Hot-rolled coil flattening prediction and prevention decision-making method based on data driving

Abstract

The invention provides a hot-rolled coil flattening prediction and prevention decision-making method based on data driving, and relates to the technical field of plate and strip production. The method comprises the steps that strip steel parameters and historical production data in the coiling process are collected; on the basis of the historical production data, a hot-rolled coil flattening prediction model is trained, the input of the hot-rolled coil flattening prediction model is flattening inducing factor data, and the output of the hot-rolled coil flattening prediction model is flattening preventing data; utilizing the trained hot-rolled coil flat coil prediction model to obtain a flat coil prediction result of the produced strip steel; and when the flat roll prediction result of the production strip steel is a flat roll condition, based on the flat inducing factor data corresponding to the production strip steel and the prediction result, screening out a proper flat roll prevention decision from similar strip steel which does not have a flat roll in history. The problems that in the prior art, whether frequently-changing incoming materials are flattened or not in the production process cannot be accurately predicted, and flattened rolling cannot be prevented through multi-angle adjustment can be solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of strip production, in particular to a hot-rolled coil flatness prediction and prevention decision-making method based on data driving. BACKGROUND

[0002] The hot-rolled coil shape defect is an important factor affecting the pass rate determination. After the hot-rolled coil generates the shape defect, corresponding measures must be taken for processing, which not only causes the increase of ton steel cost, but also affects the product delivery time, and more importantly, affects the production stability of the downstream process.

[0003] The hot-rolled coil flatness is a common type of hot-rolled coil shape defect, which refers to the change of the hot-rolled steel coil from a cylinder to an elliptical cylinder after being taken off line, which will cause the subsequent cold rolling process to fail to produce smoothly.

[0004] The main factors affecting the hot-rolled coil flatness are mechanical and phase change. Generally, the phase change is considered to cause the flatness: during the coiling process, there is still residual austenite in the steel coil, which will cause phase change during the cooling process. For different types of steel, different hot rolling and coiling processes will cause various forms of phase change, and the expansion amount caused by different phase changes is different. The volume change caused by the phase change will also cause interlayer slip. On the other hand, the volume shrinkage force occurs during the cooling of the steel coil. The steel material has a strong thermal expansion and cold shrinkage effect. The volume of the steel coil will shrink due to the temperature drop during cooling, and the steel coil will spontaneously show a "tight" phenomenon, which prevents the flatness. Whether the flatness occurs is the result of the competition between the two effects.

[0005] To prevent flatness, it is necessary to adjust various parameters in the coiling unit. At present, the existing technologies for preventing flatness mainly adjust the coiling temperature for specific steel types to prevent flatness. For example, patent CN102335681B proposes a coiling method for preventing hot-rolled strip flatness, which controls the coiling temperature at 500-600℃ and stays in the coiler for 20-60s to improve the flatness after coiling. Patent CN1506174A proposes a coiling method for preventing hot-rolled strip flatness, which coils the steel strip with carbon content greater than 0.25 from the finishing mill after laminar cooling, and controls the coiling temperature in the range of -10℃ to +60℃ to solve the flatness problem when the carbon content of the steel strip is greater than 0.25.

[0006] It is worth mentioning that in actual production, water spraying is often used during the coiling process of the strip. The water spraying will affect the phase change of the steel coil and form an internal support structure in the inner circle, which can prevent flatness to some extent.

[0007] In summary, the above techniques can improve the flatness problem of a specific steel grade to a certain extent, but cannot cope with the problem of frequent changes in incoming materials in actual production. Based on this, the present application proposes a hot rolling coil flatness prediction and decision-making method based on data driving, which can cope with the problem of frequent incoming materials, and prevent flatness by adjusting more dimensions of the unit parameters. SUMMARY

[0008] To solve the above technical problems existing in the prior art, the embodiments of the present application provide a hot rolling coil flatness prediction and prevention decision-making method based on data driving. The technical solution is as follows: A hot rolling coil flatness prediction and prevention decision-making method based on data driving, the method comprises: S1, collecting strip steel parameters and historical production data in the coiling process, and processing to obtain flatness inducing data set; S2, based on the flatness inducing data set, training a prediction hot rolling coil flatness model, wherein the input of the prediction hot rolling coil flatness model is strip steel grade data and strip steel production process parameters, and the output is whether the coil is flat or not; when the accuracy of the prediction hot rolling coil flatness model reaches 95% or more, the training is completed; The strip steel grade data includes steel grade, thickness, width and winding degree; the strip steel production process parameter data includes coiling temperature, coiling tension, cooling water temperature, cooling water intensity and cooling water flow; S3, using the trained prediction hot rolling coil flatness model, inputting the strip steel grade data of the to-be-produced strip steel and the strip steel production process parameters of the to-be-produced strip steel, and outputting the flatness prediction result of the to-be-produced strip steel; S4, when the flatness prediction result of the to-be-produced strip steel is flat, based on the flatness inducing factor data corresponding to the to-be-produced strip steel and the flatness prediction result, selecting appropriate strip steel production process parameter data from the production process parameter data of the strip steel with the same strip steel grade data as the to-be-produced strip steel and without flatness in history for production, which is used as the flatness prevention decision.

[0009] The S1 collects strip steel parameters and historical production data in the coiling process, and processes to obtain flatness inducing data set, comprising: S11, collecting the corresponding strip steel parameters and coiling unit historical production data and historical flatness condition data in the coiling process according to the strip steel number and steel grade; S12, detecting the mutation of the collected coiling unit historical production data in the coiling process, removing the mutation value, and taking the coiling unit historical production data in the coiling process after removing the mutation value as the unit flatness inducing data; S13, taking the collected strip steel parameters as incoming flatness inducing data; S14, correspond the collected historical flatness data, the unit flatness data, and the incoming material flatness data one by one according to the strip number and the steel grade, and form an integrated historical data set; S15, remove the data in the obtained historical data set, and obtain a historical data set after removing the data; S16, based on the historical data set after removing the data, establish a random forest algorithm model, calculate the weight coefficient of each index on causing flatness, and select the index with a weight coefficient exceeding 5% as a flatness inducing factor, The index is each item specification data of incoming material and each item production process parameter data in the coiling process. S17, select the data belonging to the flatness inducing factor from the historical data set after removing the data, and form a flatness inducing data set.

[0010] The strip parameters in S11 include strip thickness, strip width and strip winding degree. The historical production data of the coiling unit in the coiling process includes coiling temperature, coiling tension, cooling water temperature during coiling cooling, cooling water flow and cooling water jet intensity.

[0011] The mutation value in S12 is a variable value exceeding the original value by 150%-200%.

[0012] In S15, the data in the obtained historical data set is removed, specifically: for the same grade strip, when the differences between the strip thickness, strip width, strip winding degree, coiling temperature, coiling tension, cooling water temperature during coiling cooling, cooling water flow and cooling water jet intensity are each less than a preset difference threshold, and the flatness is the same, it is considered that the data corresponding to the strip is repeated, at this time only one strip data is retained, and a data set after removing the data is obtained.

[0013] In S16, the weight coefficient of each index is calculated based on the data set after removing the data, and each index is selected according to the weight coefficient, and the selected index is used as a flatness inducing factor, specifically: if the weight coefficient of the index exceeds a preset weight threshold, it is determined that it belongs to the flatness inducing factor; if the weight coefficient of the index does not exceed the preset weight threshold, it is determined that it does not belong to the flatness inducing factor, the index is the strip parameter and the historical production data in the coiling process, that is, the strip thickness, the strip width, the strip winding degree, the coiling temperature, the coiling tension, the cooling water temperature during coiling cooling, the cooling water flow and the cooling water jet intensity.

[0014] The prediction hot rolling coil flatness model in S2 is an existing genetic algorithm model; the training process includes: S21, according to the produced strip parameters, the coiling tension range adjusted by the coiler, the cooling capacity of the cooling system and the production plan, the upper and lower limit ranges of the strip parameters, the coiling tension, the cooling water temperature, the cooling water intensity and the cooling water flow are set respectively; This step is used to build a genetic algorithm model for realizing flat coil prediction, so as to prevent unsuitable strip from participating in the algorithm model and affecting normal production. S22, a flat coil prediction model for hot rolled coil is constructed, specifically, an existing genetic algorithm model is used, the flatting factors in the flatting data set are input, whether the flat coil is predicted by the model is output, and whether the flat coil is real in the flatting data set is compared, when the prediction accuracy is more than 95%, the training is completed.

[0015] The strip parameters of the strip to be produced and the unit production data are input into the trained flat coil prediction model for hot rolled coil in S3, and the strip flat coil or non-flat coil is output. When the flat coil of the steel plate is predicted, it is determined that the current strip cannot pass through the plate without changing the flatting factors of the unit. When the flat coil of the steel plate is predicted, it is determined that the current strip cannot pass through the plate without changing the flatting factors of the unit.

[0016] After preventing the flatness of the strip to be produced, the following judgment is further included: whether the flatness-prevented strip to be produced can pass through the plate stably; if the strip can pass through the plate stably, it is determined that the flat coil prediction model is accurate and does not need to be further corrected; if the strip cannot pass through the plate stably, the unit flatting data corresponding to the current strip at this time and the incoming flatting data are recorded and expanded to the flatting data set, and the expanded flatting data set is used for next flat coil prediction and flatness prevention analysis.

[0017] On the other hand, the present application provides an electronic device for a data-driven hot rolled coil flat coil prediction and decision method, comprising a processor and a storage medium; wherein the storage medium stores at least one instruction, which is loaded and executed by the processor to realize the above-mentioned method.

[0018] The technical scheme provided by the embodiment of the present application has at least the following beneficial effects: The above-described solution addresses the issue of flattening after hot-rolled coil unwinding. Based on historical production data, it analyzes factors inducing flattening and predicts the flattening status of the strip steel to be produced after coiling from multiple angles and dimensions. By comparing the flattening prediction with historical anti-flattening strategies, it develops anti-flattening strategies for the strip steel to be produced. This addresses the difficulty of existing technologies in covering multiple influencing factors and the frequent changes in hot-rolled coil incoming materials, thereby ensuring stable flattening of the strip steel after coiling and unwinding. Furthermore, compared to existing technologies, this invention considers more factors that may induce flattening, resulting in higher prediction accuracy. Moreover, this invention links the flattening prediction results with anti-flattening strategies, achieving a connection between flattening prediction and anti-flattening decision-making, which is more conducive to preventing hot-rolled coil flattening problems. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart of a data-driven hot-rolled coil flat coil prediction and prevention decision-making method provided in an embodiment of the present invention. Detailed Implementation

[0021] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0022] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0023] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0024] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0025] This invention provides a data-driven method for predicting and preventing hot-rolled coil flatness. For example... Figure 1 The flowchart shown illustrates a data-driven method for predicting and preventing hot-rolled coil flat coils. This method may include the following steps: S1, collect strip parameters and historical production data in the coiling process, and process to obtain a flat data set; S2, based on the flat data set, train a prediction hot-rolled flat coiling model, wherein the input of the prediction hot-rolled flat coiling model is strip product specification data and strip production process parameters, and the output is whether to flat coil; when the accuracy of the prediction hot-rolled flat coiling model reaches 95% or more, the training is completed; The strip product specification data includes steel grade, thickness, width and coiling degree; the strip production process parameter data includes coiling temperature, coiling tension, cooling water temperature, cooling water intensity and cooling water flow; S3, using the trained prediction hot-rolled flat coiling model, input the strip product specification data of the to-be-produced strip and the strip production process parameters of the to-be-produced strip, and output the flat coiling prediction result of the to-be-produced strip; S4, when the flat coiling prediction result of the to-be-produced strip is flat coiling, based on the flat coiling factor data corresponding to the to-be-produced strip and the flat coiling prediction result, the production process parameter data of the strip is selected from the production process parameter data of the strip with the same strip product specification data as the to-be-produced strip and without flat coiling in history, which is used for production as a flat prevention decision.

[0026] The following will be described in conjunction with specific embodiments.

[0027] Embodiment 1

[0028] The present application provides a technical solution: a hot-rolled flat coiling prediction and decision method based on data driving, comprising the following steps: S1, flat data acquisition and classification sub-process: Collecting strip parameters and historical production data in the coiling process; The historical production data includes flat inducing factors and flat coiling conditions in the historical production process; the historical production data is corresponded according to the strip number and the steel grade, that is, the data of the unit can be one-to-one corresponding to each strip.

[0029] Specifically, in this embodiment, S1 includes the following steps: S1-1, acquisition and preliminary analysis of unit flat data: According to the strip number and the steel grade, collect the coiling unit historical production data and flat coiling condition data of the corresponding coil; The coiling unit historical production data includes coiling temperature, coiling tension, cooling water temperature during coiling, cooling water flow, cooling water jet intensity, etc. For the collected data, the time domain variation law of each variable is analyzed preliminarily to determine whether there is a mutation value of 150% to 200% of the original value. If so, delete the value, calculate the true value of the point according to the difference value of the two points before and after the deletion point, and the difference value formula used is ; The historical production data of the coiler unit after removing the mutation value is used as the unit induction data; S1-2, collection and preliminary analysis of incoming material induction data: According to the strip number and steel grade, the corresponding strip parameters are collected; Among them, the storage medium strip parameters include strip thickness , strip width , and strip winding degree ; The strip thickness , strip width , and strip winding degree are used as incoming material induction data; S1-3, the collected historical flat winding data, unit induction data, incoming material induction data and environmental induction data are corresponded one by one according to the coil number and steel grade to form an integrated historical data set; S1-4, compare the strip data of the same grade, width, thickness and winding degree, analyze the flat winding condition, and if the flat winding condition is the same, only one data is retained in this series of data to avoid the influence of repeated data on the flat winding induction index weight analysis, and the de-duplicated data set is obtained; S1-5, based on the de-duplicated data set and the flat winding condition, a random forest algorithm model is established to calculate the weight coefficient of each index.

[0030] The storage medium indexes include: strip thickness, strip width, strip winding degree, coiling temperature, coiling tension, cooling water temperature during cooling, cooling water flow, and cooling water jet intensity; S1-6, screen each index according to the weight coefficient, and use the screened index as the flatness induction factor; if the weight coefficient of the index is more than 5%, it is determined that the index is a flatness induction factor; if the weight coefficient of the index is less than 3%, it is determined that the index is not a flatness induction factor; S1-7, organize the flatness induction factors to form a flatness induction index set, select the data belonging to the flatness induction factors from the de-duplicated data set to form a flatness induction data set.

[0031] S2, flatness prediction sub-process, Based on the storage medium historical production data, the flatness prediction sub-process is trained; The storage medium flatness prediction model is a genetic algorithm model, the input is the flatness induction factor data, and the output is whether it is flat or not. Using the trained model to obtain the flat coil prediction result of the to-be-produced strip steel: Specifically, in the present embodiment, S2 includes the following steps: S2-1, according to the production strip steel parameters, according to the adjustable coiling tension range of the coiler, the cooling capacity of the cooling system and the production plan, the base parameter fluctuation interval of each flatness inducing factor is set respectively, which is used to build a genetic algorithm model for realizing flat coil prediction, to prevent inappropriate strip steel from participating in the algorithm model and affecting normal production; S2-2, build a genetic algorithm model, and train the constructed genetic algorithm prediction model based on whether the strip steel is flat or not; wherein the output of the genetic algorithm of the storage medium is whether the strip steel is flat or not.

[0032] S2-3, using the trained flat coil prediction model to predict the to-be-produced strip steel: Using the trained flat coil prediction model to predict the flat coil condition of the to-be-produced strip steel; when it is predicted that the steel plate will be flat, it is determined that the current strip steel cannot be passed under the condition that the inducing factors of the unit are not changed; when it is predicted that the steel plate will not be flat, it is determined that the current strip steel can be passed.

[0033] S3, flat prevention action decision sub-process: When it is predicted that the to-be-produced strip steel will appear flat, based on the flat inducing factor data and the flat condition data corresponding to the to-be-produced strip steel, the strip steel with similar flat inducing factors and without flat from the historical production data is selected as the flat prevention measure by adjusting the inducing factors of the unit; Specifically, in the present embodiment, it includes the following steps: S3-1, extracting the data without flat from the historical production data set after deduplication in S1 as the flat prevention strategy set; S3-2, according to the parameters of the to-be-produced strip steel and the flat prevention strategy set, find the data similar in flat prevention factors, specifically, in the same kind of strip steel, the strip steel thickness , strip steel width , strip steel degree of winding The least square error of each parameter in the strategy set is similar, and the expression is:

[0034] Wherein is the thickness of the to-be-produced strip steel, is the width of the to-be-produced strip steel, is the winding degree of the to-be-produced strip steel, is the strip steel thickness of one of the strip steels in the strategy set with the same steel grade as the to-be-produced strip steel, is the strip steel width of one of the strip steels in the strategy set with the same steel grade as the to-be-produced strip steel, to collect the strip coils of one of the strip steels of the same steel grade as the strip steel to be produced in the strategy set; S3-3, after finding the most similar strip steel, extracting the mill flattening factor data thereof, and selecting the data suitable for the mill on site, such data can be referred to as optimal mill flattening data, and adjusting the flattening factor data on site to approximate the optimal mill flattening factor.

[0035] In summary, the embodiment provides a hot rolling flattening coil prediction and decision method based on data driving. The method can analyze the influence weight of mill flattening data and incoming flattening data according to historical production data through a random forest model, combine the historical actual flattening coil situation, establish a genetic prediction model, and then predict the flattening coil prediction situation of the incoming strip steel. At the same time, the flattening prevention strategy set is matched to obtain the flattening prevention strategy of the strip steel to be produced in advance. The method can run in parallel with the control system of the mill during the coiling process, and provides protection for the stable production scheduling of the hot rolling mill.

[0036] Embodiment 2

[0037] The embodiment uses an actual application example to illustrate the implementation process of the method of the present application, including: Based on a hot rolling production line of a certain steel plant, the historical data of the production line for two quarters is sorted and obtained, a total of 17369 strip steel data. Based on the above coil data, the flattening data collection and classification steps are performed, specifically including: S1-1, collecting historical production data of the hot rolling production line; Among them, the storage medium historical production data is corresponding according to the coil number and steel grade, that is, the data of the two mills can be one-to-one corresponding to each coil. According to the process of S1-1 to S1-3 in the above embodiment 1, the repeated data 2462 groups are screened out, and finally a data set containing 14907 coils is obtained. Among them, part of the collected data is as shown in Table 1.

[0038] Table 1 Collected data

[0039] (Table 1 continues horizontally)

[0040] S1-2, relying on the above data set, establishing a random forest algorithm model with whether the coil is flattened as the target value, and calculating the weight coefficient of each index. The calculation result is as shown in Table 2.

[0041] Table 2 Weight coefficient of each index

[0042] Through analysis, thickness, width, coiling temperature, coiling tension distribution, cooling water temperature, cooling water flow, and cooling water intensity are flattening factors. Further collate the data set of the factors of causing flatness, as shown in Table 3.

[0043] Table 3 Data of the data set of causing flatness after further collation

[0044] (Table 3 continued horizontally)

[0045] Subsequently, the flatness prediction sub-process is executed, specifically including: S2-1, according to the coiler set and the cooling capacity during coiling, production plan, respectively set the parameter fluctuation range of each flatness causing factor, as shown in Table 4.

[0046] Table 4 Parameter fluctuation range of each flatness causing factor

[0047] S2-2, construct a genetic prediction model, and combine the flatness of each coil number and steel grade mark to simulate training.

[0048] S2-3, use the trained genetic training model to predict the flatness of the to-be-produced strip steel.

[0049] Among them, the predicted flatness of the strip steel is shown in Table 5.

[0050] Table 5 Prediction of flatness of to-be-produced strip steel

[0051] (Table 5 continued horizontally)

[0052] According to the analysis result, the last two coils will not be flat, and the coiling can continue; the first coil will be flat, and the anti-coiling measure needs to be taken.

[0053] Further, for the flatness coil, the anti-flat action decision sub-process is executed, specifically including: S3-1, extract the data that does not occur flatness from the historical production data set after deduplication in S1 as the anti-flat strategy set.

[0054] S3-2, according to the parameters of the to-be-produced strip steel and the anti-flat strategy set, find the data similar in anti-flat factors.

[0055] S3-3, after finding the most similar strip steel, extracting its machine set coiling factor data, and selecting data suitable for the on-site machine set, such data can be called optimal machine set coiling data, adjusting the on-site coiling factor data to approximate the optimal machine set coiling factor. Specifically: the coiling temperature is increased to 670℃±10℃, the coiling tension, cooling water flow, cooling water intensity, and cooling water temperature do not need to be adjusted.

[0056] S3-4, performing the anti-flat action, determining that no flat roll occurs after unloading, and then the model does not need to be further corrected.

[0057] Example 3

[0058] The embodiment provides an electronic device, and a storage medium electronic device includes a processor and a storage medium; wherein the storage medium processor and the storage medium can be connected through a communication bus; the storage medium stores at least one instruction, and the storage medium processor loads and executes the instruction to realize the method of the first embodiment. In addition, the storage medium electronic device can further include a transceiver, the storage medium processor and the storage medium transceiver can be connected through a communication bus, and the storage medium transceiver is used for communicating with other devices.

[0059] Next, the various components of the electronic device will be specifically introduced: The processor is the control center of the electronic device, and the storage medium electronic device can include multiple processors, each of which can be a single-CPU or a multi-CPU. The processor here can be a processor or a general term for multiple processing elements. For example, the processor is one or more central processing units (CPU), which can also be other general-purpose processors, application specific integrated circuits (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application, such as one or more microprocessors (digital signal processors, DSP), or one or more field programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor. The processor can perform various functions of the storage medium electronic device by running or executing software programs stored in the storage medium and calling data stored in the storage medium.

[0060] In an embodiment of the application, the storage medium processor can further include one or more CPUs.

[0061] The storage medium is used to store a software program for implementing the solution of the application, and is controlled by the storage medium processor to perform the implementation, and the specific implementation can refer to the method embodiments described above, and thus will not be described here.

[0062] Alternatively, the storage medium can be a read-only memory (ROM) or other type of static storage device that can store static information and instructions, a random access memory (RAM) or other type of dynamic storage device that can store information and instructions, and can also be an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compact discs, laser discs, optical discs, digital versatile discs, Blu-ray discs, etc.), magnetic disc storage medium or other magnetic storage device, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and capable of being accessed by a computer, but not limited to this. The storage medium can be integrated with the processor, or can exist independently, and is coupled with the processor through the interface circuit of the storage medium electronic device, and the embodiments of the application do not make specific limitations here.

[0063] The storage medium transceiver can include a receiver and a transmitter. The storage medium receiver is used to realize the receiving function, and the storage medium transmitter is used to realize the transmitting function. The storage medium transceiver can be integrated with the storage medium processor, or can exist independently, and is coupled with the storage medium processor through the interface circuit of the storage medium electronic device, and the embodiments of the application do not make specific limitations here.

[0064] In addition, it should be noted that the actual device can include more or fewer components than those in the embodiment, or combine certain components, or different component arrangements. In addition, the technical effects achieved by the electronic device when executing the method of the first embodiment described above can refer to the technical effects of the storage medium of the first embodiment described above, and thus will not be described here.

[0065] The computer readable storage medium can be a ROM, a random access memory, a CD-ROM, a magnetic tape, a floppy disk, and an optical data storage device, etc. The instructions stored therein can be loaded and executed by the processor in the terminal to perform the above method.

[0066] The above merely illustrates the specific embodiments of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of the changes or replacements within the technical range disclosed by the present application, which should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A data-driven method for predicting and preventing hot-rolled coil flat coil defects, characterized in that, The method includes: S1. Collect strip steel parameters and historical production data during the coiling process, and process them to obtain the flattening dataset; S2. Based on the induced flattening dataset, train the model for predicting hot-rolled coil flattening. The input to the model is strip steel specification data and strip steel production process parameters, and the output is whether the coil is flattened. Training is complete when the accuracy of the model for predicting hot-rolled coil flattening reaches 95% or more. S3. Using the trained prediction model for hot-rolled coil flat coils, input the strip steel specification data and the strip steel production process parameters, and output the prediction results for the flat coils of the strip steel to be produced. S4. When the prediction result of the flattening of the strip steel to be produced is that flattening will occur, based on the data of the flattening inducing factors and the flattening prediction result of the strip steel to be produced, suitable strip steel production process parameters are selected from the production process parameters of strip steel that has never had flattening and whose strip steel specifications are the same as those of the strip steel to be produced, and used for production, so as to make anti-flattening decisions.

2. The data-driven hot-rolled coil flat coil prediction and prevention decision-making method according to claim 1, characterized in that, The S1 step involves collecting strip steel parameters and historical production data from the coiling process, and processing this data to obtain a flattening dataset, including: S11. Collect the corresponding strip parameters and historical production data and historical flat coil data of the coiling unit during the coiling process, according to the strip number and steel grade. S12. Perform mutation detection on the historical production data of the winding unit during the winding process, remove mutation values, and use the historical production data of the winding unit during the winding process after removing mutation values ​​as the unit's flattening data. S13. Use the collected strip steel parameters as the data for flattening incoming material; S14. The collected historical flattening data, unit flattening data, and incoming material flattening data are matched one by one according to the strip number and steel grade to form an integrated historical dataset; S15. Remove duplicate data from the obtained historical dataset to obtain the deduplicated historical dataset. S16. Based on the deduplicated historical dataset, establish a random forest algorithm model, calculate the weight coefficient of each indicator on the factors causing flattening, and filter the indicators according to the weight coefficients, selecting those with a weight coefficient exceeding 5% as factors inducing flattening. The indicators are the specifications of incoming materials and the process parameters of various production processes during the winding process; S17. Select data belonging to the flattening factor from the deduplicated historical dataset to form the flattening dataset.

3. The data-driven hot-rolled coil flat coil prediction and prevention decision-making method according to claim 2, characterized in that, The strip parameters in S11 include strip thickness, strip width, and strip coiling. Historical production data of the winding unit during the winding process includes winding temperature, winding tension, cooling water temperature during winding cooling, cooling water flow rate, and cooling water spray intensity.

4. The data-driven hot-rolled coil flat coil prediction and prevention decision-making method according to claim 2, characterized in that, In S12, the mutation value is a variable value that exceeds the original value by 150%-200%.

5. The data-driven hot-rolled coil flat coil prediction and prevention decision-making method according to claim 1, characterized in that, The strip steel specification data in S2 includes steel grade, thickness, width, and coil size; the strip steel production process parameter data includes coiling temperature, coiling tension, cooling water temperature, cooling water intensity, and cooling water flow rate.

6. The data-driven hot-rolled coil flat coil prediction and prevention decision-making method according to claim 1, characterized in that, The prediction model for hot-rolled coil flattened coil in S2 is an existing genetic algorithm model. The training process includes: S21. Based on the actual strip steel parameters, the coiling tension range adjusted by the coiler, the cooling capacity of the cooling system, and the production plan, set the upper and lower limits of strip steel parameters, coiling tension, cooling water temperature, cooling water intensity, and cooling water flow rate respectively. S22. Construct a model to predict whether hot-rolled coils will flatten. Specifically, use an existing genetic algorithm model, input the flattening factor parameters in the flattening dataset, output the model's prediction of whether the coil will flatten, and compare it with the actual flattening in the flattening dataset. When the prediction accuracy reaches more than 95%, the training is complete.

7. The data-driven hot-rolled coil flat coil prediction and prevention decision-making method according to claim 1, characterized in that, In S3, the strip parameters and unit production data of the strip to be produced are input into the trained predictive hot-rolled coil flat coil model, and the output is either flat coil or non-flat coil of strip steel. When the flattening of the steel plate is predicted, it is determined that the current strip cannot be passed through the plate without changing the flattening factors of the unit. If it is predicted that the steel plate will not be flattened and rolled, then it is determined that the current strip steel can pass through the plate.

Images