CN112100745B - Automobile girder steel mechanical property prediction method based on LDA theory - Google Patents
Automobile girder steel mechanical property prediction method based on LDA theory Download PDFInfo
- Publication number
- CN112100745B CN112100745B CN202010966431.9A CN202010966431A CN112100745B CN 112100745 B CN112100745 B CN 112100745B CN 202010966431 A CN202010966431 A CN 202010966431A CN 112100745 B CN112100745 B CN 112100745B
- Authority
- CN
- China
- Prior art keywords
- data
- sample set
- data sample
- parameters
- mechanical property
- 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
- 229910000831 Steel Inorganic materials 0.000 title claims abstract description 53
- 239000010959 steel Substances 0.000 title claims abstract description 53
- 238000000034 method Methods 0.000 title claims abstract description 50
- 238000004519 manufacturing process Methods 0.000 claims abstract description 45
- 238000005096 rolling process Methods 0.000 claims abstract description 32
- 238000012549 training Methods 0.000 claims abstract description 26
- 230000008569 process Effects 0.000 claims abstract description 16
- 239000002994 raw material Substances 0.000 claims description 9
- 239000000126 substance Substances 0.000 claims description 9
- 230000006870 function Effects 0.000 claims description 7
- 229910000742 Microalloyed steel Inorganic materials 0.000 claims description 6
- 238000009825 accumulation Methods 0.000 claims description 6
- 238000004458 analytical method Methods 0.000 claims description 6
- 238000012706 support-vector machine Methods 0.000 claims description 5
- 238000004422 calculation algorithm Methods 0.000 claims description 3
- 229910052799 carbon Inorganic materials 0.000 claims description 3
- 238000010079 rubber tapping Methods 0.000 claims description 3
- 229910052720 vanadium Inorganic materials 0.000 claims description 3
- 238000004140 cleaning Methods 0.000 claims description 2
- 238000010219 correlation analysis Methods 0.000 claims description 2
- 230000000694 effects Effects 0.000 claims description 2
- 238000000605 extraction Methods 0.000 claims description 2
- 238000010801 machine learning Methods 0.000 claims description 2
- 238000012163 sequencing technique Methods 0.000 claims description 2
- 238000005516 engineering process Methods 0.000 description 6
- 238000009826 distribution Methods 0.000 description 5
- 238000012545 processing Methods 0.000 description 5
- 238000004364 calculation method Methods 0.000 description 4
- 238000001514 detection method Methods 0.000 description 4
- 238000011161 development Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 239000011159 matrix material Substances 0.000 description 4
- 239000010955 niobium Substances 0.000 description 4
- 238000011160 research Methods 0.000 description 4
- 238000007418 data mining Methods 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 230000009467 reduction Effects 0.000 description 2
- 238000012216 screening Methods 0.000 description 2
- 238000007619 statistical method Methods 0.000 description 2
- 239000010936 titanium Substances 0.000 description 2
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 description 1
- 229910000677 High-carbon steel Inorganic materials 0.000 description 1
- 229910001209 Low-carbon steel Inorganic materials 0.000 description 1
- NINIDFKCEFEMDL-UHFFFAOYSA-N Sulfur Chemical compound [S] NINIDFKCEFEMDL-UHFFFAOYSA-N 0.000 description 1
- RTAQQCXQSZGOHL-UHFFFAOYSA-N Titanium Chemical compound [Ti] RTAQQCXQSZGOHL-UHFFFAOYSA-N 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 238000009749 continuous casting Methods 0.000 description 1
- 238000005520 cutting process Methods 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000012217 deletion Methods 0.000 description 1
- 230000037430 deletion Effects 0.000 description 1
- 238000005098 hot rolling Methods 0.000 description 1
- 238000009776 industrial production Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 229910052758 niobium Inorganic materials 0.000 description 1
- GUCVJGMIXFAOAE-UHFFFAOYSA-N niobium atom Chemical compound [Nb] GUCVJGMIXFAOAE-UHFFFAOYSA-N 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 239000002436 steel type Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 229910052717 sulfur Inorganic materials 0.000 description 1
- 239000011593 sulfur Substances 0.000 description 1
- 229910052719 titanium Inorganic materials 0.000 description 1
- LEONUFNNVUYDNQ-UHFFFAOYSA-N vanadium atom Chemical compound [V] LEONUFNNVUYDNQ-UHFFFAOYSA-N 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/15—Vehicle, aircraft or watercraft design
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
- G06F18/2132—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods based on discrimination criteria, e.g. discriminant analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/17—Mechanical parametric or variational design
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
- G06N20/10—Machine learning using kernel methods, e.g. support vector machines [SVM]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2113/00—Details relating to the application field
- G06F2113/26—Composites
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2113/00—Details relating to the application field
- G06F2113/28—Fuselage, exterior or interior
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/14—Force analysis or force optimisation, e.g. static or dynamic forces
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Geometry (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Software Systems (AREA)
- Computer Hardware Design (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Mathematical Analysis (AREA)
- Evolutionary Biology (AREA)
- Computational Mathematics (AREA)
- Medical Informatics (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- Bioinformatics & Computational Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Automation & Control Theory (AREA)
- Computing Systems (AREA)
- Aviation & Aerospace Engineering (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- General Factory Administration (AREA)
Abstract
The invention provides a method for predicting mechanical properties of automobile girder steel based on an LDA theory, and relates to the technical field of rolling production of automobile girder steel. The invention provides a method for establishing a predictive model of mechanical properties (yield strength, tensile strength and elongation) of steel. A large amount of historical production data can be generated in the production process of the hot-rolled high-strength steel, and a model training data sample set is constructed according to the historical production data; the training data sample set comprises carrying characteristic attributes (process and component parameters) and corresponding mechanical performance parameters; and (3) establishing a mechanical property prediction model by using the training data sample set, and finally inputting characteristic attribute parameters (process and component parameters) of the prediction data sample set into the mechanical property prediction model to obtain the mechanical property of the prediction data sample.
Description
Technical Field
The invention relates to the technical field of rolling production of automobile girder steel, in particular to an automobile girder steel mechanical property prediction method based on an LDA theory.
Background
In recent years, with the rapid development of computer technology and the wide application of distributed control systems in the steel industry, a large amount of data generated in the steel production process is collected and stored, and the data comprehensively reflects the internal links in the steel production process, so that the distributed control system has great application value. But how to mine useful information from a large amount of data to realize accurate prediction and stability control of product performance becomes an important trend of the development of hot-rolled strip production technology.
The linear discriminant theory LDA is a supervised learning data dimension reduction technique, that is, each sample of its dataset is class-output. The support vector machine regression method SVM is established based on a nonlinear mapping theory, and an inner product function is utilized to replace nonlinear mapping to a high-dimensional space. The core of the SVM method is the idea of maximizing classification margin, and the optimal hyperplane for feature space division is the target of the SVM. The modeling method is based on the fact that the original data sample set is mined, and a high-precision high-strength steel mechanical property online prediction model is built.
At present, development of a hot rolled strip steel mechanical property prediction model has become an important problem in the field of steel rolling. The prior China patent 200410061324.2 discloses a microstructure and mechanical property forecasting system of a hard wire product, and the patent researches a temperature, structure and property forecasting system of a continuous casting billet for directly rolling a high-carbon steel wire rod, and establishes a microstructure and mechanical property forecasting model system of the hard wire product; the Chinese patent 02109026.2 discloses a method for predicting the tissue evolution and performance of strip steel in the rolling process, and the method researches the tissue evolution process of the strip steel in the rolling process and obtains a software model for predicting the tissue evolution and the performance of the strip steel; chinese patent 200710052007.8 discloses a method for predicting the organization and mechanical properties of hot rolled Nb-containing strip steel, which is based on a physical metallurgical model, combines simulation experiments with industrial production data and establishes a mechanical property prediction model of the Nb-containing strip steel.
From the metallurgical mechanism perspective, the above patent analyzes the tissue evolution process in the strip steel rolling process and establishes a theoretical model for mechanical property prediction. Research on hot rolled strip steel mechanical property prediction technology is getting more and more attention from scientific researchers. With the continuous development of production technology, a plurality of low-carbon steel mechanical property prediction models based on neural networks, multi-objective optimization algorithms and the like are established on the basis of collecting a large amount of historical production data. However, for high-strength steel, a targeted prediction model is established due to large fluctuation of mechanical properties, and the model is used for realizing online prediction of the mechanical properties of the high-strength steel through real-time acquisition of production data. The predicted value is used for replacing the detection value, so that the sampling times can be reduced, the shipment time of the strip steel products is shortened, and the time and the cost are saved.
Disclosure of Invention
Aiming at the problems existing in the prior art, the invention provides the method for predicting the mechanical properties of the automobile girder steel based on the LDA theory, which can effectively improve the mechanical property detection efficiency, reduce the detection cost and realize the high efficiency and convenience of the automobile girder steel detection efficiency.
The technical scheme adopted by the invention is as follows:
an automobile girder steel mechanical property prediction method based on an LDA theory comprises the following steps:
step 1, acquiring historical production data of hot-rolled microalloyed steel;
building a data platform for the production data of the automobile girder steel by means of a hot continuous rolling production line to obtain the historical production data of the hot-rolled microalloyed steel;
the historical production data comprise raw material parameters, process parameters and measured mechanical properties; wherein the raw material parameters comprise chemical components, raw material thickness and tapping temperature; the technological parameters comprise rolling temperature, rolling force, final rolling thickness, final rolling temperature, final rolling speed and coiling temperature of each rolling mill;
step 2, constructing a model training data sample set based on historical production data of the hot rolled microalloyed steel;
carrying out characteristic attribute analysis on the historical production data by adopting a linear discriminant theory to obtain a training data sample set;
step 2.1, performing data cleaning work on historical production data;
firstly, incomplete data samples in historical production data are removed, the influence of characteristic attributes on mechanical properties is analyzed, and parameters with the accumulated contribution rate to the mechanical properties being more than 90% are screened out according to correlation analysis, so that a data sample set with complete characteristic attributes is obtained; the characteristic attribute comprises chemical components and process parameters; the chemical components comprise C, si, mn, ti, V, nb; the process parameters include finish rolling temperature (FDT), coiling Temperature (CT) and intermediate billet thickness (RDT);
the mechanical properties include yield strength, tensile strength, and elongation;
step 2.2, obtaining a training data sample set according to the data sample set with complete characteristic attributes;
2.2.1, determining information entropy and mutual information value of the data sample set with complete characteristic attribute by utilizing a linear discriminant theory;
2.2.2, calculating the chemical components and the process parameters of the original production data as well as the information entropy between the yield strength, the tensile strength and the elongation, sequencing the information entropy values from large to small, calculating the characteristic attribute accumulation contribution rate, and selecting the parameters with the characteristic attribute accumulation contribution rate of the data sample more than 98% as input parameters for model establishment;
the linear discriminant theory is a feature extraction method, which projects a high-dimensional data sample into an optimal discrimination vector space to achieve the effects of extracting classification information and compressing feature space dimensions, and ensures that a mode sample has the largest inter-class distance and the smallest intra-class distance in a new subspace after projection, namely, the model has optimal separability in the space.
Step 3, utilizing characteristic attributes contained in the training data sample set as input parameters of a prediction model, and utilizing yield strength, tensile strength and elongation corresponding to the characteristic attributes as output parameters of the prediction model, so as to obtain a mechanical property prediction model corresponding to the automobile girder steel; collecting production data with the same data structure as a predicted data sample set, and obtaining predicted values of yield strength, tensile strength and elongation percentage through the predicted data sample set;
the mechanical property prediction model is established based on a machine learning algorithm, a support vector machine regression (SVM) method is used, the SVM has the characteristic of fitting regression, an inner product function of the SVM is utilized for dividing an optimal hyperplane, and the mechanical property prediction model of the automobile girder steel is established based on the SVM.
The beneficial effects of adopting above-mentioned technical scheme to produce lie in:
the invention provides an automobile girder steel mechanical property prediction method based on an LDA theory, which takes high-strength steel production data as a research object, combines the production process and component characteristics of a factory, establishes an online prediction model for the production data of steel types, and realizes the real-time prediction of the high-strength steel mechanical property.
According to the invention, the mechanical property parameter prediction model of the steel is established by analyzing historical production data in the conventional steel production process, and the mechanical property is obtained according to the process, the component parameters and the mechanical property prediction model of the steel. Therefore, the established online prediction model for the mechanical properties of the hot-rolled strip steel has the characteristics of high calculation speed, no need of cutting samples, labor and material cost saving and the like, and can effectively improve the determination efficiency of the 610L automobile girder steel performance parameters.
Drawings
FIG. 1 is a flow chart of an overall method for predicting mechanical properties of automobile girder steel according to the invention;
FIG. 2 is a graph showing comparison of linear discriminant analysis results in accordance with an embodiment of the present invention;
FIG. 3 is a schematic diagram of the attribute values and accumulated contribution rates of the present invention;
FIG. 4 is a schematic diagram and calculation flow of a support vector regression method according to an embodiment of the present invention;
in the figure, (a) -a computational flow chart of a support vector regression method; (b) -a computational schematic diagram of a support vector regression method;
FIG. 5 is a graph of statistical analysis of accuracy of prediction results of mechanical properties according to an embodiment of the present invention;
in the figure, (a) -a yield strength prediction accuracy graph and (b) -a yield strength relative error distribution graph; (c) -a tensile strength prediction accuracy map, (d) -a tensile strength versus error map; (e) -an elongation prediction accuracy map, (f) -an elongation versus error map.
Detailed Description
The following describes the embodiments of the present invention in detail with reference to the drawings.
An automobile girder steel mechanical property prediction method based on LDA theory, as shown in figure 1, comprises the following steps:
step 1: the construction of the data platform is the basis of the later data mining, and the selection of the proper data is necessary to fully consider the requirement of the later data mining and the existing potential requirement, so that the establishment of the proper data platform is important. In the embodiment, a 2250 hot rolling production line is used for acquiring historical production data in the 610L automobile girder steel production process. And constructing a data platform by adopting an Oracle database technology, storing data acquired in the production process into a database in real time, and then matching historical production data according to steel coil numbers through the database technology to obtain an original data table. The collected historical production data includes: raw material parameters, process parameters and measured mechanical properties; wherein, the raw material parameters include: chemical composition, raw material thickness and tapping temperature; the technological parameters include: rolling temperature, rolling force, final rolling thickness, final rolling temperature, final rolling speed and coiling temperature of each rolling mill; the training data sample set comprises components of characteristic attributes carried by the data sample, technological parameters and mechanical performance parameters corresponding to the specific attributes carried by the data sample.
As shown in table 1, the data provided by the examples of this patent were analyzed by preprocessing.
TABLE 1 distribution of Attribute parameters for 610L high-strength Steel data samples
Step 2: because incomplete data information data samples possibly exist in the historical production data, the original sample set is preprocessed in order to ensure that the training data sample set has higher data information quality. And then analyzing the data distribution condition of the data sample set, and laying an important foundation for the subsequent model parameter screening, outlier processing and parameter rule analysis.
As shown in fig. 2 and 3, the embodiment of the present patent provides a linear discriminant theory method for data processing, and performs statistics on the attribute accumulation contribution rate of the data sample features.
Step 3: as for specific data, when the correspondence between the input parameters and the output parameters of the model is studied by using the established model, the model may be predicted to deviate seriously from the facts, so high-quality modeling data is the basis of industrial big data mining. For the collected data sample set, firstly, data sample information normalization and integrity processing is needed, data information characteristic attribute deletion and noise data are subjected to preliminary screening to obtain an intermediate data sample set, and then the intermediate data sample is processed by using a linear discriminant analysis method. Thereby obtaining a training data set with less data dimension and high quality sample characteristic information. Taking the data processing result of the elongation as an example, 24 input parameters of the original data sample are subjected to linear discriminant analysis processing, and the data characteristic attribute with the characteristic attribute accumulation of the data sample more than 98% is selected as the input parameter for model establishment, so that the characteristic attribute parameters of the training data sample are reduced to 14.
As shown in fig. 4 and 5, the embodiment of the present patent provides a mechanical performance modeling method, and performs statistical analysis of model accuracy, where fig. 4 (a) is a calculation flowchart of a support vector regression method; (b) a computational schematic diagram of a support vector regression method; FIG. 5 (a) is a graph of yield strength prediction accuracy, and (b) is a graph of yield strength versus error; (c) A tensile strength prediction accuracy graph, and (d) a tensile strength relative error distribution graph; (e) An elongation prediction accuracy map, and (f) an elongation relative error distribution map.
Step 4: and constructing a mechanical property prediction model based on a support vector machine regression model, and training the prediction model by using a training sample set to obtain a trained prediction model. The training data set original characteristic attribute set is as follows: a finish rolling inlet temperature (FET), a finish rolling average temperature value (FDT), a finish rolling average thickness value (FDH), a coiling average temperature value (CT), a carbon content (C), a sulfur content (S), a niobium content (Nb), a vanadium content (V), a titanium content (Ti) and the like.
The prediction data sample set comprises characteristic attributes (components and process parameters) and mechanical performance parameters corresponding to the characteristic attributes, the characteristic attributes (components and process parameters) of the prediction data sample set are taken as input parameters of a model, the yield strength of the steel is predicted through the prediction model, meanwhile, the actual yield strength of the steel is detected, the fault tolerance value is +/-6, the accuracy of the statistical yield strength prediction model can reach 92.2%, the accuracy is high, and good prediction of the yield strength of the steel can be achieved. The prediction accuracy of the tensile strength is up to 97.7%, and the prediction accuracy of the elongation is up to 94.9%. The accuracy of the mechanical property parameter prediction according to the mechanical property parameter prediction method is higher, and the mechanical property parameter prediction can be well realized. And calculating information entropy of the data attribute according to the data sample set, calculating conditional entropy of each attribute and the mechanical property target value on the basis of the information entropy, determining the conditional relation of each attribute information, and finally determining the training data sample set required by model input.
For a given d-dimensional data sample setFor the ith sample, N is the sample data size, x i Is an attribute parameter, y i The specific calculation steps of the step 2.2.2 are as follows:
step S1: calculating an intra-class divergence matrix S w :
x is sample data in class i, u i Is the mean value of the i-th sample;
step S2: calculating an inter-class divergence matrix S b :
Wherein m is i For the number of samples of class i, u is the mean vector of all samples;
step S3: computing a matrixFor->Singular value decomposition is carried out to obtain singular value lambda i And the corresponding specialSign vector omega i ,i=1,2,…,N-1;
Step S4: taking a feature vector corresponding to a singular value with a large dimension to be reduced to form a projection matrix W;
step S5: calculating each sample x in the sample set i Projection z in new low-dimensional space i ;
z i =W T x i
Step S6: obtaining a sample set after dimension reduction
A600 MPa-level automobile girder steel mechanical property prediction model is established based on a support vector regression method, the support vector machine regression method is a supervised learning method, and on the premise of knowing the category of training points, the corresponding relation between the training points and the category is established. Based on this, the category corresponding to the new sample point is predicted. The SVM is a method for learning and predicting a data sample set, and can better solve the over-learning problem which can not be solved by a neural network.
The specific steps of establishing the mechanical property prediction model are as follows:
step D1: let training data sample set t= { (x) i ,y i ),(x 2 ,y 2 ),…,(x N ,y N )};
Step D2: selecting a kernel function K and a penalty parameter C >0, constructing and solving a convex quadratic programming problem;
wherein alpha is i 、α j Is Lagrangian multiplier, y i 、y j Is the mechanical property target value. j is the j sample of {1,2,3, … }, and the Lagrangian multiplier optimal solution is obtained
Step D3: select alpha * Is a component of (a)Satisfy condition->Calculate->
Wherein,,is the ith Lagrangian multiplier, b * For component->Corresponding intercept;
step D4: training decision function f (x):
wherein K (x, x i ) The Gaussian kernel function is adopted, and the expression is as follows:
where σ is the standard deviation of the sample data, therefore, the expression of the training decision function is:
in this embodiment, a 610L mechanical property prediction model of the automobile girder steel is established, and a training data sample set carrying specific properties of sample information and carrying mechanical property target parameters corresponding to specific properties of an original sample data set is used to obtain a mechanical property prediction model after training.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present invention, and not for limiting the same; although the invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some or all of the technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit of the corresponding technical solutions, which are defined by the scope of the appended claims.
Claims (2)
1. A method for predicting mechanical properties of automobile girder steel based on LDA theory is characterized by comprising the following steps: the method comprises the following steps:
step 1, acquiring historical production data of hot-rolled microalloyed steel;
building a data platform for the production data of the automobile girder steel by means of a hot continuous rolling production line to obtain the historical production data of the hot-rolled microalloyed steel;
the historical production data comprise raw material parameters, process parameters and measured mechanical properties; wherein the raw material parameters comprise chemical components, raw material thickness and tapping temperature; the technological parameters comprise rolling temperature, rolling force, final rolling thickness, final rolling temperature, final rolling speed and coiling temperature of each rolling mill;
step 2, constructing a model training data sample set based on historical production data of the hot rolled microalloyed steel;
carrying out characteristic attribute analysis on the historical production data by adopting a linear discriminant theory to obtain a training data sample set;
step 2.1, performing data cleaning work on historical production data;
firstly, incomplete data samples in historical production data are removed, the influence of characteristic attributes on mechanical properties is analyzed, and parameters with the accumulated contribution rate to the mechanical properties being more than 90% are screened out according to correlation analysis, so that a data sample set with complete characteristic attributes is obtained; the characteristic attribute comprises chemical components and process parameters; the chemical components comprise C, si, mn, ti, V, nb; the process parameters comprise a finish rolling temperature FDT, a coiling temperature CT and a thickness RDT of the intermediate blank;
the mechanical properties include yield strength, tensile strength, and elongation;
step 2.2, obtaining a training data sample set according to the data sample set with complete characteristic attributes;
2.2.1, determining information entropy and mutual information value of the data sample set with complete characteristic attribute by utilizing a linear discriminant theory;
2.2.2, calculating the chemical components and the process parameters of the original production data as well as the information entropy between the yield strength, the tensile strength and the elongation, sequencing the information entropy values from large to small, calculating the characteristic attribute accumulation contribution rate, and selecting the parameters with the characteristic attribute accumulation contribution rate of the data sample more than 98% as input parameters for model establishment;
the linear discriminant theory is a feature extraction method, which projects a high-dimensional data sample into an optimal discrimination vector space to achieve the effects of extracting classification information and compressing feature space dimensions, and ensures that a mode sample has the largest inter-class distance and the smallest intra-class distance in a new subspace after projection, namely, the model has optimal separability in the space;
step 3, utilizing characteristic attributes contained in the training data sample set as input parameters of a prediction model, and utilizing yield strength, tensile strength and elongation corresponding to the characteristic attributes as output parameters of the prediction model, so as to obtain a mechanical property prediction model corresponding to the automobile girder steel; and collecting production data with the same data structure as a predicted data sample set, and obtaining predicted values of yield strength, tensile strength and elongation through the predicted data sample set.
2. The method for predicting the mechanical properties of the automobile girder steel based on the LDA theory according to claim 1, wherein the mechanical property prediction model in the step 3 is established based on a machine learning algorithm, a support vector machine regression method SVM is used, the SVM has the characteristic of fitting regression, an inner product kernel function of the SVM is utilized for dividing an optimal hyperplane, and a 610L automobile girder steel mechanical property prediction model is established based on the model.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010966431.9A CN112100745B (en) | 2020-09-15 | 2020-09-15 | Automobile girder steel mechanical property prediction method based on LDA theory |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010966431.9A CN112100745B (en) | 2020-09-15 | 2020-09-15 | Automobile girder steel mechanical property prediction method based on LDA theory |
Publications (2)
Publication Number | Publication Date |
---|---|
CN112100745A CN112100745A (en) | 2020-12-18 |
CN112100745B true CN112100745B (en) | 2023-10-13 |
Family
ID=73758736
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010966431.9A Active CN112100745B (en) | 2020-09-15 | 2020-09-15 | Automobile girder steel mechanical property prediction method based on LDA theory |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN112100745B (en) |
Families Citing this family (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113076629B (en) * | 2021-03-19 | 2024-06-11 | 张家港荣盛特钢有限公司 | Control method for heat treatment process of high-strength wire rod |
CN113118458B (en) * | 2021-04-20 | 2023-04-07 | 江西省科学院应用物理研究所 | Prediction method for tensile property of metal component formed by selective laser melting |
CN114065629A (en) * | 2021-11-17 | 2022-02-18 | 中冶赛迪重庆信息技术有限公司 | Wire rod mechanical property prediction system |
CN115415542B (en) * | 2022-07-19 | 2024-05-03 | 闽都创新实验室 | Prediction method for performance of duplex stainless steel 3D printing piece based on response surface method |
CN115034121B (en) * | 2022-08-11 | 2022-10-25 | 太原科技大学 | Strip steel process regulation and control method based on organization performance intelligent prediction model |
CN116682505B (en) * | 2023-06-14 | 2023-11-21 | 东北大学 | HRB400E steel mechanical property prediction method based on quantile regression forest |
CN116822367A (en) * | 2023-07-03 | 2023-09-29 | 东北大学 | IF steel mechanical property prediction method based on LSTM model |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105740467A (en) * | 2016-03-07 | 2016-07-06 | 东北大学 | Mining method for C-Mn steel industry big data |
CN111191822A (en) * | 2019-12-18 | 2020-05-22 | 河钢股份有限公司承德分公司 | Steel parameter prediction method and terminal equipment |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106971026B (en) * | 2017-03-08 | 2020-06-16 | 武汉科技大学 | Micro-alloy steel mechanical property forecasting method based on global additive model |
-
2020
- 2020-09-15 CN CN202010966431.9A patent/CN112100745B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105740467A (en) * | 2016-03-07 | 2016-07-06 | 东北大学 | Mining method for C-Mn steel industry big data |
CN111191822A (en) * | 2019-12-18 | 2020-05-22 | 河钢股份有限公司承德分公司 | Steel parameter prediction method and terminal equipment |
Non-Patent Citations (5)
Title |
---|
High-Dimensional Data-Driven Optimal Design for Hot Strip Rolling of Microalloyed Steel;Cao guangming;《 Wiley Online Library》;第89卷(第7期);1800015 * |
基于LDA的工科科研主题识别与可视化研究;陆阳琪;;江苏科技信息(第22期);26-29 * |
基于均匀设计响应面法的桥梁地震易损性分析;李立峰;李辉辉;徐开铎;胡思聪;;公路交通科技(第11期);104-113 * |
基于大数据的力学性能预测与工艺参数筛选;吴思炜;刘振宇;周晓光;史乃安;;钢铁研究学报;28(第12期);4-7 * |
热轧高强钢氧化动力学和氧化铁皮结构控制;曹光明等;《材料热处理学报》;第35卷(第11期);161-167 * |
Also Published As
Publication number | Publication date |
---|---|
CN112100745A (en) | 2020-12-18 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN112100745B (en) | Automobile girder steel mechanical property prediction method based on LDA theory | |
Ji et al. | A machine learning and genetic algorithm-based method for predicting width deviation of hot-rolled strip in steel production systems | |
CN111241657B (en) | Cold-rolled strip steel plate shape prediction method based on deep learning | |
US11755976B2 (en) | Method for predicting oxygen load in iron and steel enterprises based on production plan | |
CN113326960B (en) | Subway traction energy consumption prediction method based on particle swarm optimization LSTM | |
CN111444942B (en) | Intelligent forecasting method and system for silicon content of blast furnace molten iron | |
CN110782067B (en) | Sintering end point fluctuation range prediction method based on fuzzy information granulation | |
CN114897227A (en) | Multi-steel-grade mechanical property forecasting method based on improved random forest algorithm | |
CN111250548B (en) | Board convexity prediction method based on kernel partial least square combined support vector machine | |
CN109726517A (en) | A kind of equipment method for predicting residual useful life based on multivariable associated data | |
Hmamouche et al. | A causality based feature selection approach for multivariate time series forecasting | |
CN115169453A (en) | Hot continuous rolling width prediction method based on density clustering and depth residual error network | |
CN116484745A (en) | Design method of hot-rolled low-alloy steel for physical metallurgy guidance industry big data mining | |
CN115860211A (en) | Casting blank quality prediction method based on local online modeling | |
Zhao et al. | Prediction of mechanical properties of cold rolled strip based on improved extreme random tree | |
CN112818030B (en) | Self-adaptive mining method for association rules of distributed stream data | |
CN115034437A (en) | Hot rolled plate convexity prediction method based on improved XGboost | |
CN112418522B (en) | Industrial heating furnace steel temperature prediction method based on three-branch integrated prediction model | |
Yan et al. | Rolling force prediction during FGC process of tandem cold rolling based on IQGA-WNN ensemble learning | |
Sun et al. | Application of time series based SVM model on next-day electricity price forecasting under deregulated power market | |
CN116825253B (en) | Method for establishing hot rolled strip steel mechanical property prediction model based on feature selection | |
Dong et al. | Just-in-time learning-based soft sensor for mechanical properties of strip steel via multi-block weighted semisupervised models | |
CN115034554A (en) | Continuous casting billet quality prediction method and system based on time sequence network | |
CN115392104A (en) | Method for predicting mechanical property of cold-rolled continuous annealing strip steel based on annealing process | |
CN112003887A (en) | Cloud-edge collaborative deep learning device for industrial internet time sequence data prediction |
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 |