Plate shape prediction method and device for plate

Abstract

The application provides a plate shape prediction method and device for a medium plate, and relates to the technical field of medium plate hot rolling, and the method comprises the following steps: obtaining historical industrial data stored by a medium plate production line, and preprocessing the historical industrial data; combining the preprocessed data and plate shape measurement data of the medium plate into an application data set; extracting characteristic factors influencing the plate shape based on the application data set; taking each process link of a medium plate production process as an independent unit, and taking the plate shape measurement data as a dependent variable, so as to construct a medium plate plate shape prediction model based on the combination of a process mechanism model and the characteristic factors and the process link; taking process data and plate shape measurement data in the production process of the medium plate as the input of the medium plate plate shape prediction model, and using the output data of the medium plate plate shape prediction model to predict the plate shape of the medium plate. The technical problem that production data is separated from actual production, so that the model application cannot fully guide the actual production is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of medium plate hot rolling, and in particular to a medium plate shape prediction method and device. BACKGROUND

[0002] Medium plate is an important product in the steel industry and an important symbol of the level of the steel industry. Medium plate products are widely used in infrastructure construction, ship and automobile manufacturing, energy storage and other fields.

[0003] Plate shape problems have always been a problem that restricts the quality of medium plate products and are also a key problem in related research. In actual production, serious plate shape problems can directly lead to the batch of products being discarded as waste, causing serious economic losses. Therefore, plate shape prediction of plate products in production and timely adjustment of production processes based on this have important significance for cost savings, improving yield and improving product quality. At present, mining the information value contained in industrial big data has attracted the attention of researchers, and with the vigorous development of big data technology, using digital and intelligent methods to monitor and predict plate shape problems in the production process of medium plate will bring higher precision, more convenient operating environment and higher economic benefits.

[0004] Under the production background of multiple processes and small batch, the complex hot rolling process of medium plate and the numerous index orders result in an exceptionally large amount of industrial production data. However, due to the separation of industrial production data and models, the research on plate shape problems by industrial production data has not effectively guided actual production. Therefore, a medium plate shape prediction method is needed to solve the separation of production data in each production link and actual production. SUMMARY

[0005] Therefore, the present application provides a medium plate shape prediction method and device to solve at least one of the above problems.

[0006] In order to achieve the above purpose, the present application adopts the following scheme:

[0007] According to a first aspect of the present application, a plate shape prediction method for a medium plate is provided, the method comprising: obtaining historical industrial data retained by a medium plate production line, preprocessing the historical industrial data to obtain preprocessed data; combining the preprocessed data and plate shape measurement data of the medium plate into an application data set; extracting characteristic factors affecting plate shape based on the application data set by a machine learning method; taking each process link of a medium plate production process as an independent unit, taking the plate shape measurement data as a dependent variable, and constructing a medium plate plate shape prediction model based on the characteristic factors and a process mechanism model of the process link; taking process data and plate shape measurement data in a medium plate production process as input of the medium plate plate shape prediction model, and using output data of the medium plate plate shape prediction model to predict the plate shape of the medium plate.

[0008] Preferably, in the above method of the embodiment of the present application, preprocessing the historical industrial data to obtain preprocessed data comprises: obtaining historical industrial data retained by a medium plate production and supporting instruments and equipment; re-arranging the industrial data of different process links according to a unified format, and cleaning and processing error values and missing values to obtain preprocessed data.

[0009] Preferably, in the above method of the embodiment of the present application, combining the preprocessed data and the plate shape measurement data of the medium plate into an application data set comprises: matching and combining the preprocessed data and the plate shape measurement data of the medium plate according to the slab number to form an application data set.

[0010] Preferably, in the above method of the embodiment of the present application, extracting characteristic factors affecting plate shape based on the application data set by a machine learning method comprises: extracting characteristic factors by a factor analysis technology of a machine learning method based on the application data set, specifically comprising: setting the information amount reflected by the selected characteristic value to be more than 95% of the total information amount, calculating the characteristic value and characteristic vector of the correlation matrix of all independent variables, then calculating the factor loading matrix, and finally linearly expressing the common factor vector with the independent variable vector according to the factor model, the common factor vector being the characteristic factor.

[0011] Preferably, in the above method of the embodiment of the present application, each process link of the plate production process is taken as an independent unit, the plate shape measurement data is taken as the dependent variable, and a plate shape prediction model based on the combination of the characteristic factors and the process mechanism model of the process link is constructed, including: taking each process link of the plate production process as an independent unit, adding the characteristic factor variables in the form of a polynomial to the process mechanism model of each process link to form a plurality of process link fitting models; linking the process link fitting models to construct a simultaneous equation set to form a plate shape prediction model; dividing the application data set into two sub-data sets, and the sample sizes of the sub-data sets are 75% and 25% of the original data sample size respectively, while ensuring that the data set corresponding to each steel grade is divided into sub-data sets with sample sizes of 75% and 25%; taking the sub-data set with a sample size of 75% as the training set of the model to fit the coefficients corresponding to the variables other than the process mechanism model in the process link fitting model, and taking the sub-data set with a sample size of 25% as the test set to test the prediction accuracy of the plate shape prediction model; if the prediction accuracy meets the requirements, outputting the plate shape prediction model, if the prediction accuracy does not meet the requirements, adjusting the parameters other than the process mechanism model in the fitting model for retraining until the prediction accuracy requirements are met.

[0012] Preferably, in the above method of the embodiment of the present application, the process mechanism model includes one or more of a rolling mechanism model, a cooling mechanism model, a straightening mechanism model or a shearing mechanism model.

[0013] According to a second aspect of the present application, a plate shape prediction device for a plate is provided, the device comprising: a historical data acquisition unit configured to acquire historical industrial data retained by a plate production line; a preprocessing unit configured to preprocess the historical industrial data to obtain preprocessed data; a data merging unit configured to merge the preprocessed data and plate shape measurement data of the plate into an application data set; a feature extraction unit configured to extract characteristic factors affecting the plate shape based on the application data set by a machine learning method; a model construction unit configured to take each process link of the plate production process as an independent unit, take the plate shape measurement data as the dependent variable, and construct a plate shape prediction model based on the combination of the characteristic factors and the process mechanism model of the process link; and a plate shape prediction unit configured to take process data and plate shape measurement data in the plate production process as input of the plate shape prediction model, and use output data of the plate shape prediction model to predict the plate shape of the plate.

[0014] Preferably, the historical data acquisition unit in the above-described apparatus of the present invention is specifically used to: acquire historical industrial data stored in the production of medium and heavy plates and related instruments and equipment; the preprocessing unit is specifically used to: rearrange the industrial data stored in different process steps according to a unified format, and clean and process the erroneous and missing values ​​to obtain preprocessed data.

[0015] Preferably, the data merging unit in the above-described apparatus of the present invention is specifically used to: match and combine the preprocessed data with the plate shape measurement data of the medium-thick plate according to the slab number, and merge them into an application dataset.

[0016] Preferably, the feature extraction unit in the above-described apparatus of the present invention includes: a setting module, used to set the amount of information reflected by the selected feature values ​​to account for more than 95% of the total information; and a calculation model, used to first calculate the relevant...

[0017] The matrix has eigenvalues ​​and eigenvectors, and then the factor loading matrix is ​​calculated; the feature extraction module is used to linearly express the common factor vector using the independent variable vector according to the factor model, and the common factor vector is the feature factor.

[0018] Preferably, the model building unit in the above-described apparatus of the present invention includes: a model fitting module, used to take each process step of the medium-thick plate production process as an independent unit, and add characteristic factor variables in the form of polynomials to the process mechanism model of each process step to form multiple process step fitting models; a simultaneous equation system construction module, used to combine the fitting models of each process step to construct a simultaneous equation system to form a plate shape prediction model; a subset dataset construction module, used to divide the application dataset into two subset datasets, the sample sizes of the subset datasets being 75% and 25% of the original data sample size, respectively, while ensuring that the dataset corresponding to each steel grade is divided into subset datasets with sample sizes of 75% and 25%; a training and testing module, used to use the subset dataset with a sample size of 75% as the training set of the model to fit the coefficients corresponding to the variables other than the process mechanism model in the fitting models of each process step, and use the subset dataset with a sample size of 25% as the test set to test the prediction accuracy of the plate shape prediction model; and a model output module, used to output the medium-thick plate shape prediction model when the prediction accuracy meets the requirements.

[0019] Preferably, the process mechanism model in the above-described apparatus of the present invention includes one or more of the following mechanism models: rolling mechanism model, cooling mechanism model, straightening mechanism model, or shearing mechanism model.

[0020] According to a third aspect of the present invention, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the steps of the above-described method.

[0021] According to a fourth aspect of the present application, there is provided a computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of the above method.

[0022] According to a fifth aspect of the present application, there is provided a computer program product comprising computer programs / instructions which, when executed by a processor, implements the steps of the above method.

[0023] From the above technical solution, the plate shape prediction method and device provided by the present application can serve the improvement of the process mechanism model, can find out the parameter variables which have not been considered in the process mechanism model and affect the actual plate shape, and can be evaluated to improve the prediction accuracy, so that the data can reflect and guide the actual production. In addition, the scheme of fitting the models of each process link together to construct a simultaneous equation set to form a plate shape prediction model in the present application can connect the plate shape problems of each link together, and a relatively complete plate shape prediction model is constructed according to the production process. The plate shape prediction accuracy is further improved. It can be seen that the present application solves the technical problem that the production data is disconnected from the actual production, which leads to that the model application cannot fully guide the actual production. BRIEF DESCRIPTION OF DRAWINGS

[0024] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or prior art description will be briefly introduced as follows. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor. In the drawings:

[0025] Figure 1 is a flowchart of a plate shape prediction method for a medium plate provided by an embodiment of the present application;

[0026] Figure 2 is a flowchart of pre-processing of historical industrial data provided by an embodiment of the present application;

[0027] Figure 3 is a flowchart of feature factor extraction using factor analysis technology provided by an embodiment of the present application;

[0028] Figure 4 is a flowchart of construction of a plate shape prediction model for a medium plate provided by an embodiment of the present application;

[0029] Figure 5 is a principle diagram of a plate shape prediction method for a medium plate provided by an embodiment of the present application;

[0030] Figure 6is a structural schematic diagram of a plate shape prediction device for a medium plate provided by an embodiment of the present application.

[0031] Figure 7 is a structural schematic diagram of a feature extraction unit provided by an embodiment of the present application.

[0032] Figure 8 is a structural schematic diagram of a model construction unit provided by an embodiment of the present application.

[0033] Figure 9 is a schematic diagram of an electronic device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0034] To make the objectives, technical solutions and advantages of the embodiments of the present application clearer, further detailed descriptions of the embodiments of the present application are given below with reference to the drawings. Here, the schematic embodiments of the present application and their descriptions are used to explain the present application but not to limit the present application.

[0035] Based on the complex hot rolling process of the medium plate and the numerous index orders in the prior art, the associated industrial production data is extremely complex, and due to the separation of the industrial production data and the model, the research on the plate shape problem by the industrial production data has not been able to effectively guide the actual production, and the present application is dedicated to solving the technical problem.

[0036] As shown in Figure 1 is a flowchart of a plate shape prediction method for a medium plate provided by an embodiment of the present application, and the method comprises the following steps:

[0037] Step S101: Obtain the historical industrial data retained by the medium plate production line, and pre-process the historical industrial data to obtain pre-processed data.

[0038] Preferably, as shown in Figure 2 , the step specifically can comprise the following sub-steps:

[0039] Step S1011: Obtain the historical industrial data retained by the medium plate production and the supporting instrument equipment.

[0040] The historical industrial data retained here can be, for example, the data of the last year, and of course, in order to increase the data amount, historical data retained for several years can also be selected. Specifically, the industrial data retained can include process flow data, monitoring and detection data at each link, equipment operation parameter data, operation data generated by worker intervention, and worker-recorded production-related measurement data.

[0041] Step S1012: re-arranging the industrial retained data of different process links according to a unified format. The unified format re-arrangement here includes re-unified arrangement of data format and re-unified arrangement of data logic. This is to facilitate the subsequent data modeling needs, and the dimensions of part of the process link data may be different. To build a model, these data of different structures need to be unified into one structure.

[0042] Step S1013: obtaining pre-processed data after cleaning and processing the error values and missing values therein.

[0043] Through the above pre-processing steps, the influence deviation of the model as a whole caused by some special factors or random interference factors can be effectively avoided, thereby improving the accuracy and efficiency of the model.

[0044] Step S102: combining the pre-processed data and the plate shape measurement data of the medium plate into an application data set.

[0045] The plate shape measurement data here also belongs to historical data, and its selection range can be consistent with the industrial retained data, such as selecting the plate shape measurement data of the last year. Since the plate shape measurement data and the pre-processed industrial retained data have not established a corresponding relationship, in order to correspond and associate the plate shape measurement data of different products with the production process, it needs to be combined. Preferably, the pre-processed data and the plate shape measurement data of the medium plate can be matched and combined according to the slab number to form an application data set.

[0046] Step S103: extracting the characteristic factors affecting the plate shape based on the application data set through a machine learning method.

[0047] Preferably, the embodiment can use the factor analysis technology in the machine learning method to extract the characteristic factors, as shown in FIG. 8. Figure 3 The specific process of using the factor analysis technology to extract the characteristic factors is as follows:

[0048] Step S1031: setting the information amount reflected by the common factor to more than 95% of the total information amount. The more than 95% here means greater than or equal to 95%, and of course, the value can also be set according to the needs of the person skilled in the art.

[0049] Step S1032: calculating the eigenvalues and eigenvectors of the correlation matrix of all independent variables.

[0050] Step S1033: calculating the factor loading matrix.

[0051] Step S1034: linearly expressing the common factor vector with independent variables according to the factor loading matrix, and the common factor vector is the characteristic factor.

[0052] Specific to the embodiment, the specific way of the above factor analysis technique for extracting characteristic factors can further include the following steps:

[0053] First, all the pretreated industrial remaining data are standardized. Assuming that there are n indicators to be standardized and m pieces of steel plate data, let x ij be the value of the jth indicator of the ith piece of steel plate data, and the standardized indicator is as shown in the following formula (1):

[0054]

[0055] wherein respectively represent the sample mean and sample standard deviation of the jth indicator.

[0056] The correlation coefficient matrix R=(r ab ) n×n is calculated by the following formula (2):

[0057]

[0058] wherein r ii =1, r ij =r ji , and r ab is the correlation coefficient of the a th indicator and the b th indicator. The eigenvalues λ1≥λ2≥λ3≥…≥λ n ≥0 of the correlation coefficient matrix R are calculated, as well as the corresponding eigenvectors u1, u2, u3,…u n , which constitute the factor loading matrix A as shown in the following formula (3):

[0059]

[0060] which is written in vector mode as shown in the following formula (4):

[0061] x=AF+ε (4)

[0062] wherein F is the common factor vector, i.e., the above characteristic factor, and the number of common factors is determined according to the variance contribution rate, so that the cumulative contribution rate exceeds 95%. x can be linearly represented by the common factor F. F is converted into a function of x as shown in the following formula (5):

[0063] F=A T R -1 ·x (5)

[0064] wherein A T is the transpose of the A vector, and R -1 is the inverse of the correlation coefficient matrix.

[0065] Step S104: taking each process link of the plate production process as an independent unit, taking the plate shape measurement data as the dependent variable, and constructing a plate shape prediction model based on the combination of the characteristic factors and the process mechanism model of the process link.

[0066] Preferably, the process mechanism model in this step S104 can include one or more of a rolling mechanism model, a cooling mechanism model, a straightening mechanism model, or a shearing mechanism model. Through this step S104, the plate shape problems of each link can be linked together, and a relatively complete plate shape prediction model can be constructed according to the production process.

[0067] Preferably, as shown in FIG. 4, this step can further include the following sub-steps: Figure 4

[0068] Step S1041: taking each process link of the plate production process as an independent unit, adding the characteristic factor variables to the process mechanism model of each process link in the form of a polynomial to form a plurality of process link fitting models.

[0069] Step S1042: simultaneously solving the process link fitting models to construct a simultaneous equation system to form a plate shape prediction model.

[0070] Step S1043: dividing the application data set into two sub-data sets, the sample sizes of the sub-data sets being 75% and 25% of the original data sample size, respectively, while ensuring that the data set corresponding to each steel grade is divided into sub-data sets with sample sizes of 75% and 25%.

[0071] Step S1044: taking the sub-data set with a sample size of 75% as the training set of the model to fit the coefficients corresponding to the variables other than the process mechanism model in the process link fitting models, and taking the sub-data set with a sample size of 25% as the test set to test the prediction accuracy of the plate shape prediction model.

[0072] Step S1045: if the prediction accuracy meets the requirements, outputting the plate shape prediction model of the plate, and if the prediction accuracy does not meet the requirements, adjusting the parameters other than the process mechanism model in the fitting model for retraining until the prediction accuracy requirements are met.

[0073] The steps of constructing the plate shape prediction model described above will be further described based on a specific embodiment as follows:

[0074] Taking the plate crown calculation of the rolling link as an example, the following equation (6) of the crown mechanism model is considered:

[0075]

[0076] wherein ​is the effect of rolling force on the loaded roll gap, Roll is the rolling force, is the effect of bending force on the loaded roll gap, T is the bending force, K roll and K T are the lateral stiffnesses related to the rolling force and the bending force respectively, ω W is the worn roll profile, ω0 is the original roll profile, E ω is the width-related influence coefficient, E H ω H is the effect of the hot roll profile, E c ω c is the effect of continuously variable crown rolling on the roll gap, C0 is the mill entry plate crown, K C is the crown transfer coefficient, K C C0 represents the effect of the mill entry crown on the mill exit crown.

[0077] The common factor vector F extracted in the above formula (5) is added to the process mechanism model in the form of a polynomial to obtain the following formula (7)

[0078]

[0079] where F is the common factor vector and β is the corresponding coefficient vector, which needs to be estimated by relying on actual data to close the above model.

[0080] In this way, the influence of some special factors in the original process mechanism model or in the production process will be reflected on the β·F term, thereby realizing the improvement of the fitting accuracy of the mechanism model and making the theory and the actual more consistent.

[0081] Other process links use the same modeling method to construct the fitting model, and the fitting models of each link are connected in the form of a system of simultaneous equations to form a plate shape prediction model.

[0082] Let S Roll =f(X), S Cool =g(Y), S Straight =h(Z), and S shear =p(W) be the mechanism models of the rolling link, the cooling link, the straightening link, and the shearing link about the plate shape, respectively, the common factor variable is added to each equation, and the four equations are solved simultaneously to have the following formula (8):

[0083]

[0084] where the independent variables X, Y, Z, and W in different equations have certain correlation in equipment or time, and for the common factor F, the corresponding coefficient matrices α, β, γ, and δ in different equations are also different, and the construction of the system of simultaneous equations will greatly improve the fitting degree of the actual data and the theoretical mechanism through the adjustment of these coefficients.

[0085] The plate shape of the final product is formed by superimposing the plate shapes of different process links, which can be regarded as a linear combination of S Roll , S Cool , S Straight and S shear , and the final plate shape equation can be obtained by a regression equation as follows (9):

[0086] S = θ0 + θ1S Roll + θ2S Cool + θ3S Straight + θ4S Shear + ε (9)

[0087] Solving the coefficients θ0, θ1, θ2, θ3 and θ4 to obtain the exact solution, thereby realizing the prediction of the plate shape.

[0088] The training set in the above step S1044 is used to fit the coefficients corresponding to the variables in the process link fitting model except the process mechanism model, that is, α, β, γ and δ in the above formula (8) and θ0, θ1, θ2, θ3 and θ4 in the final plate shape equation (9).

[0089] In addition, the training set in the above step S1044 is used to test the prediction accuracy of the plate shape prediction model, and in step S1045, when the prediction accuracy does not reach the requirement, the parameters in the fitting model except the process mechanism model are adjusted, and specifically, the high order terms of the common factor can be added in the process link fitting model, for example, considering the plate shape equation S Roll = f(X) + αF, here we expand the vector to obtain the following formula (10) for easy understanding:

[0090]

[0091] Where q is the number of common factors. In the above formula (10), the quadratic term of the common factor variable is added to obtain formula (11):

[0092]

[0093] Then the plate shape prediction model is retrained until the prediction accuracy requirement is met.

[0094] Step S105: Taking the process data and plate shape measurement data in the plate production process as inputs of the plate shape prediction model of the plate, and using the output data of the plate shape prediction model of the plate to predict the plate shape of the plate.

[0095] As shown in Figure 5 the principle schematic diagram of the plate shape prediction method provided by the embodiment of the application, the plate shape prediction method provided by the embodiment of the application comprises the following steps: Figure 5It can be seen that the application firstly establishes a database through data collection (including industrial remaining data and plate shape measurement data) and data preprocessing, then extracts characteristic factors affecting plate shape by using factor analysis technology, and then builds a plate shape prediction model based on the characteristic factors, specifically by equation association of mechanism model and characteristic factors to build a plate shape prediction model. Meanwhile, the collected data is divided into a training set and a test set, the training set is used to train the model coefficients except the mechanism model, the test set is used to test whether the plate shape prediction model meets the prediction accuracy requirement, if it meets, the plate shape prediction model is output for plate shape prediction of the plate, if it does not meet, the function form is adjusted and then the training is continued.

[0096] It can be seen from the above technical solution that the plate shape prediction method of the present application makes the data mining technology serve the improvement of the process mechanism model, can find out the parameter variables that have not been considered in the process mechanism model and affect the actual plate shape, and evaluate them to improve the prediction accuracy, so that the data reflects and guides the actual production. In addition, the scheme of fitting the models of each process link and associating them to build a system of simultaneous equations to form a plate shape prediction model makes the plate shape problems of each link connected together, and a relatively complete plate shape prediction model is built according to the production process. The plate shape prediction accuracy is further improved. It can be seen that the present application solves the technical problem that the production data is disconnected from the actual production, which leads to that the model application cannot fully guide the actual production.

[0097] As shown in Figure 6 The structure schematic diagram of a plate shape prediction device provided by the embodiment of the present application, the device comprises: a historical data acquisition unit 610, a preprocessing unit 620, a data merging unit 630, a feature extraction unit 640, a model building unit 650 and a plate shape prediction unit 660, which are sequentially connected.

[0098] The historical data acquisition unit 610 is used for acquiring the historical industrial remaining data remaining in the plate production line.

[0099] The preprocessing unit 620 is used for preprocessing the historical industrial remaining data to obtain preprocessing data.

[0100] As an embodiment of the present application, preferably, the historical data acquisition unit 610 can be specifically used for acquiring the historical industrial remaining data remaining in the plate production and the supporting instrument equipment. The preprocessing unit 620 can be specifically used for rearranging the industrial remaining data of different process links according to a unified format, and cleaning and processing the error values and missing values to obtain preprocessing data.

[0101] The data merging unit 630 is used for merging the preprocessing data and the plate shape measurement data of the plate into an application data set.

[0102] As an embodiment of the present application, preferably, the data merging unit 630 is specifically configured to: match and combine the preprocessed data and the plate shape measurement data of the plate in plate number to form an application data set.

[0103] The feature extraction unit 640 is configured to extract the characteristic factors affecting the plate shape based on the application data set by a machine learning method.

[0104] As an embodiment of the present application, preferably, as shown in Figure 7 The feature extraction unit 640 can include a setting module 641, a calculation module 642 and a feature extraction module 643.

[0105] The setting module 641 is configured to set the information amount reflected by the selected characteristic value to account for more than 95% of the total information amount.

[0106] The calculation module 642 is configured to first calculate the eigenvalues and eigenvectors of all independent variable correlation matrices, and then calculate the factor loading matrix.

[0107] The feature extraction module 643 is configured to linearly express the common factor vector with the independent variable vector according to the factor model, and the common factor vector is the characteristic factor.

[0108] The model construction unit 650 is configured to take each process link of the plate production process as an independent unit, and take the plate shape measurement data as the dependent variable, to construct a plate shape prediction model based on the combination of the characteristic factors and the process mechanism model of the process link.

[0109] As an embodiment of the present application, preferably, the process mechanism model includes one or more of a rolling mechanism model, a cooling mechanism model, a straightening mechanism model or a shearing mechanism model.

[0110] As an embodiment of the present application, preferably, as shown in Figure 8 The model construction unit 650 can specifically include a model fitting module 651, a system of simultaneous equations construction module 652, a sub-data set construction module 653, a training test module 654 and a model output module 655.

[0111] The model fitting module 651 is configured to take each process link of the plate production process as an independent unit, and add the characteristic factor variable to the process mechanism model of each process link in the form of a polynomial to form a plurality of process link fitting models.

[0112] The system of simultaneous equations construction module 652 is configured to simultaneously solve the process link fitting models to construct a system of simultaneous equations to form a plate shape prediction model.

[0113] The sub-data set construction module 653 is configured to divide the application data set into two sub-data sets, and the sample sizes of the two sub-data sets are 75% and 25% of the original data sample size respectively, while ensuring that the data set corresponding to each steel grade is divided into sub-data sets with sample sizes of 75% and 25%.

[0114] The training test module 654 is configured to use the sub-data set with a sample size of 75% as a training set of the model, so as to fit the coefficients corresponding to the variables in the fitting model of each process link except the process mechanism model, and use the sub-data set with a sample size of 25% as a test set, so as to test the prediction accuracy of the plate shape prediction model.

[0115] The model output module 655 is configured to output the plate shape prediction model of the plate when the prediction accuracy reaches the requirement.

[0116] The plate shape prediction unit 660 is configured to use the process data and the plate shape measurement data in the production process of the plate as inputs of the plate shape prediction model of the plate, and use the output data of the plate shape prediction model of the plate to predict the plate shape of the plate.

[0117] The detailed description of each unit can be referred to the related description in the foregoing method embodiment, and will not be described here.

[0118] As can be seen from the technical solution, the plate shape prediction device of the plate provided by the application can use the data mining technology to improve the process mechanism model, find out the parameter variables that have not been considered in the process mechanism model and affect the actual plate shape, and evaluate the parameter variables to improve the prediction accuracy, so that the data can reflect the actual production and guide the actual production. In addition, the scheme of simultaneously solving the fitting models of each process link to construct a simultaneous equation set to form a plate shape prediction model can connect the plate shape problems of each link together, and construct a relatively complete plate shape prediction model according to the production process, thereby further improving the plate shape prediction accuracy. It can be seen that the application solves the technical problem that the production data is disconnected from the actual production, and the model application cannot fully guide the actual production.

[0119] Figure 9 FIG. 1 is a schematic diagram of an electronic device provided by an embodiment of the application. Figure 9 The electronic device shown is a general data processing device, which includes a general computer hardware structure, and at least includes a processor 801 and a memory 802. The processor 801 and the memory 802 are connected through a bus 803. The memory 802 is suitable for storing one or more instructions or programs executable by the processor 801. The one or more instructions or programs are executed by the processor 801 to implement the steps in the plate shape prediction method of the plate.

[0120] The processor 801 can be a stand-alone microprocessor or a set of one or more microprocessors. The processor 801 performs the processing of data and the control of other devices by executing commands stored in the memory 802, thereby implementing the method flow of the embodiments of the present application as described above. The bus 803 connects the above components together, and connects the above components to a display controller 804 and a display device, and an input / output (I / O) device 805. The input / output (I / O) device 805 can be a mouse, a keyboard, a modem, a network interface, a touch input device, a body sense input device, a printer, and other devices known in the art. Typically, the input / output (I / O) device 805 is connected to the system through the input / output (I / O) controller 806.

[0121] The memory 802 can store software components, such as an operating system, communication modules, interaction modules, and application programs. Each of the above modules and application programs corresponds to a set of executable program instructions for completing one or more functions and the methods described in the embodiments of the present application.

[0122] The embodiments of the present application also provide a computer readable storage medium, which stores a computer program. The computer program is executed by a processor to implement the steps of the plate shape prediction method for a medium plate.

[0123] The embodiments of the present application also provide a computer program product, which includes computer programs / instructions. The computer programs / instructions are executed by a processor to implement the steps of the plate shape prediction method for a medium plate.

[0124] In summary, the plate shape prediction method and device for a medium plate provided by the present application make the data mining technology serve the improvement of the process mechanism model. The parameter variables that have not been considered in the process mechanism model and affect the actual plate shape can be found out and evaluated to improve the prediction accuracy, so that the data can reflect and guide the actual production. In addition, the scheme of fitting the models of each process flow link together, constructing a system of simultaneous equations, and forming a plate shape prediction model makes the plate shape problems of each link be connected together, and a relatively complete plate shape prediction model is constructed according to the production process flow. The plate shape prediction accuracy is further improved. It can be seen that the present application solves the technical problem that the production data is disconnected from the actual production, which leads to the model application being unable to fully guide the actual production.

[0125] The preferred embodiments of the application are described above with reference to the accompanying drawings. Many features and advantages of the embodiments are apparent from the detailed specification, and it is therefore intended by the appended claims to cover all such features and advantages of the embodiments within their true spirit and scope. Further, since numerous modifications and changes will readily occur to those skilled in the art, it is not desired to limit the embodiments of the present application to the exact construction and operation

[0126] Those skilled in the art will appreciate that embodiments of the present application can be devised for a method, a system, or a computer program product. Accordingly, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application can take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage devices, etc.) embodying computer-readable program code.

[0127] The present application is described herein with reference to the flowchart illustrations and / or block diagrams of the methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general purpose computer, special purpose computer, embedded processing system or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart Figure 1 one or more functions specified in the flowchart block or blocks. Figure 1 means for performing each of the functions specified in the flowchart block or blocks.

[0128] These computer program instructions can also be stored in a computer- readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the flowchart Figure 1 one or more functions specified in the flowchart block or blocks. Figure 1 means for performing each of the functions specified in the flowchart block or blocks.

[0129] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the flowchart Figure 1 one or more functions specified in the flowchart block or blocks. Figure 1 means for performing each of the functions specified in the flowchart block or blocks.

[0130] The above-described specific embodiments further illustrate the objects, technical solutions and advantages of the present application. It should be understood that the above-described specific embodiments are merely for the purpose of illustrating the present application, and are not intended to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A plate shape prediction method for a medium-thickness plate, characterized by, The method comprises: acquiring historical industrial data retained by a plate production line, and preprocessing the historical industrial data to obtain preprocessed data; combining the preprocessed data and plate shape measurement data of the plate into an application data set; extracting feature factors affecting plate shape based on the application data set by a machine learning method; taking each process link of a plate production process as an independent unit, taking the plate shape measurement data as a dependent variable, and constructing a plate shape prediction model based on the feature factors and the process links and a process mechanism model; using process data and plate shape measurement data in the plate production process as inputs of the plate shape prediction model, and using output data of the plate shape prediction model to predict plate shape of the plate; the plate shape prediction model based on the feature factors and the process links and the process mechanism model comprises: taking each process link of a plate production process as an independent unit, adding feature factor variables to the process mechanism model of each process link in the form of a polynomial to form a plurality of process link fitting models; combining the process link fitting models to construct a simultaneous equation system to form a plate shape prediction model; dividing the application data set into two sub-data sets, with sample sizes of the sub-data sets being 75% and 25% of the sample size of the original data, and ensuring that the data set corresponding to each steel grade is divided into sub-data sets with sample sizes of 75% and 25%; using the sub-data set with a sample size of 75% as a training set of the model to fit coefficients corresponding to variables other than the process mechanism model in the process link fitting model, and using the sub-data set with a sample size of 25% as a test set to test the prediction accuracy of the plate shape prediction model; if the prediction accuracy meets the requirements, outputting the plate shape prediction model, and if the prediction accuracy does not meet the requirements, adjusting parameters other than the process mechanism model in the fitting model for retraining until the prediction accuracy meets the requirements.

2. The plate shape prediction method for a medium thick plate according to claim 1, wherein the preprocessing of the historical industrial data comprises: acquiring historical industrial data retained in plate production and supporting instruments and equipment; rearranging the industrial data of different process links according to a unified format, and cleaning and processing error values and missing values in the data to obtain preprocessed data.

3. The plate shape prediction method for a medium thick plate according to claim 1, wherein the combination of the preprocessed data and the plate shape measurement data into an application data set comprises: matching and combining the preprocessed data and the plate shape measurement data according to the slab number to form an application data set.

4. The plate shape prediction method for a medium thick plate according to claim 1, wherein the extraction of feature factors affecting plate shape based on the application data set by a machine learning method comprises: The feature factor extraction is performed based on the application data set by using a factor analysis technique of a machine learning method, and specifically includes: setting that information amount reflected by the selected characteristic value accounts for more than 95% of the total information amount, calculating characteristic values and characteristic vectors of a correlation matrix of all independent variables, then calculating a factor loading matrix, and finally linearly expressing a common factor vector with an independent variable vector according to a factor model, the common factor vector being the feature factor.

5. The plate shape prediction method for a medium thick plate according to Claim 1, wherein The process mechanism model comprises one or more of a rolling mechanism model, a cooling mechanism model, a straightening mechanism model or a shearing mechanism model.

6. A medium-thick plate shape prediction device, characterized in that, The device comprises: a historical data acquisition unit configured to acquire historical industrial data retained by a heavy plate production line; a preprocessing unit configured to preprocess the historical industrial data to obtain preprocessed data; a data merging unit configured to merge the preprocessed data and plate shape measurement data of the heavy plate into an application data set; a feature extraction unit configured to extract a feature factor affecting the plate shape based on the application data set by using a machine learning method; a model construction unit configured to construct a heavy plate plate shape prediction model based on the feature factor and a process mechanism model of each process link of a heavy plate production process, with each process link being an independent unit and the plate shape measurement data being a dependent variable; a plate shape prediction unit configured to input process data and plate shape measurement data in a heavy plate production process into the heavy plate plate shape prediction model, and to predict the plate shape of the heavy plate by using output data of the heavy plate plate shape prediction model; the model construction unit comprises: a model fitting module configured to add the feature factor variable into the process mechanism model of each process link in a polynomial form to form a plurality of process link fitting models; a simultaneous equation set construction module configured to construct a simultaneous equation set by simultaneously solving the process link fitting models to form the plate shape prediction model; a sub-data set construction module configured to divide the application data set into two sub-data sets, with sample amounts of the sub-data sets being 75% and 25% of a sample amount of the original data respectively, and ensuring that the data set corresponding to each steel grade is divided into the sub-data sets with sample amounts of 75% and 25%; a training and testing module configured to use the sub-data set with the sample amount of 75% as a training set of the model to fit coefficients corresponding to variables other than the process mechanism model in the process link fitting models, and use the sub-data set with the sample amount of 25% as a test set to test prediction accuracy of the plate shape prediction model; a model output module configured to output the heavy plate plate shape prediction model when the prediction accuracy meets a requirement.

7. The plate shape prediction device for a medium-thickness plate according to Claim 6, wherein The historical data acquisition unit is specifically configured to acquire historical industrial data retained by a heavy plate production process and matching instruments and equipment; and the preprocessing unit is specifically configured to rearrange the industrial data of different process links according to a unified format, and clean and process error values and missing values in the industrial data to obtain the preprocessed data.

8. The plate shape prediction device for a medium thick plate according to claim 6, wherein The data merging unit is specifically configured to match and combine the preprocessed data and the plate shape measurement data of the heavy plate according to a slab number to form the application data set.

9. The plate shape prediction device for a medium thick plate according to claim 6, wherein The feature extraction unit comprises: A setting module is configured to set the information amount reflected by the selected feature values to be more than 95% of the total information amount; A calculation model is configured to calculate the eigenvalues and eigenvectors of all independent variable correlation matrices first, and then calculate the factor loading matrix; A feature extraction module is configured to linearly express the common factor vector with the independent variable vector according to the factor model, wherein the common factor vector is the characteristic factor.

10. The plate shape prediction device for a medium thick plate according to claim 6, wherein The process mechanism model comprises one or more of a rolling mechanism model, a cooling mechanism model, a straightening mechanism model, or a shearing mechanism model.

11. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the computer program to implement the steps of the method of any one of claims 1 to 5.

12. A computer readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the method of any one of claims 1 to 5.

13. A computer program product comprising computer programs / instructions, characterized in that, The computer program / instructions are executed by the processor to implement the steps of the method of any one of claims 1 to 5.

Images