Parameter determination method and device, equipment, medium and program product

Abstract

The embodiment of the invention discloses a parameter determination method and device, equipment, a medium and a program product, and the method comprises the steps that the current rolling pass of a strip and the current rolling parameters of a rolling mill are obtained; the current rolling pass is used for representing the current number of times of rolling the strip by the rolling mill; determining strip shape information of the strip; a target model corresponding to the current rolling pass is determined from the multiple models; different models in the plurality of models correspond to different training modes; and by means of the target model, based on the strip shape information of the strip and the current rolling parameters of the rolling mill, strip shape control parameters, used for rolling the strip, of the rolling mill are determined. Therefore, compared with the prior art that the roll bending force is adjusted manually, the strip shape control parameters of the rolling mill are determined through the model, and the accuracy of determining the strip shape control parameters of the rolling mill can be improved.

Description

Technical Field

[0001] The present application relates to the field of metallurgical technology, and in particular to a parameter determination method, device, equipment, medium and program product. Background Art

[0002] At present, the shape control of the strip during the rolling process will directly affect the production efficiency and the final product quality. Poor shape control will lead to strip quality problems, and in serious cases, it will lead to production accidents and damage to mechanical equipment. After the strip is bitten, the operator needs to observe the strip shape at all times to manually adjust the bending roll force. Shape control depends on the operator's experience, and manual operation will inevitably lead to errors and lags. Summary of the invention

[0003] In view of this, embodiments of the present application at least provide a parameter determination method, apparatus, device, medium, and program product.

[0004] The technical solution of the embodiment of the present application is implemented as follows:

[0005] In a first aspect, an embodiment of the present application provides a parameter determination method, the method comprising:

[0006] Obtain the current rolling pass of the strip and the current rolling parameters of the rolling mill; the current rolling parameter current rolling pass is used to characterize the current number of times the current rolling parameter rolling mill rolls the strip;

[0007] Determine the strip shape information of the current rolling parameters;

[0008] Determine a target model corresponding to the current rolling parameters and the current rolling pass from the multiple models; different models in the multiple models of the current rolling parameters have different corresponding training methods;

[0009] The current rolling parameter target model is used to determine the shape control parameters of the current rolling parameter rolling mill for rolling the current rolling parameter strip based on the shape information of the current rolling parameter strip and the current rolling parameters of the current rolling parameter rolling mill.

[0010] In a second aspect, an embodiment of the present application provides a model training method, the method comprising:

[0011] Acquire a roll simulation model of a rolling mill and a strip simulation model of a strip;

[0012] Based on the current rolling parameter roll simulation model and the current rolling parameter strip simulation model, the rolling process of the current rolling parameter strip is simulated to obtain the simulation parameters;

[0013] Training the initial model based on current rolling parameter simulation parameters to obtain a first model;

[0014] Based on the real parameters obtained by rolling the strip by the rolling mill under the current rolling parameters, the first model of the current rolling parameters is trained to obtain at least one second model;

[0015] Among them, at least one second model of the current rolling parameters and the target model in the first model of the current rolling parameters are used to determine the plate shape control parameters of the current rolling parameter mill for rolling the current rolling parameter strip based on the plate shape information of the current rolling parameter strip and the current rolling parameters of the current rolling parameter mill; the current rolling parameter target model is determined based on the current rolling pass; the current rolling pass of the current rolling parameters is used to characterize the current number of times the current rolling parameter mill rolls the strip.

[0016] In a third aspect, an embodiment of the present application provides a parameter determination device, the parameter determination device comprising:

[0017] The first acquisition module is used to acquire the current rolling pass of the strip and the current rolling parameters of the rolling mill; the current rolling parameter current rolling pass is used to characterize the current number of times the current rolling parameter rolling mill rolls the strip;

[0018] An information determination module is used to determine the strip shape information of the current rolling parameters;

[0019] A model determination module is used to determine a target model corresponding to the current rolling parameters and the current rolling pass from multiple models; different models in the multiple models of the current rolling parameters correspond to different training methods;

[0020] The parameter determination module is used to determine the plate shape control parameters of the current rolling parameter rolling mill for rolling the current rolling parameter strip based on the plate shape information of the current rolling parameter strip and the current rolling parameters of the current rolling parameter rolling mill using the current rolling parameter target model.

[0021] In a fourth aspect, an embodiment of the present application provides a model training device, the model training device comprising:

[0022] A second acquisition module is used to acquire a roller simulation model of a rolling mill and a strip simulation model of a strip;

[0023] A simulation module is used to simulate the rolling process of the strip based on the roller simulation model and the strip simulation model to obtain simulation parameters;

[0024] A first training module, used for training the initial model based on simulation parameters to obtain a first model;

[0025] A second training module is used to train the first model based on real parameters obtained by rolling the strip by the rolling mill to obtain at least one second model;

[0026] Among them, at least one second model and the target model in the current rolling parameter first model are used to determine the plate shape control parameters of the current rolling parameter rolling mill for rolling the current rolling parameter strip based on the plate shape information of the current rolling parameter strip and the current rolling parameters of the current rolling parameter rolling mill; the current rolling parameter target model is determined based on the current rolling pass of the current rolling parameters; the current rolling pass of the current rolling parameters is used to characterize the current number of times the current rolling parameter rolling mill rolls the strip.

[0027] In a fifth aspect, an embodiment of the present application provides a computer device, including a memory and a processor, wherein the memory stores a computer program that can be executed on the processor, and when the processor executes the program, some or all of the steps in the above method are implemented.

[0028] In a sixth aspect, an embodiment of the present application provides a computer-readable storage medium having a computer program stored thereon, which implements some or all of the steps in the above method when executed by a processor.

[0029] In a seventh aspect, an embodiment of the present application provides a program product, which includes a computer program or instructions, and when the computer program or instructions are executed by a processor, some or all of the steps in the above method are implemented.

[0030] In the embodiment of the present application, the current rolling pass of the strip and the current rolling parameters of the rolling mill are first obtained, and then the corresponding target model is determined from multiple models through the current rolling pass. Finally, the target model is used to determine the plate shape control parameters of the rolling mill for rolling the strip based on the plate shape information of the strip and the current rolling parameters of the rolling mill. In this way, compared with the manual adjustment of the bending roll force in the related art, determining the plate shape control parameters of the rolling mill through the model can improve the accuracy of determining the plate shape control parameters of the rolling mill. In addition, the above-mentioned target model is determined based on the current rolling pass. Therefore, by determining the plate shape control parameters of the rolling mill of the current rolling pass through the target model corresponding to the current rolling pass, the accuracy of determining the plate shape control parameters of the rolling mill can also be improved.

[0031] It should be understood that the above general description and the following detailed description are merely exemplary and explanatory, and are not intended to limit the technical solutions of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] The drawings herein are incorporated into the specification and constitute a part of the specification. These drawings illustrate embodiments consistent with the present application and are used together with the specification to illustrate the technical solution of the present application.

[0033] Figure 1 A schematic diagram of an implementation process of a parameter determination method provided in an embodiment of the present application Figure 1 ;

[0034] Figure 2 A schematic diagram of an implementation process of a parameter determination method provided in an embodiment of the present application Figure 2 ;

[0035] Figure 3 A schematic diagram of an implementation process of a parameter determination method provided in an embodiment of the present application Figure 3 ;

[0036] Figure 4 A schematic diagram of an implementation process of a parameter determination method provided in an embodiment of the present application Figure 4 ;

[0037] Figure 5 A schematic diagram of the implementation process of a model training method provided in an embodiment of the present application Figure 1 ;

[0038] Figure 6 A schematic diagram of the implementation process of a model training method provided in an embodiment of the present application Figure 2 ;

[0039] Figure 7 A schematic diagram of the implementation process of a model training method provided in an embodiment of the present application Figure 3 ;

[0040] Figure 8 A schematic diagram of the implementation process of a model training method provided in an embodiment of the present application Figure 4 ;

[0041] Fig. 9 A schematic diagram of a strip rolling method according to an embodiment of the present application;

[0042] Fig.10 A schematic diagram of the implementation process of a model training method provided in an embodiment of the present application Figure 5 ;

[0043] Fig.11 A schematic diagram of a roller model provided in an embodiment of the present application;

[0044] Fig.12 A schematic diagram of the implementation process of a model training method provided in an embodiment of the present application Figure 6 ;

[0045] Fig.13 A schematic diagram of the implementation process of a model training method provided in an embodiment of the present application Figure 7 ;

[0046] Fig.14 A schematic diagram of the implementation process of a model training method provided in an embodiment of the present application Figure 8 ;

[0047] Fig.15A schematic diagram of an implementation method of a parameter determination method provided in an embodiment of the present application;

[0048] Fig.16 A schematic diagram of the structure of a parameter determination device provided in an embodiment of the present application;

[0049] Fig.17 A schematic diagram of the structure of a model training device provided in an embodiment of the present application;

[0050] Fig.18 A hardware entity diagram of a computer device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0051] In order to make the purpose, technical solutions and advantages of the present application clearer, the technical solutions of the present application are further elaborated in detail below in conjunction with the drawings and embodiments. The described embodiments should not be regarded as limiting the present application. All other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present application.

[0052] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments, but it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict. The terms "first / second / third" involved are merely used to distinguish similar objects and do not represent a specific order for the objects. It is understood that "first / second / third" may be interchanged in a specific order or sequence where permitted, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein.

[0053] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein are only for the purpose of describing this application and are not intended to limit this application.

[0054] During the hot rolling production process, due to the progress of the rolling process, the head data (such as rolling force, bending roll force) set by the secondary model in the hot rolling control system can no longer meet the plate shape control requirements of the entire strip. The preset value of the secondary model is only the average value of the head section of the strip. However, from the strip biting to the completion of rolling, the rolling force changes dynamically throughout the process, and the rolling force change is not linearly related to the bending roll force change. Therefore, the traditional setting value method will cause the plate shape value of the entire strip to differ greatly in the length direction. The plate shape difference will cause the following effects:

[0055] 1. The shape of the strip varies greatly along the length direction, which affects subsequent processing;

[0056] 2. Poor head shape will seriously affect the threading process and may cause threading failure;

[0057] 3. If the plate shape is not good, it will damage the mechanical equipment;

[0058] 4. Poor plate shape will affect the surface quality of the strip.

[0059] In summary, online real-time control of plate shape is very necessary in production. The current solution is mostly for operators to manually adjust to improve the plate shape. This method relies on the operator's experience and reaction speed, and the entire rolling process requires manual adjustment, which can no longer meet the current high-speed production rhythm and increasingly stringent product quality requirements.

[0060] In order to solve the above technical problems, an embodiment of the present application provides a parameter determination method, which can be executed by a processor of a computer device. The computer device may refer to a device with data processing capabilities such as a server. In some embodiments, the above computer device may be a device in a secondary process control system in a rolling production line. After determining the plate shape control parameters of the rolling mill, the computer device may send the plate shape control parameters of the rolling mill to the controller of the rolling mill, and the controller may control the rollers of the rolling mill to roll the strip based on the plate shape control parameters of the rolling mill.

[0061] Figure 1 A schematic diagram of an implementation process of a parameter determination method provided in an embodiment of the present application Figure 1 ,like Figure 1 As shown, the method includes the following steps S101 to S104:

[0062] Step S101, obtaining the current rolling pass of the strip and the current rolling parameters of the rolling mill.

[0063] Here, the current rolling pass is used to characterize the current number of times the rolling mill rolls the strip. The current number represents the number of times the rolling mill rolls the strip, and a complete rolling process of the strip belongs to one rolling pass. It can be understood that in order to achieve the final plate shape of the strip, the rolling mill needs to roll the strip multiple times, and the shape of the strip is different after each rolling, so it is necessary to determine the corresponding plate shape control parameters according to the current rolling pass. For example, when the current rolling pass is 3, it means that the rolling mill rolls the strip for the third time.

[0064] The current rolling parameters of the rolling mill refer to the rolling force parameters currently applied by the rolling mill to the plate. It is understandable that, after the plate enters the rolling rollers of the rolling mill, the shape and temperature of the plate will continue to change with the rolling process. Therefore, even in the same rolling pass, the current rolling parameters of the rolling mill will continue to change due to the changes in the shape and temperature of the plate. Therefore, it is necessary to continuously obtain the current rolling parameters of the rolling mill to determine the plate shape control parameters corresponding to the current rolling parameters. In this way, compared with the related art of obtaining the set average rolling force of the head section of the strip, obtaining the current rolling parameters of the rolling mill can improve the accuracy of determining the plate shape control parameters.

[0065] In some embodiments, the current rolling parameters of the rolling mill may be acquired according to a preset cycle, and after the plate shape control parameters of the current cycle are determined, the current rolling parameters of the rolling mill are acquired in the next cycle. For example, the cycle may be 200 ms.

[0066] Step S102, determining the shape information of the strip.

[0067] Here, the strip shape information indicates the strip convexity value, that is, the curvature of the strip cross-section. It is understandable that the size of the convexity value directly affects the running state of the strip. Too large a convexity value may cause uneven force on the strip, increase wear, and affect production efficiency. Therefore, it is necessary to determine the strip shape information.

[0068] In some embodiments, the strip shape information may include the target convexity information corresponding to the current rolling pass and the real-time convexity information of the strip. It is understandable that one of the goals of rolling the strip is to make the strip reach the final convexity value. Since the final convexity value cannot be reached through one rolling, it is necessary to first determine the target convexity information corresponding to each rolling pass, and then make the convexity value of the strip gradually approach the final convexity value through multiple rolling passes. In addition, in the early stage of rolling the strip, since the gap between the current convexity value of the strip and the final convexity value is large, it is not necessary to detect the real-time convexity information of the strip, and the target convexity information corresponding to the current rolling pass can be obtained at this time. In the later stage of strip rolling, the convexity value of the strip is very close to the final convexity value, and the real-time convexity information of the strip can be obtained at this time. Therefore, the strip shape information can be determined by determining the current rolling period of the strip.

[0069] In some embodiments, the strip shape information can be determined by the current rolling pass of the strip. It can be understood that, because the total number of rolling passes of the strip is certain, the current rolling period of the strip can be determined by the current rolling pass of the strip. For example, if the total number of rolling passes is 3 and the current rolling pass of the strip is 3, it can be determined that the current rolling period of the strip belongs to the late rolling period, so the real-time convexity information of the strip can be obtained.

[0070] Step S103, determining a target model corresponding to the current rolling pass from a plurality of models; different models in the plurality of models correspond to different training methods.

[0071] In an embodiment of the present application, multiple different models can be trained in advance based on different training methods, and then after the current rolling pass is obtained, the corresponding target model can be determined from the multiple models through the current rolling pass.

[0072] It is understandable that different strip shape information is obtained for different rolling passes, and the strip shape information is used as the input parameter of the target model. When the input parameters are different, the target model will also change accordingly. Therefore, it is necessary to determine the corresponding target model according to the current rolling pass so that the input parameters of the model correspond to the model.

[0073] In some embodiments, the training methods corresponding to different models refer to the differences in sample data sets used for training. Exemplarily, the different sample data sets may include: a sample data set containing real parameters and a sample data set containing simulation parameters.

[0074] It is understandable that, because it is not necessary to detect the convexity value of the strip in the early stage of rolling, it is impossible to obtain the real parameters of the rolling process. At this time, a model can be obtained by training a sample data set including simulation parameters, and then the plate shape control parameters of the strip in the early stage of rolling can be obtained by the model trained based on the simulation parameters. Similarly, because the real parameters of the rolling process can be obtained in the later stage of rolling of the strip, a model can be obtained by training a sample data set including real parameters, and then the plate shape control parameters of the strip in the later stage of rolling can be obtained by the model trained based on the real parameters. In this way, the plate shape control parameters corresponding to the current rolling pass can be determined.

[0075] In some embodiments, the above-mentioned multiple models may be a deep neural network (DNN). Exemplarily, the deep neural network may be a fully connected neural network (FCNN).

[0076] Step S104: using the target model, based on the strip shape information and the current rolling parameters of the rolling mill, determine the strip shape control parameters of the rolling mill for rolling the strip.

[0077] In the embodiment of the present application, the strip shape information and the current rolling parameters of the rolling mill can be input into the target model to obtain the strip shape control parameters of the rolling mill for rolling the strip.

[0078] Here, the plate shape control parameters of the strip can represent the plate shape control parameters of the rolling mill in the current cycle. It can be understood that the rolling time of a rolling pass of the strip is relatively long. During the rolling time of a rolling pass, the current rolling parameters of the rolling mill will change due to changes in the shape and temperature of the strip, so the plate shape control parameters of the rolling mill output by the target model will also change. Therefore, in the current rolling pass, the target model can output different plate shape control parameters as the current rolling parameters of the rolling mill are continuously changed. Exemplarily, the plate shape control parameters of the above-mentioned rolling mill can be the bending roll force of the rolling mill.

[0079] In the embodiment of the present application, the above-mentioned real parameters and simulation parameters used for training the model include the rolling parameters and plate shape control parameters of the rolling mill. That is to say, whether it is a model trained by real parameters or a model trained by simulation parameters, the corresponding plate shape control parameters of the rolling mill can be determined by the current rolling parameters of the rolling mill. In this way, the present application constructs the relationship between rolling parameters and plate shape control parameters through a network model, which can more accurately determine the plate shape control parameters of the rolling mill for rolling strips compared with the linear relationship between rolling force change and bending roll force change in the related art.

[0080] In some embodiments, after the computer device determines the plate shape control parameter of the rolling mill, the plate shape control parameter can be sent to the controller of the rolling mill, and the controller can control the rollers of the rolling mill to roll the strip based on the plate shape control parameter. Exemplarily, the rolling process can be a hot rolling process.

[0081] In the embodiment of the present application, the current rolling pass of the strip and the current rolling parameters of the rolling mill are first obtained, and then the corresponding target model is determined from multiple models through the current rolling pass. Finally, the target model is used to determine the plate shape control parameters of the rolling mill for rolling the strip based on the plate shape information of the strip and the current rolling parameters of the rolling mill. In this way, compared with the manual adjustment of the bending roll force in the related art, determining the plate shape control parameters of the rolling mill through the model can improve the accuracy of determining the plate shape control parameters of the rolling mill. In addition, the above-mentioned target model is determined based on the current rolling pass. Therefore, by determining the plate shape control parameters of the rolling mill of the current rolling pass through the target model corresponding to the current rolling pass, the accuracy of determining the plate shape control parameters of the rolling mill can also be improved.

[0082] Figure 2 A schematic diagram of an implementation process of a parameter determination method provided in an embodiment of the present application Figure 2 , the method can be executed by a processor of a computer device. Figure 1 , Figure 1 S103 in can be updated to S201 and S202, which will be combined Figure 2 The steps shown are explained.

[0083] Step S201, based on the type information of the rolling mill and the current rolling pass, determine the stage information of the current rolling pass in the rolling process of the strip.

[0084] Here, the type information of the rolling mill may indicate the number of stands of the rolling mill. For example, the type information of the rolling mill may be a single-stand reciprocating rolling mill, indicating that the number of stands of the rolling mill is 1; the type information of the rolling mill may be a 3-stand continuous rolling mill, indicating that the number of stands of the rolling mill is 3; the type information of the rolling mill may be a 4-stand continuous rolling mill, indicating that the number of stands of the rolling mill is 4.

[0085] The stage information of the current rolling pass is used to indicate the position stage of the current rolling pass in the rolling process in time. The rolling process can be divided into the head stage, the middle stage and the tail stage, and each stage can include at least one rolling pass. In actual scenarios, when the rolling process of the strip is in the tail stage, the detection device at the tail of the rolling mill will collect the real-time plate shape information of the strip. In other words, the stage information of the current rolling pass can be used to determine whether the real-time plate shape information of the strip can be obtained.

[0086] In some embodiments, the rolling pass range corresponding to different stages can be determined first by the type information of the rolling mill and the total rolling passes of the strip. For example, if the total rolling passes are 5 and the rolling mill is a 3-stand continuous rolling mill, the rolling pass range corresponding to the tail stage can be determined to be 3 to 5, that is, the last 3 passes of the total rolling passes can obtain the real-time plate shape information of the strip; the rolling pass range corresponding to the non-tail stage (i.e., the head stage and the middle stage) is 1 and 2. After determining the rolling pass range corresponding to different stages, the stage information of the current rolling pass can be determined based on the current rolling pass and the rolling pass range corresponding to different stages. For example, if the current rolling pass is 3, and the rolling pass range corresponding to the tail stage is 3 to 5, it can be determined that the stage information of the current rolling pass in the rolling process of the strip is the tail stage; if the current rolling pass is 1, and the rolling pass range corresponding to the tail stage is 3 to 5, it can be determined that the stage information of the current rolling pass in the rolling process of the strip is the non-tail stage.

[0087] Step S202: determining a target model from a plurality of models based on stage information of a current rolling pass in the rolling process of the strip.

[0088] In the embodiment of the present application, the correspondence between different stage information and different models can be obtained, and the target model can be determined from multiple correspondences through the stage information of the current rolling pass.

[0089] In the embodiment of the present application, the stage information of the current rolling pass in the rolling process of the strip can be determined first by the type information of the rolling mill and the current rolling pass; then, based on the stage information of the current rolling pass in the rolling process of the strip, the target model can be determined from multiple models. In this way, by determining the stage information of the current rolling pass in the rolling process of the strip, the rolling progress of the strip can be accurately determined, so that the target model corresponding to the current rolling progress can be obtained, and then the accuracy of determining the plate shape control parameters of the rolling mill can be improved.

[0090] Figure 3 A schematic diagram of an implementation process of a parameter determination method provided in an embodiment of the present application Figure 3 The method can be executed by a processor of a computer device. The above multiple models include a first model obtained by training based on simulation parameters and a second model obtained by training based on real parameters; the above S202 can be updated to S301 and S302, combining Figure 3 The steps shown are explained.

[0091] Step S301, when the stage information of the current rolling pass is a non-tail stage in the rolling process of the strip, a first model among multiple models is determined as a target model.

[0092] In an embodiment of the present application, when the stage information of the current rolling pass is a non-tail stage in the rolling process of the strip, it means that the strip is currently in the initial rolling stage. At this time, the detection device will not collect the real-time plate shape information of the strip, that is, the first model obtained based on the simulation parameter training can be used to determine the plate shape control parameters of the rolling mill.

[0093] In some embodiments, the rolling process of the strip can be simulated based on the roller simulation model and the strip simulation model to obtain simulation parameters; and the initial model can be trained based on the simulation parameters to obtain the first model.

[0094] Step S302: when the stage information of the current rolling pass is the tail stage in the rolling process of the strip, the second model among the multiple models is determined as the target model.

[0095] In an embodiment of the present application, when the stage information of the current rolling pass is the tail stage in the rolling process of the strip, it means that the strip is currently in the final stage of rolling. At this time, the detection device will collect the real-time plate shape information of the strip, that is, the second model obtained by training based on the real parameters can be used to determine the plate shape control parameters of the rolling mill.

[0096] In some embodiments, the first model is trained based on real parameters obtained by rolling the strip by a rolling mill to obtain at least one second model.

[0097] In other embodiments, the initial model may also be trained based on real parameters obtained by rolling the strip by the rolling mill to obtain at least one second model.

[0098] In the embodiment of the present application, when the stage information of the current rolling pass is the non-tail stage in the rolling process of the strip, the first model obtained by training based on simulation parameters among multiple models is determined as the target model; when the stage information of the current rolling pass is the tail stage in the rolling process of the strip, the second model obtained by training based on real parameters among multiple models is determined as the target model. In this way, by determining the target model through the stage information of the current rolling pass, the target model can output the plate shape control parameters corresponding to the stage information of the current rolling pass, thereby improving the accuracy of the rolling mill rolling the strip.

[0099] Figure 4 A schematic diagram of an implementation process of a parameter determination method provided in an embodiment of the present application Figure 4 , the method may be executed by a processor of a computer device. Figure 4 As shown, the parameter determination method can also be implemented through step S401, and the above step S104 can be implemented through step S402:

[0100] Step S401: determining a target second model in at least one second model based on the alloy type of the strip and target thickness information of the strip.

[0101] Here, the target thickness information of the strip refers to the thickness information of the strip expected to be obtained after rolling, that is, the target thickness information of the strip is the target value. In order to make the thickness information of the strip finally obtained meet the target thickness information, it is necessary to determine the corresponding rolling process, and different rolling processes correspond to different plate shape control parameters. At the same time, different alloy types correspond to different rolling processes, even if the target thickness information is the same. Therefore, in the process of training the model to obtain the second model, it is necessary to add the alloy type of the strip and the target thickness information of the strip to the sample data, so as to obtain at least one second model. The different second models correspond to different alloy types and / or different target thickness information.

[0102] In the embodiment of the present application, each second model corresponds to its own alloy type and target thickness information. Therefore, the target second model can be determined in at least one second model through the alloy type of the strip and the target thickness information of the strip.

[0103] Step S402: using the target second model, based on the strip shape information and the current rolling parameters of the rolling mill, determine the strip shape control parameters of the rolling mill for rolling the strip.

[0104] In the embodiment of the present application, the plate shape information of the strip and the current rolling parameters of the rolling mill can be input into the target second model, and the target second model outputs the plate shape control parameters of the rolling mill for rolling the strip.

[0105] In the embodiment of the present application, the target second model is determined in at least one second model by the alloy type of the strip and the target thickness information of the strip, and then the target second model is used to determine the plate shape control parameters of the rolling mill for rolling the strip based on the plate shape information of the strip and the current rolling parameters of the rolling mill. In this way, by determining the target second model by the alloy type of the strip and the target thickness information of the strip, the target second model can output the plate shape control parameters corresponding to the alloy type of the strip and the target thickness information of the strip, thereby improving the accuracy of the rolling mill rolling the strip.

[0106] In some embodiments, the above step S102 may be implemented by step S1021 or step S1022:

[0107] Step S1021, when the stage information of the current rolling pass is the tail stage in the rolling process of the strip, obtain the real-time convexity information of the strip and determine the target width information of the strip, or determine the target convexity information of the strip and the target width information of the strip.

[0108] In the embodiment of the present application, when the stage information of the current rolling pass is determined based on the type information of the rolling mill and the current rolling pass, it means that the current rolling pass is at the end of the strip rolling process. At this time, when determining the strip shape information, two situations may occur:

[0109] The first case: the real-time convexity information of the strip can be obtained, because the collection device for collecting the real-time convexity information of the strip is at the exit of the rolling mill, and the collection device is a certain distance away from the exit of the rolling mill, so this case means that the head section of the strip has gone out of the exit of the rolling mill and the head reaches the collection device, so the collection device can collect the real-time convexity information of the strip. Therefore, in this case, the real-time convexity information of the strip is obtained and the target width information of the strip is determined.

[0110] The second situation: the real-time convexity information of the strip cannot be obtained. This situation refers to the head section of the strip going out from the exit of the rolling mill, but the head section has not yet reached the collection device, so the collection device cannot collect the real-time convexity information of the strip. Therefore, in this case, the target convexity information of the strip and the target width information of the strip are determined.

[0111] In the embodiment of the present application, the target width information of the strip corresponds to the current rolling pass. It is understandable that the goal of rolling the strip includes not only making the strip reach the final convexity value, but also making the strip reach the final width. Because the final width cannot be achieved through one rolling, it is necessary to first determine the target width information corresponding to each rolling pass, and then through multiple rolling passes, the width of the strip gradually meets the final width. Therefore, the target width information of each rolling pass can be determined based on the final width information. Thereby, the target width information of the current rolling pass can be determined based on the target width information corresponding to multiple rolling passes.

[0112] In the embodiment of the present application, the method of determining the target convexity information of the strip and the method of determining the target width information of the strip, that is, determining the target convexity information of each rolling pass according to the final convexity information. Thus, the target convexity information of the current rolling pass can be determined from the target convexity information corresponding to multiple rolling passes.

[0113] In some embodiments, after obtaining the real-time convexity information of the strip and determining the target width information of the strip, the above step S104 can be implemented by step S1041:

[0114] Step S1041, using the second model, based on the target width information of the strip, the real-time convexity information of the strip and the current rolling parameters of the rolling mill, determine the plate shape control parameters of the rolling mill for rolling the strip.

[0115] In the embodiment of the present application, the second model can be any model trained based on real parameters obtained by rolling the strip by a rolling mill.

[0116] In some embodiments, a target second model may be first determined in at least one second model based on the alloy type of the strip and the target thickness information of the strip, and then the target second model may be used to determine the plate shape control parameters of the rolling mill for rolling the strip based on the target width information of the strip, the real-time convexity information of the strip and the current rolling parameters of the rolling mill.

[0117] In some embodiments, when the rolling mill hot-rolls the strip, because the rollers will expand thermally, in order to improve the accuracy of determining the plate shape control parameters, the thermal expansion values ​​of the rollers can be obtained, and then a second model is used to determine the plate shape control parameters of the rolling mill for rolling the strip based on the target width information of the strip, the real-time convexity information of the strip, the current rolling parameters of the rolling mill, and the thermal expansion values ​​of the rollers.

[0118] In some embodiments, after determining the target convexity information of the strip and the target width information of the strip, the above step S104 can be implemented by step S1042:

[0119] Step S1042, using the second model, based on the target width information of the strip, the target convexity information of the strip and the current rolling parameters of the rolling mill, determine the plate shape control parameters of the rolling mill for rolling the strip.

[0120] In the embodiment of the present application, the second model can be any model trained based on real parameters obtained by rolling the strip by a rolling mill.

[0121] In some embodiments, a target second model may be first determined in at least one second model based on the alloy type of the strip and the target thickness information of the strip, and then the target second model may be used to determine the plate shape control parameters of the rolling mill for rolling the strip based on the target width information of the strip, the target convexity information of the strip and the current rolling parameters of the rolling mill.

[0122] In some embodiments, when the rolling mill hot-rolls the strip, because the rollers will expand thermally, in order to improve the accuracy of determining the plate shape control parameters, the thermal expansion values ​​of the rollers can be obtained, and then a second model is used to determine the plate shape control parameters of the rolling mill for rolling the strip based on the target width information of the strip, the target convexity information of the strip, the current rolling parameters of the rolling mill, and the thermal expansion values ​​of the rollers.

[0123] Step S1022, when the stage information of the current rolling pass is a non-tail stage in the rolling process of the strip, determine the target convexity information of the strip and the target width information of the strip.

[0124] In the embodiment of the present application, when the stage information of the current rolling pass is determined based on the type information of the rolling mill and the current rolling pass as the non-tail stage of the strip rolling process, it means that the current rolling pass is in the initial stage of the strip rolling. At this time, because the current strip convexity value is far from the final target convexity value, the acquisition device will not collect the current strip convexity value. Therefore, the target convexity information corresponding to the current rolling pass can be used as the strip shape information and input into the model.

[0125] In some embodiments, after determining the target convexity information of the strip and the target width information of the strip, the above step S104 can be implemented by step S1043:

[0126] Step S1043, using the first model, based on the target convexity information of the strip, the target width information of the strip and the current rolling parameters of the rolling mill, determine the plate shape control parameters of the rolling mill for rolling the strip.

[0127] In the embodiment of the present application, when the stage information of the current rolling pass is the tail stage of the strip rolling process, the real-time convexity information of the strip is obtained and the target width information of the strip is determined, or the target convexity information of the strip and the target width information of the strip are determined; when the stage information of the current rolling pass is the non-tail stage of the strip rolling process, the target convexity information of the strip and the target width information of the strip are determined. In this way, different input parameters of the model can be determined based on the different stages of the current rolling pass, thereby improving the accuracy of the model in determining the strip shape control parameters.

[0128] Figure 5 A schematic diagram of the implementation process of a model training method provided in an embodiment of the present application Figure 1 , the method may be executed by a processor of a computer device. Figure 5 As shown, the model training method is implemented through steps S501 to S504:

[0129] Step S501, obtaining a roller simulation model of a rolling mill and a strip simulation model of a strip.

[0130] In the embodiment of the present application, a roller simulation model can be constructed according to the size parameters of the roller of the rolling mill. For example, the size parameters of the roller may include the diameter, length, and neck data of the roller. Similarly, a strip simulation model can be constructed according to the size parameters of the strip.

[0131] Step S502: Based on the roller simulation model and the strip simulation model, simulate the rolling process of the strip to obtain simulation parameters.

[0132] In the embodiment of the present application, the rolling process of the strip can be simulated by finite element analysis (FEA).

[0133] In some embodiments, during the finite element analysis of the roller simulation model, the constraint conditions may be continuously changed so as to obtain simulation parameters in various situations.

[0134] Step S503: training the initial model based on the simulation parameters to obtain a first model.

[0135] In the embodiment of the present application, a sample data set for training the first model can be constructed based on the simulation parameters, and then the initial model can be trained by the sample data set. The sample data set for training the first model includes sample data and label data. The sample data can be the current rolling parameters of the roller simulation model during the simulated rolling process and the plate shape information of the strip simulation model; the label data can be the plate shape control parameters of the roller simulation model.

[0136] Exemplarily, the initial model may be a fully connected neural network, the intermediate hidden layer data of the fully connected neural network may be a single layer or multiple layers, and the number of neurons in each layer may be 18 to 25.

[0137] Step S504: training the first model based on real parameters obtained by rolling the strip by the rolling mill to obtain at least one second model.

[0138] Here, the training of the first model can be regarded as the self-learning process of the first model in the actual rolling process. The self-learning time of the first model can be when the strip is rolled, at which time the real parameters obtained by the rolling mill when rolling the strip can be obtained, so that the self-learning process of the first model can be completed based on the real parameters.

[0139] In the embodiment of the present application, different corresponding second models can be generated for strips of different materials and different target width information of the strips.

[0140] Among them, at least one second model and the target model in the first model are used to determine the plate shape control parameters of the rolling mill for rolling the strip based on the plate shape information of the strip and the current rolling parameters of the rolling mill; the target model is determined based on the current rolling pass; the current rolling pass is used to characterize the current number of times the rolling mill rolls the strip.

[0141] In the embodiment of the present application, based on the roller simulation model and the strip simulation model, the rolling process of the strip is simulated to obtain simulation parameters; the initial model is trained based on the simulation parameters to obtain a first model; based on the real parameters obtained by the rolling mill rolling the strip, the first model is trained to obtain at least one second model. In this way, different models can be obtained by training the model with different types of parameters (i.e., simulation parameters and real parameters), so that the plate shape control parameters of the rolling mill can be determined in the actual production process using a model that meets the current scenario, thereby improving the accuracy of the rolling mill rolling the strip.

[0142] Figure 6 A schematic diagram of the implementation process of a model training method provided in an embodiment of the present application Figure 2 , the method may be executed by a processor of a computer device. Figure 6 As shown, the above step S502 can be implemented through steps S601 to S603:

[0143] Step S601, determining at least one constraint condition of the roller simulation model during the simulated rolling process.

[0144] Here, the constraint conditions include a plurality of constraint parameters, which include rolling force parameters of the roll simulation model, size parameters of the strip simulation model, and bending roll force parameters of the roll simulation model.

[0145] In the embodiment of the present application, based on multiple constraint parameters of the constraint condition, at least one constraint condition may include at least one of the following:

[0146] The rolling force parameters of the roll simulation model are fixed, the size parameters of the strip simulation model are fixed, and the bending force parameters of the roll simulation model are variable;

[0147] The rolling force parameters of the roll simulation model are fixed, the size parameters of the strip simulation model are varied, and the bending force parameters of the roll simulation model are fixed;

[0148] The rolling force parameters of the roll simulation model vary, the size parameters of the strip simulation model are fixed, and the bending roll force parameters of the roll simulation model are fixed.

[0149] Among them, for the first constraint condition, the rolling force parameter of the roll simulation model and the size parameter of the strip simulation model can be given. Exemplarily, the rolling force parameter can be 10000KN, and the size parameter of the strip simulation model can be the width of the strip simulation model, such as 1500mm. Then the bending roll force parameter of the roll simulation model is continuously changed within the preset bending roll force range. Exemplarily, the preset bending roll force range is -800KN to 800KN, and it can be changed from -800KN to 800KN. In some embodiments, the size of the bending roll force can be controlled each time it is changed, for example, the size of the bending roll force is changed each time without exceeding 10KN, and it can be changed from -800KN to 800KN, which can be -800KN, -790KN, -780KN, ..., 750KN, 760KN, 770KN, 780KN, 790KN, 800KN.

[0150] For the second constraint, the rolling force parameters of the roll simulation model and the bending roll force parameters of the roll simulation model can be given. Exemplarily, the rolling force parameter can be 10000KN, and the bending roll force parameter can be 500KN. Then the size parameters of the strip simulation model are continuously changed within the preset size parameter range. Exemplarily, the size parameter is the width of the strip simulation model, and the preset size parameter range is 1000mm to 2000mm, which can be changed from 1000mm to 2000mm. In some embodiments, the size of each change in the size parameter can be controlled, for example, each time the size of the size parameter is changed by 10mm, it changes from 1000mm to 2000mm, which can be 1000mm, 1010mm, 1020mm, 1030mm, ..., 1970mm, 1980mm, 1990m, 2000mm. For the third constraint, the size parameters of the strip simulation model and the bending roll force parameters of the roll simulation model can be given. Exemplarily, the size parameter is the width of the strip simulation model, the width is 1500mm, and the bending roll force parameter can be 500KN. Then the rolling force parameter of the roll simulation model is continuously changed within the preset rolling force parameter range. Exemplarily, the preset rolling force parameter range is 1000KN to 50000KN, and it can be changed from 1000KN to 50000KN. In some embodiments, the size of each change of the rolling force parameter can be controlled. For example, each time the size of the rolling force parameter is changed by 1000KN, it changes from 1000KN to 50000KN, which can be 1000Kn, 2000KN, 3000KN, ..., 47000KN, 48000KN, 49000KN, 50000KN.

[0151] It can be seen from the above examples that each constraint condition contains multiple constraint sub-conditions, so that various rolling conditions in real scenes can be simulated as much as possible, and different plate shape information of the strip simulation model under different rolling scenes can be obtained.

[0152] Step S602, for each constraint condition, based on the roller simulation model and the strip simulation model, simulate the rolling process of the strip under the constraint condition to obtain the plate shape information of the strip simulation model.

[0153] Step S603: determining a constraint parameter in at least one constraint condition and the plate shape information of the strip simulation model corresponding to each constraint condition as simulation parameters.

[0154] In the embodiment of the present application, under each constraint condition, based on the roller simulation model and the strip simulation model, the constraint parameters in the constraint condition can be used to simulate the process of rolling the strip by the rolling mill, and then after the rolling is completed, the plate shape information of the strip simulation model is collected. The plate shape information can be the convexity value of the strip simulation model after rolling.

[0155] In an embodiment of the present application, when the constraint conditions change, finite element calculations can be performed on the roller simulation model and the strip simulation model to obtain force and deformation data on the contact surface between the roller simulation model and the strip simulation model in the length direction, and finally obtain the plate shape information of the strip simulation model.

[0156] In the embodiment of the present application, by determining at least one constraint condition, the strip shape information generated by the strip simulation model under different constraint conditions after simulated rolling can be obtained. In this way, different strip shape information of the plate under different rolling parameters can be obtained in a virtual environment, so that a variety of sample data with large differences can be obtained, thereby improving the accuracy of the training model.

[0157] Figure 7 A schematic diagram of the implementation process of a model training method provided in an embodiment of the present application Figure 3 , the method may be executed by a processor of a computer device. Figure 7 As shown, the above step S503 can be implemented through step S701 and step S702:

[0158] Step S701, using the rolling force parameters of the roller simulation model, the size parameters of the strip simulation model and the plate shape information of the strip simulation model corresponding to at least one constraint condition as sample data, and the bending roll force parameters of the roller simulation model corresponding to at least one constraint condition as label data, to construct a first sample data set.

[0159] Step S702: training an initial model based on a first sample data set to obtain a first model.

[0160] It can be understood that the purpose of training the model is to construct parameters between the rolling force parameters and the bending roll parameters of the rolling mill through the model. Therefore, the rolling force parameters of the rolling roll simulation model, the dimensional parameters of the strip simulation model and the plate shape information of the strip simulation model can be used as sample data, and the bending roll force parameters corresponding to each sample data can be used as label data to form a first sample data set.

[0161] In an embodiment of the present application, the sample data in the first sample data set can be input into the initial model, and the ReLU and Tanh functions are used as activation functions, the least squares method is used as the loss function, and the gradient descent method is used to train the initial model. When the gap between the output of the initial model and the bending roll force parameter as the label data is within a preset range, or the number of iterations of the initial model meets the preset conditions, it is determined that the initial model training is completed and the first model is obtained.

[0162] Figure 8 A schematic diagram of the implementation process of a model training method provided in an embodiment of the present application Figure 4, the method may be executed by a processor of a computer device. Figure 8 As shown, the above step S504 can be implemented through steps S801 to S803:

[0163] Step S801, after the rolling mill completes rolling of the strip, obtain the size parameters of the strip, the alloy type of the strip, the real-time shape information of the strip, the rolling force parameters of the rolling mill, and the bending roll force parameters of the rolling mill.

[0164] Here, the dimension parameters of the strip may include the strip thickness and the strip width.

[0165] Step S802, constructing a second sample data set with the size parameters of the strip, the alloy type of the strip, the real-time shape information of the strip, the rolling force parameters of the rolling mill as sample data, and the bending roll force parameters of the rolling mill as label data.

[0166] Step S803: training the first model based on the second sample data set to obtain at least one second model.

[0167] In an embodiment of the present application, the first model may be trained after the strip rolling is completed, so as to obtain the size parameters of the strip during the rolling process, the alloy type of the strip, the real-time plate shape information of the strip, and the rolling force parameters of the rolling mill.

[0168] In an embodiment of the present application, the sample data in the second sample data set can be input into the first model, and the Tanh function is used as the activation function, the least squares method is used as the loss function, and the gradient descent method is used to train the first model. When the difference between the output of the first model and the bending roll force parameter as the label data is within a preset range, or the number of iterations of the first model meets the preset conditions, it is determined that the training of the first model is completed and the second model is obtained.

[0169] In some embodiments, Fig. 9 As shown, the strip rolling system includes a host computer 901 and a rolling mill controller 902, and the host computer 901 and the rolling mill controller 902 are connected to each other in communication. The host computer 901 can be the execution subject of the parameter determination method and the model training method provided in the above embodiment;

[0170] Based on the above-mentioned strip rolling system, an embodiment of the present application provides a strip rolling method, which is implemented by the following steps:

[0171] Step 1: The host computer trains the initial model to obtain multiple models.

[0172] In the embodiment of the present application, step 1 can be implemented by the model training method provided in the above embodiment.

[0173] Step 2: The host computer sends a parameter acquisition request to the controller of the rolling mill.

[0174] Step 3: The controller of the rolling mill responds to the parameter acquisition request and sends the current rolling pass of the strip and the current rolling parameters of the rolling mill to the host computer.

[0175] Step 4: The host computer uses the target model corresponding to the current rolling pass among the multiple models to determine the shape control parameters of the rolling mill for rolling the strip based on the shape information of the strip and the current rolling parameters of the rolling mill.

[0176] In the embodiment of the present application, the host computer can receive the current rolling pass of the strip and the current rolling parameters of the rolling mill sent by the controller of the rolling mill, and then use the target model corresponding to the current rolling pass in multiple models to determine the plate shape control parameters of the rolling mill for rolling the strip based on the plate shape information of the strip and the current rolling parameters of the rolling mill. Among them, step 4 can be implemented by the parameter determination method provided in the above embodiment.

[0177] Step 5: The host computer sends the plate shape control parameters of the rolling mill to the controller of the rolling mill.

[0178] Step 6: The controller of the rolling mill controls the rollers of the rolling mill to roll the strip based on the plate shape control parameters of the rolling mill.

[0179] In the embodiment of the present application, steps 2 to 6 are executed in a loop until the strip rolling is completed.

[0180] In some embodiments, after the strip rolling is completed, a second model among the multiple models is trained through self-learning based on real parameters during the rolling process.

[0181] The following describes the application of the method provided in the embodiment of the present application in a practical scenario. Fig.10 As shown, it mainly involves steps S901 to S906.

[0182] Step S901: construct a roll model based on the size information of the roll, and perform finite element meshing on the roll model.

[0183] In the embodiment of the present application, a 3D roll model is established based on the roll diameter, roll length, and roll neck data, and the roll model is meshed by finite element. The meshing requirement for the roll model to be in contact with the roll model and the roll model to be in contact with the strip is not greater than 5mm*5mm, otherwise the finite element calculation accuracy cannot reach the practical application. The finite element meshing result of the roll model can be found in Fig.11 ,like Fig.11 As shown, a plurality of finite element grids are distributed on the surface of the roll model 11 .

[0184] Step S902, determining the strip shape data under the constraints of the rolling force and the strip width and changing the bending roll force.

[0185] In the present application embodiment, Fig.12 As shown, the above step S902 can be implemented through steps S9021 to S9024:

[0186] Step S9021, giving rolling force.

[0187] Exemplarily, the rolling force may be set to 10000 KN.

[0188] Step S9022, specifying the strip width.

[0189] In the embodiment of the present application, the strip width can be given according to the product outline of the production. For example, the strip width can be 1500 mm.

[0190] Step S9023, within the bending roll force range, a given bending roll force is set from the minimum bending roll force to the maximum bending roll force.

[0191] In the embodiment of the present application, the bending roll force ranges from -800KN to 800KN, and the magnitude of the bending roll force is changed by no more than 10KN each time, that is, the given bending roll force is -800KN, -790KN, -780KN, ..., 750KN, 760KN, 770KN, 780KN, 790KN, 800KN.

[0192] Step S9024, performing finite element calculation on the roll model based on the rolling force, strip width and bending roll force to obtain strip shape data.

[0193] In the embodiment of the present application, each time the bending roll force data is given, a finite element calculation is performed to obtain the force and deformation data on the contact surface of the working roll with the strip in the length direction, and finally the strip shape data is calculated.

[0194] Step S903, under the constraint conditions of constraining the rolling force and the bending roll force and changing the strip width, determine the strip shape data.

[0195] In the present application embodiment, Fig.13 As shown, the above step S903 can be implemented through steps S9031 to S9034:

[0196] Step S9031, giving rolling force.

[0197] Exemplarily, the rolling force may be set to 10000 KN.

[0198] Step S9032, giving the bending roll force.

[0199] Exemplarily, the bending roll force may be set to 500 KN.

[0200] Step S9033, within the width range, the strip width is given from the minimum width to the maximum width.

[0201] In the embodiment of the present application, the width range is 1000mm to 2000mm, and the width is changed in steps of 10mm, that is, the given strip width is 1000mm, 1010mm, 1020mm, 1030mm,..., 1970mm, 1980mm, 1990mm, 2000mm.

[0202] Step S9034, performing finite element calculation on the roll model based on the rolling force, strip width and bending roll force to obtain strip shape data.

[0203] In the embodiment of the present application, each time the strip width is given, a finite element calculation is performed to obtain the force and deformation data on the contact surface of the working roll with the strip in the length direction, and finally the strip shape data is calculated.

[0204] Step S904, determining the strip shape data under the constraints of constraining the strip width and the bending roll force and changing the rolling force.

[0205] In the present application embodiment, Fig.14 As shown, the above step S904 can be implemented through steps S9041 to S9044:

[0206] Step S9041, specifying the strip width.

[0207] In the embodiment of the present application, the strip width can be given according to the product outline of the production. For example, the strip width can be 1500 mm.

[0208] Step S9042, giving the bending roll force.

[0209] Exemplarily, the bending roll force may be set to 500 KN.

[0210] Step S9043, within the rolling force range, a rolling force is given from a minimum rolling force to a maximum rolling force.

[0211] In the embodiment of the present application, if the maximum rolling force of the rolling mill is 50000KN, the rolling force varies in the range of 1000KN to 50000KN, with the applied rolling force changing in steps of 1000KN, such as 1000Kn, 2000KN, 3000KN,..., 47000KN, 48000Kn, 49000KN, 50000KN.

[0212] Step S9044, performing finite element calculation on the roll model based on the rolling force, strip width and bending roll force to obtain strip shape data.

[0213] In the embodiment of the present application, each time a rolling force is given, a finite element calculation is performed to obtain the force and deformation data on the contact surface of the working roll with the strip in the length direction, and finally the strip shape data is calculated.

[0214] Step S905, based on the rolling force, bending roll force, strip width and strip shape data obtained by finite element calculation, the relationship between the four variables is fitted through a fully connected neural network.

[0215] In the embodiment of the present application, the input variables of the fully connected neural network are rolling force, strip width, strip convexity data, the intermediate hidden layer data is single-layer or multi-layer, the number of neurons in each layer is 18 to 25, the bending roll force is the output, the network is trained and the training results are saved.

[0216] In the embodiment of the present application, the process of training the fully connected neural network is: rolling force, width, strip convexity (i.e. strip shape data) are used as input, and the bias is set to 1, and the bending roll force is output. The hidden layer can be single-layer or multi-layer, with 18 to 25 neurons in each layer, and 10 sets of data are used each time, and 6000 iterations are used for training. ReLU and Tanh functions are used as activation functions, and the least squares method is used as the loss function. The gradient descent method is used to train the network. The ultimate goal of training the network is to obtain the value of the bending roll force according to the change of the input quantity. The trained network is stored in binary form as a file for subsequent program calls.

[0217] Step S906: Perform online neural network training on the trained network.

[0218] In an embodiment of the present application, after the strip begins to bite, the measurement values ​​of the first-level feedback are stored in the memory as a data group with a period of 100ms to 200ms, including rolling force, bending roll force, alloy type, strip thickness, strip width, and strip shape data. When the strip tail is thrown off, the model is triggered to perform online neural network training.

[0219] In the embodiment of the present application, the input data of the online neural network training are rolling force, alloy type, strip thickness, strip width, strip shape (strip convexity data), roller thermal convexity, and the output is the total set bending roll force. The alloy type is used for automatic classification by the artificial intelligence (AI) model, and the AI ​​model automatically calls the corresponding network for different alloy types. The training network is trained with 10 groups of data collected in the length direction of the strip as a group, and iterates 5000 times. The activation function is Tanh, the loss function is the least squares method, and the gradient descent is used for iterative calculation. The calculated results are saved as binary text. Different alloy types and different strip thicknesses correspond to different models.

[0220] In some embodiments, the present application also provides a parameter determination method, which includes the following steps:

[0221] Step S1001, judging whether there is plate shape data (real-time convexity information in the above embodiment) fed back by a measuring instrument for the pass according to the input pass number (the current rolling pass in the above embodiment).

[0222] In the embodiment of the present application, when there is no feedback value from the measuring instrument, step S1002 is executed; when there is feedback value from the measuring instrument, step S1003 is executed.

[0223] Step S1002, the actual rolling force of the first-level feedback (the current rolling parameter in the above embodiment), the predicted strip width corresponding to the pass (the target width information in the above embodiment), and the target convexity value of the current pass (the target convexity information of the strip in the above embodiment) are input into the first network.

[0224] Step S1003, the alloy type of the strip, the predicted strip width corresponding to the pass (the target width information of the strip in the above embodiment), the first-level feedback rolling force (the current rolling parameter in the above embodiment), the strip thickness (the target thickness information of the strip in the above embodiment), the thermal convexity of the roll and the strip convexity fed back by the measuring instrument (the real-time convexity information of the strip) are input into the second network.

[0225] In an embodiment of the present application, after the first network (the first model in the above embodiment) or the second network (the second model in the above embodiment) outputs the bending roll force, the bending roll force is sent to the controller of the rolling mill in a cycle of 200 ms to control the rolling mill to perform rolling.

[0226] like Fig.15As shown, the pass number 10 is used to determine whether the network used is the first network 11 or the second network 12; wherein, the input of the first network 11 is the actual rolling force 13 of the first-level feedback, the predicted strip width 14 corresponding to the pass, and the target convexity value 15 of the current pass; the input of the second network 12 is the alloy type 16 of the strip, the predicted strip width 17 corresponding to the pass, the rolling force 18 of the first-level feedback, the strip thickness 19, the thermal convexity of the roll 20, and the strip convexity 21 fed back by the measuring instrument; the outputs of the first network 11 and the second network 12 are both the bending roll force 22.

[0227] Based on the foregoing embodiments, an embodiment of the present application provides a parameter determination device, which includes the units included and the modules included in the units, which can be implemented by a processor in a computer device; of course, it can also be implemented by a specific logic circuit; in the implementation process, the processor can be a central processing unit (CPU), a microprocessor (MPU), a digital signal processor (DSP) or a field programmable gate array (FPGA), etc.

[0228] Fig.16 A schematic diagram of the structure of a parameter determination device provided in an embodiment of the present application is shown in FIG. Fig.16 As shown, the parameter determination device 1500 includes: a first acquisition module 1501, an information determination module 1502, a model determination module 1503 and a parameter determination module 1504, wherein:

[0229] The first acquisition module 1501 is used to acquire the current rolling pass of the strip and the current rolling parameters of the rolling mill; the current rolling pass is used to represent the current number of times the rolling mill rolls the strip;

[0230] An information determination module 1502 is used to determine the strip shape information;

[0231] The model determination module 1503 is used to determine the target model corresponding to the current rolling pass from multiple models; different models in the multiple models correspond to different training methods;

[0232] The parameter determination module 1504 is used to determine the shape control parameters of the rolling mill for rolling the strip based on the shape information of the strip and the current rolling parameters of the rolling mill by using the target model.

[0233] In some embodiments, the model determination module 1503 is also used to determine the stage information of the current rolling pass in the rolling process of the strip based on the type information of the rolling mill and the current rolling pass; and determine the target model from multiple models based on the stage information of the current rolling pass in the rolling process of the strip.

[0234] In some embodiments, the model determination module 1503 is also used to determine the first model among multiple models trained based on simulation parameters as the target model when the stage information of the current rolling pass is the non-tail stage in the rolling process of the strip; and to determine the second model among multiple models trained based on real parameters as the target model when the stage information of the current rolling pass is the tail stage in the rolling process of the strip.

[0235] In some embodiments, the model determination module 1503 is also used to determine a target second model in at least one second model based on the alloy type of the strip and the target thickness information of the strip; the parameter determination module 1504 is also used to determine the plate shape control parameters of the rolling mill for rolling the strip based on the plate shape information of the strip and the current rolling parameters of the rolling mill using the target second model.

[0236] In some embodiments, the information determination module 1502 is also used to obtain real-time convexity information of the strip and determine the target width information of the strip when the stage information of the current rolling pass is the tail stage in the rolling process of the strip, or to determine the target convexity information of the strip and the target width information of the strip; and to determine the target convexity information of the strip and the target width information of the strip when the stage information of the current rolling pass is the non-tail stage in the rolling process of the strip.

[0237] In some embodiments, when the stage information of the current rolling pass is the tail stage in the rolling process of the strip, the parameter determination module 1504 is further used to determine the flatness control parameters of the rolling mill based on the target width information of the strip, the target convexity information of the strip and the current rolling parameters of the rolling mill using the second model; or, determine the flatness control parameters of the rolling mill based on the target width information of the strip, the real-time convexity information of the strip and the current rolling parameters of the rolling mill using the second model; the parameter determination module 1504 is further used to determine the flatness control parameters of the rolling mill based on the target width information of the strip, the target convexity information of the strip and the current rolling parameters of the rolling mill using the first model;

[0238] In some embodiments, the parameter determination device 1500 also includes a sending module and a receiving module; the sending module is used to send a parameter acquisition request to the controller of the rolling mill; the receiving module is used to receive the current rolling pass of the strip and the current rolling parameters of the rolling mill sent by the controller of the rolling mill; the sending module is also used to send the plate shape control parameters of the rolling mill to the controller of the rolling mill; the controller of the rolling mill is used to control the rolling mill rollers of the rolling mill to roll the strip based on the plate shape control parameters of the rolling mill.

[0239] Fig.17 A schematic diagram of the structure of a model training device provided in an embodiment of the present application is shown in FIG. Fig.17 As shown, the model training device 1600 includes: a second acquisition module 1601, a simulation module 1602, a first training module 1603 and a second training module 1604, wherein:

[0240] The second acquisition module 1601 is used to acquire a roller simulation model of a rolling mill and a strip simulation model of a strip;

[0241] A simulation module 1602 is used to simulate the rolling process of the strip based on the roller simulation model and the strip simulation model to obtain simulation parameters;

[0242] A first training module 1603 is used to train the initial model based on simulation parameters to obtain a first model;

[0243] A second training module 1604 is used to train the first model based on real parameters obtained by rolling the strip by the rolling mill to obtain at least one second model;

[0244] Among them, at least one second model and the target model in the first model are used to determine the plate shape control parameters of the rolling mill for rolling the strip based on the plate shape information of the strip and the current rolling parameters of the rolling mill; the target model is determined based on the current rolling pass; the current rolling pass is used to characterize the number of rolling passes of the rolling mill on the strip.

[0245] In some embodiments, the simulation module 1602 is also used to determine at least one constraint condition of the roller simulation model during the simulated rolling process; for each constraint condition, based on the roller simulation model and the strip simulation model, the strip rolling process is simulated under the constraint condition to obtain the plate shape information of the strip simulation model; the constraint parameters of at least one constraint condition and the plate shape information of the strip simulation model corresponding to each constraint condition are determined as simulation parameters.

[0246] In some embodiments, the constraint parameters of the constraint condition include rolling force parameters of the roll simulation model, size parameters of the strip simulation model, and bending roll force parameters of the roll simulation model; at least one constraint condition includes at least one of the following: the rolling force parameters of the roll simulation model are fixed, the size parameters of the strip simulation model are fixed, and the bending roll force parameters of the roll simulation model vary; the rolling force parameters of the roll simulation model are fixed, the size parameters of the strip simulation model vary, and the bending roll force parameters of the roll simulation model are fixed; the rolling force parameters of the roll simulation model vary, the size parameters of the strip simulation model are fixed, and the bending roll force parameters of the roll simulation model are fixed.

[0247] In some embodiments, the above-mentioned first training module 1603 is also used to construct a first sample data set with the rolling force parameters of the roller simulation model, the size parameters of the strip simulation model and the plate shape information of the strip simulation model corresponding to at least one constraint condition as sample data, and the bending roll force parameters of the roller simulation model corresponding to at least one constraint condition as label data; train the initial model based on the first sample data set to obtain the first model.

[0248] In some embodiments, the second training module 1604 is also used to obtain the size parameters of the strip, the alloy type of the strip, the real-time plate shape information of the strip, the rolling force parameters of the rolling mill and the bending roll force parameters of the rolling mill after the rolling mill completes rolling of the strip; the size parameters of the strip, the alloy type of the strip, the real-time plate shape information of the strip, and the rolling force parameters of the rolling mill are used as sample data, and the bending roll force parameters of the rolling mill are used as label data to construct a second sample data set; based on the second sample data set, the first model is trained to obtain the second model.

[0249] The description of the above device embodiment is similar to the description of the above method embodiment, and has similar beneficial effects as the method embodiment. In some embodiments, the functions or modules included in the device provided in the embodiment of the present application can be used to execute the method described in the above method embodiment. For technical details not disclosed in the device embodiment of the present application, please refer to the description of the method embodiment of the present application for understanding.

[0250] It should be noted that in the embodiment of the present application, if the above-mentioned parameter determination method is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the embodiment of the present application is essentially or the part that contributes to the relevant technology can be embodied in the form of a software product, which is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods of each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a U disk, a mobile hard disk, a read-only memory (ROM), a disk or an optical disk. In this way, the embodiment of the present application is not limited to any specific hardware, software or firmware, or any combination of hardware, software, and firmware.

[0251] An embodiment of the present application provides a computer device, including a memory and a processor, wherein the memory stores a computer program that can be run on the processor, and when the processor executes the program, some or all of the steps in the above method are implemented.

[0252] The embodiment of the present application provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, some or all of the steps in the above method are implemented. The computer-readable storage medium can be transient or non-transient.

[0253] An embodiment of the present application provides a computer program, including a computer-readable code. When the computer-readable code is run in a computer device, a processor in the computer device executes some or all of the steps for implementing the above method.

[0254] The embodiment of the present application provides a computer program product, which includes a non-transitory computer-readable storage medium storing a computer program, and when the computer program is read and executed by a computer, some or all of the steps in the above method are implemented. The computer program product can be implemented specifically by hardware, software or a combination thereof. In some embodiments, the computer program product is specifically embodied as a computer storage medium, and in other embodiments, the computer program product is specifically embodied as a software product, such as a software development kit (SDK) and the like.

[0255] It should be noted here that the description of the various embodiments above tends to emphasize the differences between the various embodiments, and the same or similar aspects can be referenced to each other. The description of the above device, storage medium, computer program and computer program product embodiments is similar to the description of the above method embodiment, and has similar beneficial effects as the method embodiment. For technical details not disclosed in the embodiments of the device, storage medium, computer program and computer program product of this application, please refer to the description of the method embodiment of this application for understanding.

[0256] Fig.18 A hardware entity diagram of a computer device provided in an embodiment of the present application is shown in FIG. Fig.18 As shown, the hardware entity of the computer device 1700 includes: a processor 1701, a communication interface 1702 and a memory 1703, wherein:

[0257] When the processor 1701 executes the program, the steps of any of the above methods are implemented. The processor 1701 generally controls the overall operation of the computer device 1700.

[0258] The communication interface 1702 enables the computer device to communicate with other terminals or servers through a network.

[0259] The memory 1703 is configured to store instructions and applications executable by the processor 1701, and can also cache data to be processed or processed by the processor 1701 and each module in the computer device 1700 (for example, image data, audio data, voice communication data, and video communication data), which can be implemented by flash memory (FLASH) or random access memory (Random Access Memory, RAM). Data can be transmitted between the processor 1701, the communication interface 1702, and the memory 1703 through the bus 1704.

[0260] An embodiment of the present application provides a computer storage medium, which stores one or more programs. The one or more programs can be executed by one or more processors to implement the steps of the parameter determination method of any of the above embodiments.

[0261] It should be noted here that the description of the above storage medium and device embodiments is similar to the description of the above method embodiments, and has similar beneficial effects as the method embodiments. For technical details not disclosed in the storage medium and device embodiments of this application, please refer to the description of the method embodiments of this application for understanding.

[0262] The processor may be at least one of an Application Specific Integrated Circuit (ASIC), a Digital Signal Processor (DSP), a Digital Signal Processing Device (DSPD), a Programmable Logic Device (PLD), a Field Programmable Gate Array (FPGA), a Central Processing Unit (CPU), a controller, a microcontroller, and a microprocessor. It is understandable that the electronic device that implements the functions of the processor may also be other, and the embodiments of the present application are not specifically limited.

[0263] The above-mentioned computer storage medium / memory can be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), a magnetic random access memory (FRAM), a flash memory (Flash Memory), a magnetic surface memory, an optical disk, or a compact disc read-only memory (CD-ROM) and the like; it can also be various terminals including one or any combination of the above-mentioned memories, such as mobile phones, computers, tablet devices, personal digital assistants, etc.

[0264] The above is only an implementation method of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A parameter determination method, characterized in that: The parameter determination method comprises: Acquire the current rolling pass of the strip and the current rolling parameters of the rolling mill; the current rolling pass is used to characterize the current number of times the rolling mill rolls the strip; Determining the plate shape information of the strip; Determining a target model corresponding to the current rolling pass from a plurality of models; different models in the plurality of models correspond to different training methods; The target model is used to determine the flatness control parameters of the rolling mill for rolling the strip based on the flatness information of the strip and the current rolling parameters of the rolling mill.

2. The parameter determination method according to claim 1, characterized in that: Determining a target model corresponding to the current rolling pass from a plurality of models includes: Based on the type information of the rolling mill and the current rolling pass, determining the stage information of the current rolling pass in the rolling process of the strip; The target model is determined from a plurality of models based on stage information of the current rolling pass in the rolling process of the strip.

3. The parameter determination method according to claim 2, characterized in that: The multiple models include a first model trained based on simulation parameters and a second model trained based on real parameters; The determining the target model from a plurality of models based on the stage information of the current rolling pass in the rolling process of the strip includes: In a case where the stage information of the current rolling pass is a non-tail stage in the rolling process of the strip, determining the first model among the multiple models as the target model; In a case where the stage information of the current rolling pass is the tail stage in the rolling process of the strip, the second model among the plurality of models is determined as the target model.

4. The parameter determination method according to claim 3, characterized in that: The parameter determination method further comprises: determining a target second model in at least one of the second models based on the alloy type of the strip and the target thickness information of the strip; The method of using the target model to determine the shape control parameters of the rolling mill for rolling the strip based on the shape information of the strip and the current rolling parameters of the rolling mill includes: The target second model is used to determine the flatness control parameters of the rolling mill for rolling the strip based on the flatness information of the strip and the current rolling parameters of the rolling mill.

5. The parameter determination method according to any one of claims 2 to 4, characterized in that: The step of determining the plate shape information of the strip includes: When the stage information of the current rolling pass is the tail stage in the rolling process of the strip, real-time crown information of the strip is obtained and target width information of the strip is determined, or target crown information of the strip and target width information of the strip are determined; In a case where the stage information of the current rolling pass is a non-tail stage in the rolling process of the strip, target crown information of the strip and target width information of the strip are determined.

6. The parameter determination method according to claim 5, characterized in that: In a case where the stage information of the current rolling pass is the tail stage in the rolling process of the strip, the step of using the target model to determine the shape control parameters of the rolling mill for rolling the strip based on the shape information of the strip and the current rolling parameters of the rolling mill includes: Determine the flatness control parameters of the rolling mill using the second model based on the target width information of the strip, the target convexity information of the strip and the current rolling parameters of the rolling mill; or, Determining the flatness control parameters of the rolling mill using the second model based on the target width information of the strip, the real-time convexity information of the strip, and the current rolling parameters of the rolling mill; In a case where the stage information of the current rolling pass is a non-tail stage in the rolling process of the strip, the step of using the target model to determine the shape control parameters of the rolling mill for rolling the strip based on the shape information of the strip and the current rolling parameters of the rolling mill includes: The first model is used to determine the flatness control parameters of the rolling mill based on the target width information of the strip, the target crown information of the strip, and the current rolling parameters of the rolling mill.

7. The parameter determination method according to any one of claims 1 to 6, characterized in that: The obtaining of the current rolling pass of the strip and the current rolling parameters of the rolling mill includes: Sending a parameter acquisition request to a controller of the rolling mill; Receiving the current rolling pass of the strip and the current rolling parameters of the rolling mill sent by the controller of the rolling mill; The method further comprises: The plate shape control parameters of the rolling mill are sent to the controller of the rolling mill; the controller of the rolling mill is used to control the rollers of the rolling mill to roll the strip based on the plate shape control parameters of the rolling mill.

8. A model training method, characterized in that: The model training method comprises: Acquire a roll simulation model of a rolling mill and a strip simulation model of a strip; Based on the roller simulation model and the strip simulation model, simulating the rolling process of the strip to obtain simulation parameters; Training the initial model based on the simulation parameters to obtain a first model; Based on real parameters obtained by rolling the strip by the rolling mill, the first model is trained to obtain at least one second model; Among them, the at least one second model and the target model in the first model are used to determine the plate shape control parameters of the rolling mill for rolling the strip based on the plate shape information of the strip and the current rolling parameters of the rolling mill; the target model is determined based on the current rolling pass; the current rolling pass is used to characterize the current number of times the rolling mill rolls the strip.

9. The model training method according to claim 8, characterized in that: The step of simulating the rolling process of the strip based on the roller simulation model and the strip simulation model to obtain simulation parameters includes: Determining at least one constraint condition of the roll simulation model during the simulated rolling process; For each of the constraints, based on the roller simulation model and the strip simulation model, the rolling process of the strip is simulated under the constraints to obtain the plate shape information of the strip simulation model; The constraint parameters of the at least one constraint condition and the plate shape information of the strip simulation model corresponding to each constraint condition are determined as the simulation parameters.

10. The model training method according to claim 9, characterized in that: The constraint parameters of the constraint conditions include rolling force parameters of the roll simulation model, size parameters of the strip simulation model, and bending roll force parameters of the roll simulation model; The at least one constraint condition includes at least one of the following: The rolling force parameters of the roll simulation model are fixed, the size parameters of the strip simulation model are fixed, and the bending roll force parameters of the roll simulation model are changed; The rolling force parameters of the roll simulation model are fixed, the size parameters of the strip simulation model are changed, and the bending roll force parameters of the roll simulation model are fixed; The rolling force parameters of the roll simulation model vary, the size parameters of the strip simulation model are fixed, and the bending roll force parameters of the roll simulation model are fixed.

11. The model training method according to claim 10, characterized in that: The step of training the initial model based on the simulation parameters to obtain a first model includes: The rolling force parameters of the roll simulation model, the size parameters of the strip simulation model and the plate shape information of the strip simulation model respectively corresponding to the at least one constraint condition are used as sample data, and the bending roll force parameters of the roll simulation model respectively corresponding to the at least one constraint condition are used as label data to construct a first sample data set; The initial model is trained based on the first sample data set to obtain the first model.

12. The model training method according to claim 8, characterized in that: The first model is trained based on the real parameters obtained by rolling the strip by the rolling mill to obtain the second model, comprising: After the rolling mill completes rolling of the strip, obtaining the size parameters of the strip, the alloy type of the strip, the real-time plate shape information of the strip, the rolling force parameters of the rolling mill, and the bending roll force parameters of the rolling mill; The size parameters of the strip, the alloy type of the strip, the real-time shape information of the strip, and the rolling force parameters of the rolling mill are used as sample data, and the bending roll force parameters of the rolling mill are used as label data to construct a second sample data set; Based on the second sample data set, the first model is trained to obtain the second model.

13. A parameter determination device, characterized in that: The parameter determination device comprises: A first acquisition module is used to acquire the current rolling pass of the strip and the current rolling parameters of the rolling mill; the current rolling pass is used to represent the current number of times the rolling mill rolls the strip; An information determination module, used to determine the plate shape information of the strip; A model determination module, used to determine a target model corresponding to the current rolling pass from a plurality of models; different models in the plurality of models correspond to different training methods; A parameter determination module is used to determine the plate shape control parameters of the rolling mill for rolling the strip by using the target model based on the plate shape information of the strip and the current rolling parameters of the rolling mill.

14. A model training device, characterized in that: The model training device comprises: A second acquisition module is used to acquire a roller simulation model of a rolling mill and a strip simulation model of a strip; A simulation module, used for simulating the rolling process of the strip based on the roller simulation model and the strip simulation model to obtain simulation parameters; A first training module, used for training the initial model based on the simulation parameters to obtain a first model; A second training module, configured to train the first model based on real parameters obtained by the rolling mill on the strip, to obtain at least one second model; Among them, the at least one second model and the target model in the first model are used to determine the plate shape control parameters of the rolling mill for rolling the strip based on the plate shape information of the strip and the current rolling parameters of the rolling mill; the target model is determined based on the current rolling pass; the current rolling pass is used to characterize the current number of times the rolling mill rolls the strip.

15. A computer device comprising a memory and a processor, wherein the memory stores a computer program executable on the processor, wherein: When the processor executes the program, the steps in the method of any one of claims 1 to 7 or the steps in the method of any one of claims 8 to 12 are implemented.

16. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 or the steps of the method according to any one of claims 8 to 12 are implemented.

17. A program product comprising a computer program or instructions, characterized in that When the computer program or instruction is executed by a processor, the steps of the method described in any one of claims 1 to 7 or the steps of the method described in any one of claims 8 to 12 are implemented.

Images