A head-tail optimized shearing method for thick and wide plate motherboard based on performance prediction

Abstract

The present invention relates to a method for optimizing the shearing of the head and tail of a wide and thick plate mother plate based on performance prediction. The method is based on industrial big data of the steel plate production process, and after the steel plate production is completed, the performance of the wide and thick plate mother plate along the length direction is predicted. A comparative analysis is performed based on the predicted performance indicators, and an indicator with large performance fluctuation is selected as the performance indicator for shear optimization. The performance of the performance indicator from beginning to end is considered, and the most uniform part of the performance indicator is selected, and the shearing amount of the mother plate head and tail is calculated to achieve optimized shearing of the head and tail of the steel plate. The present invention can achieve optimized shearing of the head and tail of the steel plate based on the prediction of the full-length mechanical properties in the length direction, reduce the re-judgment caused by the head and tail performance exceeding the standard value, reduce the same-plate performance difference of the mother plate, and achieve the purpose of optimized shearing control.

Description

Technical Field

[0001] The invention relates to a method for optimizing the head and tail shearing of a wide and thick plate mother plate based on performance prediction, and belongs to the field of wide and thick plate production control methods. Background Art

[0002] As one of the most important products of the steel industry, wide and thick plates play an important role in the development of the national economy. They are generally used in shipbuilding, construction, engineering machinery, bridges, military industry, welded pipes, containers, etc. Wide and thick plates are mostly produced by thermomechanical rolling (TMCP). The rolling process is a multi-pass reciprocating rolling process. During the production process, wide and thick plates have three-dimensional heat dissipation at the head and tail. The temperature of the head and tail of the steel plate is low. When the steel plate is controlled to cool, the head of the steel plate enters the water cooling system first, and the tail enters the water cooling system later. Therefore, there is a temperature difference between the head, tail and the middle part of the steel plate during the rolling process. Especially for TMCP (Thermo Mechanical Control Process) steel plates, the cooling steel plate enters the water cooling system at different times and temperatures along the length direction, resulting in fluctuations in the organization and mechanical properties of the steel plate in the length direction after cooling, and large deviations in the mechanical properties of the head and tail, and even some performances exceed the standard range. After sampling at the head and tail, the customer re-tested the mechanical properties and found them unqualified.

[0003] In the entire production process of wide and thick plates, the shearing process of the rolled mother plate is a key step, which has an important impact on the quality and cost of the final product. At present, the head and tail shearing of wide and thick plates is generally calculated according to the same amount of cutting at the head and tail to ensure that the size of the mother plate after shearing meets the size requirements of the contract order. Some optimized shearing algorithms mainly consider the surface quality and shape of the mother plate after rolling, such as a medium and thick plate optimized shearing control method based on online surface quality evaluation (CN 110216319A), a wide and thick plate optimized shearing method based on shape recognition (CN 112894489 A), and a hot-rolled strip optimized shearing method based on machine vision (CN 113198854 B). All of them use machine recognition technology to identify surface defects and the surface shape of the steel plate for optimized shearing, without considering the performance fluctuations of the head and tail of the steel plate. The fluctuations in the mechanical properties of the head and tail are likely to cause the performance of the steel plate to be inconsistent, and the mechanical properties of the same steel plate have large deviations, which affects the subsequent production and processing of customers. Patent CN117655108 A provides an online optimized shearing method and system for rolled large plates, which optimizes shearing mainly by monitoring the rolling process and shearing process parameters. CN 112836305 A provides a method for calculating the shearing strategy for wide and thick plates, which collects the contour data points of the wide and thick plate mother plate and the length and width of the wide and thick plate mother plate through a profiler, and calculates the shearing strategy of the head cutting length, tail cutting length, and width of the wide and thick plate mother plate. All of these optimization patents are mainly optimized based on surface defects, surface quality, plane shape, and head and tail contours, etc., without considering the performance differences of the same length direction of the steel plate. There are temperature changes along the length direction during the production of wide and thick plates, and the mechanical properties of the wide and thick plates in the length direction also fluctuate, and even the head and tail mechanical properties do not meet the standard requirements. How to reduce performance fluctuations by optimizing shearing is also a problem that plagues enterprises.

[0004] At the same time, the mechanical properties test of wide and thick plates is done after the steel plates are produced. Cut samples are sent to the testing center for processing and testing. The cut samples are taken at the head or the middle of the steel plate. The sample test represents a certain position of the mother plate, which cannot reflect the overall performance of the mother plate in the longitudinal direction of the mother plate. In addition, the testing center generally selects one sample for mechanical properties test, which cannot fully reflect the mechanical properties of the mother plate along the length direction. There is a certain lag in performance testing. The shearing of the head and tail of the wide and thick plate mother plate is generally carried out online. Therefore, the mother plate is completed without comprehensive testing of the mechanical properties of the steel plate.

[0005] With the rapid development of computer technology, communication technology, and cloud computing technology, the application of industrial big data in the steel manufacturing industry has been promoted. On the basis of performance prediction technology based on traditional physical metallurgical models, industrial big data has been introduced. The performance prediction model of industrial big data is closer to production practice. For this reason, the real-time prediction technology of mechanical properties of wide and thick plates based on industrial big data has attracted the attention of production companies. How to apply performance prediction technology to the production line to guide production and optimize process parameters is a common problem faced by scientific research institutions and production companies. Summary of the invention

[0006] The present invention provides a method for optimizing the head and tail shearing of a wide and thick plate motherboard based on performance prediction, which solves the problems of large deviation in mechanical properties of the same plate or inconsistency in mechanical properties of the head and tail caused by performance fluctuations of the head and tail of the wide and thick plate motherboard.

[0007] The technical solution adopted by the present invention to solve its technical problem is:

[0008] A method for optimizing the head and tail shearing of a wide and thick plate mother plate based on performance prediction, wherein the wide and thick plate is a wide and thick plate rolled by a single-stand or double-stand rolling mill, and specifically comprises the following steps:

[0009] Step S1: Based on the MES system, historical wide and thick plate rolling data is obtained, wherein the rolling data includes the slab width W, length L and thickness H, the width w, length l and thickness h after rolling; and the target length after the head and tail of the wide and thick plate mother plate are cut to size. , determine the cut-off head arc , determine the tail arc to be cut off And the length of the steel plate after rolling l;

[0010] Step S2: After rolling is completed, select a sampling length in the length direction of the steel plate and select n Sampling points, real-time recording of the final rolling surface temperature of each sampling point , The surface temperature of the steel plate after water cooling starts And the final cooling surface temperature of the steel plate after water cooling , i =1,2,…… n ;

[0011] The sampling length is the minimum cutting length after deducting the head and tail of the steel plate. The sampling length is divided into several sections, and the sampling points are distributed in each section.

[0012] Step S3: The data obtained in step S1 and step S2 are used to form a sample set, a neural network model is constructed based on the sample set, and the neural network model is trained to finally obtain a performance prediction model; the mechanical properties data of the wide and thick plate mother plate in the cooling process after rolling are predicted in real time by the performance prediction model;

[0013] Step S4: Predict based on the performance prediction model m Mechanical performance index n Performance data from m Among the mechanical performance indicators, determine the one with the largest performance fluctuation, defined as k , the performance index k As a calculation index for optimizing shearing;

[0014] Step S5: Continue to obtain the target length of the thick and wide plate motherboard after cutting the head and tail from the MES system , determine the cut-off head arc , determine the tail arc to be cut off And the length of the steel plate after rolling l;

[0015] Step S6: According to the selected k performance indicators, from wide and thick motherboard n Select continuous u A numerical value, that is u A series of performance indicators;

[0016] By calculation u The standard deviation of the performance indicators is selected as the minimum continuous u The performance index standard is set as follows: v Start to v + u -1 ends, then the target length is determined v Specific value of

[0017] Step S7: Combine step S6 and step S7 to determine the total length of the head cut And the tail cutting length , cut the total length of the head obtained And the tail cutting length Send to the head shear control system, send a shearing command to the wide and thick plate motherboard, and complete the cutting of the head and tail;

[0018] Furthermore, in step S1, the width w and thickness h after rolling are measured and processed by a width gauge and a thickness gauge respectively, and the length l after rolling is calculated as follows: ;

[0019] Furthermore, in step S2, the sampling length is set to S, then , n Take the integer;

[0020] Further, in step S3, the neural network model includes an input layer and an output layer, the input items of the input layer used for training the learning samples include the chemical composition of the wide and thick plate mother plate, the historical wide and thick plate rolling data and the production process parameters, and the expected output items of the output layer include the mechanical property data of the wide and thick plate mother plate in the cooling process after rolling;

[0021] Furthermore, in step S4, the specific steps of determining the calculation index for optimizing shearing are:

[0022] Step S41, obtain n performance data of m mechanical performance indicators according to the performance prediction model to obtain a matrix:

[0023] ;

[0024] Step S42, calculate the average value of each performance indicator, the calculation formula is: ,in , j represents the jth performance index, i represents the i-th calculated performance data, n represents the number of sampling points, and the average value of each mechanical performance index is obtained as follows: ;

[0025] Step S43, calculating the standard deviation of each mechanical performance index, the calculation formula is: ,in, ; The standard deviation formula of mechanical performance index is obtained as follows: ;

[0026] Step S44, calculating the dispersion coefficient by the average value of each mechanical performance index obtained in step S42 and the standard deviation of each mechanical performance index obtained in step S43, the calculation formula is: , the discrete coefficient formula is: ;

[0027] Step S45, calculate step S44 to obtain m The discrete coefficients of the mechanical performance indicators are sorted, and the maximum discrete coefficient is , then k The performance index fluctuates the most. k As a calculation index for optimizing shearing;

[0028] Furthermore, the specific steps in step S6 are:

[0029] Step S61, the obtained u Among the serialized performance indicators, , where the target length of the thick plate after cutting the head and tail Obtained from the MES system, u Take an integer, and ;

[0030] Step S62, according to the selected k performance indicators, and obtain the k The average value of performance indicators , standard deviation ;

[0031] Step S63, from the wide and thick board motherboard k performance indicators n Performance data , select Continuous u Step S64, calculate u The standard deviation of the performance index is ;

[0032] Step S64, for the continuous u Performance index standards Sort and select the smallest consecutive u Performance index standards ,in, , , finally determine the target length after cutting and rolling v The minimum internal performance standard deviation is ;

[0033] Further, in step S7, the head arc to be cut off is obtained from the MES system through step S5. Then, the head cutting length is calculated. , the calculation formula is , then the total length of the head cutting for , tail cutting length for .

[0034] Through the above technical solution, compared with the prior art, the present invention has the following beneficial effects:

[0035] 1. The head-to-tail optimized shearing method of a wide and thick plate mother plate based on performance prediction provided by the present invention is based on the analysis of industrial big data in the steel plate production process, which can better understand the laws and characteristics of steel plate production and provide a solid foundation for subsequent performance prediction and optimization;

[0036] 2. The method for optimizing the head and tail shearing of a wide and thick plate motherboard based on performance prediction provided by the present invention fully considers the fluctuation of the longitudinal performance of the motherboard during the actual production process, and predicts the performance of the motherboard along the length direction after the steel plate production is completed, so as to fully understand the performance of the motherboard at different positions, and provide an accurate reference for the subsequent shearing optimization;

[0037] 3. The method for optimizing the shearing of the mother plate of a wide and thick plate based on performance prediction provided by the present invention performs comparative analysis according to the predicted performance indicators, especially selects an indicator with the largest performance fluctuation as the performance indicator for shearing optimization, and optimizes it, so as to effectively improve the overall quality of the steel plate; at the same time, considering the performance of the performance indicator from the beginning to the end, the performance change of the mother plate in the length direction can be more comprehensively understood;

[0038] 4. The method for optimizing the head and tail shearing of wide and thick plate mother plates based on performance prediction provided by the present invention selects the most uniform part of the performance index in the length direction after considering the performance of the performance index with the largest fluctuation, reasonably calculates the shearing amount of the head and tail of the mother plate, removes the parts with poor performance at the head and tail of the mother plate, optimizes the shearing of the head and tail of the steel plate, reduces the revision caused by the performance exceeding the standard value of the head and tail, reduces the performance difference of the same plate of the mother plate, and achieves the purpose of optimized shearing control. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] The present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0040] Figure 1 It is a schematic diagram of the head and tail cutting of a wide and thick plate mother plate provided by the present invention. DETAILED DESCRIPTION

[0041] The present invention will now be described in further detail with reference to the accompanying drawings. The specific dimensions used in this embodiment are only for illustrating the technical solution and do not limit the protection scope of the present invention.

[0042] As described in the background technology, the shearing process of traditional rolled large plates is usually based on a pre-set uniform shearing of the head and tail to ensure that the size of the mother plate after shearing meets the size requirements of the contract order (here is an explanation for the mother plate, the so-called mother plate is also called a rolled large plate, wide and thick plates are generally delivered in fixed lengths, and can be rolled in multiple lengths. After a slab is rolled into a mother plate, it is cut into two or three small plates according to the contract order and delivered), and the shearing optimization mainly considers the plane shape and surface quality of the steel plate, and fails to fully consider the fluctuation of the longitudinal performance of the mother plate in the actual production process. In this way, the steel plate at the shearing position is easy to cause the mechanical properties of the head and tail of the steel plate to exceed the standard, or the difference between the same plate of the steel plate does not meet the customer's requirements. Secondly, with regard to the performance detection of the steel plate, during the production process of the mother plate, samples are taken from the head, middle, and tail for mechanical property inspection, which cannot fully reflect the performance changes of the steel plate in the full length direction, and there are deviations. Some customers require uniform performance of the steel plate. For example, some customers of pipeline steel require that the performance difference of the same plate be controlled within 40MPa. This practice is easy to cause the problem of inconsistent performance of the head and tail of the steel plate after shearing. At the same time, traditional sampling adopts batch sampling, taking one sample for 30-50 tons of steel plates for performance spot check, which cannot reflect the performance of other motherboards in the same batch.

[0043] Therefore, in order to solve the above problems, the present application provides a method for optimizing the head and tail shearing of wide and thick plate mother plates based on performance prediction, which fully considers the fluctuation of the longitudinal performance of the mother plate in the actual production process, and at the same time, the performance changes in the full length direction of the steel plate are fully reflected, solving the problem of inconsistent performance of the head and tail of the steel plate after shearing.

[0044] In simple terms, the entire optimized shearing method is to sample a motherboard according to length, record relevant parameters, use the performance prediction model to predict the mechanical properties of the steel plate at that point, and optimize the shearing amount at the head and tail by comparing the mechanical properties of all points.

[0045] It should also be noted that the present application is aimed at the shearing of wide and thick plate mother plates. Generally, in order to distinguish coils, hot-rolled plates are divided into two categories: medium and thick plates and coils. The so-called wide and thick plates are also called medium and thick plates. The rolling mills in the production conditions on which the present application is based are usually single-stand or double-stand rolling mills. Therefore, the optimized shearing method provided by the present application is the mother plate of wide and thick plates formed by rolling with single-stand or double-stand rolling mills, which are collectively referred to as wide and thick plate mother plates below.

[0046] Next, the optimized shearing method provided is specifically described, including the following steps:

[0047] Based on the MES system, historical wide and thick plate rolling data is obtained, and the rolling data includes the slab width W, length L and thickness H, the width w, length l and thickness h after rolling; and also includes the target length after the head and tail of the wide and thick plate mother plate are cut to size. , determine the cut-off head arc , determine the tail arc to be cut off And the length of the steel plate after rolling l;

[0048] The width w and thickness h are the online measurement data after rolling. The width w and thickness h after rolling are measured by the width gauge and thickness gauge respectively and obtained after data processing. The calculation formula of the length l after rolling is: .

[0049] Step S2: After rolling is completed, a sampling length S is selected in the length direction of the steel plate, and a sampling length S is selected. n Sampling points, real-time recording of the final rolling surface temperature of each sampling point , The surface temperature of the steel plate after water cooling starts And the final cooling surface temperature of the steel plate after water cooling , i =1,2,…… n , And take an integer; all the surface temperatures mentioned above are sampling temperatures, which can be obtained by measuring and recording with a pyrometer. The sampling length S is the minimum cutting length of the steel plate after deducting the head and tail. The sampling length is divided into several sections, and the sampling points are distributed in each section. A temperature value is recorded in each section.

[0050] Next is step S3. All kinds of data obtained in step S1 and step S2 are basic process data for performance prediction. These data are used to construct a performance prediction sample set for a wide and thick plate mother plate. A neural network model is constructed based on the sample set, and the neural network model is trained to finally obtain a performance prediction model. The mechanical property data of the wide and thick plate mother plate in the cooling process after rolling is obtained in real time through the performance prediction model.

[0051] It should be noted here that the method of constructing a neural network model is a relatively common and mature means, but this application has determined that the results obtained by this prediction model have the highest accuracy after screening tests, so this model is applied to this application. Of course, the input and output of the model are customized. Specifically, the neural network model includes an input layer and an output layer. The input items of the input layer used to train the learning samples include the chemical composition of the wide and thick plate mother plate, historical wide and thick plate rolling data, and production process parameters. The expected output items of the output layer include the mechanical properties data of the wide and thick plate mother plate during the cooling process after rolling. That is, according to the chemical composition, specifications, and actual production process parameters recorded in the length direction of the wide and thick plate mother plate that has completed the cooling process after rolling, the neural network model is used to predict the mechanical properties of the wide and thick plate mother plate that has completed the cooling process after rolling in real time.

[0052] The subsequent steps are the most important innovations of this application. Step S4: Based on the performance prediction model, the prediction is obtained. m Mechanical performance index n Performance data from m Among the mechanical performance indicators, determine the one with the largest performance fluctuation, defined as k , the performance index k As a calculation index for optimizing shearing;

[0053] In this step, the specific steps for determining the calculation index for optimized shearing are:

[0054] Step S41: predicting based on the performance prediction model m Mechanical performance index n The performance data is obtained as a matrix:

[0055] ;

[0056] Step S42, calculate the average value of each performance indicator, the calculation formula is: ,in , j represents the jth performance index, i represents the i-th calculated performance data, n represents the number of sampling points, and the average value of each mechanical performance index is obtained as follows: ;

[0057] Step S43, calculating the standard deviation of each mechanical performance index, the calculation formula is: ,in, ; The standard deviation formula of mechanical performance index is obtained as follows: ;

[0058] Step S44, calculating the dispersion coefficient by the average value of each mechanical performance index obtained in step S42 and the standard deviation of each mechanical performance index obtained in step S43, the calculation formula is: , the discrete coefficient formula is: ;

[0059] Step S45, calculate step S44 to obtain m The discrete coefficients of the mechanical performance indicators are sorted, and the maximum discrete coefficient is , then k The performance index fluctuates the most. k As a calculation index for optimizing shearing.

[0060] The main purpose of this step is to select the performance indicator with the largest dispersion coefficient from tensile strength, yield strength, impact and other properties to better represent the performance requirements of the wide and thick plate motherboard.

[0061] Step S5: Continue to obtain the target length of the thick and wide plate motherboard after cutting the head and tail from the MES system , determine the cut-off head arc , determine the tail arc to be cut off And the length of the rolled steel plate l.

[0062] Step S6: According to the selected k performance indicators, from wide and thick motherboard n Select continuous u A numerical value, that is u A series of performance indicators;

[0063] By calculation u The standard deviation of the performance indicators is selected as the minimum continuous u The performance index standard is set as follows: v Start to v + u -1 ends, then the target length is determined v Specific value of

[0064] The specific steps are:

[0065] Step S61, the obtained u Among the serialized performance indicators, , where the target length of the thick plate after cutting the head and tail Obtained from the MES system, u Take an integer, and ;

[0066] Step S62, according to the selected k performance indicators, and obtain the k The average value of performance indicators , standard deviation ;

[0067] Step S63, from the wide and thick board motherboard k performance indicators n Performance data , select Continuous u Step S64, calculate u The standard deviation of the performance index is ;

[0068] Step S64, for the continuous u Performance index standards Sort and select the smallest consecutive u Performance index standards ,in, , , finally determine the target length after cutting and rolling v The minimum internal performance standard deviation is .

[0069] In step S61-step S64, the wide and thick board motherboard k performance indicators n Performance data Obtained by matrix method in step S41, k The average value of performance indicators , Standard Deviation The same calculation method is used in steps S42 and S43 to finally determine the self v Start to v+u- Point 1 is the most uniform after performance comparison u points, the length of this segment is the length we need, 1 to v This distance is the length removed, that is, v Click on the distance in front to remove it.

[0070] Step S7: Combine step S6 and step S7 to determine the total length of the head cut And the tail cutting length , cut the total length of the head obtained And the tail cutting length It is sent to the head shear control system, and a shearing command is sent to the wide and thick plate motherboard to complete the cutting of the head and tail.

[0071] In order to verify the feasibility of the head-to-tail optimized shearing method of wide and thick plate mother plates based on performance prediction provided by this application, this application was applied to a domestic wide and thick plate production line. The production line is equipped with a manufacturing execution system (MES), a process control system, a basic automation system and an inspection and testing system. It can automatically calculate the rolling procedure and automatically roll steel according to the set values, and can collect and record the final rolling surface temperature in the length direction of the steel plate, the surface temperature at the beginning of water cooling, and the final cooling surface temperature at the end of water cooling. Based on the existing data, the temperature recording and performance prediction of the wide and thick plate mother plates in the length direction are realized.

[0072] The pipeline steel X70M product produced by the production line was selected, and the actual values ​​of the main chemical components are shown in Table 1.

[0073] Table 1 Actual chemical composition of X70M

[0074]

[0075] Slab Billet: , the blank size data is obtained from the MES system,

[0076] Size of wide and thick plate after rolling: , where thickness Measured by a thickness gauge, Measured by the width gauge. The target length of the customer's cut-to-length steel plate is 12500mm. The mother plate is rolled at 3 times the length. The target length after cutting off the head and tail , obtained from the MES system according to steel grade and thickness layer , . Sampling length s=60mm, sampling number n=674, u=625.

[0077] The recorded final rolling surface temperature is expressed as , Cooling surface temperature , Final cooling surface temperature As shown in Table 2.

[0078] Table 2 Recorded surface temperatures

[0079]

[0080] Table 3 Predicted X70M performance target requirements

[0081]

[0082] According to historical mechanical properties test data, the performance indicators of X70M with the largest fluctuations in the length direction are yield strength and tensile strength, elongation and impact energy of the steel plate, which meet customer requirements and have a certain margin. In this example, yield strength and tensile strength are selected for performance calculation, so the performance index data m =2, the yield strength is , the tensile strength is , i=1,2……,674.

[0083] Table 4 Predicted mechanical properties

[0084]

[0085] Calculating Yield Strength The mean ;

[0086] Calculate tensile strength The mean ;

[0087] Calculating Yield Strength Standard Deviation ;

[0088] Calculate tensile strength Standard Deviation ;

[0089] Calculation of Yield Strength Coefficient of Variation ;

[0090] Calculation of tensile strength coefficient of dispersion .

[0091] , the dispersion coefficient of tensile strength is greater than that of yield strength. Therefore, tensile strength is selected as the first k performance indicators.

[0092] From the motherboard n = 674 tensile strength values Select Continuous u = 625 tensile strength values, calculate the standard deviation of 625 consecutive tensile properties:

[0093] , , , , , , , , ;

[0094] Sort the 50 calculated tensile strength standard deviations, the smallest one is ,get v =27;

[0095] Make sure to cut off the head arc After that, the head cutting length , then the total length of the head cutting is ,

[0096] Determine the tail cutting length .

[0097] In combination with the above embodiments, the present application is based on the industrial big data of the steel plate production process, and predicts the performance of the mother plate along the length direction after the steel plate production is completed. A comparative analysis is performed based on the predicted performance indicators, and indicators with large performance fluctuations are selected as performance indicators for shear optimization. The performance of the performance indicator from beginning to end is considered, and the most uniform part of the performance indicator is selected. The shearing amount of the head and tail of the mother plate is calculated to achieve optimized shearing of the head and tail of the steel plate, reduce the revision caused by the performance of the head and tail exceeding the standard value, reduce the performance difference of the same plate of the mother plate, and achieve the purpose of optimized shearing control.

[0098] It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as generally understood by those skilled in the art to which this application belongs. It should also be understood that terms such as those defined in common dictionaries should be understood to have meanings consistent with the meanings in the context of the prior art, and will not be interpreted with idealized or overly formal meanings unless defined as herein.

[0099] The meaning of "and / or" described in this application means that the situations where each exists alone or both exist at the same time are included.

[0100] The term “connection” as used in this application may mean a direct connection between components or an indirect connection between components via other components.

[0101] Based on the above ideal embodiments of the present invention, the relevant staff can make various changes and modifications without departing from the technical concept of the present invention through the above description. The technical scope of the present invention is not limited to the contents of the specification, and its technical scope must be determined according to the scope of the claims.

Claims

1. A method for optimizing the head and tail shearing of a wide and thick plate motherboard based on performance prediction, characterized in that: The wide and thick plates targeted are those rolled by single-stand or double-stand medium and thick plate mills, which specifically include the following steps: Step S1: Based on the MES system, historical wide and thick plate rolling data is obtained, wherein the rolling data includes the slab width W, length L and thickness H, the width w, length l and thickness h after rolling; and the target length after the head and tail of the wide and thick plate mother plate are cut to size. , determine the cut-off head arc , determine the tail arc to be cut off And the length of the steel plate after rolling l; Step S2: After rolling is completed, select a sampling length in the length direction of the steel plate and select n Sampling points, real-time recording of the final rolling surface temperature of each sampling point , The surface temperature of the steel plate after water cooling starts And the final cooling surface temperature of the steel plate after water cooling , i =1,2,…… n ; The sampling length is the minimum cutting length after deducting the head and tail of the steel plate. The sampling length is divided into several sections, and the sampling points are distributed in each section. Step S3: The data obtained in step S1 and step S2 are used to form a sample set, a neural network model is constructed based on the sample set, and the neural network model is trained to finally obtain a performance prediction model; the mechanical properties data of the wide and thick plate mother plate in the cooling process after rolling are predicted in real time by the performance prediction model; Step S4: Predict based on the performance prediction model m Mechanical performance index n Performance data from m Among the mechanical performance indicators, determine the one with the largest performance fluctuation, defined as k , the performance index k As a calculation index for optimizing shearing; Step S5: Continue to obtain the target length of the thick and wide plate motherboard after cutting the head and tail from the MES system , determine the cut-off head arc , determine the tail arc to be cut off And the length of the steel plate after rolling l; Step S6: According to the selected k performance indicators, from wide and thick motherboard n Select continuous u A numerical value, that is u A series of performance indicators; By calculation u The standard deviation of the performance indicators is selected as the minimum continuous u The performance index standard is set as follows: v Start to v + u -1 ends, then the target length is determined v Specific value of Step S7: Combine step S6 and step S7 to determine the total length of the head cut And the tail cutting length , cut the total length of the head obtained And the tail cutting length It is sent to the head shear control system, and a shearing command is sent to the wide and thick plate motherboard to complete the cutting of the head and tail.

2. The method for optimizing the head and tail cutting of a wide and thick plate motherboard based on performance prediction according to claim 1 is characterized in that: In step S1, the width w and thickness h after rolling are measured and processed by a width gauge and a thickness gauge respectively, and the length l after rolling is calculated as follows: .

3. The method for optimizing the head and tail cutting of a wide and thick plate motherboard based on performance prediction according to claim 1 is characterized in that: In step S2, the sampling length is set to S, then , n Take the integer.

4. The method for optimizing the head and tail cutting of a wide and thick plate motherboard based on performance prediction according to claim 1 is characterized in that: In step S3, the neural network model includes an input layer and an output layer, the input layer is used to train the learning samples, and the input items include the chemical composition of the wide and thick plate mother plate, the historical wide and thick plate rolling data and the production process parameters, and the expected output items of the output layer include the mechanical properties data of the wide and thick plate mother plate in the cooling process after rolling.

5. The method for optimizing the head and tail cutting of a wide and thick plate motherboard based on performance prediction according to claim 1 is characterized in that: In step S4, the specific steps of determining the calculation index for optimizing shearing are: Step S41, predicting based on the performance prediction model m Mechanical performance index n The performance data is obtained as a matrix: ； Step S42, calculate the average value of each performance indicator, the calculation formula is: ,in , j represents the jth performance index, i represents the i-th calculated performance data, n represents the number of sampling points, and the average value of each mechanical performance index is obtained as follows: ; Step S43, calculating the standard deviation of each mechanical performance index, the calculation formula is: ,in, ; The standard deviation formula of mechanical performance index is obtained as follows: ; Step S44, calculating the dispersion coefficient by the average value of each mechanical performance index obtained in step S42 and the standard deviation of each mechanical performance index obtained in step S43, the calculation formula is: , the discrete coefficient formula is: ; Step S45, calculate step S44 to obtain m The discrete coefficients of the mechanical performance indicators are sorted, and the maximum discrete coefficient is , then k The performance index fluctuates the most. k As a calculation index for optimizing shearing.

6. The method for optimizing the head and tail cutting of a wide and thick plate motherboard based on performance prediction according to claim 1 is characterized in that: The specific steps in step S6 are: Step S61, the obtained u Among the serialized performance indicators, , where the target length of the thick plate after cutting the head and tail Obtained from the MES system, u Take an integer, and ; Step S62, according to the selected k performance indicators, and obtain the k The average value of performance indicators , standard deviation ; Step S63, from the wide and thick board motherboard k performance indicators n Performance data , select Continuous u Step S64, calculate u The standard deviation of the performance index is ; Step S64, for the continuous u Performance index standards Sort and select the smallest consecutive u Performance index standards ,in, , , finally determine the target length after cutting and rolling v The minimum internal performance standard deviation is .

7. The method for optimizing the head and tail cutting of a wide and thick plate motherboard based on performance prediction according to claim 1 is characterized in that: In step S7, the head arc to be cut off is obtained from the MES system through step S5. Then, the head cutting length is calculated. , the calculation formula is , then the total length of the head cutting for , tail cutting length for .

Images