Method for automatically setting thickness of intermediate billet through neural network model
Automatically set the intermediate blank thickness through the neural network model, the problem of complex setting of intermediate blank thickness in steel rolling production is solved, the production efficiency and the working conditions of the operator are improved, and the appropriate setting of the intermediate blank thickness and the matching of the production rhythm is achieved. .
Patent Information
- Application Number
- CN202411221592.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-02
- Publication Date
- 2024-11-08
- Estimated Expiration
- 2044-09-02
AI Technical Summary
In steel rolling production, the thickness of the intermediate billet is complex and depends on the experience of the operator, resulting in low production efficiency and high labor intensity. It is difficult for the existing technology to effectively match the rhythm of rough rolling and finishing mills.
Using an automatic setting method based on neural network model, a computer database is established by collecting historical data. The input layer includes operator characteristics, steel type, steel plate width/thickness, and intermediate blank length. The output layer is the thickness deviation of the intermediate blank, which is automatically given. The reference value is transmitted to the operating table in real time to reduce the labor intensity of the operator.
It improves the rolling production rhythm, reduces the labor intensity of operators, and standardizes operating habits to ensure the appropriate setting of the thickness of the intermediate blank, and improves the product yield.
Smart Images

Figure CN118904906A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of steel rolling production, and in particular relates to a method for automatically setting the thickness of an intermediate billet through a neural network model. Background Art
[0002] The controlled rolling and cooling technology (TMCP) is currently widely used in the production of medium and thick plates, and has played a positive role in the production of pipeline steel, high-strength structural steel, offshore platform steel, ship plates, etc. The slab is rolled into an intermediate billet of a certain thickness and width by the rough rolling mill to meet the requirements of the subsequent finishing process. The specific thickness setting of the intermediate billet is related to the requirements of the finished strip thickness, product structure and performance. The rough rolling parameters are calculated by presetting the roll gap according to the target intermediate billet thickness, so that the size of the intermediate billet at the last rolling of the rough rolling mill meets the requirements.
[0003] Generally speaking, the target thickness of the intermediate billet required by the process is generally a range rather than a single value. The thickness of the intermediate billet will affect the rolling parameters of the previous and subsequent rough rolling and finishing rolling. In production, the operator will select the appropriate intermediate billet thickness within the target range according to their rough rolling and finishing process operating habits, so that the entire rolling rhythm can be more efficient and smooth. Within the adjustable range, determining the most appropriate intermediate billet thickness deviation is conducive to improving rolling production efficiency. However, in actual production, the operator's experience judgment and manual adjustment of the deviation will greatly increase the operator's ability requirements and workload. At present, the research on the optimal target size of the intermediate billet in the existing technology is mainly based on steel plate performance, equipment status and other aspects.
[0004] For example, the Chinese patent with publication number CN117259449 A discloses a method and system for controlling the thickness of the intermediate billet in rough rolling, which mainly solves the problem of large deviation between the actual thickness of the intermediate billet in the final rough rolling pass and the target value. The patent establishes a system to ensure that the actual thickness of the intermediate billet hits the target value. The system combines the wear state of the mill roll gap with the actual situation of the current rolling pass, and corrects the set value of the roll gap of the next pass in real time to ensure that the actual thickness of the intermediate billet hits the target value.
[0005] Another example is the Chinese patent with publication number CN114101344 A, which discloses a method for adjusting the thickness of the intermediate billet during medium and thick plate rolling. From the perspective of product quality, this patent provides a method for adjusting the thickness of the intermediate billet during rolling of thick-gauge steel plates. According to the specific pre-rolling slab size and steel plate thickness, an empirical formula is used to calculate a more accurate upper and lower limit control range of the intermediate billet thickness, optimize the control range of the intermediate billet thickness, avoid the problem of unstable steel plate performance caused by too wide a control range of the intermediate billet thickness, especially the low yield strength of the finished steel plate, and ultimately improve the product yield rate.
[0006] Another example is the Chinese patent with publication number CN116475244 A, which discloses a method and device for adjusting the thickness of an intermediate billet, by obtaining the set value of the current rolled steel and the target value of the intermediate billet thickness of the current rolled steel; if the quality inspection result; the intermediate billet result of the previous rolled steel does not meet the first preset condition, or the intermediate billet thickness set value of the previous rolled steel and the intermediate billet thickness of the current rolled steel are obtained to obtain the intermediate billet thickness adjustment value; according to the difference between the intermediate billet thickness adjustment value and the upper target value does not meet the second preset condition, the intermediate billet thickness set value of a rolled steel and the intermediate billet thickness target value of the next rolled steel are determined to adjust the intermediate billet thickness set value of the next rolled steel, so as to adjust the intermediate billet thickness of the next rolled steel. In this way, the operator is provided with an opportunity to adjust the intermediate billet thickness setting according to the on-site situation when there is a quality problem or the work changes. When the quality problem occurs on the steel rolling site, the intermediate billet setting value is adjusted to avoid batch quality defective products.
[0007] Since the factors affecting the thickness of the intermediate billet are relatively complex, it is necessary to set the thickness of the intermediate billet in combination with the current rough rolling and finishing rolling rhythm and the operator's habits while meeting the prerequisite process requirements. It is difficult to develop a related intermediate billet thickness adaptive mathematical model. Therefore, the current matching of the rough rolling and finishing mill rolling rhythms requires the operator to carefully and timely adjust the intermediate billet thickness deviation based on production experience, which greatly increases the operator's ability requirements and workload. Summary of the invention
[0008] The present invention provides an automatic operation method for producing medium and thick plates based on a steel rolling control system. The method can automatically provide and set intermediate billet thickness deviation parameters according to current steel plate information, effectively improving the rolling production rhythm and reducing the labor intensity of operators.
[0009] In order to achieve the above-mentioned purpose, the present application adopts the following technical solution: a method for setting the thickness of the intermediate billet based on a neural network model, characterized in that it includes the following steps.
[0010] S1: Collect historical data of steel plate specification parameters and corresponding intermediate billet thickness deviation parameters and establish a computer background database, and establish a neural network model based on the database; wherein the input layer of the neural network model is operator characteristics, steel type, steel plate width / thickness, intermediate billet length and slab thickness, and the output layer is the intermediate billet thickness deviation manually set by the operator.
[0011] S2: During the rough rolling production process, the current steel plate rolling parameters are collected. If the specification parameters of the current steel plate exist in the database, the neural network model automatically gives the reference value of the intermediate billet thickness deviation parameter, and executes step S3. If the specification parameters of the current steel plate do not exist in the database, the operator intervenes and manually inputs the intermediate billet thickness deviation parameter.
[0012] S3: The reference value of the intermediate billet thickness deviation parameter is automatically set and transmitted to the operating console in real time. If the operator does not change the reference value, the production setting is performed according to the reference value; if the operator intervenes and modifies the reference value, the production setting is performed according to the parameter value input by the operator for the last time.
[0013] Furthermore, the neural network model will continue to calculate the rolling parameters of each steel plate and improve the neural network model through self-learning.
[0014] Furthermore, the steps of establishing the neural network model are as follows.
[0015] (1) Collect historical data of relevant processes and establish a large database for neural network training. The specific process parameters include: operator type, steel grade, steel plate width / thickness, intermediate billet length, slab thickness, and intermediate billet thickness deviation manually set by the operator.
[0016] (2) Set 90% of the data groups in the database as training groups and 10% of the data groups as experimental groups.
[0017] (3) Without manually setting the initial weight values and bias items, a neural network self-learning model is established through the training group data, where the activation function is the Sigmoid function and the hidden layer is set to two layers.
[0018] (4) The accuracy and calculation speed of the neural network model are verified by the experimental group, and the error between the model output deviation and the actual deviation of the experimental group is less than 10%. If the accuracy and calculation speed of the neural network model meet the requirements of online production, the model is successfully established. If the accuracy and calculation speed of the neural network model cannot meet the requirements of online production, the weight values of the input layer parameters can be manually set according to production experience, and the bias items can be set according to the actual conditions of the intermediate billet length and steel plate width / thickness according to the characteristics of the activation function, and the model is retrained. After the training is completed, repeat step (4).
[0019] Furthermore, the setting logic of the bias item and the weight value is determined by the operator according to actual production experience data and specific parameters of the database.
[0020] The beneficial effects of the present invention are as follows: the present invention adopts a regression analysis method based on big data, establishes a machine learning model based on a historical production database, and gives a suitable reference value of the intermediate billet thickness deviation in combination with the current process parameters, and automatically sets it, which effectively improves the rolling production rhythm, reduces the labor intensity of operators, and can also standardize the operating habits of operators. BRIEF DESCRIPTION OF THE DRAWINGS
[0021] Figure 1This is the intermediate billet thickness deviation neural network model established in the present invention.
[0022] Figure 2 This is an example diagram of the neural network model correction method.
[0023] Figure 3 This is an example of a neural network model dataset.
[0024] Figure 4 is the Sigmoid activation function curve.
[0025] Figure 5 An example of data transformation for the model input layer.
[0026] Figure 6 The interface is displayed when the operator enters parameters.
[0027] Figure 7 The interface is displayed when the reference value is automatically loaded. DETAILED DESCRIPTION
[0028] In order to make the technical solution and advantages of the present invention clearer, the technical solution will be described clearly and completely below in combination with specific process data and corresponding drawings to reflect the advantages of the present invention.
[0029] The present invention provides a method for setting the thickness of an intermediate billet based on a neural network model, comprising the following steps.
[0030] S1: Collect historical data of steel plate specification parameters and corresponding intermediate billet thickness deviation parameters and establish a computer background database, and establish a neural network model based on the database; wherein the input layer of the neural network model is operator characteristics, steel type, steel plate width / thickness, intermediate billet length and slab thickness, and the output layer is the intermediate billet thickness deviation manually set by the operator.
[0031] S2: During the rough rolling production process, the current steel plate rolling parameters are collected. If the specification parameters of the current steel plate exist in the database, the neural network model automatically gives the reference value of the intermediate billet thickness deviation parameter, and executes step S3. If the specification parameters of the current steel plate do not exist in the database, the operator intervenes and manually inputs the intermediate billet thickness deviation parameter.
[0032] S3: The reference value of the intermediate billet thickness deviation parameter is automatically set and transmitted to the operating console in real time. If the operator does not change the reference value, the production setting is performed according to the reference value; if the operator intervenes and modifies the reference value, the production setting is performed according to the parameter value input by the operator for the last time.
[0033] Specifically, the technical solution of the present invention is to collect the intermediate billet thickness deviation values set under different processes in historical production, and establish a neural network model according to the relevant process parameter types, and the parameters involved include operator habits, steel type, steel plate width / thickness, intermediate billet length, slab thickness, etc. The basic design rules of the background neural network model are: operator type, steel grade, target width range of rolled steel plate, target thickness range of rolled steel plate, and intermediate billet length are used as input index conditions, and the output data includes the intermediate billet thickness deviation offset.
[0034] The first step is to collect historical data of related processes and establish a large database for neural network training. The specific process parameters include: operator type, steel type, steel plate width / thickness, intermediate billet length, slab thickness, and intermediate billet thickness deviation manually set by the operator. Among them, operators are divided into 4 categories according to their operating habits, and are referred to by parameters 1, 2, 3, and 4 respectively.
[0035] Step 2: Use random sampling to set 90% of the data groups in the database as training groups and 10% of the data groups as experimental groups.
[0036] Step 3: Without manually setting the initial weight value and bias term of the input layer, a neural network self-learning model (with the activation function being the Sigmoid function) is established through the training group data, with 2 hidden layers.
[0037] Sigmoid activation function: ; This function can map a real number to the interval (0,1). It is often used as the activation function of the input layer in neural networks. It works better when the input value features are complex or the absolute value is close to 0.
[0038] Step 4: Verify its accuracy and calculation speed through the experimental group. The error between the model output deviation and the actual deviation of the experimental group is less than 10%. The verification results are shown in Table 1.
[0039] Table 1 Error statistics of the experimental group
[0040] According to Table 1, the number of groups with an error less than 10% only accounts for 52.5%. The model needs to be corrected. The specific operation is: manually set the weights before retraining the model, and reduce the weight values of steel type and steel plate width to 0.3 based on production experience. Set bias items for steel plate thickness / width and intermediate billet length. The reason for selecting intermediate billet length and steel plate width as bias items is that when the absolute value in the input layer is greater than 5, the "gradient disappearance" phenomenon is likely to occur after activation of the Sigmoid function (such as Figure 4 ).
[0041] The input layer bias formula is:
[0042] in, is the average value of the data, , is the maximum / minimum value of the data. After the input layer parameters are biased by the above formula, the range is determined to be between [-5,5], which is conducive to distinguishing the particularity of the input layer during self-learning (such as Figure 5 ).
[0043] Next, the training set with adjusted weights and biased processing was used for training, and the experimental group was used to verify the accuracy of the modified neural network model. The error between the model output deviation and the actual deviation of the experimental group was less than 10%. The verification results are shown in Table 2. The number of groups with an error less than 10% accounted for 78.9%, which basically met the production requirements.
[0044] Table 2 Error statistics of the experimental group
[0045] In addition, to ensure that the neural network model in the background can be discovered and stopped in time when an abnormal situation occurs, the reference value given by the neural network will be sent to the operation console synchronously, so that the operator can be informed and adjusted in time. If the intermediate billet handover thickness function input box or button on the screen is a number directly entered by the operator or a switch directly selected, including the value adjusted based on the reference value, then the display remains the same as the original WinCC, such as Figure 6 shown.
[0046] If an input box or button on the screen is a reference value or switch given by the L2 background program of the automation system, a yellow frame will be added to the input box or button on the WinCC screen to provide a prompt, such as Figure 7 As shown, the operator is informed that this data is no longer the original input value, but a reference value automatically adjusted by the background. Please confirm it. When the operator modifies the value of this input box again, the yellow box disappears.
[0047] If the operator does not adjust the value of this input box, the yellow box will remain, and the value will be automatically updated by the L2 background until the corresponding reference value function button is turned off or there is no corresponding reference value for the next piece of steel (for example, after changing the specification, there is no parameter recommendation script for the new specification), then the yellow box disappears and the value in this input box is restored to the last value directly entered manually.
[0048] The above are only specific implementations 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. Therefore, the protection scope of the present application should be based on the protection scope of the claims.
Claims
1. A method for setting the thickness of an intermediate billet based on a neural network model, characterized in that: The steps include: S1: Collect historical data of steel plate specification parameters and corresponding intermediate billet thickness deviation parameters and establish a computer background database, and establish a neural network model based on the database; wherein the input layer of the neural network model is operator characteristics, steel type, steel plate width / thickness, intermediate billet length and slab thickness, and the output layer is the intermediate billet thickness deviation manually set by the operator; S2: During the rough rolling production process, the current steel plate rolling parameters are collected. If the specification parameters of the current steel plate exist in the database, the neural network model automatically gives a reference value of the intermediate billet thickness deviation parameter, and step S3 is executed; If the specification parameters of the current steel plate do not exist in the database, the operator intervenes and manually inputs the intermediate billet thickness deviation parameters; S3: The reference value of the intermediate billet thickness deviation parameter is automatically set and transmitted to the operating console in real time. If the operator does not change the reference value, the production setting is performed according to the reference value; if the operator intervenes and modifies the reference value, the production setting is performed according to the parameter value input by the operator for the last time.
2. The method for setting the intermediate billet thickness based on a neural network model according to claim 1, characterized in that: The neural network model will continue to calculate the rolling parameters of each steel plate and improve the neural network model through self-learning.
3. The method for setting the intermediate billet thickness based on a neural network model according to claim 1, characterized in that: The steps of establishing the neural network model are as follows: (1) Collect historical data of relevant processes and establish a large database for neural network training. Specific process parameters include: operator type, steel grade, steel plate width / thickness, intermediate billet length, slab thickness, and intermediate billet thickness deviation manually set by the operator; (2) 90% of the data groups in the database are set as training groups, and 10% of the data groups are set as experimental groups; (3) Without manually setting the initial weight value and bias term, a neural network self-learning model is established through the training group data, where the activation function is the Sigmoid function and the hidden layer is set to two layers; (4) Verify the accuracy and calculation speed of the neural network model through the experimental group, and the error between the model output deviation and the actual deviation of the experimental group is less than 10%; If the accuracy and calculation speed of the neural network model meet the requirements of online production, the model is successfully established; If the accuracy and calculation speed of the neural network model cannot meet the requirements of online production, the weight values of the input layer parameters can be manually set based on production experience, and the bias items can be set according to the actual conditions of the intermediate billet length and steel plate width / thickness based on the characteristics of the activation function. The model can be retrained and step (4) can be repeated after the training is completed.
4. The method for setting the intermediate billet thickness based on a neural network model according to claim 3, characterized in that: The setting logic of the bias item and weight value is determined by the operator based on actual production experience data and specific parameters of the database.
Citation Information
Patent Citations
Temperature-waiting thickness adjusting method for intermediate billet rolled by thick steel plate
CN114101344A
Method and device for adjusting thickness of intermediate billet
CN116475244A
Rough rolling intermediate billet thickness control method and system
CN117259449A
Reverse optimization method of steel plate rolling and cooling controlling-process temperature institution
CN101391268A
Method and device for setting convexity of intermediate billet and computer equipment
CN114918261A
Cited By
Self-adaptive rolling method of hot rolling and roughing rolling unit
CN120515825A
