical polishing processing application comprises chemical mechanical polishing process example optimization and chemical mechanical polishing process scheme reasoning;
the chemical mechanical polishing processing problem definition is to modularly input basic information required for a subsequent intelligent process decision process through a process problem description part corresponding to an example library, wherein the basic information comprises sapphire size, thickness, material removal rate, surface roughness, subsurface damage and flatness; generating a process problem definition file after inputting basic information, splicing decision and application evaluation residual information in an instance recycling stage, forming a new process instance, and storing the new process instance in a chemical mechanical polishing processing knowledge base;
the chemical mechanical polishing process example is preferably realized by a comprehensive confidence factor based on the global similarity, confidence and activity calculated by three decision theory, analytic hierarchy process and nearest neighbor algorithm;
the reasoning of the chemical mechanical polishing process scheme is realized based on a prediction model combining heterogeneous integrated learning of a decision tree and a deep neural network and particle swarm optimization;
developing a chemical mechanical polishing processing example preferably, and if the processing effect of the example which does not meet the threshold setting or the example which is obtained by the preference cannot meet the processing requirement in the pre-experiment, predicting a chemical mechanical polishing processing scheme; in the optimization process of the chemical mechanical polishing processing example, a method of combining subjective and objective weights is adopted, wherein three decision theories are selected to calculate objective weight values of all attributes, a chromatographic analysis method is selected to calculate subjective weight values of all the attributes, and a linear weighting principle is adopted to obtain comprehensive characteristic attribute weight values; calculating the local similarity value of the attribute between the new process problem and each instance by adopting a nearest neighbor algorithm, then calculating the overall similarity value of the new process problem and each instance by combining the weight of each characteristic attribute, arranging the values in descending order, and outputting the instance number meeting the set threshold value through instance preference; when an instance meeting a threshold cannot be searched or the searched instance cannot meet the processing requirement, a process scheme prediction module is started, a process optimization algorithm based on heterogeneous integrated learning of a decision tree and a deep neural network algorithm is adopted to optimize a combined weight value of machine learning and deep learning by adopting a particle swarm algorithm, an optimal learning prediction result is obtained, and a final processing scheme is formed by combining instance optimization;
the intelligent decision method for the sapphire chemical mechanical polishing processing technology comprises the following steps:
step one, building a sapphire chemical mechanical polishing intelligent process decision system frame based on a 7R model;
step two, a perfect chemical mechanical polishing database is established based on the database structure;
thirdly, carrying out example optimization by combining three decision theories, an analytic hierarchy process and comprehensive confidence factors of global similarity, confidence and activity calculated by a nearest neighbor algorithm;
establishing a prediction model combining heterogeneous integrated learning of a decision tree and a deep neural network and optimization of a particle swarm algorithm to obtain an optimal polishing process scheme output;
and fifthly, developing a sapphire chemical mechanical polishing processing technology decision system based on Qt 4.8.7 and SQLite 3 and applying the system to a domestic numerical control polishing machine tool.
2. The method of claim 1, wherein the 7R model in step one includes instance expression, instance reuse, instance retrieval, instance reasoning, instance modification, instance reclamation, and instance reclassification.
3. A sapphire chemical mechanical polishing process intelligent decision system applying the sapphire chemical mechanical polishing process intelligent decision method according to any one of claims 1-2, characterized in that the sapphire chemical mechanical polishing process intelligent decision system comprises: the system comprises a user layer, a system algorithm layer, a data management layer, a data operation layer and a data storage layer; the method is used for inputting the process problem information of the workpiece type and the material type and importing new process problems through texts; the system executes comprehensive confidence factor calculation, and a user receives recommended examples in a threshold value from the existing example database or acquires a processing technological scheme by utilizing heterogeneous integrated learning;
the user layer comprises a computer end, a processing machine tool terminal, a man-machine interaction interface and a user end;
the system algorithm layer comprises a grinding process example optimization algorithm and a grinding process scheme reasoning algorithm; the grinding process example optimization algorithm comprises a rough set, a three-branch decision, a CRITIC method and an AHP method, and the grinding process scheme reasoning algorithm comprises a neural network and a support vector machine;
the data management layer comprises basic data management, experience data management, decision data management and data security management; the basic data management comprises a machine tool library, a material library, a polishing solution library, a polishing pad library and a finishing library, the experience data management comprises an instance library, an algorithm library, a model library and a rule library, the decision data management comprises processing technology information, technology instance reasoning, technology intelligent reasoning and technology scheme output, and the data security management comprises user management, authority management, data backup and security log;
the data operation layer comprises data access/file operation;
the data storage layer comprises a basic database, a process knowledge base and a user database.
4. A computer device, characterized in that the computer device comprises a memory and a processor, the memory storing a computer program, which when executed by the processor, causes the processor to execute the steps of the intelligent decision method for the sapphire chemical mechanical polishing process according to any of claims 1-2.
5. A computer readable storage medium storing a computer program which, when executed by a processor, causes the processor to perform the steps of the intelligent decision method of a sapphire chemical mechanical polishing process according to any of claims 1-2.
6. An information data processing terminal, wherein the information data processing terminal is used for realizing the intelligent decision system of the sapphire chemical mechanical polishing processing technology according to claim 3.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210816729.0A CN115310623B (en) | 2022-07-12 | 2022-07-12 | Intelligent decision-making method, system and terminal for sapphire chemical mechanical polishing processing technology |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210816729.0A CN115310623B (en) | 2022-07-12 | 2022-07-12 | Intelligent decision-making method, system and terminal for sapphire chemical mechanical polishing processing technology |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115310623A CN115310623A (en) | 2022-11-08 |
| CN115310623B true CN115310623B (en) | 2023-05-12 |
Family
ID=83856136
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210816729.0A Active CN115310623B (en) | 2022-07-12 | 2022-07-12 | Intelligent decision-making method, system and terminal for sapphire chemical mechanical polishing processing technology |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115310623B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117067096B (en) * | 2023-10-18 | 2023-12-15 | 苏州博宏源机械制造有限公司 | Automatic control system and method for double-sided grinding and polishing equipment based on parameter optimization |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104155914A (en) * | 2014-09-01 | 2014-11-19 | 湘潭大学 | CMP process intelligent decision making system for polishing carbide blade |
| CN109799787A (en) * | 2019-01-10 | 2019-05-24 | 湖南科技大学 | Smart camshaft grinding process software database system based on digital control system |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5917726A (en) * | 1993-11-18 | 1999-06-29 | Sensor Adaptive Machines, Inc. | Intelligent machining and manufacturing |
| CN109858775A (en) * | 2019-01-10 | 2019-06-07 | 湖南科技大学 | The integrated application method of smart camshaft grinding process software database system based on digital control system |
| CN109857395B (en) * | 2019-01-10 | 2022-09-09 | 湖南科技大学 | Integrated application method of intelligent camshaft grinding process software database system based on open numerical control system |
| CN110216553B (en) * | 2019-05-22 | 2022-03-25 | 湖南科技大学 | Intelligent decision cloud service method and system for numerical control grinding of camshaft |
-
2022
- 2022-07-12 CN CN202210816729.0A patent/CN115310623B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104155914A (en) * | 2014-09-01 | 2014-11-19 | 湘潭大学 | CMP process intelligent decision making system for polishing carbide blade |
| CN109799787A (en) * | 2019-01-10 | 2019-05-24 | 湖南科技大学 | Smart camshaft grinding process software database system based on digital control system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115310623A (en) | 2022-11-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Sharp et al. | A survey of the advancing use and development of machine learning in smart manufacturing | |
| Köksal et al. | A review of data mining applications for quality improvement in manufacturing industry | |
| Roy et al. | Best order sort: a new algorithm to non-dominated sorting for evolutionary multi-objective optimization | |
| Xu et al. | Hybrid feature selection for wafer acceptance test parameters in semiconductor manufacturing | |
| CN115310623B (en) | Intelligent decision-making method, system and terminal for sapphire chemical mechanical polishing processing technology | |
| Karthik et al. | Prognostic Kalman Filter Based Bayesian Learning Model for Data Accuracy Prediction. | |
| He et al. | Modeling and analyses of energy consumption for machining features with flexible machining configurations | |
| Huang et al. | The rough set based approach to generic routing problems: case of reverse logistics supplier selection | |
| Wu et al. | Research on segmenting e-commerce customer through an improved k-medoids clustering algorithm | |
| Zhang et al. | Hierarchical multistrategy genetic algorithm for integrated process planning and scheduling | |
| Ouared et al. | Deepcm: Deep neural networks to improve accuracy prediction of database cost models | |
| Mahiri et al. | Data-driven sustainable smart manufacturing: A conceptual framework | |
| Scheinert et al. | On the potential of execution traces for batch processing workload optimization in public clouds | |
| He et al. | Integrated carbon footprint with cutting parameters for production scheduling | |
| Kashkoush et al. | An integer programming model for discovering associations between manufacturing system capabilities and product features | |
| Esmaeili et al. | Variable reduction for multi-objective optimization using data mining techniques; application to aerospace structures | |
| Zhang et al. | Credit evaluation of SMEs based on GBDT-CNN-LR hybrid integrated model | |
| Telegraphi et al. | A mathematical model for the sustainable design of a cellular manufacturing system in the tactical planning of a closed-loop supply chain featuring alternative routings and outsourcing option | |
| Zhang | Enterprise Supply Chain Risk Assessment Based on the Support Vector Machine Algorithm and Fuzzy Model | |
| Puangpontip et al. | On Using Deep Learning for Business Analytics: At what cost? | |
| Su et al. | Efficient machine layout design method with a fuzzy set theory within a bay in a TFT-LCD plant | |
| Wu et al. | Application of Improved Feature Pre-processing Method in Prevention and Control of Electricity Charge Risk | |
| Chou et al. | An interactive method for multi-criteria dispatching problems with unknown preference functions | |
| Ailisto et al. | Benefits of Machine Learning in the Manufacturing Industry | |
| Zhang | Synergistic advantages of deep learning and reinforcement learning in economic forecasting |
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 |