Multi-scale quality control system and method for large-size slab intelligent hot rolling

Abstract

The present application belongs to the technical field of metal rolling, and particularly relates to a multi-scale quality control system and method for large-size slab intelligent hot rolling. In order to realize quality abnormality control and process coordination and optimization control in the rolling process of strip and plate, a microstructure model based on crystallographic principles and physical principles is established, the required material parameters are obtained through experiments, and the rolling process organization prediction is realized by means of computer simulation calculation. By establishing a macro-micro evolution model of the slab rolling, the model is embedded into an intelligent control system to realize process optimization based on organization prediction. The reliable model and system can further widen the application range of the prediction model, improve the accuracy and reliability of the prediction, and also provide strong data support for intelligent production of rolling process and ideas and methods for the intelligent development of strip and plate.

Description

Technical Field

[0001] This invention belongs to the field of metal rolling technology, specifically relating to a multi-scale quality control system and method for intelligent hot rolling of large-size slabs. Background Technology

[0002] Plate and strip steel are important products in the steel industry, reflecting to some extent the continuity of production and economies of scale of enterprises. In recent years, driven by the "dual carbon" target, plate and strip rolling technology has developed towards high-end product performance, advanced production technology, and green and intelligent production lines. The hot rolling process of plate and strip steel involves heating and thermal deformation, which macroscopically changes the shape of the slab and microscopically involves material thermal stress and microstructure transformation. The nonlinearity, time-varying nature, and complex thermo-mechanical coupling of the hot rolling process pose significant challenges to the diagnosis of product quality anomalies and the stability of hot rolling production.

[0003] Researching integrated intelligent control of strip and sheet rolling quality is a development direction for realizing intelligent production. Following the requirements of process autonomy, collaborative diagnosis and optimization, principle-data correlation operation, and virtual-real collaborative control, it is essential to establish relevant material databases, process model databases, and geometric model databases. The quality of strip and sheet products is inextricably linked to their microstructure characteristics; product performance can be improved by controlling the microstructure and refining grain size. Traditional microstructure observation and analysis often rely on experimental research, which is generally time-consuming, costly, and difficult to cover all process conditions.

[0004] Improving model accuracy and dynamic adaptation to complex working conditions, and achieving coordinated optimization at all levels of the process, can lead to improved product quality and production efficiency, while reducing costs and emissions. Therefore, quality anomaly control and process coordination optimization control in the strip rolling process are urgent problems to be solved. Summary of the Invention

[0005] This invention addresses the aforementioned problems by providing a multi-scale quality control system and method for intelligent hot rolling of large-size slabs. By integrating digital twins of macroscopic deformation conditions and microscopic scale models and comparing production data, it achieves production quality control, anomaly diagnosis, and reverse optimization of hot rolling process parameters.

[0006] To achieve the above objectives, the present invention employs the following technical solution:

[0007] This invention provides a multi-scale quality control system for intelligent hot rolling of large-size slabs, comprising:

[0008] The virtual computing module performs numerical simulation calculations on the slab rolling process based on numerical simulation software and outputs the simulation calculation results.

[0009] The quality control module records the technical parameters of the slab rolling production process and compares them with the simulation results of the virtual calculation module.

[0010] The quality anomaly diagnosis module analyzes the differences in quality status based on the process parameters of the slab rolling process, the simulation calculation results of the virtual calculation module, and the comparison results of the quality control module. It optimizes the process parameters and feeds the optimized process parameters back to the slab rolling process for control.

[0011] Furthermore, the process parameters for the slab rolling process include material parameters, slab dimensions, hot rolling process parameters, and hot rolling equipment parameters.

[0012] Furthermore, the virtual calculation module utilizes the finite element analysis software ABAQUS. The process parameters of the slab rolling process are all imported through the preprocessing module of the finite element analysis software ABAQUS. The hot deformation stress model and the average grain size model of the slab material are written by ABAQUS Python.

[0013] The thermal deformation stress model is updated as follows:

[0014]

[0015] Wherein, α(ε,T), And n(ε,T) are the material parameters of the billet. R is the activation energy for thermal deformation, R is the ideal gas constant, and T is the absolute temperature.

[0016] The microstructure average grain size model is updated as follows:

[0017]

[0018] Where d0 is the initial grain size, d t The average grain size after t seconds is given, where t is the holding time, K and n are material-related constants, and Q is the average grain size after t seconds. b This is the activation energy for grain boundary self-diffusion.

[0019] Furthermore, the quality anomaly diagnosis module specifically comprises:

[0020] The slab thickness b, temperature T, rolling force F, and rolling speed v during the slab rolling process are collected as parameters and differentially calculated with the results of the virtual calculation module to calculate the slab thickness change Δb, temperature change ΔT, rolling force change ΔF, and rolling speed Δv.

[0021] The parameters p = [Δb, ΔT, ΔF, Δv] after differencing are standardized using the following formula:

[0022]

[0023] Where p' is the parameter after standardization;

[0024] The standardized parameters p' = [Δb', ΔT', ΔF', Δv'] are input into the quality anomaly diagnosis model, which outputs rolling temperature anomalies, pressure anomalies, speed anomalies, and equipment vibration. This allows for reverse optimization of the rolling process parameters, which are then fed back into the slab rolling process. The quality anomaly diagnosis model employs a BP neural network, comprising an input layer, a hidden layer, and an output layer. The transfer equation of the hidden layer is f1(·), and the output of the hidden layer is:

[0025]

[0026] If the transfer equation of the output layer is f2(·), then the output of the output layer is:

[0027]

[0028] Where i takes the values ​​1, 2, 3, or 4, i.e., x1 = Δb', x2 = ΔT', x3 = ΔF', x4 = Δv', n is the number of input layer nodes, q is the number of hidden layer nodes, m is the number of output layer nodes, and v is the number of hidden layer nodes. ki The weights w represent the weights from input layer node i to hidden layer node k. jk It represents the weights from hidden layer node k to output layer node j.

[0029] This invention also provides a multi-scale quality control method for intelligent hot rolling of large-size slabs, comprising the following steps:

[0030] Step 1: Real-time acquisition of process parameters during slab rolling;

[0031] Step 2: Based on the real-time collected process parameters of the billet rolling process, perform numerical simulation calculations on the process parameters;

[0032] Step 3: Compare the real-time collected process parameters of the billet rolling process with the simulation calculation results of Step 2;

[0033] Step 4: Based on the real-time acquisition of process parameters during slab rolling, the results of simulation calculation in Step 2, and the comparison results in Step 3, quality status difference analysis is performed to optimize process parameters. The optimized process parameters are then fed back to the slab rolling process for control.

[0034] Furthermore, the process parameters for the slab rolling process in step 1 include material parameters, slab dimensions, hot rolling process parameters, and hot rolling equipment parameters.

[0035] Furthermore, step 2 specifically includes:

[0036] Using the finite element analysis software ABAQUS, the process parameters of the slab rolling process are all imported through the preprocessing module of the finite element analysis software ABAQUS, and the hot deformation stress model and the average grain size model of the microstructure are written by ABAQUSPython.

[0037] The thermal deformation stress model is updated as follows:

[0038]

[0039] Wherein, α(ε,T), And n(ε,T) are the material parameters of the billet. R is the activation energy for thermal deformation, R is the ideal gas constant, and T is the absolute temperature.

[0040] The microstructure average grain size model is updated as follows:

[0041]

[0042] Where d0 is the initial grain size, d t The average grain size after t seconds is given, where t is the holding time, K and n are material-related constants, and Q is the average grain size after t seconds. b This is the activation energy for grain boundary self-diffusion.

[0043] Furthermore, step 4 specifically includes:

[0044] The slab thickness b, temperature T, rolling force F, and rolling speed v during the slab rolling process are collected as parameters and compared with the simulation calculation results in step 1 to calculate the slab thickness change Δb, temperature change ΔT, rolling force change ΔF, and rolling speed Δv.

[0045] The parameters p = [Δb, ΔT, ΔF, Δv] after differencing are standardized using the following formula:

[0046]

[0047] Where p' is the parameter after standardization;

[0048] The standardized parameters p' = [Δb', ΔT', ΔF', Δv'] are input into the quality anomaly diagnosis model, which outputs rolling temperature anomalies, pressure anomalies, speed anomalies, and equipment vibration. This allows for reverse optimization of the rolling process parameters, which are then fed back into the slab rolling process. The quality anomaly diagnosis model employs a BP neural network, comprising an input layer, a hidden layer, and an output layer. The transfer equation of the hidden layer is f1(·), and the output of the hidden layer is:

[0049]

[0050] If the transfer equation of the output layer is f2(·), then the output of the output layer is:

[0051]

[0052] Where i takes the values ​​1, 2, 3, or 4, i.e., x1 = Δb', x2 = ΔT', x3 = ΔF', x4 = Δv', n is the number of input layer nodes, q is the number of hidden layer nodes, m is the number of output layer nodes, and v is the number of hidden layer nodes. ki The weights w represent the weights from input layer node i to hidden layer node k. jk It represents the weights from hidden layer node k to output layer node j.

[0053] Compared with the prior art, the present invention has the following advantages:

[0054] 1) By establishing a microstructure model based on crystallographic and physical principles, obtaining the required material parameters through experiments, and using computer simulation calculations, the microstructure prediction of the rolling process can be achieved. By establishing a macro-microscopic evolution model of billet rolling and embedding the model into an intelligent control system, process optimization based on microstructure prediction can be realized. Reliable models and systems can further broaden the applicability of prediction models, improve the accuracy and reliability of predictions, and provide strong data support for intelligent production in rolling processes, offering ideas and methods for the intelligent development of plate and strip materials.

[0055] 2) The quality control system and method of the present invention can realize the visualization of the microscale of the slab during the rolling process. By using digital twins, the stress field, strain field and average grain size of the rolling process can be visualized and mapped in real time, so as to intuitively understand the rolling process and predict slab quality problems in a timely manner.

[0056] 3) The system of the present invention has a wide range of industrial samples in the database, reliable sources of production samples and simulation samples, simple calculations and fast convergence speed in the quality anomaly diagnosis module, and achieves the goal of effective quality control by a few key rules. Attached Figure Description

[0057] Figure 1 This is a schematic diagram of the multi-scale quality control system for intelligent hot rolling of large-size slabs according to the present invention.

[0058] Figure 2 This is a schematic diagram of a BP neural network. Detailed Implementation

[0059] To further illustrate the technical solution of the present invention, the present invention will be further described below through embodiments.

[0060] like Figure 1As shown in this embodiment, a multi-scale quality control system for intelligent hot rolling of large-size slabs mainly consists of three parts: a virtual computing module, a quality control module, and a quality anomaly diagnosis module. Production process data is provided to these three parts: macroscopic production process data, microscopic crystallographic and physical principles of thermal deformation stress models, and microstructure models. Combined with a data-driven quality anomaly diagnosis model, this system enables prediction, quality feedback, and reverse optimization of the rolling process for large-size slabs. It provides strong data support for intelligent rolling processes and offers ideas and methods for the intelligent development of plate and strip materials.

[0061] Among them, 1) Virtual calculation module: Based on numerical simulation software, it performs numerical simulation calculations on the slab rolling process and outputs the simulation calculation results;

[0062] The process parameters for the slab rolling process in this embodiment include material parameters, slab dimensions, hot rolling process parameters, and hot rolling equipment parameters. The material parameters include thermophysical parameters, hot deformation constitutive model, and hot deformation grain model.

[0063] The virtual computing module uses the finite element analysis software ABAQUS. The process parameters of the slab rolling process are all imported through the preprocessing module of the finite element analysis software ABAQUS. The hot deformation stress model and the average grain size model of the microstructure of the slab material are written by ABAQUSPython.

[0064] The thermal deformation stress model is updated as follows:

[0065]

[0066] Wherein, α(ε,T), n(ε,T) are the material parameters of the casting billet, obtained from experimental calculations. R is the activation energy for thermal deformation, R is the ideal gas constant, and T is the absolute temperature.

[0067] The microstructure average grain size model is updated as follows:

[0068]

[0069] Where d0 is the initial grain size, d t The average grain size after t seconds is given, where t is the holding time, K and n are material-related constants, and Q is the average grain size after t seconds. b This is the activation energy for grain boundary self-diffusion.

[0070] 2) Quality control module: Records the technical parameters of the slab rolling production process and compares them with the simulation calculation results of the virtual calculation module;

[0071] The quality control module in this embodiment is implemented based on the existing online integrated quality control system. The quality control system includes quality judgment rules, quality evaluation methods, rolling process weight table, and identification of slab and equipment characteristic parameters.

[0072] 3) Quality Anomaly Diagnosis Module: Based on the process parameters of the slab rolling process, the simulation calculation results of the virtual calculation module, and the comparison results of the quality control module, the module performs quality status difference analysis, optimizes the process parameters, and feeds back the optimized process parameters to the slab rolling process for regulation.

[0073] In this embodiment, the quality anomaly diagnosis module collects the slab thickness b, temperature T, rolling force F, and rolling speed v during the slab rolling process as parameters, performs differential calculations with the results of the virtual calculation module, and calculates the slab thickness change Δb, temperature change ΔT, rolling force change ΔF, and rolling speed Δv.

[0074] The parameters p = [Δb, ΔT, ΔF, Δv] after differencing are standardized using the following formula:

[0075]

[0076] Where p' is the parameter after standardization;

[0077] The standardized parameters p' = [Δb', ΔT', ΔF', Δv'] are input into the quality anomaly diagnosis model, which outputs rolling temperature anomalies, pressure anomalies, speed anomalies, and equipment vibration. This allows for reverse optimization of the rolling process parameters, which are then fed back into the slab rolling process. The quality anomaly diagnosis model employs a BP neural network (e.g., BP neural network). Figure 2 As shown), it contains an input layer, a hidden layer, and an output layer. The transfer equation of the hidden layer is f1(·), and the output of the hidden layer is:

[0078]

[0079] If the transfer equation of the output layer is f2(·), then the output of the output layer is:

[0080]

[0081] Where i takes the values ​​1, 2, 3, or 4, i.e., x1 = Δb', x2 = ΔT', x3 = ΔF', x4 = Δv', n is the number of input layer nodes, q is the number of hidden layer nodes, m is the number of output layer nodes, and v is the number of hidden layer nodes. ki The weights w represent the weights from input layer node i to hidden layer node k. jk It represents the weights from hidden layer node k to output layer node j.

[0082] This embodiment of a multi-scale quality control method for intelligent hot rolling of large-size slabs includes the following steps:

[0083] Step 1: Real-time acquisition of process parameters during slab rolling. Process parameters during slab rolling include material parameters, slab dimensions, hot rolling process parameters, and hot rolling equipment parameters. Material parameters include thermophysical parameters, hot deformation constitutive model, and hot deformation grain model.

[0084] Step 2: Based on the real-time collected process parameters of the billet rolling process, perform numerical simulation calculations on the process parameters;

[0085] Numerical simulation calculations were performed using the finite element analysis software ABAQUS. The process parameters of the slab rolling process were all imported through the preprocessing module of the finite element analysis software ABAQUS, and the hot deformation stress model and the average grain size model of the microstructure were written by ABAQUS Python.

[0086] The thermal deformation stress model is updated as follows:

[0087]

[0088] Wherein, α(ε,T), And n(ε,T) are the material parameters of the billet. R is the activation energy for thermal deformation, R is the ideal gas constant, and T is the absolute temperature.

[0089] The microstructure average grain size model is updated as follows:

[0090]

[0091] Where d0 is the initial grain size, d t The average grain size after t seconds is given, where t is the holding time, K and n are material-related constants, and Q is the average grain size after t seconds. b This is the activation energy for grain boundary self-diffusion.

[0092] Step 3: Compare the real-time collected process parameters of the billet rolling process with the simulation calculation results of Step 2;

[0093] Step 4: Based on the real-time acquisition of process parameters during slab rolling, the results of simulation calculation in Step 2, and the comparison results in Step 3, perform quality status difference analysis to optimize process parameters. The optimized process parameters are then fed back to the slab rolling process for control.

[0094] First, the slab thickness b, temperature T, rolling force F, and rolling speed v during the slab rolling process are collected as parameters. The difference calculation is performed with the results of the simulation calculation in step 1 to calculate the slab thickness change Δb, temperature change ΔT, rolling force change ΔF, and rolling speed Δv.

[0095] Then, the differencing parameters p = [Δb, ΔT, ΔF, Δv] are standardized using the following formula:

[0096]

[0097] Where p' is the parameter after standardization;

[0098] Finally, the standardized parameters p' = [Δb', ΔT', ΔF', Δv'] are input into the quality anomaly diagnosis model, which outputs rolling temperature anomalies, pressure anomalies, speed anomalies, and equipment vibration. This allows for reverse optimization of the rolling process parameters, which are then fed back into the slab rolling process. The quality anomaly diagnosis model uses a BP neural network, comprising an input layer, a hidden layer, and an output layer. The transfer equation of the hidden layer is f1(·), and the output of the hidden layer is:

[0099]

[0100] If the transfer equation of the output layer is f2(·), then the output of the output layer is:

[0101]

[0102] Where i takes the values ​​1, 2, 3, or 4, i.e., x1 = Δb', x2 = ΔT', x3 = ΔF', x4 = Δv', n is the number of input layer nodes (n = 4), q is the number of hidden layer nodes (q = 6), m is the number of output layer nodes (m = 4), and v ki The weights w represent the weights from input layer node i to hidden layer node k. jk It represents the weights from hidden layer node k to output layer node j.

[0103] The foregoing has shown and described the main features and advantages of the present invention. It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the invention can be implemented in other specific forms without departing from its spirit or essential characteristics. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, it is intended that all variations falling within the meaning and scope of equivalents of the claims be included within the present invention.

[0104] Furthermore, it should be understood that although this specification describes embodiments, not every embodiment contains only one independent technical solution. This narrative style is merely for clarity. Those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A multiscale quality control system for large size slab intelligent hot rolling, characterized in that, include: The virtual computing module performs numerical simulation calculations on the slab rolling process based on numerical simulation software and outputs the simulation calculation results. The quality control module records the technical parameters of the slab rolling production process and compares them with the simulation results of the virtual calculation module. The quality anomaly diagnosis module analyzes the differences in quality status based on the process parameters of the slab rolling process, the simulation calculation results of the virtual calculation module, and the comparison results of the quality control module. It optimizes the process parameters and feeds the optimized process parameters back to the slab rolling process for regulation. The virtual computing module uses the finite element analysis software ABAQUS. The process parameters of the slab rolling process are all imported through the preprocessing module of the finite element analysis software ABAQUS. The hot deformation stress model and the average grain size model of the microstructure of the slab material are written by ABAQUS Python. The thermal deformation stress model is updated as follows: wherein , and is a casting material parameter, is a thermal deformation activation energy, R is the ideal gas constant and T is the absolute temperature; The microstructure average grain size model is updated as follows: where d0 is the initial grain size, d t is the average grain size after t seconds, t is the holding time, K, n are material dependent constants, Q b is the grain boundary self-diffusion activation energy; The quality anomaly diagnosis module is specifically as follows: The slab thickness b, temperature T, rolling force F, and rolling speed v during the slab rolling process are collected as parameters and differentially calculated with the results of the virtual calculation module to calculate the slab thickness change Δb, temperature change ΔT, rolling force change ΔF, and rolling speed Δv. The parameters p = [Δb, ΔT, ΔF, Δv] after differencing are standardized using the following formula: wherein p ' is a parameter after standardization processing; The parameter p ' =[Δb ' , ΔT ' , ΔF ' , Δv ' ] after standardization is input into a quality abnormality diagnosis model, and rolling temperature abnormality, pressure abnormality, speed abnormality and equipment vibration are output, and then rolling process parameter reverse optimization is carried out, and the optimized process parameters are fed back to the slab rolling process; the quality abnormality diagnosis model adopts a BP neural network, and contains an input layer, a hidden layer and an output layer, wherein the transfer equation of the hidden layer is , and the output of the hidden layer is The transfer equation for the output layer is The output of the output layer is then where i takes 1, 2, 3, 4, i.e. x1=Δb ' , x2=ΔT ' , x3=ΔF ' , x4=Δv ' , n is the number of input layer nodes, q is the number of hidden layer nodes, m is the number of output layer nodes, v ki is the weight of input layer node i to hidden layer node k, w jk is the weight of hidden layer node k to output layer node j.

2. A multiscale quality control system for smart hot rolling of large size slabs as claimed in claim 1 wherein, The process parameters for the slab rolling process include material parameters, slab dimensions, hot rolling process parameters, and hot rolling equipment parameters.

3. A multi-scale quality control method for intelligent hot rolling of large-size slabs, characterized in that, Includes the following steps: Step 1: Real-time acquisition of process parameters during slab rolling; Step 2: Based on the real-time collected process parameters of the billet rolling process, perform numerical simulation calculations on the process parameters; Step 3: Compare the real-time collected process parameters of the billet rolling process with the simulation calculation results of Step 2; Step 4: Based on the real-time acquisition of process parameters during slab rolling, the results of simulation calculation in Step 2, and the comparison results in Step 3, perform quality status difference analysis to optimize process parameters. The optimized process parameters are then fed back to the slab rolling process for control. Step 2 specifically involves: Using the finite element analysis software ABAQUS, the process parameters of the slab rolling process are all imported through the preprocessing module of the finite element analysis software ABAQUS, and the hot deformation stress model and the average grain size model of the microstructure are written by ABAQUS Python. The thermal deformation stress model is updated as follows: in, , and For the parameters of the billet material, R is the activation energy for thermal deformation, R is the ideal gas constant, and T is the absolute temperature. The microstructure average grain size model is updated as follows: where d0 is the initial grain size, d t is the average grain size after t seconds, t is the holding time, K, n are material dependent constants, Q b is the grain boundary self-diffusion activation energy; Step 4 specifically involves: The slab thickness b, temperature T, rolling force F, and rolling speed v during the slab rolling process are collected as parameters and compared with the simulation calculation results in step 1 to calculate the slab thickness change Δb, temperature change ΔT, rolling force change ΔF, and rolling speed Δv. The parameters p = [Δb, ΔT, ΔF, Δv] after differencing are standardized using the following formula: wherein p ' is a parameter after standardization processing; The standardized parameter p ' =[Δb ' , ΔT ' , ΔF ' , Δv ' The input quality anomaly diagnosis model outputs rolling temperature anomalies, pressure anomalies, speed anomalies, and equipment vibration anomalies, and then performs reverse optimization of rolling process parameters. The optimized process parameters are then fed back into the slab rolling process. The quality anomaly diagnosis model uses a BP neural network, which includes an input layer, a hidden layer, and an output layer. The transfer equation of the hidden layer is as follows: The output of the hidden layer is then: The transfer equation of the output layer is Then the output of the output layer is: where i takes 1, 2, 3, 4, i.e. x1=Δb ' , x2=ΔT ' , x3=ΔF ' , x4=Δv ' , n is the number of input layer nodes, q is the number of hidden layer nodes, m is the number of output layer nodes, v ki is the weight of input layer node i to hidden layer node k, w jk is the weight of hidden layer node k to output layer node j.

4. The multi-scale quality control method for intelligent hot rolling of large-size slabs according to claim 3, characterized in that, The process parameters for the slab rolling process in step 1 include material parameters, slab dimensions, hot rolling process parameters, and hot rolling equipment parameters.

Images