H-shaped steel straightness prediction method, H-shaped steel straightness control method, H-shaped steel straightness prediction system, medium and electronic equipment

Through neural network, predicting the initial and residual stress distribution of H-shaped steel, combining process parameters, a post-standard straightness model was constructed, which solved the problem of failure to effectively control the straightness of H-shaped steel in the existing technology, and achieved high-precision flatness Straightness control and multi-spec adaptability improve product quality and production efficiency.

CN119007884AActive Publication Date: 2024-11-22CISDI HEAVY MACHINERY

Patent Information

Application Number
CN202411030155.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-30
Publication Date
2024-11-22
Estimated Expiration
2044-07-30

AI Technical Summary

Technical Problem

The existing H-shaped steel straightening process fails to effectively consider the impact of residual stresses on straightness during rolling and cooling, resulting in the product being bending, twisting or warping in the length direction, and the control method relies on manual experience. Lack of intelligent analysis, it is difficult to ensure the stability of product straightness and adapt to multiple specifications.

Method used

The neural network model is used to predict the initial straightness and residual stress distribution, and combined with rolling, cooling and straightening process parameters, a prediction model is built to predict the final straightness of H-shaped steel, and the process parameters are adjusted according to the difference matching control strategy to achieve Accurate control of the straightness of H-shaped steel.

Benefits of technology

Through deep fusion of rolling, cooling and straightening process parameters, high-precision control of the straightness of H-shaped steel is achieved, suitable for a variety of specifications, improving the straightness stability and production adaptability of the product, reducing cutting losses. Improved metal yield and product quality.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119007884A_ABST
    Figure CN119007884A_ABST
Patent Text Reader

Abstract

The invention relates to an H-shaped steel straightness prediction method and system, a control method and system, a medium and electronic equipment, and belongs to the technical field of H-shaped steel straightening. The method comprises the steps of data acquisition, initial straightness prediction and residual stress prediction, and the predicted initial straightness, predicted residual stress distribution data and second data are input into a straightened straightness prediction model to obtain predicted final straightness at multiple positions of the head, the tail and the middle of a straightened H-shaped steel finished product. According to the target straightness required by production, comparing and predicting the difference between the final straightness and the target straightness, and matching a corresponding control strategy from a strategy library according to the difference; and according to the control strategy obtained through matching, working parameters of all components of the rolling production line and / or the cooling production line and / or the straightening production line are controlled. According to the method, rolling, cooling and straightening processes are deeply fused with straightness parameters, the influence of residual stress distribution on the straightening process is fully considered, and high straightness control over the H-shaped steel rolled piece is achieved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention belongs to the technical field of H-beam straightening, and relates to an H-beam straightness prediction method, a control method and system, a medium and an electronic device. Background Art

[0002] H-shaped steel is the core supporting material for grand projects such as green prefabricated buildings, large steel structure systems, skyscrapers, and bridges across rivers and seas. In particular, small and medium-sized H-shaped steel are mainstream products in the market. Its production scale continues to expand. According to statistics in 2020, the annual output has reached 51.74 million tons, and exceeded 60 million tons the following year, accounting for more than 70% of the total output of steel sections, demonstrating the industry's booming development trend.

[0003] With the continuous expansion of H-beam application areas and the surge in market demand, the requirements for the diversification and refinement of steel size specifications are increasing, and steel companies are facing the dual challenges of improving production efficiency and product quality. The production process of H-beam mainly includes three links: rolling, cooling and straightening, among which straightening has become a key step in determining product quality. Straightening technology is not only related to the final shape of the product, but also an important indicator to measure the advancement of production technology.

[0004] To meet this challenge, domestic and foreign companies have adopted advanced multi-roller straightening equipment and continuous straightening technology, and transformed the traditional mechanical pressure adjustment device into a hydraulic one, which greatly improved the adjustment accuracy and flexibility. At the same time, the in-depth exploration of straightening theory has promoted the optimization and upgrading of straightening equipment and the scientific formulation of process solutions. For example, the application of cantilever straightening machines has become a new favorite in the industry with its convenient roll change operation, good adaptability and high production efficiency.

[0005] However, the existing straightening process focuses on the standardized application of the reduction procedure, but often ignores the potential impact of residual stress generated during rolling and cooling on the straightness of the steel. Residual stress may cause H-shaped steel to bend, twist or warp in the length direction. These uneven phenomena not only affect the appearance quality of the component, but also may reduce its bearing capacity and performance. During the rolling process, H-shaped steel will be subjected to strong plastic deformation, which may lead to uneven stress distribution inside the material. Rolling parameters such as rolling temperature, rolling speed, rolling force, etc. can affect the residual stress distribution. Reasonable control of the rolling process can effectively reduce the generation of residual stress. During the cooling process, the temperature of the H-shaped steel will drop rapidly, resulting in thermal stress inside the material. The choice of cooling rate and cooling method has a significant impact on the residual stress. The residual stress generated during the rolling and cooling process cannot be avoided, and the subsequent straightening process is to use mechanical methods to make the H-shaped steel elastically or plastically deform to eliminate or reduce the residual stress inside it. The existing straightening process does not consider the influence of residual stress when selecting parameters such as roller diameter, roller shape, and straightening force of the straightening machine, and its control over the straightness of the finished product is very limited.

[0006] Moreover, the analysis of straightening data and the setting of straightening parameters are highly dependent on manual experience, lacking intelligent analysis and utilization, resulting in frequent process adjustments and difficulty in ensuring the stability of product straightness. In addition, the existing control methods are limited in universality, making it difficult to fully cover a variety of H-beam specifications, not adaptable to new sample data, and requiring high parameter accuracy and data quality, increasing operational difficulty and cost. Summary of the invention

[0007] In view of this, the purpose of the present invention is to provide a method for predicting the flatness of H-beam, a control method and system, a medium and an electronic device, comprehensively analyze the influence of rolling, cooling and straightening processes on flatness, especially consider the influence and utilization of residual stress, and propose a control strategy for rolling, cooling and straightening processes to achieve high flatness production of H-beam.

[0008] To achieve the above object, the present invention provides a method for predicting the straightness of an H-beam, comprising the following steps:

[0009] Acquire first data and second data;

[0010] Inputting the first data into an initial flatness prediction model to obtain a predicted initial flatness of the H-beam before straightening;

[0011] Inputting the first data into a residual stress prediction model to obtain predicted residual stress distribution data of the H-beam before straightening;

[0012] The predicted initial flatness, the predicted residual stress distribution data and the second data are input into the straightness prediction model after straightening to obtain the predicted final straightness at the head, tail and multiple positions in the middle of the H-beam finished product after straightening.

[0013] The first data includes the first sub-data measured when the H-beam billet is at the entrance position of the final rolling pass and the second sub-data measured when the H-beam finished product is at the exit position of the final rolling pass; the first sub-data includes the external dimension sample data at multiple positions of the flange and web of the H-beam billet, the pass reduction data at multiple positions, the running speed sample data, and the temperature field distribution sample data.

[0014] The second sub-data includes sample data of the outer dimensions at multiple positions of the flange and web of the H-beam finished product, sample data of the running speed, sample data of the temperature field distribution from the mill outlet to the cooling bed inlet, and the chemical composition of the H-beam.

[0015] The second data includes sample data of the outer dimensions of the finished H-beam flange and web at multiple positions, various process parameters of the straightening production line, fixed length or multiple length, and chemical composition of the H-beam.

[0016] When performing initial flatness prediction, the first sub-data is input into the initial flatness prediction model to obtain the predicted initial flatness at the head, tail and multiple positions in the middle of the H-beam before straightening.

[0017] When predicting the residual stress, the second sub-data is input into the residual stress prediction model to obtain the predicted residual stress distribution data at the head, tail and multiple positions in the middle of the H-beam before straightening.

[0018] The construction process of the initial flatness prediction model includes:

[0019] Acquire a first historical production data set including rolling and cooling process parameters and measured initial straightness of H-beam before straightening;

[0020] Training a first neural network based on a first historical production data set to obtain an initial flatness prediction model;

[0021] Correct the initial flatness prediction model and determine the output results.

[0022] Correct the initial flatness prediction model and determine the output results including:

[0023] Performing residual sequence processing on multiple groups of first initial flatness prediction values ​​output by the initial flatness prediction model and the measured initial flatness of the corresponding data group to obtain multiple groups of initial flatness deviation values;

[0024] Inputting multiple groups of initial straightness deviation values ​​into a back propagation neural network to obtain initial straightness deviation correction values;

[0025] Compensating the initial flatness deviation correction value to the first initial flatness prediction value to obtain a second initial flatness prediction value;

[0026] The second initial flatness prediction value is output as the predicted initial flatness.

[0027] Optionally, the first historical production data set is the rolling and cooling process parameters and the measured initial straightness of the H-beam before straightening at the production site for at least half a year, or the rolling and cooling process parameters and the measured initial straightness of the H-beam before straightening of an H-beam production line with an output exceeding 500,000 tons.

[0028] Training a first neural network based on a first historical production data set includes:

[0029] Constructing a first primitive model based on running time series through recurrent neural networks, stacked recurrent neural networks, and recurrent convolutional neural networks;

[0030] Dividing the first historical production data set into a first training set and a first test set according to a set ratio;

[0031] Select multiple factors that affect the initial straightness of H-beam from the first historical production data set as the input layer, and use the initial straightness of H-beam as the output layer;

[0032] The training function of the first original model is selected to obtain an initial flatness prediction model.

[0033] Optionally, the factors affecting the initial flatness of the H-shaped steel are the layout parameters and working parameters of the components of the rolling production line and the cooling production line, and the number of neurons in the input layer is the number of selected factors affecting the initial flatness.

[0034] If the number of data samples in the first training set is greater than 100,000, the square loss function is selected as the training function. The square loss function is L(f(x), y) = (f(x)-y) 2 ; If the number of data samples in the first training set is less than 100,000, select the absolute value loss function as the training function, and the absolute value loss function is L(Y,f(x))=|Yf(x)|.

[0035] Optionally, the factors affecting the initial straightness of H-beam in the first historical production data set include: the position of the universal rolling mill in the finishing mill group, the deformation amount of the universal rolling mill pass, the rolling temperature of the universal rolling mill, the position of the edge rolling mill, the deformation amount of the edge rolling mill pass, the position of the universal finishing mill, the deformation amount of the universal finishing mill pass, the position of the water tank after finishing rolling, the water volume of the water tank after finishing rolling, the water pressure of the water tank after finishing rolling, the number of nozzles of the water tank after finishing rolling, the water pressure of the water tank after finishing rolling, the final rolling speed of the rolled piece, the final rolling temperature of the rolled piece, the section modulus of the rolled piece, the distance from the mill outlet to the cooling bed entrance, and the temperature of the cooling bed on the rolled piece.

[0036] Optionally, the first historical production data set is divided into a first training set and a first test set in a ratio of 8:2, the first training set is used to train and optimize the initial flatness prediction model, and the first test set is used to verify the results of the initial flatness prediction model.

[0037] The construction process of the residual stress prediction model includes:

[0038] A second historical production data set including rolling and cooling process parameters and predicted residual stress distribution data of H-beam before straightening is obtained; and a second neural network and a residual stress prediction model are trained based on the second historical production data set.

[0039] Optionally, the second historical production data set is the rolling, cooling process parameters and predicted residual stress distribution data of H-beam before straightening for at least half a year at the production site, or the rolling, cooling process parameters and predicted residual stress distribution data of H-beam before straightening for an H-beam production line with an output exceeding 500,000 tons.

[0040] Training the second neural network based on the second historical production data set includes:

[0041] A second primitive model based on the running time series is constructed by using a recurrent neural network, a stacked recurrent neural network, and a recurrent convolutional neural network;

[0042] Dividing the second historical production data set into a second training set and a second test set according to a set ratio;

[0043] Select multiple factors that affect the residual stress of H-beam before straightening from the second historical production data set as the input layer, and use the residual stress of H-beam before straightening as the output layer;

[0044] The training function of the second original model is selected to obtain the residual stress prediction model.

[0045] Optionally, the factors affecting the residual stress of the H-beam before straightening are the layout parameters and operating parameters of the components of the rolling production line and the cooling production line, and the chemical composition of the H-beam, and the number of neurons in the input layer is the number of selected factors affecting the residual stress of the H-beam before straightening.

[0046] If the number of data samples in the first training set is greater than 100,000, the square loss function is selected as the training function. The square loss function is L(f(x), y) = (f(x)-y) 2 ; If the number of data samples in the first training set is less than 100,000, select the absolute value loss function as the training function, and the absolute value loss function is L(Y,f(x))=|Yf(x)|.

[0047] Optionally, the factors affecting the residual stress of H-beam before straightening in the second historical production data set include: C content, Si content, Mn content, S content, P content, Cr content, Nb content, V content, Ti content of the H-beam, the position of the universal rolling mill of the finishing mill, the deformation of the universal rolling mill pass, the rolling temperature of the universal rolling mill, the position of the edge rolling mill, the deformation of the edge rolling mill pass, the position of the universal finishing mill, the deformation of the universal finishing mill pass, the position of the water tank after finishing rolling, the water volume of the water tank after finishing rolling, the water pressure of the water tank after finishing rolling, the number of nozzles of the water tank after finishing rolling, the water pressure of the water tank after finishing rolling, the final rolling speed of the rolled piece, the final rolling temperature of the rolled piece, the cooling speed of the rolled piece, the distance from the mill outlet to the cooling bed entrance, the upper cooling bed temperature of the rolled piece, and the lower cooling bed temperature of the mill.

[0048] The process of constructing the straightness prediction model after straightening includes:

[0049] Acquire a third historical production data set including straightening process parameters and final straightness of the H-beam after straightening;

[0050] Training a third neural network based on a third historical production data set to obtain a straightness prediction model after straightening;

[0051] Correct the straightness prediction model after straightening and determine the output result;

[0052] Among them, the straightness prediction model after straightening is corrected and the output results are determined to include:

[0053] Performing residual sequence processing on multiple groups of first final flatness prediction values ​​output by the straightening straightness prediction model and the measured final flatness of the corresponding data group to obtain multiple groups of final flatness deviation values;

[0054] Inputting multiple groups of final straightness deviation values ​​into a back propagation neural network to obtain final straightness deviation correction values;

[0055] Compensating the final flatness deviation correction value to the first final flatness prediction value to obtain a second final flatness prediction value;

[0056] The second final flatness prediction value is output as the final flatness.

[0057] Optionally, the third historical production data set is the straightening process parameters and the measured final straightness of the H-beam after straightening at the production site for at least half a year, or the straightening process parameters and the measured final straightness of the H-beam after straightening of an H-beam production line with an output exceeding 500,000 tons.

[0058] Training the third neural network based on the third historical production data set includes:

[0059] A third primitive model based on running time series is constructed through recurrent neural networks, stacked recurrent neural networks, and recurrent convolutional neural networks;

[0060] Dividing the third historical production data set into a third training set and a third test set according to a set ratio;

[0061] Select multiple factors that affect the final straightness of the H-beam from the third historical production data set as the input layer, and use the final straightness of the H-beam as the output layer;

[0062] The training function of the third original model is selected to obtain a straightness prediction model after straightening.

[0063] Optionally, the factors affecting the final straightness of the H-beam are the layout parameters and working parameters of the components of the straightening production line, the chemical composition content and the number of fixed lengths or multiples of the H-beam, and the number of neurons in the input layer is the number of selected factors affecting the final straightness of the H-beam.

[0064] If the number of data samples in the third training set is greater than 100,000, the square loss function is selected as the training function. The square loss function is L(f(x), y) = (f(x)-y) 2 ; If the number of data samples in the third training set is less than 100,000, the absolute value loss function is selected as the training function, and the absolute value loss function is L(Y,f(x))=|Yf(x)|.

[0065] Optionally, the factors affecting the straightness of the rolled piece after straightening in the third historical production data set include: C content, Si content, Mn content, S content, P content, Cr content, Nb content, V content, Ti content of H-shaped steel, straightening temperature, straightening speed, number of straightening rolls, reduction amount of each straightening roll, number of fixed lengths or number of multiple lengths.

[0066] The present invention also provides a method for controlling the straightness of an H-shaped steel, comprising the following steps:

[0067] Using the above prediction method, the predicted final straightness of the H-beam is obtained;

[0068] According to the target flatness required by production, the difference between the predicted final flatness and the target flatness is compared.

[0069] Difference = (|predicted final flatness - target initial flatness| ÷ target initial flatness) × 100%;

[0070] Match the corresponding control strategy from the strategy library according to the difference;

[0071] According to the control strategy obtained by matching, the operating parameters of the components of the rolling production line and / or the cooling production line and / or the straightening production line are controlled.

[0072] Optionally, the strategy library includes multiple control strategies, and the priority of each control strategy is determined according to its control efficiency. The control efficiency is negatively correlated with the time required for the final flatness to reach the target flatness after adopting the control strategy, and the control efficiency is positively correlated with the priority of the control strategy.

[0073] Optionally, the strategy library includes at least a first-class control strategy with a higher priority and a second-class control strategy with a lower priority, the first-class control strategy is a combined control strategy for controlling at least two working parameters related to the rolling process and / or the cooling process and / or the straightening process, and the second-class control strategy is a single control strategy for controlling one working parameter related to the rolling process or the cooling process or the straightening process.

[0074] When matching control strategies, if the difference is greater than the preset difference, the first type of control strategy is matched; if the difference is less than the preset difference, the second type of control strategy is matched.

[0075] The present invention also provides an H-beam flatness control system, comprising:

[0076] A data acquisition module, used to acquire first data related to the H-beam rolling and cooling process and second data related to the H-beam straightening process;

[0077] A first prediction module is used to input the first data into an initial flatness prediction model to obtain the predicted initial flatness at the head, tail and multiple positions in the middle of the H-beam before straightening;

[0078] A second prediction module is used to input the first data into a residual stress prediction model to obtain predicted residual stress distribution data of the H-beam finished product before straightening;

[0079] The third prediction module is used to input the predicted initial flatness, the predicted residual stress distribution data and the second data into the straightness prediction model after straightening to obtain the predicted final straightness at the head, tail and multiple positions in the middle of the H-beam finished product after straightening;

[0080] The control module is used to compare the predicted final flatness with the target flatness required by production, obtain the difference between the two, match the corresponding control strategy from the strategy library according to the difference, and control the working parameters of each component of the rolling production line and / or cooling production line and / or straightening production line according to the corresponding control strategy.

[0081] Optionally, the data acquisition module is also used to obtain a first historical production data set including rolling, cooling process parameters and the measured initial straightness of the H-beam before straightening, a second historical production data set including rolling, cooling process parameters and predicted residual stress distribution data of the H-beam before straightening, and a third historical production data set including straightening process parameters and the final straightness of the H-beam after straightening.

[0082] Optionally, the first historical production data set is the rolling and cooling process parameters of the production site for at least half a year and the measured initial straightness of the H-beam before straightening, or the rolling and cooling process parameters of an H-beam production line with an output exceeding 500,000 tons and the measured initial straightness of the H-beam before straightening; the second historical production data set is the rolling and cooling process parameters of the production site for at least half a year and the predicted residual stress distribution data of the H-beam before straightening, or the rolling and cooling process parameters of an H-beam production line with an output exceeding 500,000 tons and the predicted residual stress distribution data of the H-beam before straightening; the third historical production data set is the straightening process parameters of the production site for at least half a year and the measured final straightness of the H-beam after straightening, or the straightening process parameters of an H-beam production line with an output exceeding 500,000 tons and the measured final straightness of the H-beam after straightening.

[0083] Optionally, the first prediction module is also used to construct an initial flatness prediction model. The first prediction module trains a first neural network based on a first historical production data set, takes multiple factors affecting the initial flatness of the H-beam in the first historical production data set as an input layer, and takes the initial flatness of the H-beam as an output layer to obtain an initial flatness prediction model, and corrects the initial flatness prediction model to determine the output result.

[0084] Optionally, the second prediction module is also used to construct a residual stress prediction model. The second prediction module trains a second neural network based on a second historical production data set, takes multiple factors affecting the residual stress of the H-beam before straightening in the second historical production data set as the input layer, and takes the residual stress of the H-beam before straightening as the output layer to obtain a residual stress prediction model.

[0085] Optionally, the third prediction module is also used to construct a straightness prediction model after straightening. The third prediction module trains a third neural network based on a third historical production data set, takes multiple factors affecting the final flatness of the H-beam in the third historical production data set as the input layer, and takes the final flatness of the H-beam as the output layer to obtain a straightness prediction model after straightening, and corrects the straightness prediction model after straightening to determine the output result.

[0086] The present invention also provides an electronic device, comprising: a processor, and a memory connected to the processor in communication; the memory stores computer execution instructions; the processor executes the computer execution instructions stored in the memory to implement the aforementioned H-beam flatness control method.

[0087] The present invention also provides a computer-readable storage medium, in which computer-executable instructions are stored, and the computer-executable instructions are executed by a processor to implement the aforementioned H-beam flatness control method.

[0088] The present invention also provides a computer program product, comprising a computer program / instruction, which implements the aforementioned H-beam flatness control method when executed by a processor.

[0089] The beneficial effects of the present invention are:

[0090] The H-beam flatness prediction and control method provided by the present invention obtains an initial flatness prediction model, a residual stress prediction model and a straightness prediction model after straightening by training a neural network based on a historical production data set, thereby obtaining a predicted initial and predicted residual stress distribution. In the control process, the predicted initial flatness, predicted residual stress distribution data, and the sample data of the external dimensions at multiple positions of the flange and web of the H-beam finished product, various process parameters of the straightening production line, the fixed length or multiple lengths, and the chemical composition of the H-beam are input into the straightness prediction model after straightening to obtain the predicted final straightness of the head, tail and multiple positions in the middle of the H-beam finished product after straightening. Further compare the difference between the predicted final flatness and the target flatness, match the appropriate control strategy to control and adjust the working parameters of the various components of the rolling, cooling and straightening production lines, and complete the control of the flatness of the H-beam rolled piece.

[0091] In addition, the present invention deeply integrates the rolling, cooling and straightening processes with parameters related to straightness, fully considers the influence of the residual stress distribution of H-beam after rolling and cooling on the straightening process, and realizes the control of the straightness of H-beam rolled products. It has strong universality, can cover a large range of H-beam products, has strong adaptability to newly emerging sample data, and is more convenient to implement. Furthermore, since the size and chemical composition of H-beam are also input as influencing factors into each prediction model during the prediction and control process, the control method of the present invention can flexibly respond to the production needs of various types of steel from ordinary carbon structural steel to high-strength structural steel, covering a full range of specifications (100mm×50mm to 400mm×200mm) such as wide flange, middle flange, narrow flange, and thin wall, and has significant production adaptability.

[0092] In addition, when constructing the straightness prediction model after straightening, the present invention also takes into account the change in straightness of the H-beam finished product after straightening when it is further cut into fixed-length or double-length products. That is, the present invention takes into account the additional impact of fixed-length or double-length products on straightness after the straightening process is completed, so as to control the straightening process in advance to ensure the high straightness of double-length and fixed-length products. This makes it less likely for the straightened H-beam product to bend again after being fixed-length or double-length, reduces cutting loss, and improves metal yield and product quality.

[0093] Other advantages, objectives and features of the present invention will be described in the following description to some extent, and to some extent, will be obvious to those skilled in the art based on the following examination and study, or can be taught from the practice of the present invention. The objectives and other advantages of the present invention can be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0094] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be described in detail below in conjunction with the accompanying drawings, wherein:

[0095] Figure 1 Schematic diagram of the process of predicting the straightness of H-beam provided by the present invention

[0096] Figure 2 A schematic diagram of a process for constructing an initial straightness prediction model in the H-beam straightness prediction method provided by the present invention;

[0097] Figure 3 A schematic diagram of a flow chart of constructing a residual stress prediction model in the H-beam straightness prediction method provided by the present invention;

[0098] Figure 4 A schematic diagram of a process for constructing a straightness prediction model after straightening in the H-beam straightness prediction method provided by the present invention;

[0099] Figure 5 A schematic diagram of the process of the H-beam straightness control method provided by the present invention;

[0100] Figure 6 This is a schematic structural diagram of the H-beam straightness control system provided by the present invention.

[0101] Reference numerals:

[0102] 100 - data acquisition module; 200 - first prediction module; 300 - second prediction module; 400 - third prediction module; 500 - control module. DETAILED DESCRIPTION

[0103] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments only illustrate the basic concept of the present invention in a schematic manner, and the following embodiments and features in the embodiments can be combined with each other without conflict.

[0104] Among them, the drawings are only used for illustrative explanations, and they only represent schematic diagrams rather than actual pictures, and should not be understood as limitations on the present invention. In order to better illustrate the embodiments of the present invention, some parts of the drawings may be omitted, enlarged or reduced, and do not represent the size of actual products. For those skilled in the art, it is understandable that some well-known structures and their descriptions in the drawings may be omitted.

[0105] The same or similar numbers in the drawings of the embodiments of the present invention correspond to the same or similar parts; in the description of the present invention, it should be understood that if the terms "upper", "lower", "left", "right", "front", "rear", etc. indicate the orientation or position relationship, they are based on the orientation or position relationship shown in the drawings, which is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operate in a specific orientation. Therefore, the terms describing the position relationship in the drawings are only used for illustrative purposes and cannot be understood as limiting the present invention. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to specific circumstances.

[0106] Example 1

[0107] The present invention provides a method for predicting the straightness of H-beam. Figure 1 , including the following steps:

[0108] Step S1, data acquisition: acquiring first data related to the H-beam rolling process and second data related to the H-beam straightening process;

[0109] Step S2, initial flatness prediction: inputting the first data into an initial flatness prediction model to obtain the predicted initial flatness at the head, tail and multiple positions in the middle of the H-beam before straightening;

[0110] Step S3, residual stress prediction: inputting the first data into the residual stress prediction model to obtain the predicted residual stress distribution data of the H-beam finished product before straightening;

[0111] Step S4, final flatness prediction: input the predicted initial flatness, the predicted residual stress distribution data and the second data into the straightness prediction model after straightening to obtain the predicted final straightness at the head, tail and multiple positions in the middle of the H-beam finished product after straightening.

[0112] The data acquired in step S1 includes:

[0113] The first data includes first sub-data measured when the H-beam billet is at the entrance position of the final rolling pass and second sub-data measured when the H-beam finished product is at the exit position of the final rolling pass;

[0114] The first sub-data includes sample data of the external dimensions at multiple positions of the flange and web of the H-beam billet, pass reduction data at multiple positions, running speed sample data, and temperature field distribution sample data;

[0115] The second sub-data includes sample data of the dimensions of the finished flange and web at multiple locations of the H-beam, sample data of the running speed, sample data of the temperature field distribution from the mill outlet to the cooling bed inlet, and the chemical composition of the H-beam;

[0116] The second data includes sample data of the outer dimensions of the finished H-beam flange and web at multiple positions, various process parameters of the straightening production line, fixed length or multiple length, and chemical composition of the H-beam.

[0117] Step S2 further comprises: when performing initial flatness prediction, inputting the first sub-data into an initial flatness prediction model to obtain the predicted initial flatness at the head, tail and multiple positions in the middle of the H-beam before straightening;

[0118] Step S3 further includes: when predicting residual stress, inputting the second sub-data into a residual stress prediction model to obtain predicted residual stress distribution data of the H-beam finished product before straightening.

[0119] Step S2 includes:

[0120] Step S21: Constructing an initial flatness prediction model

[0121] Step S22: input the first data into a residual stress prediction model to obtain predicted residual stress distribution data of the H-beam finished product before straightening.

[0122] like Figure 2 , step S21 comprises:

[0123] Step S211, obtaining a first historical production data set including rolling process parameters and the measured initial straightness of the H-beam before straightening;

[0124] Step S212: training a first neural network based on a first historical production data set to obtain an initial flatness prediction model;

[0125] Step S213: calibrate the initial flatness prediction model and determine the output result.

[0126] The first historical production data set in step S211 is the rolling process parameters of the production site for at least half a year and the measured initial straightness of the H-beam before straightening, or the rolling process parameters of an H-beam production line with an output of more than 500,000 tons and the measured initial straightness of the H-beam before straightening.

[0127] Table 1 shows some of the first historical production data (for H-beams with specifications of H340×250)

[0128]

[0129] Step S212 includes:

[0130] Step S2121, constructing a first original model based on the running time series through a recurrent neural network, a stacked recurrent neural network and a recurrent convolutional neural network;

[0131] Step S2122: dividing the first historical production data set into a first training set and a first test set according to a set ratio;

[0132] Step S2123, selecting multiple factors affecting the initial straightness of the H-beam from the first historical production data set as the input layer, and taking the initial straightness of the H-beam as the output layer;

[0133] Step S2124: Select the training function of the first original model to obtain an initial flatness prediction model.

[0134] In step S2122, the first historical production data set is divided into a first training set and a first test set in a ratio of 8:2. The first training set is used to train and optimize the initial flatness prediction model, and the first test set is used to verify the results of the initial flatness prediction model.

[0135] In step S2123, the factors affecting the initial flatness of the H-beam are the layout parameters and working parameters of the components of the rolling production line and the cooling production line, and the number of neurons in the input layer is the number of selected factors affecting the flatness.

[0136] The factors affecting the initial straightness of H-beam in the first historical production data set include: the position of the universal mill of the finishing mill, the deformation of the universal mill pass, the rolling temperature of the universal mill, the position of the edge rolling mill, the deformation of the edge rolling mill pass, the position of the universal finishing mill, the deformation of the universal finishing mill pass, the position of the water tank after finishing rolling, the water volume of the water tank after finishing rolling, the water pressure of the water tank after finishing rolling, the number of nozzles of the water tank after finishing rolling, the water pressure of the water tank after finishing rolling, the final rolling speed of the rolled piece, the final rolling temperature of the rolled piece, the section modulus of the rolled piece, the distance from the mill outlet to the cooling bed entrance, and the temperature of the cooling bed on the rolled piece.

[0137] In step S2123, the factors affecting the initial straightness of H-beam are selected from the first historical production data set: the position of the universal mill of the final rolling mill group, the deformation of the universal mill pass, the rolling temperature of the universal mill, the position of the edge rolling mill, the deformation of the edge rolling mill pass, the position of the universal finishing mill, the deformation of the universal finishing mill pass, the position of the water tank after finishing rolling, the water volume of the water tank after finishing rolling, the water pressure of the water tank after finishing rolling, the number of nozzles of the water tank after finishing rolling, the water pressure of the water tank after finishing rolling, the final rolling speed of the rolled piece, the final rolling temperature of the rolled piece, the section modulus of the rolled piece, the distance from the mill outlet to the cooling bed entrance, and the temperature of the cooling bed on the rolled piece. At this time, the number of neurons in the input layer is 17.

[0138] In step S2124, when selecting a training function, if the number of data samples in the first training set is greater than 100,000, the square loss function is selected as the training function, and the square loss function is L(f(x), y) = (f(x) - y) 2 ; If the number of data samples in the first training set is less than 100,000, select the absolute value loss function as the training function, and the absolute value loss function is L(Y,f(x))=|Yf(x)|.

[0139] Step S213 includes:

[0140] Step S2131, performing residual sequence processing on multiple groups of first initial flatness prediction values ​​output by the initial flatness prediction model and the measured initial flatness of the corresponding data group to obtain multiple groups of initial flatness deviation values;

[0141] Step S2132, inputting multiple groups of initial flatness deviation values ​​into a back propagation neural network to obtain initial flatness deviation correction values;

[0142] Step S2133, compensating the initial flatness deviation correction value to the first initial flatness prediction value to obtain a second initial flatness prediction value;

[0143] Step S2134: output the second initial flatness prediction value as the predicted initial flatness.

[0144] Step S3 includes:

[0145] Step S31, constructing a residual stress prediction model;

[0146] Step S32: input the first data into a residual stress prediction model to obtain predicted residual stress distribution data of the H-beam finished product before straightening.

[0147] like Figure 3 , step S31 comprises:

[0148] Step S311, obtaining a second historical production data set including rolling process parameters and predicted residual stress distribution data of H-beam before straightening;

[0149] Step S312: training a second neural network and a residual stress prediction model based on a second historical production data set.

[0150] The second historical production data set in step S311 is the rolling process parameters and the predicted residual stress distribution data of H-beam before straightening at the production site for at least half a year, or the rolling process parameters and the predicted residual stress distribution data of H-beam before straightening of an H-beam production line with an output exceeding 500,000 tons.

[0151] Table 2 Part of the second historical production data display (for H-beam with specification of H340×250)

[0152]

[0153] Step S312 includes:

[0154] Step S3121, constructing a second original model based on the running time series through a recurrent neural network, a stacked recurrent neural network and a recurrent convolutional neural network;

[0155] Step S3122, dividing the second historical production data set into a second training set and a second test set according to a set ratio;

[0156] Step S3123, selecting multiple factors that affect the residual stress of the H-beam before straightening from the second historical production data set as the input layer, and taking the residual stress of the H-beam before straightening as the output layer;

[0157] Step S3124: Select the training function of the second original model to obtain a residual stress prediction model.

[0158] In step S3122, if the number of data samples in the first training set is greater than 100,000, the square loss function is selected as the training function, and the square loss function is L(f(x), y) = (f(x)-y) 2 ; If the number of data samples in the first training set is less than 100,000, select the absolute value loss function as the training function, and the absolute value loss function is L(Y,f(x))=|Yf(x)|.

[0159] In step S3123, the factors affecting the residual stress of the H-beam before straightening are the layout parameters and working parameters of the components of the rolling production line and the cooling production line, as well as the chemical composition of the H-beam. The number of neurons in the input layer is the number of selected factors affecting the residual stress of the H-beam before straightening.

[0160] The factors affecting the residual stress of H-beam before straightening in the second historical production data set include: C content, Si content, Mn content, S content, P content, Cr content, Nb content, V content, and Ti content of H-beam, the position of the universal rolling mill of the final rolling mill, the deformation of the universal rolling mill pass, the rolling temperature of the universal rolling mill, the position of the edge rolling mill, the deformation of the edge rolling mill pass, the position of the universal finishing mill, the deformation of the universal finishing mill pass, the position of the water tank after finishing rolling, the water volume of the water tank after finishing rolling, the water pressure of the water tank after finishing rolling, the number of nozzles of the water tank after finishing rolling, the water pressure of the water tank after finishing rolling, the final rolling speed of the rolled piece, the final rolling temperature of the rolled piece, the cooling speed of the rolled piece, the distance from the mill outlet to the cooling bed entrance, the upper cooling bed temperature of the rolled piece, and the lower cooling bed temperature of the mill.

[0161] In step S3123, the factors affecting the residual stress of the H-beam before straightening are selected as follows: C content, Si content, Mn content, S content, P content, Cr content, Nb content, V content, and Ti content of the H-beam, the position of the universal mill of the final rolling mill, the deformation of the universal mill pass, the rolling temperature of the universal mill, the position of the edge rolling mill, the deformation of the edge rolling mill pass, the position of the universal finishing mill, the deformation of the universal finishing mill pass, the position of the water tank after finishing rolling, the water volume of the water tank after finishing rolling, the water pressure of the water tank after finishing rolling, the number of nozzles of the water tank after finishing rolling, the water pressure of the water tank after finishing rolling, the final rolling speed of the rolled piece, the final rolling temperature of the rolled piece, the cooling speed of the rolled piece, the distance from the mill outlet to the cooling bed entrance, the upper cooling bed temperature of the rolled piece, and the lower cooling bed temperature of the mill. At this time, the number of neurons in the input layer is 27.

[0162] Step S4 includes:

[0163] Step S41, constructing a straightness prediction model after straightening;

[0164] Step S42, inputting the predicted initial flatness, the predicted residual stress distribution data and the second data into a straightness prediction model after straightening to obtain the predicted final straightness at the head, tail and multiple positions in the middle of the H-beam finished product after straightening.

[0165] like Figure 4 , step S41 comprises:

[0166] Step S411, obtaining a third historical production data set including straightening process parameters and final straightness of the H-beam after straightening;

[0167] Step S412: training a third neural network based on a third historical production data set to obtain a straightness prediction model after straightening;

[0168] Step S413, calibrating the straightness prediction model after straightening and determining the output result.

[0169] Step S412 includes:

[0170] Step S4121, constructing a third original model based on the running time series through a recurrent neural network, a stacked recurrent neural network and a recurrent convolutional neural network;

[0171] Step S4122: dividing the third historical production data set into a third training set and a third test set according to a set ratio;

[0172] Step S4123, selecting multiple factors affecting the final straightness of the H-beam from the third historical production data set as the input layer, and taking the final straightness of the H-beam as the output layer;

[0173] Step S4124: Select the training function of the third original model to obtain a straightness prediction model after straightening.

[0174] The third historical production data set in step S411 is the straightening process parameters and the measured final straightness of the H-beam after straightening at the production site for at least half a year, or the straightening process parameters and the measured final straightness of the H-beam after straightening of an H-beam production line with an output exceeding 500,000 tons.

[0175] Table 3 Part III Historical Production Data Display (for H-beam with specification of H340×250)

[0176]

[0177] In step S4123, the factors affecting the final straightness of the H-beam are the layout parameters and working parameters of the components of the straightening production line, the chemical composition content and the number of fixed lengths or multiples of the H-beam. The number of neurons in the input layer is the number of selected factors affecting the final straightness of the H-beam.

[0178] The factors affecting the straightness of the rolled piece after straightening in the third historical production data set include: C content, Si content, Mn content, S content, P content, Cr content, Nb content, V content, Ti content of H-shaped steel, straightening temperature, straightening speed, number of straightening rollers, reduction of each straightening roller, number of fixed lengths or multiple lengths.

[0179] In step S4123, the factors affecting the straightness of the rolled product after straightening are selected from the third historical production data set: C content, Si content, Mn content, S content, P content, Cr content, Nb content, V content, Ti content of H-beam, straightening temperature, straightening speed, number of straightening rollers, the pressure of each straightening roller of the nine-roller straightening machine (9 straightening rollers have 8 pressure data), fixed length or multiple length. At this time, the number of neurons in the input layer is 21.

[0180] When selecting a training function in step S4124, if the number of data samples in the third training set is greater than 100,000, the square loss function is selected as the training function, and the square loss function is L(f(x), y) = (f(x) - y) 2 ; If the number of data samples in the third training set is less than 100,000, the absolute value loss function is selected as the training function, and the absolute value loss function is L(Y,f(x))=|Yf(x)|.

[0181] Step S413 includes:

[0182] Step S4131, performing residual sequence processing on multiple groups of first final flatness prediction values ​​output by the straightening straightness prediction model and the measured final flatness of the corresponding data group to obtain multiple groups of final flatness deviation values;

[0183] Step S4132, inputting multiple groups of final flatness deviation values ​​into the back propagation neural network to obtain final flatness deviation correction values;

[0184] Step S4133, compensating the final flatness deviation correction value to the first final flatness prediction value to obtain a second final flatness prediction value;

[0185] Step S4134: output the second final flatness prediction value as the final flatness.

[0186] Example 2

[0187] The present invention also provides a method for controlling the straightness of an H-shaped steel. Figure 5 , including the following steps:

[0188] Step S1, using the prediction method provided in Example 1 to obtain the predicted final straightness of the H-beam;

[0189] Step S2: Compare the difference between the predicted final flatness and the target flatness according to the target flatness required by production.

[0190] Step S3, matching the corresponding control strategy from the strategy library according to the difference degree;

[0191] Step S4: Control and / or adjust the operating parameters of the components of the rolling and straightening production line according to the control strategy obtained by matching.

[0192] The calculation formula of the difference in step S2 is: difference = (|predicted final flatness - target initial flatness| ÷ target initial flatness) × 100%;

[0193] The strategy library in step S3 includes multiple control strategies, and the priority of each control strategy is determined according to its control efficiency. The control efficiency is negatively correlated with the time required for the final flatness to reach the target flatness after adopting the control strategy, and the control efficiency is positively correlated with the priority of the control strategy.

[0194] The strategy library in step S3 includes at least a first-class control strategy with a higher priority and a second-class control strategy with a lower priority. The first-class control strategy is a combined control strategy for controlling at least two working parameters related to the rolling process and / or the straightening process, and the second-class control strategy is a single control strategy for controlling one working parameter related to the rolling process or the straightening process.

[0195] When matching the control strategies in step S3, if the difference is greater than the preset difference, the first type of control strategy is matched; if the difference is less than the preset difference, the second type of control strategy is matched.

[0196] Example 3

[0197] The present invention also provides an H-beam flatness control system, such as Figure 6 , including a data acquisition module 100, which is used to obtain first data related to the H-beam rolling and cooling process and second data related to the H-beam straightening process; a first prediction module 200, which is used to input the first data into the initial flatness prediction model to obtain the predicted initial flatness of the head, tail and multiple positions in the middle of the H-beam before straightening; a second prediction module 300, which is used to input the first data into the residual stress prediction model to obtain the predicted residual stress distribution data of the finished H-beam before straightening; a third prediction module 400, which is used to input the predicted initial flatness, the predicted residual stress distribution data and the second data into the straightness prediction model after straightening to obtain the predicted final flatness of the head, tail and multiple positions in the middle of the finished H-beam after straightening; a control module 500, which is used to compare the predicted final flatness with the target flatness required by production to obtain the difference between the two, match the corresponding control strategy from the strategy library according to the difference, and control the working parameters of each component of the rolling production line and / or the cooling production line and / or the straightening production line according to the corresponding control strategy.

[0198] The data acquisition module 100 is also used to obtain a first historical production data set including rolling, cooling process parameters and the measured initial straightness of the H-beam before straightening, a second historical production data set including rolling, cooling process parameters and predicted residual stress distribution data of the H-beam before straightening, and a third historical production data set including straightening process parameters and the final straightness of the H-beam after straightening.

[0199] The first historical production data set is the rolling and cooling process parameters of the production site for at least half a year and the measured initial straightness of the H-beam before straightening, or the rolling and cooling process parameters of the H-beam production line with an output of more than 500,000 tons and the measured initial straightness of the H-beam before straightening; the second historical production data set is the rolling and cooling process parameters of the production site for at least half a year and the predicted residual stress distribution data of the H-beam before straightening, or the rolling and cooling process parameters of the H-beam production line with an output of more than 500,000 tons and the predicted residual stress distribution data of the H-beam before straightening; the third historical production data set is the straightening process parameters of the production site for at least half a year and the measured final straightness of the H-beam after straightening, or the straightening process parameters of the H-beam production line with an output of more than 500,000 tons and the measured final straightness of the H-beam after straightening.

[0200] The first prediction module 200 is also used to construct an initial flatness prediction model. The first prediction module 200 trains a first neural network based on a first historical production data set, takes multiple factors affecting the initial flatness of the H-beam in the first historical production data set as an input layer, and takes the initial flatness of the H-beam as an output layer to obtain an initial flatness prediction model, and corrects the initial flatness prediction model to determine the output result.

[0201] The second prediction module 300 is also used to construct a residual stress prediction model. The second prediction module 300 trains a second neural network based on a second historical production data set, takes multiple factors affecting the residual stress of the H-beam before straightening in the second historical production data set as the input layer, and takes the residual stress of the H-beam before straightening as the output layer to obtain a residual stress prediction model.

[0202] The third prediction module 400 is also used to construct a straightness prediction model after straightening. The third prediction module 400 trains a third neural network based on a third historical production data set, takes multiple factors affecting the final straightness of the H-beam in the third historical production data set as the input layer, and takes the final straightness of the H-beam as the output layer to obtain a straightness prediction model after straightening, and corrects the straightness prediction model after straightening to determine the output result.

[0203] In some embodiments, the present invention further provides an electronic device, including a processor and a memory communicatively connected to the processor; the memory stores computer-executable instructions.

[0204] In this embodiment, the memory may include a random access memory (RAM), and may also include a non-volatile memory (non-volatile memory), such as at least one disk storage. The above-mentioned processor may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it may also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, discrete gates or transistor logic devices, and discrete hardware components.

[0205] In some embodiments, the present invention further provides a computer program product, including a computer program / instruction, which, when executed by a processor, implements the H-beam flatness control method described in Example 2.

[0206] In some embodiments, the present invention further provides a computer-readable storage medium, in which computer execution instructions are stored, and the computer execution instructions are executed by a processor to implement the H-beam straightness control method described in Example 2.

[0207] The computer-readable storage medium in this embodiment can be understood by ordinary technicians in this field: all or part of the steps of implementing the above-mentioned method embodiments can be completed by hardware related to the computer program. The aforementioned computer program can be stored in a computer-readable storage medium. When the program is executed, the steps of the above-mentioned method embodiments are executed; and the aforementioned storage medium includes: ROM, RAM, magnetic disk or optical disk and other media that can store program codes.

[0208] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution, which should be included in the scope of the claims of the present invention.

Claims

1. A method for predicting the straightness of an H-beam, characterized in that: The following steps are involved: Acquire first data and second data; Inputting the first data into an initial flatness prediction model to obtain a predicted initial flatness of the H-beam before straightening; Inputting the first data into a residual stress prediction model to obtain predicted residual stress distribution data of the H-beam before straightening; The predicted initial flatness, the predicted residual stress distribution data and the second data are input into a straightness prediction model after straightening to obtain the predicted final flatness at the head, tail and multiple positions in the middle of the H-beam.

2. The prediction method according to claim 1, characterized in that: The first data includes first sub-data measured when the H-beam billet is at the entrance position of the final rolling pass and second sub-data measured when the H-beam finished product is at the exit position of the final rolling pass; The first sub-data includes sample data of the external dimensions at multiple positions of the flange and web of the H-beam billet, pass reduction data at multiple positions, running speed sample data, and temperature field distribution sample data; The second sub-data includes sample data of the dimensions of the finished flange and web at multiple positions of the H-beam, sample data of the running speed, sample data of the temperature field distribution from the mill outlet to the cooling bed inlet, and the chemical composition of the H-beam; The second data includes sample data of the outer dimensions of the finished H-beam flange and web at multiple positions, various process parameters of the straightening production line, fixed length or multiple length, and chemical composition of the H-beam.

3. The prediction method according to claim 2, characterized in that: When performing initial flatness prediction, the first sub-data is input into the initial flatness prediction model to obtain the predicted initial flatness at the head, tail and multiple positions in the middle of the H-beam before straightening; When performing residual stress distribution prediction, the second sub-data is input into the residual stress prediction model to obtain predicted residual stress distribution data at the head, tail and multiple positions in the middle of the H-beam before straightening.

4. The prediction method according to claim 1, characterized in that: The construction process of the initial flatness prediction model includes: Acquire a first historical production data set including rolling and cooling process parameters and measured initial straightness of H-beam before straightening; Training a first neural network based on the first historical production data set to obtain the initial flatness prediction model; Correcting the initial flatness prediction model and determining an output result; Wherein, correcting the initial flatness prediction model and determining the output result comprises: Performing residual sequence processing on multiple groups of first initial flatness prediction values ​​output by the initial flatness prediction model and the measured initial flatness of the corresponding data group to obtain multiple groups of initial flatness deviation values; Inputting the multiple groups of initial flatness deviation values ​​into a back propagation neural network to obtain initial flatness deviation correction values; Compensating the initial flatness deviation correction value to the first initial flatness prediction value to obtain a second initial flatness prediction value; The second initial flatness prediction value is output as the predicted initial flatness.

5. The prediction method according to claim 4, characterized in that: The first historical production data set is the rolling and cooling process parameters and the measured initial straightness of the H-beam before straightening for at least half a year at the production site, or the rolling and cooling process parameters and the measured initial straightness of the H-beam before straightening for an H-beam production line with an output of more than 500,000 tons; wherein, The factors affecting the initial straightness of H-shaped steel in the first historical production data set include: the position of the universal rolling mill of the finishing mill group, the deformation amount of the universal rolling mill pass, the rolling temperature of the universal rolling mill, the position of the edge rolling mill, the deformation amount of the edge rolling mill pass, the position of the universal finishing mill, the deformation amount of the universal finishing mill pass, the position of the water tank after finishing rolling, the water volume of the water tank after finishing rolling, the water pressure of the water tank after finishing rolling, the number of nozzles of the water tank after finishing rolling, the water pressure of the water tank after finishing rolling, the final rolling speed of the rolled piece, the final rolling temperature of the rolled piece, the section modulus of the rolled piece, the distance from the mill outlet to the cooling bed entrance, and the temperature of the cooling bed on the rolled piece.

6. The prediction method according to claim 5, characterized in that: Training a first neural network based on the first historical production data set includes: Constructing a first primitive model based on running time series through recurrent neural networks, stacked recurrent neural networks, and recurrent convolutional neural networks; Dividing the first historical production data set into a first training set and a first test set according to a set ratio; Selecting multiple factors affecting the initial straightness of the H-beam from the first historical production data set as an input layer, and using the initial straightness of the H-beam as an output layer; Selecting the training function of the first original model to obtain the initial flatness prediction model; in, The factors affecting the initial flatness of the H-beam are the layout parameters and working parameters of the components of the rolling production line and the cooling production line, and the number of neurons in the input layer is the number of factors selected to affect the initial flatness; If the number of data samples in the first training set is greater than 100,000, the square loss function is selected as the training function, and the square loss function is L(f(x), y) = (f(x)-y) 2 ; If the number of data samples in the first training set is less than 100,000, the absolute value loss function is selected as the training function, and the absolute value loss function is L(Y, f(x)) = |Yf(x)|.

7. The prediction method according to claim 1, characterized in that: The construction process of the residual stress prediction model includes: Acquire a second historical production data set including rolling and cooling process parameters and predicted residual stress distribution data of H-beam before straightening; The second neural network is trained based on the second historical production data set to obtain the residual stress prediction model; wherein, The second historical production data set is the rolling, cooling process parameters and predicted residual stress distribution data of H-beam before straightening at the production site for at least half a year, or the rolling, cooling process parameters and predicted residual stress distribution data of H-beam before straightening of an H-beam production line with an output of more than 500,000 tons.

8. The prediction method according to claim 7, characterized in that: Training a second neural network based on the second historical production data set includes: A second primitive model based on the running time series is constructed by using a recurrent neural network, a stacked recurrent neural network, and a recurrent convolutional neural network; Dividing the second historical production data set into a second training set and a second test set according to a set ratio; Selecting multiple factors affecting the residual stress of the H-beam before straightening from the second historical production data set as an input layer, and taking the residual stress of the H-beam before straightening as an output layer; Selecting the training function of the second original model to obtain a residual stress prediction model; in, The factors affecting the residual stress of the H-beam before straightening are the layout parameters and working parameters of the components of the rolling production line and the cooling production line, and the chemical composition of the H-beam; The number of neurons in the input layer is the number of factors selected to affect the residual stress of the H-beam before straightening; If the number of data samples in the first training set is greater than 100,000, the square loss function is selected as the training function, and the square loss function is L(f(x), y) = (f(x)-y) 2 ; If the number of data samples in the first training set is less than 100,000, the absolute value loss function is selected as the training function, and the absolute value loss function is L(Y, f(x)) = |Yf(x)|.

9. The prediction method according to claim 10, characterized in that: The factors affecting the residual stress of H-beam before straightening in the second historical production data set include: C content, Si content, Mn content, S content, P content, Cr content, Nb content, V content, and Ti content of the H-beam, the position of the universal rolling mill of the finishing mill, the deformation amount of the universal rolling mill pass, the rolling temperature of the universal rolling mill, the position of the edge rolling mill, the deformation amount of the edge rolling mill pass, the position of the universal finishing mill, the deformation amount of the universal finishing mill pass, the position of the water tank after finishing rolling, the water volume of the water tank after finishing rolling, the water pressure of the water tank after finishing rolling, the number of nozzles of the water tank after finishing rolling, the water pressure of the water tank after finishing rolling, the final rolling speed of the rolled piece, the final rolling temperature of the rolled piece, the cooling speed of the rolled piece, the distance from the mill outlet to the cooling bed entrance, the upper cooling bed temperature of the rolled piece, and the lower cooling bed temperature of the mill.

10. The prediction method according to claim 1, characterized in that: The process of constructing the straightness prediction model after straightening includes: Acquire a third historical production data set including straightening process parameters and final straightness of the H-beam after straightening; Training a third neural network based on the third historical production data set to obtain the straightness prediction model after straightening; Correcting the straightness prediction model after straightening to determine the output result; Wherein, correcting the straightness prediction model after straightening and determining the output result includes: Performing residual sequence processing on multiple groups of first final flatness prediction values ​​output by the straightening straightness prediction model and the measured final flatness of the corresponding data group to obtain multiple groups of final flatness deviation values; Inputting the multiple groups of final flatness deviation values ​​into a back propagation neural network to obtain final flatness deviation correction values; Compensating the final flatness deviation correction value to the first final flatness prediction value to obtain a second final flatness prediction value; The second final flatness prediction value is output as the final flatness.

11. The prediction method according to claim 10, characterized in that: The third historical production data set is the straightening process parameters and the measured final straightness of the H-beam after straightening at the production site for at least half a year, or the straightening process parameters and the measured final straightness of the H-beam after straightening of an H-beam production line with an output of more than 500,000 tons; wherein, The factors affecting the straightness of the rolled piece after straightening in the third historical production data set include: C content, Si content, Mn content, S content, P content, Cr content, Nb content, V content, Ti content of H-shaped steel, straightening temperature, straightening speed, number of straightening rollers, reduction amount of each straightening roller, number of fixed lengths or number of multiple lengths.

12. The prediction method according to claim 11, characterized in that: Training a third neural network based on the third historical production data set includes: A third primitive model based on running time series is constructed through recurrent neural networks, stacked recurrent neural networks, and recurrent convolutional neural networks; Dividing the third historical production data set into a third training set and a third test set according to a set ratio; Selecting multiple factors affecting the final straightness of the H-beam from the third historical production data set as an input layer, and taking the final straightness of the H-beam as an output layer; Selecting the training function of the third original model to obtain the straightness prediction model after straightening; in, The factors affecting the final straightness of the H-beam are the layout parameters and working parameters of the components of the straightening production line, the chemical composition content and the number of cuts or multiples of the H-beam, and the number of neurons in the input layer is the number of factors selected to affect the final straightness of the H-beam; If the number of data samples in the third training set is greater than 100,000, the square loss function is selected as the training function, and the square loss function is L(f(x), y) = (f(x) - y) 2 ; If the number of data samples in the third training set is less than 100,000, the absolute value loss function is selected as the training function, and the absolute value loss function is L(Y, f(x)) = |Yf(x)|.

13. A method for controlling the straightness of an H-beam, characterized in that: The following steps are involved: Using the prediction method described in claims 1 to 12, the predicted final straightness of the H-beam is obtained; According to the target flatness required by production, the difference between the predicted final flatness and the target flatness is compared, Difference = (|predicted final flatness - target initial flatness| ÷ target initial flatness) × 100%; Matching a corresponding control strategy from a strategy library according to the difference; According to the control strategy obtained by matching, the operating parameters of the components of the rolling production line and / or the cooling production line and / or the straightening production line are controlled; in, The strategy library includes multiple control strategies, and the respective priorities of each control strategy are determined according to their control efficiency. The control efficiency is negatively correlated with the time required for the final flatness to reach the target flatness after adopting the control strategy, and the control efficiency is positively correlated with the priority of the control strategy.

14. The control method according to claim 13, characterized in that: The strategy library includes at least a first-class control strategy with a higher priority and a second-class control strategy with a lower priority; The first type of control strategy is a combined control strategy for controlling at least two working parameters related to the rolling process and / or the cooling process and / or the straightening process; The second type of control strategy is a single control strategy for controlling a working parameter related to a rolling process, a cooling process, or a straightening process; wherein, When matching the control strategy, if the difference is greater than the preset difference, the first type of control strategy is matched; When matching the control strategies, if the difference is less than a preset difference, the second type of control strategy is matched.

15. An H-beam flatness control system, characterized in that: include: A data acquisition module, used to acquire first data related to the H-beam rolling and cooling process and second data related to the H-beam straightening process; A first prediction module, used for inputting the first data into an initial flatness prediction model to obtain the predicted initial flatness at the head, tail and multiple positions in the middle of the H-beam before straightening; A second prediction module is used to input the first data into a residual stress prediction model to obtain predicted residual stress distribution data of the H-beam finished product before straightening; A third prediction module is used to input the predicted initial flatness, the predicted residual stress distribution data and the second data into a straightness prediction model after straightening to obtain the predicted final straightness at the head, tail and multiple positions in the middle of the H-beam finished product after straightening; A control module is used to compare the predicted final flatness with the target flatness required by production, obtain the difference between the two, match the corresponding control strategy from the strategy library according to the difference, and control the working parameters of each component of the rolling production line and / or the cooling production line and / or the straightening production line according to the corresponding control strategy.

16. The control system according to claim 15, characterized in that: The data acquisition module is also used to obtain a first historical production data set including rolling, cooling process parameters and the measured initial straightness of the H-beam before straightening, a second historical production data set including rolling, cooling process parameters and predicted residual stress distribution data of the H-beam before straightening, and a third historical production data set including straightening process parameters and the final straightness of the H-beam after straightening; wherein, The first historical production data set is the rolling and cooling process parameters and the measured initial straightness of the H-beam before straightening at the production site for at least half a year, or the rolling and cooling process parameters and the measured initial straightness of the H-beam before straightening of an H-beam production line with an output of more than 500,000 tons; The second historical production data set is the rolling and cooling process parameters and the predicted residual stress distribution data of H-beam before straightening for at least half a year at the production site, or the rolling and cooling process parameters and the predicted residual stress distribution data of H-beam before straightening for an H-beam production line with an output of more than 500,000 tons; The third historical production data set is the straightening process parameters and the measured final straightness of the H-beam after straightening at the production site for at least half a year, or the straightening process parameters and the measured final straightness of the H-beam after straightening of an H-beam production line with an output exceeding 500,000 tons.

17. The control system according to claim 16, characterized in that: The first prediction module is also used to construct the initial flatness prediction model. The first prediction module trains a first neural network based on the first historical production data set, uses multiple factors affecting the initial flatness of the H-beam in the first historical production data set as an input layer, and uses the initial flatness of the H-beam as an output layer to obtain an initial flatness prediction model, and corrects the initial flatness prediction model to determine an output result; The second prediction module is also used for constructing the residual stress prediction model. The second prediction module trains a second neural network based on the second historical production data set, takes multiple factors affecting the residual stress of the H-beam before straightening in the second historical production data set as the input layer, and takes the residual stress of the H-beam before straightening as the output layer, so as to obtain the residual stress prediction model; The third prediction module is also used to construct the straightness prediction model after straightening. The third prediction module trains the third neural network based on the third historical production data set, takes multiple factors affecting the final straightness of the H-beam in the third historical production data set as the input layer, and takes the final straightness of the H-beam as the output layer to obtain the straightness prediction model after straightening, and corrects the straightness prediction model after straightening to determine the output result.

18. An electronic device, characterized in that: include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executable instructions; The processor executes the computer-executable instructions stored in the memory to implement the control method according to any one of claims 13 to 14.

19. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, which are used to implement the control method according to any one of claims 13 to 14 when executed by a processor.

20. A computer program product comprising a computer program / instructions, characterized in that: When the computer program / instructions are executed by a processor, the control method according to any one of claims 13 to 14 is implemented.

Citation Information

Patent Citations

  • Steel plate straightening method and device, terminal equipment and storage medium

    CN115446150A

  • Plate and strip rolling process quality comprehensive prediction and process regulation and control method

    CN116140374A

  • Straightener plate shape prediction and straightening process control method

    CN116441354A

  • Method and device for measuring shot peening strengthening residual stress based on neural network

    CN117288365A

  • Hot-rolled strip steel flattening strip shape control method and system based on strip shape distribution characteristics

    CN117655116A

Cited By

  • Controlled rolling method for improving rolling straightness of medium plate

    CN120502589A