Method, device and storage medium for predicting or controlling silicon steel iron loss

Abstract

This invention provides a method, apparatus, and storage medium for predicting or controlling iron loss in silicon steel coils. The prediction method includes: S1, determining a set of process parameter variables used to predict iron loss in the entire silicon steel coil production process; specifically, using PLS combined with recursive variable elimination to determine the aforementioned set of process parameter variables; S2, training a neural network based on the selected variable set and the obtained historical dataset to obtain a prediction model for iron loss; S3, based on the selected variable set, obtaining the actual values ​​of process parameters for currently completed processes and the historical median of process parameters for subsequent uncompleted processes during the silicon steel production process; S4, inputting the data obtained in step S3 into the prediction model for iron loss to predict iron loss. Using the above technical solution, online prediction of iron loss in silicon steel can be achieved more accurately.

Description

Technical Field

[0001] This invention relates to the field of silicon steel production control, and in particular to a method, apparatus and storage medium for predicting or controlling silicon steel losses. Background Technology

[0002] Silicon steel possesses characteristics such as high magnetic permeability, low coercivity, and high resistivity, and is mainly used as a magnetic material in motors, transformers, electrical appliances, and electrical instruments. Core loss, or iron loss for short, is the most important quality indicator of silicon steel, directly determining its performance in cold-rolled applications. Low iron loss in silicon steel can save significant amounts of electrical energy, extend the operating time of motors and transformers, and simplify cooling systems. Therefore, while ensuring safe production operations, iron loss should be minimized as much as possible to optimize the entire silicon steel production process.

[0003] The production process of silicon steel is complex, and numerous factors influence iron loss, including: chemical composition such as C, Si, Mn, P, S, Al, and N; hot rolling parameters such as entry and exit temperatures of the finishing mill, coiling temperature, and hot-rolled thickness; annealing parameters; temperatures of each furnace section; parameters of each cooling section; furnace atmosphere parameters; and drying furnace temperature parameters. Current research on silicon steel iron loss mainly focuses on the process mechanism, and these studies are mostly qualitative analyses with limited research on process parameters. Research using statistical modeling methods to predict silicon steel iron loss is scarce. Currently, there is no established mechanistic or data model of the relationship between process parameters and iron loss throughout the entire silicon steel production process, nor are there any relevant optimization control models for iron loss control in production. Summary of the Invention

[0004] Embodiments of the present invention provide a method, apparatus, and storage medium for predicting or controlling iron loss in silicon steel production, so as to perform online prediction or control optimization of iron loss during silicon steel production.

[0005] To achieve the above objectives, on the one hand, a method for predicting iron loss in silicon steel is provided, for predicting iron loss in silicon steel during the silicon steel production process, the method comprising:

[0006] S1, Determine the set of process parameter variables used to predict silicon steel loss throughout the entire silicon steel coil production process, including:

[0007] S11, Select the full process parameter variables of the steel coil as the initial variable set, obtain the full process parameter variable values ​​and corresponding iron loss values ​​for each completed steel coil, and use the obtained full process parameter variable values ​​and iron loss values ​​to establish a historical dataset.

[0008] S12, Based on the current set of variables, a regression model is established using the partial least squares (PLS) method, and the root mean square error of the regression model under the current set of variables is calculated through cross-validation.

[0009] S13, for regression models, remove the variable with the smallest absolute value of the regression coefficient of the variables composed of process parameter variables of the whole process;

[0010] S14, determine whether the number of remaining variables is equal to the predetermined minimum number of variables; if yes, proceed to step S15, otherwise, return to step S12;

[0011] S15. By comparing the root mean square error obtained by cross-validation of the regression model with different numbers of variables, the set of variables corresponding to the smallest root mean square error is selected as the set of process parameter variables for predicting silicon steel loss.

[0012] S2, Train a neural network based on the selected set of variables and historical datasets for predicting silicon steel loss to obtain a predictive model for silicon steel loss;

[0013] S3, based on the selected set of variables for predicting silicon steel losses, obtain the actual values ​​of process parameters for currently completed processes and the historical median of process parameters for subsequent uncompleted processes during the silicon steel production process;

[0014] S4. Input the obtained actual values ​​and historical medians of process parameters into the prediction model for iron loss in silicon steel to predict iron loss.

[0015] Preferably, in the method, the neural network is a three-layer BP neural network.

[0016] Preferably, in the method, the process parameters for the entire process include at least: the chemical composition content in the steelmaking process, the heating temperature and thickness in the hot rolling process, the tension and temperature in the normalizing and pickling unit process, the thickness in the rolling mill process, and the tension, temperature and current in the continuous annealing unit.

[0017] On the other hand, a method for controlling silicon steel loss is provided, for controlling silicon steel loss during silicon steel production, including:

[0018] In the silicon steel production process, the method for predicting silicon steel loss as described above is used to predict silicon steel loss.

[0019] Based on the predicted silicon steel loss, the process parameters of subsequent unfinished processes are optimized.

[0020] Preferably, the method, wherein optimizing the process parameters of subsequent unfinished processes based on the predicted silicon steel loss includes:

[0021] The particle swarm optimization algorithm is used to optimize the process parameters of subsequent unfinished processes based on the actual values ​​of the process parameters of the currently completed processes and the upper and lower limits of the process parameters of the subsequent unfinished processes.

[0022] On the other hand, an apparatus for predicting silicon steel losses is provided, including a memory and a processor, the memory storing at least one program, the at least one program being executed by the processor to implement the method for predicting silicon steel losses as described above.

[0023] In another aspect, an apparatus for controlling silicon steel loss is provided, characterized in that it includes a memory and a processor, the memory storing at least one program, the at least one program being executed by the processor to implement the method for controlling silicon steel loss as described above.

[0024] In another aspect, a computer-readable storage medium is provided, wherein at least one program is stored therein, the at least one program being executed by a processor to implement the prediction method or control method as described above.

[0025] The above technical solution has the following technical effects:

[0026] The technical solution of this invention uses PLS combined with recursive variable elimination for feature screening. This allows for the identification of a set of process parameter variables that significantly impact iron loss and can be used to predict silicon steel loss from numerous process parameters throughout the entire silicon steel production process. The identified variable set and historical data are then used to train a neural network to obtain a predictive model for silicon steel loss. This predictive model enables online prediction of silicon steel loss during the production process. Compared to traditional qualitative analysis and manual control based on production experience, the technical solution of this invention, by comprehensively considering all process parameters of silicon steel production and employing statistical data analysis to establish an iron loss prediction model, can achieve more accurate and reliable online prediction of iron loss.

[0027] In a further embodiment of the present invention, by using the results of online prediction to adjust the parameters of subsequent processes, the loss of silicon steel can be optimized and controlled, thereby improving the production quality of silicon steel. Attached Figure Description

[0028] Figure 1 This is a flowchart illustrating a method for predicting silicon steel loss according to an embodiment of the present invention;

[0029] Figure 2 This is a schematic flowchart of a method for controlling steel loss in silicon steel according to an embodiment of the present invention;

[0030] Figure 3 This is a schematic diagram of the structure of a device for predicting or controlling silicon steel losses according to an embodiment of the present invention. Detailed Implementation

[0031] To further illustrate the various embodiments, the present invention provides accompanying drawings. These drawings are part of the disclosure of the present invention, primarily used to illustrate the embodiments and to explain the operating principles of the embodiments in conjunction with the relevant descriptions in the specification. With reference to these drawings, those skilled in the art should be able to understand other possible implementations and the advantages of the present invention. Components in the drawings are not drawn to scale, and similar component symbols are generally used to represent similar components.

[0032] The present invention will now be further described in conjunction with the accompanying drawings and specific embodiments.

[0033] The technical solution of this invention predicts silicon steel loss using statistical modeling. Statistical modeling methods can be broadly categorized into linear and nonlinear methods. This invention combines linear and nonlinear methods to predict silicon steel loss. Linear methods, such as Partial Least Squares (PLS), have unique advantages in handling problems with multiple independent variables and severe multivariate correlations. Through PLS processing, a linear expression between the dependent and nonlinear variables can be obtained, allowing analysis of the influence of each variable on the dependent variable. Nonlinear methods, such as Backpropagation (BP) neural networks, do not require pre-defined formulas; the model is trained iteratively using data and possesses strong nonlinear mapping capabilities.

[0034] Given the numerous process parameters (up to hundreds) in the silicon steel production process, it's practically impossible to process data for all of them. Therefore, this invention first employs PLS combined with recursive variable elimination for feature filtering, selecting several process parameters that significantly impact silicon steel loss. Then, based on these selected parameters, a BP neural network is used to establish a silicon steel loss prediction model. During production, this model is used to predict iron loss. In a further embodiment, based on the predicted iron loss, a particle swarm optimization algorithm can be used to further optimize and control subsequent silicon steel loss.

[0035] Example 1:

[0036] Figure 1 This is a schematic flowchart illustrating a method for predicting silicon steel loss according to an embodiment of the present invention. The prediction method of this embodiment includes:

[0037] S1, determine the set of process parameter variables used to predict iron loss in silicon steel coil production; wherein, the set of process parameter variables used to predict iron loss in silicon steel is the set of process parameters that have a certain or significant impact on iron loss in silicon steel selected from many, such as hundreds, of process parameters throughout the entire process.

[0038] Specifically, the set of process parameter variables used to predict silicon steel losses is determined through the following steps:

[0039] S11, Select the full-process process parameter variables of the steel coil as the initial variable set, obtain the full-process process parameter variable values ​​and corresponding iron loss values ​​for each completed steel coil, and establish a historical dataset using the obtained full-process process parameter variable values ​​and iron loss values; wherein, the full-process process parameter variable values ​​are the input data of the steel coil, and the iron loss values ​​are the output data of the steel coil; specifically, the historical dataset can be obtained by tracking the full-process process parameters of each completed steel coil; in specific implementation, the number of full-process process parameter variables used as the initial variable set is large, possibly up to hundreds; for example, the initial variable set can be all the process parameter variables of the entire process, or it can be a selection of process parameter variables with a certain number, such as hundreds or dozens, selected from all the process parameters;

[0040] S12, based on the current variable set, a regression model is established using the partial least squares (PLS) method, and the root mean square error of the regression model under the current variable set is calculated through cross-validation. In this step, the initial set of selected variables is initially used to establish the regression model using the PLS method, and the root mean square error is calculated for the initial variable set. In the subsequent recursive algorithm, after variable removal, the current variable set here refers to the variable set after removing variables and reducing the number of variables.

[0041] S13, for the regression model, remove the variable with the smallest absolute value of the regression coefficient of the variables composed of the process parameter variables of the whole process; through this step, process parameter variables with little impact on silicon steel loss are removed.

[0042] S14, determine whether the number of remaining variables is equal to the predetermined minimum number of variables; if yes, proceed to step S15; otherwise, return to step S12; the predetermined minimum number of variables in this step is the minimum number of variables that can be used to predict iron loss in silicon steel, that is, the number of variables in the final variable set used to predict iron loss cannot be less than the predetermined number; the method of this embodiment of the invention filters feature removal variables through recursive operations in steps S12 to S13 until the predetermined minimum number of variables is reached, and then the recursive algorithm ends and no more variables are removed; for example, this minimum number of variables can be 10, 15, etc., and can be set according to the actual production situation;

[0043] S15. By comparing the root mean square error obtained by cross-validation of the regression model with different numbers of variables, the set of variables corresponding to the smallest root mean square error is selected as the set of process parameter variables for predicting silicon steel loss. This step obtains the set of variables that have the greatest impact on silicon steel loss in each set of variables, and this set of variables can be used to predict silicon steel loss.

[0044] During the variable elimination process, the variable set will change, including the number of variables in the variable set and which specific process parameter the variable is. Different variable sets correspond to different root mean square errors. For example, through the above steps, a curve can be obtained that reflects the relationship between the variable set and the root mean square error of the model. The variable set corresponding to the minimum root mean square error is taken. This variable set consists of a certain number of process parameters.

[0045] S2, Train the neural network based on the selected set of variables for predicting silicon steel loss and the historical dataset to obtain the prediction model for silicon steel loss; Here, the historical data used for model training are the historical data corresponding to the process parameter variables in the selected set of variables.

[0046] S3, Based on the selected set of variables for predicting silicon steel loss, obtain the actual values ​​of process parameters of the currently completed process and the historical median of process parameters of the subsequent uncompleted process in the silicon steel production process; In this step, the process parameters of the currently completed process and the process parameters of the subsequent uncompleted process correspond to the process parameter variables in the selected set of variables for predicting silicon steel loss.

[0047] S4. Input the obtained actual values ​​and historical medians of the above process parameters into the prediction model for iron loss of silicon steel to predict iron loss.

[0048] Preferably, the above-mentioned neural network is a three-layer BP neural network.

[0049] Preferably, the overall process parameters include at least: the chemical composition content in the steelmaking process, the heating temperature and thickness in the hot rolling process, the tension and temperature in the normalizing and pickling unit process, the thickness in the rolling mill process, and the tension, temperature, and current in the continuous annealing unit. Specifically, the chemical composition content is, for example, the content of components such as C, Si, and Mn.

[0050] Those skilled in the art will know that the overall process parameters also include many other process parameters, which will not be elaborated here.

[0051] Example 2:

[0052] Figure 2 This invention provides a method for controlling steel loss in silicon steel production, as an embodiment of the present invention. Such method is used to control steel loss during the silicon steel production process. Figure 2 This embodiment of the method for controlling silicon steel loss includes, after predicting the silicon steel loss using the prediction method of this embodiment, optimizing the process parameters of subsequent unfinished processes based on the predicted silicon steel loss. Using this embodiment, by providing optimized control suggestions for the process parameters of subsequent processes, the iron loss can be made to meet the set value. This achieves online prediction and control of silicon steel loss.

[0053] For example, based on the actual values ​​of the process parameters of the currently completed process and the upper and lower limits of the process parameters of the subsequent uncompleted processes, the particle swarm optimization algorithm is used to optimize the process parameters of the subsequent uncompleted processes to reduce iron loss. This enables optimized control of iron loss in the silicon steel production process and improves the production quality of silicon steel. Figure 2 In this context, PLSR refers to Partial Least Squares Regression, which is a regression model built using PLS.

[0054] Example 3:

[0055] The present invention also provides a device for predicting or controlling losses in silicon steel, such as... Figure 3 As shown, the device includes a processor 301, a memory 302, a bus 303, and a computer program stored in the memory 302 and executable on the processor 301. The processor 301 includes one or more processing cores. The memory 302 is connected to the processor 301 via the bus 303. The memory 302 is used to store program instructions. When the processor executes the computer program, it implements the steps in the above-described method embodiments of Embodiment 1 or Embodiment 2 of the present invention.

[0056] Furthermore, as an executable solution, the device for predicting or controlling silicon steel loss can be a computer unit, which can be a desktop computer, laptop, handheld computer, or cloud server, etc. The computer unit may include, but is not limited to, a processor and memory. Those skilled in the art will understand that the above-described computer unit structure is merely an example and does not constitute a limitation on the computer unit. It may include more or fewer components, or combine certain components, or use different components. For example, the computer unit may also include input / output devices, network access devices, buses, etc., and this embodiment of the invention does not limit this.

[0057] Furthermore, as an executable solution, the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc. The processor is the control center of the computer unit, connecting various parts of the entire computer unit via various interfaces and lines.

[0058] The memory can be used to store the computer programs and / or modules. The processor implements various functions of the computer unit by running or executing the computer programs and / or modules stored in the memory and by calling data stored in the memory. The memory may mainly include a program storage area and a data storage area. The program storage area may store the operating system and at least one application program required for a function; the data storage area may store data created based on the use of the mobile phone, etc. In addition, the memory may include high-speed random access memory and may also include non-volatile memory, such as hard disk, RAM, plug-in hard disk, smart media card (SMC), secure digital card (SD), flash card, at least one disk storage device, flash memory device, or other volatile solid-state storage device.

[0059] Example 4:

[0060] The present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the method described in the embodiments of the present invention.

[0061] If the modules / units integrated in the computer unit are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the methods of the above embodiments can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording media, USB flash drives, portable hard drives, magnetic disks, optical disks, computer memory, read-only memory (ROM), random access memory (RAM), and software distribution media, etc. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction.

[0062] Although the invention has been specifically shown and described in conjunction with preferred embodiments, those skilled in the art should understand that various changes in form and detail may be made to the invention without departing from the spirit and scope of the invention as defined in the appended claims, all of which shall be within the scope of protection of the invention.

Claims

1. A method for predicting iron loss in silicon steel production, characterized in that, include: S1, Determine the set of process parameter variables used to predict silicon steel loss throughout the entire silicon steel coil production process, including: S11, Select the full process parameter variables of the steel coil as the initial variable set, obtain the full process parameter variable values ​​and corresponding iron loss values ​​for each completed steel coil, and use the obtained full process parameter variable values ​​and iron loss values ​​to establish a historical dataset. S12, Based on the current set of variables, a regression model is established using the partial least squares (PLS) method, and the root mean square error of the regression model under the current set of variables is calculated through cross-validation. S13, For the regression model, remove the variable with the smallest absolute value of the regression coefficient in the variable set composed of the process parameter variables of the whole process; S14, determine whether the number of remaining variables is equal to the predetermined minimum number of variables; If so, proceed to step S15; otherwise, return to step S12. S15. Compare the root mean square error obtained by cross-validation of the regression model under different numbers of variables, and select the variable set corresponding to the minimum root mean square error as the process parameter variable set for predicting silicon steel loss. S2, Train a neural network based on the selected set of variables for predicting silicon steel loss and the historical dataset to obtain a prediction model for silicon steel loss; S3, based on the selected set of variables used to predict silicon steel losses, obtain the actual values ​​of process parameters for currently completed processes and the historical median of process parameters for subsequent uncompleted processes during the silicon steel production process; S4. Input the obtained actual values ​​of the process parameters and the historical median into the prediction model of silicon steel loss to predict iron loss.

2. The method according to claim 1, characterized in that, The neural network is a three-layer BP neural network.

3. The method according to claim 1, characterized in that, The process parameters for the entire process include at least: the chemical composition content in the steelmaking process, the heating temperature and thickness in the hot rolling process, the tension and temperature in the normalizing and pickling unit process, the thickness in the rolling mill process, and the tension, temperature and current in the continuous annealing unit.

4. A method for controlling steel loss in silicon steel production, characterized in that, include: In the silicon steel production process, the method described in any one of claims 1-3 is used to predict silicon steel loss; Based on the predicted silicon steel loss, the process parameters of subsequent unfinished processes are optimized.

5. The method according to claim 4, characterized in that, The optimization of process parameters for subsequent unfinished processes based on the predicted silicon steel loss includes: The particle swarm optimization algorithm is used to optimize the process parameters of subsequent unfinished processes based on the actual values ​​of the process parameters of the currently completed processes and the upper and lower limits of the process parameters of the subsequent unfinished processes.

6. A device for predicting steel loss in silicon steel, characterized in that, It includes a memory and a processor, the memory storing at least one program, the at least one program being executed by the processor to implement the method as claimed in any one of claims 1 to 3.

7. A device for controlling steel loss in silicon steel, characterized in that, It includes a memory and a processor, the memory storing at least one program, the at least one program being executed by the processor to implement the method as described in any one of claims 4 to 5.

8. A computer-readable storage medium, characterized in that, The storage medium stores at least one program segment, which is executed by a processor to implement the method as described in any one of claims 1 to 5.

Images