A method for intelligent optimization of rolling process parameters in silicon steel production

Abstract

This invention belongs to the field of silicon steel production technology. It discloses an intelligent optimization method for silicon steel rolling process parameters, including the following steps: S1: Real-time collection of process parameters, quality data, and environmental status data during the silicon steel rolling process. This invention achieves intelligent optimization recommendations for process parameters by setting a process parameter optimization prediction model. The system inputs multi-dimensional information such as the current process status score, change rate, incoming material characteristics, and environmental conditions, and outputs the optimal adjustment of each process parameter. This optimization method is more scientific and precise, avoiding interference from human factors. In terms of energy saving and consumption reduction, energy consumption is reduced by finely adjusting process parameters while ensuring quality. By setting an energy consumption prediction model, the rolling energy consumption per ton of steel is accurately predicted. By comparing the predicted energy consumption before and after optimization, the system can scientifically judge the effectiveness of energy-saving measures and avoid quality losses caused by blind adjustments.

Description

Technical Field

[0001] This invention relates to the field of silicon steel production technology, and more specifically, to a method for intelligent optimization of rolling process parameters in silicon steel production. Background Technology

[0002] Silicon steel, as an important electrical steel material, is widely used in electrical equipment such as power transformers and motors. Its magnetic properties directly determine the energy efficiency level of electrical equipment. In the production process of silicon steel, the precise control of hot rolling and cold rolling process parameters has a decisive impact on the magnetic properties, thickness uniformity, and sheet shape quality of the final product.

[0003] Currently, automatic control systems based on fixed rules are mainly used. These systems typically have fixed ranges for process parameters, and automatically adjust when a parameter exceeds the range. However, this method has significant limitations. On the one hand, multiple factors such as temperature, speed, reduction rate, and cooling are coupled during silicon steel rolling, making it difficult for simple single-variable control to handle complex interactive effects. On the other hand, fixed thresholds cannot adapt to changes in different steel grades, specifications, and equipment conditions, lacking adaptive capabilities. Currently, there is a lack of comprehensive quantitative evaluation methods for the state of the silicon steel rolling process, making it difficult to accurately judge the quality of the process. There is also a lack of early warning mechanisms for quality trends, allowing for passive adjustments only after quality problems occur. Furthermore, there is a lack of methods for synergistic optimization of process parameters and energy consumption, making it difficult to achieve energy conservation and consumption reduction while ensuring quality. Summary of the Invention

[0004] To address the problems in the background art, this invention proposes an intelligent optimization method for the rolling process parameters in silicon steel production.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a method for intelligent optimization of rolling process parameters in silicon steel production, comprising the following steps: S1: Real-time acquisition of process parameters, quality data, and environmental status data during the silicon steel rolling process.

[0006] S2: Based on the collected data, generate process status scores in four dimensions during the silicon steel rolling process, including thickness uniformity score, plate shape quality score, magnetic property trend score, and process stability score, and calculate the rate of change of each score.

[0007] S3: Compare the rate of change of each score with the threshold of each rate of change to identify abnormal patterns, perform cause analysis and process parameter adjustment strategies based on the preset abnormal pattern library, and decide whether to trigger the process parameter optimization process.

[0008] S4: Generate a comprehensive process score from the scores, set a score threshold, and decide whether to trigger the process parameter optimization process based on the comprehensive process score and the score threshold.

[0009] S5: Obtain historical process parameter optimization data and construct a process parameter optimization prediction model. Use the process parameter optimization prediction model to predict the optimized process parameters, and adjust the process parameters in the silicon steel rolling process according to the predicted optimized process parameters.

[0010] S6: Obtain historical energy consumption data and build an energy consumption prediction model. Use the energy consumption prediction model to predict the energy consumption of production equipment. Calculate the energy consumption ratio before and after the predicted process parameters are optimized. Set a ratio threshold. Decide whether to start the energy-saving optimization strategy based on the energy consumption ratio and the ratio threshold.

[0011] Furthermore, the process parameters include rolling temperature, rolling speed, reduction rate, rolling force and cooling water flow rate; the quality data include strip thickness, strip shape deviation and magnetic induction intensity; and the environmental condition data include ambient temperature, ambient humidity, incoming material thickness, incoming material temperature and target thickness. The rolling temperature is measured by an infrared thermometer, the rolling speed is the preset speed of the equipment, the reduction rate is measured by a displacement sensor, the rolling force is measured by a piezoelectric sensor, the cooling water flow rate is measured by an electromagnetic flowmeter, the strip thickness is measured by a laser thickness gauge, the strip shape deviation is measured by a strip roll, the magnetic induction intensity is measured by an online permeability meter, the ambient temperature and humidity are measured by a temperature and humidity sensor, the incoming material thickness and temperature are measured by a thickness gauge, and the target thickness is the preset target of the system.

[0012] Furthermore, the process of generating a four-dimensional process status score for silicon steel rolling based on the collected data includes: Thickness uniformity score :

[0013] In the formula, The thickness measurement value at the i-th sampling point. Let n be the target thickness, and n be the number of sampling points within the calculation window. For thickness tolerance, To calculate the standard deviation of thickness measurements within the window, This represents the maximum permissible standard deviation. Plate shape quality score :

[0014] In the formula, Let the plate shape deviation be the j-th plate shape measurement unit. The maximum allowable shape deviation for the corresponding unit is set according to the strip width, where m is the number of shape measurement units. The penalty coefficient for the changing trend. / dt is the rate of change of plate shape deviation with time, which is calculated by the central difference method; Magnetic property trend score :

[0015]

[0016]

[0017]

[0018] In the formula, The current measured magnetic flux density is... The target magnetic flux density, The current rolling temperature, This is the optimal rolling temperature for this steel grade. The current reduction rate, The optimal reduction rate is given by V, which represents the current rolling speed. For optimal rolling speed, , and The attenuation coefficient; Process stability rating :

[0019]

[0020]

[0021]

[0022]

[0023] In the formula, , , These represent the standard deviations of rolling force, speed, temperature, and reduction rate, respectively. , , , These are the average values ​​of the corresponding parameters. , , and These are the weighting coefficients.

[0024] Furthermore, the process of calculating the rate of change for each score includes: An evaluation cycle is set. At each evaluation cycle, a thickness uniformity score is obtained. The thickness uniformity scores of the first 2n evaluation cycles are taken. The 2n thickness uniformity scores are arranged in chronological order. The average of the first n thickness uniformity scores is calculated to obtain the first thickness uniformity score average. The average of the last n thickness uniformity scores is calculated to obtain the second thickness uniformity score average. The average of the second thickness uniformity score is subtracted from the average of the first thickness uniformity score to obtain the score difference. The score difference is divided by the average of the first thickness uniformity score to obtain the thickness uniformity score change rate. The plate shape quality score, magnetic property trend score, and process stability score all obtained their corresponding change rates using the methods described above.

[0025] Furthermore, the process of comparing the rate of change of each score with the threshold of each rate of change to identify abnormal patterns, performing cause analysis and process parameter adjustment strategies based on a pre-set abnormal pattern library, and deciding whether to trigger the process parameter optimization process includes: Threshold U for the rate of change of thickness uniformity score:

[0026] In the formula, R is the rate of change of the thickness uniformity score. This represents the average rate of change of thickness uniformity score over the most recent 100 calculation periods. The standard deviation of the rate of change of the thickness uniformity score. Sensitivity coefficient; The corresponding change rate thresholds for the plate shape quality score change rate, magnetic property trend score change rate, and process stability score change rate are obtained using the above methods. When the absolute value of the rate of change in the rating is greater than its corresponding rate of change threshold, an abnormal trend is observed, and the abnormal pattern needs to be identified in a timely manner. If the rate of change of thickness uniformity score is less than the negative of the threshold for the rate of change of thickness uniformity score, the abnormal pattern is determined to be thickness offset. If the rate of change of the plate shape quality score is less than the negative of the threshold for the rate of change of the plate shape quality score, the abnormal pattern is determined to be plate shape deterioration. When the rate of change of magnetic property trend score is less than the negative of the threshold of magnetic property trend score change rate, the abnormal pattern is determined to be a decrease in magnetic property. If the rate of change of the process stability score is less than the negative of the threshold for the rate of change of the process stability score, the abnormal mode is determined to be process instability. Based on the recommended adjustment strategies from the abnormal mode library, each abnormal mode has a corresponding adjustment strategy. Predefined process parameters are selected and executed according to the adjustment strategy. The rate of change of each score is monitored for x cycles after adjustment. If the initial adjustment effect is not significant, the process parameter optimization process is initiated.

[0027] Furthermore, the process of generating a comprehensive process score from the various scores, setting a scoring threshold, and deciding whether to trigger the process parameter optimization process based on the comprehensive process score and the scoring threshold includes: Overall process score P:

[0028] In the formula, , , and These are the weighting coefficients, obtained through training based on historical data; Set an appropriate comprehensive process score threshold based on historical comprehensive process score data, where historical comprehensive process score data refers to the data set of previous comprehensive process scores. Compare the comprehensive process score with the comprehensive process score threshold. If the comprehensive process score is less than the comprehensive process score threshold, the process parameter optimization process is triggered.

[0029] Furthermore, the process of acquiring historical process parameter optimization data and constructing a process parameter optimization prediction model, using the process parameter optimization prediction model to predict the optimized process parameters, and adjusting the process parameters in the silicon steel rolling process based on the predicted optimized process parameters includes: The process parameter optimization process is achieved through a process parameter optimization prediction model. Factors affecting the optimized process parameters include: the rate of change of thickness uniformity score, the rate of change of plate shape quality score, the rate of change of magnetic property trend score, the rate of change of process stability score, the comprehensive process score, the incoming material thickness, the incoming material temperature, the target thickness, the ambient temperature, and the ambient humidity. Acquire historical optimized process parameter data for production equipment and products. The historical optimized process parameter data includes the change rate of thickness uniformity score, change rate of plate shape quality score, change rate of magnetic property trend score, change rate of process stability score, comprehensive process score, incoming material thickness, incoming material temperature, target thickness, ambient temperature, ambient humidity, and historical optimized process parameters of production equipment. Based on the change rate of thickness uniformity score, change rate of plate shape quality score, change rate of magnetic property trend score, change rate of process stability score, comprehensive process score, incoming material thickness, incoming material temperature, target thickness, ambient temperature, ambient humidity and corresponding historical optimized process parameters in different historical optimized process parameter data, an optimized process parameter prediction set is generated and divided into the first training set and the first test set. A first convolutional neural network is constructed, and the following parameters from different historical optimized process parameters in the first training set are used as input data: thickness uniformity score change rate, plate shape quality score change rate, magnetic property trend score change rate, process stability score change rate, comprehensive process score, incoming material thickness, incoming material temperature, target thickness, ambient temperature and ambient humidity. The corresponding historical optimized process parameters in the first training set are used as output data of the first convolutional neural network. The first convolutional neural network is trained to obtain the first initial convolutional neural network. The first initial convolutional neural network is validated using the first test set. The first initial convolutional neural network, whose output is less than or equal to the preset first test error threshold, is used as the optimized process parameter prediction model. The real-time thickness uniformity score change rate, plate shape quality score change rate, magnetic property trend score change rate, process stability score change rate, comprehensive process score, incoming material thickness, incoming material temperature, target thickness, ambient temperature and ambient humidity of the production equipment and products are input into the optimized process parameter prediction model to obtain the predicted optimized process parameters. The predicted and optimized process parameters include the optimized rolling temperature, rolling speed, reduction rate, rolling force, and cooling water flow rate. The predicted and optimized process parameters are executed, and the change rate of each score and the corresponding comprehensive process score are monitored for y cycles after adjustment.

[0030] Furthermore, the process of acquiring historical energy consumption data and constructing an energy consumption prediction model, and then using this model to predict the energy consumption of production equipment, includes: The energy consumption refers to the energy consumed by the production equipment; Factors affecting the energy consumption of production equipment include: rolling temperature, rolling speed, reduction rate, rolling force, cooling water flow rate, strip width, incoming material thickness, incoming material temperature, and ambient temperature. Acquire historical energy consumption data of the production equipment in the production task. The historical energy consumption data includes the rolling temperature, rolling speed, reduction rate, rolling force, cooling water flow rate, strip width, incoming material thickness, incoming material temperature, ambient temperature and historical energy consumption of the production equipment in the production task. Based on the rolling temperature, rolling speed, reduction rate, rolling force, cooling water flow rate, strip width, incoming material thickness, incoming material temperature, ambient temperature and corresponding historical energy consumption of production equipment in different historical energy consumption data, an energy consumption prediction set is generated and divided into a second training set and a second test set. A second convolutional neural network is constructed. The rolling temperature, rolling speed, reduction rate, rolling force, cooling water flow rate, strip width, incoming material thickness, incoming material temperature and ambient temperature in different historical energy consumption data in the second training set are used as the input data of the second convolutional neural network, and the corresponding historical energy consumption in the second training set is used as the output data of the second convolutional neural network. The second convolutional neural network is trained to obtain the second initial convolutional neural network. The second initial convolutional neural network is validated using the second test set. The second initial convolutional neural network that outputs a second test error threshold less than or equal to the preset second test error threshold is used as the energy consumption prediction model. The real-time rolling temperature, rolling speed, reduction rate, rolling force, cooling water flow rate, strip width, incoming material thickness, incoming material temperature, and ambient temperature of the production equipment are input into the energy consumption prediction model to obtain the predicted energy consumption of the production equipment.

[0031] Furthermore, the process of calculating the energy consumption ratio based on the predicted energy consumption before and after process parameter optimization, setting a ratio threshold, and deciding whether to initiate an energy-saving optimization strategy based on the energy consumption ratio and the ratio threshold includes: Both the optimized process parameters and the unoptimized process parameters are input into the energy consumption prediction model to obtain the predicted energy consumption after optimization and the predicted energy consumption before optimization. The energy consumption ratio is obtained by dividing the predicted energy consumption after optimization by the predicted energy consumption before optimization. Set an appropriate ratio threshold based on historical energy consumption ratio data. The historical energy consumption ratio data refers to the data set of previous energy consumption ratios. Compare the energy consumption ratio with the ratio threshold. If the energy consumption ratio is less than the ratio threshold, the energy-saving optimization strategy will not be triggered and the original optimized process parameters will be maintained. If the energy consumption ratio is greater than the ratio threshold, the energy-saving optimization strategy will be triggered. Energy-saving optimization strategies: Fine-tune each process parameter until the energy consumption ratio is lower than the ratio threshold, set the lower limit of each score, and after fine-tuning, ensure that each score is higher than the lower limit of each score, and that the overall process score is higher than the overall process score threshold. Energy consumption can also be reduced by the following operations: mass-producing similar specifications of products, optimizing the production rhythm, and avoiding peak electricity consumption periods.

[0032] The technical effects and advantages of the intelligent optimization method for silicon steel production rolling process parameters of this invention are as follows: (1) By setting up a process parameter optimization prediction model, intelligent optimization recommendation of process parameters is realized. The system inputs multi-dimensional information such as the current process status score, change rate, incoming material characteristics, and environmental conditions, and outputs the optimal adjustment of each process parameter. Compared with traditional experience adjustment, this optimization method is more scientific and accurate, avoiding interference from human factors. In terms of energy saving and consumption reduction, the system reduces energy consumption by finely adjusting process parameters while ensuring quality. By setting up an energy consumption prediction model, the system considers factors such as rolling temperature, speed, reduction rate, rolling force, and cooling water flow rate, and accurately predicts the rolling energy consumption per ton of steel. By comparing the predicted energy consumption before and after optimization, the system can scientifically judge the effect of energy-saving measures and avoid quality loss caused by blind adjustment.

[0033] (2) By setting multiple scores, a comprehensive quantitative assessment of the state of silicon steel rolling process was achieved. A score change rate analysis mechanism was introduced. By calculating the change trend of each score, an early warning of quality anomalies was achieved. Traditional quality control methods usually only detect anomalies after the quality indicators exceed the specification limits. At this time, a large number of unqualified products may have already been produced. By monitoring the score change rate, the abnormal trend can be identified before the quality declines significantly, realizing the transformation from "passive response" to "active prevention". The anomaly pattern library diagnostic system further improves the accuracy of problem handling. The system intelligently identifies the anomaly type based on the combination pattern of the four score change rates and recommends corresponding adjustment strategies for different anomaly types. This diagnostic method based on multi-dimensional information fusion is more accurate and reliable than the traditional single indicator judgment. Attached Figure Description

[0034] Figure 1 This is a schematic diagram of the method flow of the present invention. Detailed Implementation

[0035] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0036] Reference Figure 1 A method for intelligent optimization of rolling process parameters in silicon steel production includes the following steps: S1: Real-time acquisition of process parameters, quality data, and environmental status data during the silicon steel rolling process.

[0037] S2: Based on the collected data, generate process status scores in four dimensions during the silicon steel rolling process, including thickness uniformity score, plate shape quality score, magnetic property trend score, and process stability score, and calculate the rate of change of each score.

[0038] S3: Compare the rate of change of each score with the threshold of each rate of change to identify abnormal patterns, perform cause analysis and process parameter adjustment strategies based on the preset abnormal pattern library, and decide whether to trigger the process parameter optimization process.

[0039] S4: Generate a comprehensive process score from the scores, set a score threshold, and decide whether to trigger the process parameter optimization process based on the comprehensive process score and the score threshold.

[0040] S5: Obtain historical process parameter optimization data and construct a process parameter optimization prediction model. Use the process parameter optimization prediction model to predict the optimized process parameters, and adjust the process parameters in the silicon steel rolling process according to the predicted optimized process parameters.

[0041] S6: Obtain historical energy consumption data and build an energy consumption prediction model. Use the energy consumption prediction model to predict the energy consumption of production equipment. Calculate the energy consumption ratio before and after the predicted process parameters are optimized. Set a ratio threshold. Decide whether to start the energy-saving optimization strategy based on the energy consumption ratio and the ratio threshold.

[0042] It should be further explained that, in the specific implementation process, the process parameters include rolling temperature, rolling speed, reduction rate, rolling force and cooling water flow rate, the quality data include strip thickness, strip shape deviation and magnetic induction intensity, and the environmental condition data include ambient temperature, ambient humidity, incoming material thickness, incoming material temperature and target thickness. The rolling temperature is measured by an infrared thermometer, the rolling speed is the preset speed of the equipment, the reduction rate is measured by a displacement sensor, the rolling force is measured by a piezoelectric sensor, the cooling water flow rate is measured by an electromagnetic flowmeter, the strip thickness is measured by a laser thickness gauge, the strip shape deviation is measured by a strip roll, the magnetic induction intensity is measured by an online permeability meter, the ambient temperature and humidity are measured by a temperature and humidity sensor, the incoming material thickness and temperature are measured by a thickness gauge, and the target thickness is the preset target of the system.

[0043] It should be further explained that, in the specific implementation process, the process of generating a four-dimensional process status score based on the collected data during the silicon steel rolling process includes: Thickness uniformity score :

[0044] In the formula, The thickness measurement value at the i-th sampling point. Let n be the target thickness, and n be the number of sampling points within the calculation window. Let n be 100. The allowable thickness deviation is set to 0.01 mm. To calculate the standard deviation of thickness measurements within the window, The maximum permissible standard deviation is taken as 0.015 mm; The measured thickness of a certain strip of steel is [0.301, 0.299, 0.300, 0.302, 0.298] mm, the target thickness is 0.300 mm, and the calculated average absolute deviation is 0.0012 mm. =0.00158mm, then It is 0.793; Plate shape quality score :

[0045] In the formula, Let the plate shape deviation be the j-th plate shape measurement unit. The maximum allowable shape deviation for the corresponding unit is set according to the strip width, where m is the number of shape measurement units, typically 25-41. The penalty coefficient for the changing trend is taken as follows: It is 0.1. / dt is the rate of change of plate shape deviation with time, which is calculated by the central difference method; The shape deviations of each unit in a certain strip are [1.5, 2.0, 1.8, 2.2, 1.6]. What is the maximum allowable shape deviation for each unit? If the value is 5 IU and the plate shape deviation change rate is 0.3 IU / s, then... It is 0.606; Magnetic property trend score :

[0046]

[0047]

[0048]

[0049] In the formula, The current measured magnetic flux density is... The target magnetic flux density is taken as 1.75T. The current rolling temperature, This is the optimal rolling temperature for this steel grade. The current reduction rate, The optimal reduction rate is given by V, which represents the current rolling speed. For optimal rolling speed, , and The attenuation coefficients are taken as 20℃, 5%, and 1m / s, respectively. If the measured current magnetic induction intensity of the strip is 1.7T, the current rolling temperature is 1150℃, the optimal rolling temperature is 1130℃, the current reduction rate is 32%, the optimal reduction rate is 30%, the current rolling speed is 5.5m / s, and the optimal rolling speed is 5m / s, then... It is 0.148; Process stability rating :

[0050]

[0051]

[0052]

[0053]

[0054] In the formula, , , These represent the standard deviations of rolling force, speed, temperature, and reduction rate, respectively. , , , These are the average values ​​of the corresponding parameters. , , and These are weighting coefficients, set to 0.3, 0.3, 0.2, and 0.2 respectively. like It is 600kN. It is 12000kN. It is 0.4 m / s. It is 5 m / s. It is 15℃. It is 1130℃. It is 1.5%. 30%, then It is 0.9483.

[0055] It should be further explained that, in the specific implementation process, the calculation of the rate of change for each score includes: An evaluation cycle is set. At each evaluation cycle, a thickness uniformity score is obtained. The thickness uniformity scores of the first 2n evaluation cycles are taken. The 2n thickness uniformity scores are arranged in chronological order. The average of the first n thickness uniformity scores is calculated to obtain the first thickness uniformity score average. The average of the last n thickness uniformity scores is calculated to obtain the second thickness uniformity score average. The average of the second thickness uniformity score is subtracted from the average of the first thickness uniformity score to obtain the score difference. The score difference is divided by the average of the first thickness uniformity score to obtain the thickness uniformity score change rate. The plate shape quality score, magnetic property trend score, and process stability score all obtained their corresponding change rates using the methods described above.

[0056] It should be further explained that, in the specific implementation process, the process of comparing the change rate of each score with the respective change rate threshold, identifying abnormal patterns, performing cause analysis and process parameter adjustment strategies based on the preset abnormal pattern library, and deciding whether to trigger the process parameter optimization process includes: Threshold U for the rate of change of thickness uniformity score:

[0057] In the formula, R is the rate of change of the thickness uniformity score. This represents the average rate of change of thickness uniformity score over the most recent 100 calculation periods. The standard deviation of the rate of change of the thickness uniformity score. The sensitivity coefficient is set to 2. The corresponding change rate thresholds for the plate shape quality score change rate, magnetic property trend score change rate, and process stability score change rate are obtained using the above methods. When the absolute value of the rate of change in the rating is greater than its corresponding rate of change threshold, an abnormal trend is observed, and the abnormal pattern needs to be identified in a timely manner. If the rate of change of thickness uniformity score is less than the negative of the threshold for the rate of change of thickness uniformity score, the abnormal pattern is determined to be thickness offset. If the rate of change of the plate shape quality score is less than the negative of the threshold for the rate of change of the plate shape quality score, the abnormal pattern is determined to be plate shape deterioration. When the rate of change of magnetic property trend score is less than the negative of the threshold of magnetic property trend score change rate, the abnormal pattern is determined to be a decrease in magnetic property. If the rate of change of the process stability score is less than the negative of the threshold for the rate of change of the process stability score, the abnormal mode is determined to be process instability. Based on the recommended adjustment strategies from the abnormal mode library, each abnormal mode has a corresponding adjustment strategy. Predefined process parameters are selected and executed according to the adjustment strategy. The rate of change of each score is monitored for x cycles after adjustment. If the initial adjustment effect is not significant, the process parameter optimization process is initiated.

[0058] It should be further explained that, in the specific implementation process, the process of generating a comprehensive process score through various scores, setting a scoring threshold, and deciding whether to trigger the process parameter optimization process based on the comprehensive process score and the scoring threshold includes: Overall process score P:

[0059] In the formula, , , and These are weighting coefficients, obtained through training based on historical data, specifically 0.3, 0.25, 0.25, and 0.2. Set an appropriate comprehensive process score threshold based on historical comprehensive process score data, where historical comprehensive process score data refers to the data set of previous comprehensive process scores. Compare the comprehensive process score with the comprehensive process score threshold. If the comprehensive process score is less than the comprehensive process score threshold, the process parameter optimization process is triggered.

[0060] It should be further explained that, in the specific implementation process, the process of obtaining historical process parameter optimization data and constructing a process parameter optimization prediction model, using the process parameter optimization prediction model to predict the optimized process parameters, and adjusting the process parameters in the silicon steel rolling process based on the predicted optimized process parameters includes: The process parameter optimization process is achieved through a process parameter optimization prediction model. Factors affecting the optimized process parameters include: the rate of change of thickness uniformity score, the rate of change of plate shape quality score, the rate of change of magnetic property trend score, the rate of change of process stability score, the comprehensive process score, the incoming material thickness, the incoming material temperature, the target thickness, the ambient temperature, and the ambient humidity. Acquire historical optimized process parameter data for production equipment and products. The historical optimized process parameter data includes the change rate of thickness uniformity score, change rate of plate shape quality score, change rate of magnetic property trend score, change rate of process stability score, comprehensive process score, incoming material thickness, incoming material temperature, target thickness, ambient temperature, ambient humidity, and historical optimized process parameters of production equipment. Based on the change rate of thickness uniformity score, change rate of plate shape quality score, change rate of magnetic property trend score, change rate of process stability score, comprehensive process score, incoming material thickness, incoming material temperature, target thickness, ambient temperature, ambient humidity and corresponding historical optimized process parameters in different historical optimized process parameter data, an optimized process parameter prediction set is generated and divided into the first training set and the first test set. A first convolutional neural network is constructed, and the following parameters from different historical optimized process parameters in the first training set are used as input data: thickness uniformity score change rate, plate shape quality score change rate, magnetic property trend score change rate, process stability score change rate, comprehensive process score, incoming material thickness, incoming material temperature, target thickness, ambient temperature and ambient humidity. The corresponding historical optimized process parameters in the first training set are used as output data of the first convolutional neural network. The first convolutional neural network is trained to obtain the first initial convolutional neural network. The first initial convolutional neural network is validated using the first test set. The first initial convolutional neural network, whose output is less than or equal to the preset first test error threshold, is used as the optimized process parameter prediction model. The real-time thickness uniformity score change rate, plate shape quality score change rate, magnetic property trend score change rate, process stability score change rate, comprehensive process score, incoming material thickness, incoming material temperature, target thickness, ambient temperature and ambient humidity of the production equipment and products are input into the optimized process parameter prediction model to obtain the predicted optimized process parameters. The predicted and optimized process parameters include the optimized rolling temperature, rolling speed, reduction rate, rolling force, and cooling water flow rate. The predicted and optimized process parameters are executed, and the change rate of each score and the corresponding comprehensive process score are monitored for y cycles after adjustment.

[0061] It should be further explained that, in the specific implementation process, the process of acquiring historical energy consumption data and building an energy consumption prediction model, and then using the energy consumption prediction model to predict the energy consumption of production equipment, includes: The energy consumption refers to the energy consumed by the production equipment; Factors affecting the energy consumption of production equipment include: rolling temperature, rolling speed, reduction rate, rolling force, cooling water flow rate, strip width, incoming material thickness, incoming material temperature, and ambient temperature. Acquire historical energy consumption data of the production equipment in the production task. The historical energy consumption data includes the rolling temperature, rolling speed, reduction rate, rolling force, cooling water flow rate, strip width, incoming material thickness, incoming material temperature, ambient temperature and historical energy consumption of the production equipment in the production task. Based on the rolling temperature, rolling speed, reduction rate, rolling force, cooling water flow rate, strip width, incoming material thickness, incoming material temperature, ambient temperature and corresponding historical energy consumption of production equipment in different historical energy consumption data, an energy consumption prediction set is generated and divided into a second training set and a second test set. A second convolutional neural network is constructed. The rolling temperature, rolling speed, reduction rate, rolling force, cooling water flow rate, strip width, incoming material thickness, incoming material temperature and ambient temperature in different historical energy consumption data in the second training set are used as the input data of the second convolutional neural network, and the corresponding historical energy consumption in the second training set is used as the output data of the second convolutional neural network. The second convolutional neural network is trained to obtain the second initial convolutional neural network. The second initial convolutional neural network is validated using the second test set. The second initial convolutional neural network that outputs a second test error threshold less than or equal to the preset second test error threshold is used as the energy consumption prediction model. The real-time rolling temperature, rolling speed, reduction rate, rolling force, cooling water flow rate, strip width, incoming material thickness, incoming material temperature and ambient temperature of the production equipment are input into the energy consumption prediction model to obtain the predicted energy consumption of the production equipment. In an embodiment of the present invention, the predicted energy consumption of the production equipment is obtained through an energy consumption prediction model. The predicted energy consumption is related to rolling temperature, rolling speed, reduction rate, rolling force, cooling water flow rate, strip width, incoming material thickness, incoming material temperature and ambient temperature. Rolling temperature affects heating energy consumption and deformation resistance; rolling speed affects motor power and energy consumption per unit time; reduction rate affects deformation work and is positively correlated with energy consumption; rolling force directly reflects deformation energy consumption; cooling water flow rate affects pumping energy consumption; strip width affects rolling force and cooling area; incoming material thickness affects total deformation; incoming material temperature affects heating energy consumption; and ambient temperature affects heat dissipation loss.

[0062] It should be further explained that, in the specific implementation process, the process of calculating the energy consumption ratio based on the predicted energy consumption before and after process parameter optimization, setting a ratio threshold, and deciding whether to activate the energy-saving optimization strategy based on the energy consumption ratio and the ratio threshold includes: Both the optimized process parameters and the unoptimized process parameters are input into the energy consumption prediction model to obtain the predicted energy consumption after optimization and the predicted energy consumption before optimization. The energy consumption ratio is obtained by dividing the predicted energy consumption after optimization by the predicted energy consumption before optimization. Set an appropriate ratio threshold based on historical energy consumption ratio data. The historical energy consumption ratio data refers to the data set of previous energy consumption ratios. Compare the energy consumption ratio with the ratio threshold. If the energy consumption ratio is less than the ratio threshold, the energy-saving optimization strategy will not be triggered and the original optimized process parameters will be maintained. If the energy consumption ratio is greater than the ratio threshold, the energy-saving optimization strategy will be triggered. Energy-saving optimization strategies: Fine-tune each process parameter until the energy consumption ratio is lower than the ratio threshold, set the lower limit of each score, and after fine-tuning, ensure that each score is higher than the lower limit of each score, and that the overall process score is higher than the overall process score threshold. Energy consumption can also be reduced by the following operations: mass-producing similar specifications of products, optimizing the production rhythm, and avoiding peak electricity consumption periods.

[0063] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0064] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for intelligent optimization of rolling process parameters in silicon steel production, characterized in that, Includes the following steps: S1: Real-time acquisition of process parameters, quality data, and environmental status data during the silicon steel rolling process. S2: Based on the collected data, generate process status scores in four dimensions during the silicon steel rolling process, including thickness uniformity score, plate shape quality score, magnetic property trend score, and process stability score, and calculate the rate of change of each score. S3: Compare the rate of change of each score with the threshold of each rate of change to identify abnormal patterns, perform cause analysis and process parameter adjustment strategies based on the preset abnormal pattern library, and decide whether to trigger the process parameter optimization process. S4: Generate a comprehensive process score from the scores, set a score threshold, and decide whether to trigger the process parameter optimization process based on the comprehensive process score and the score threshold. S5: Obtain historical process parameter optimization data and construct a process parameter optimization prediction model. Use the process parameter optimization prediction model to predict the optimized process parameters, and adjust the process parameters in the silicon steel rolling process according to the predicted optimized process parameters. S6: Obtain historical energy consumption data and build an energy consumption prediction model. Use the energy consumption prediction model to predict the energy consumption of production equipment. Calculate the energy consumption ratio before and after the predicted process parameters are optimized. Set a ratio threshold. Decide whether to start the energy-saving optimization strategy based on the energy consumption ratio and the ratio threshold.

2. The intelligent optimization method for silicon steel production rolling process parameters according to claim 1, characterized in that, The process parameters include rolling temperature, rolling speed, reduction rate, rolling force, and cooling water flow rate; the quality data include strip thickness, strip shape deviation, and magnetic induction intensity; and the environmental condition data include ambient temperature, ambient humidity, incoming material thickness, incoming material temperature, and target thickness. The rolling temperature is measured by an infrared thermometer, the rolling speed is the preset speed of the equipment, the reduction rate is measured by a displacement sensor, the rolling force is measured by a piezoelectric sensor, the cooling water flow rate is measured by an electromagnetic flowmeter, the strip thickness is measured by a laser thickness gauge, the strip shape deviation is measured by a strip roll, the magnetic induction intensity is measured by an online permeability meter, the ambient temperature and humidity are measured by a temperature and humidity sensor, the incoming material thickness and temperature are measured by a thickness gauge, and the target thickness is the preset target of the system.

3. The intelligent optimization method for silicon steel production rolling process parameters according to claim 2, characterized in that, The process of generating a four-dimensional process status score based on the collected data during silicon steel rolling includes: Thickness uniformity score : In the formula, The thickness measurement value at the i-th sampling point. Let n be the target thickness, and n be the number of sampling points within the calculation window. For thickness tolerance, To calculate the standard deviation of thickness measurements within the window, This represents the maximum permissible standard deviation. Plate shape quality score : In the formula, Let the plate shape deviation be the j-th plate shape measurement unit. The maximum allowable shape deviation for the corresponding unit is set according to the strip width, where m is the number of shape measurement units. The penalty coefficient for the changing trend. / dt is the rate of change of plate shape deviation with time, which is calculated by the central difference method; Magnetic property trend score : In the formula, The current measured magnetic flux density is... The target magnetic flux density, The current rolling temperature, This is the optimal rolling temperature for this steel grade. The current reduction rate, The optimal reduction rate is given by V, which represents the current rolling speed. For optimal rolling speed, , and The attenuation coefficient; Process stability rating : In the formula, , , These represent the standard deviations of rolling force, speed, temperature, and reduction rate, respectively. , , , These are the average values ​​of the corresponding parameters. , , and These are the weighting coefficients.

4. The intelligent optimization method for silicon steel production rolling process parameters according to claim 3, characterized in that, The process of calculating the rate of change for each score includes: An evaluation cycle is set. At each evaluation cycle, a thickness uniformity score is obtained. The thickness uniformity scores of the first 2n evaluation cycles are taken. The 2n thickness uniformity scores are arranged in chronological order. The average of the first n thickness uniformity scores is calculated to obtain the first thickness uniformity score average. The average of the last n thickness uniformity scores is calculated to obtain the second thickness uniformity score average. The average of the second thickness uniformity score is subtracted from the average of the first thickness uniformity score to obtain the score difference. The score difference is divided by the average of the first thickness uniformity score to obtain the thickness uniformity score change rate. The plate shape quality score, magnetic property trend score, and process stability score all obtained their corresponding change rates using the methods described above.

5. The intelligent optimization method for silicon steel production rolling process parameters according to claim 4, characterized in that, The process of comparing the rate of change of each score with the threshold of each rate of change to identify abnormal patterns, performing cause analysis and process parameter adjustment strategies based on a pre-set abnormal pattern library, and deciding whether to trigger the process parameter optimization process includes: Threshold U for the rate of change of thickness uniformity score: In the formula, R is the rate of change of the thickness uniformity score. This represents the average rate of change of thickness uniformity score over the most recent 100 calculation periods. The standard deviation of the rate of change of the thickness uniformity score. Sensitivity coefficient; The corresponding change rate thresholds for the plate shape quality score change rate, magnetic property trend score change rate, and process stability score change rate are obtained using the above methods. When the absolute value of the rate of change in the rating is greater than its corresponding rate of change threshold, an abnormal trend is observed, and the abnormal pattern needs to be identified in a timely manner. If the rate of change of thickness uniformity score is less than the negative of the threshold for the rate of change of thickness uniformity score, the abnormal pattern is determined to be thickness offset. If the rate of change of the plate shape quality score is less than the negative of the threshold for the rate of change of the plate shape quality score, the abnormal pattern is determined to be plate shape deterioration. When the rate of change of magnetic property trend score is less than the negative of the threshold of magnetic property trend score change rate, the abnormal pattern is determined to be a decrease in magnetic property. If the rate of change of the process stability score is less than the negative of the threshold for the rate of change of the process stability score, the abnormal mode is determined to be process instability. Based on the recommended adjustment strategies from the abnormal mode library, each abnormal mode has a corresponding adjustment strategy. Predefined process parameters are selected and executed according to the adjustment strategy. The rate of change of each score is monitored for x cycles after adjustment. If the initial adjustment effect is not significant, the process parameter optimization process is initiated.

6. The intelligent optimization method for silicon steel production rolling process parameters according to claim 5, characterized in that, The process of generating a comprehensive process score from various scores, setting a scoring threshold, and deciding whether to trigger the process parameter optimization process based on the comprehensive process score and the scoring threshold includes: Overall process score P: In the formula, , , and These are the weighting coefficients, obtained through training based on historical data; Set an appropriate comprehensive process score threshold based on historical comprehensive process score data, where historical comprehensive process score data refers to the data set of previous comprehensive process scores. Compare the comprehensive process score with the comprehensive process score threshold. If the comprehensive process score is less than the comprehensive process score threshold, the process parameter optimization process is triggered.

7. The intelligent optimization method for silicon steel production rolling process parameters according to claim 6, characterized in that, The process of acquiring historical process parameter optimization data and constructing a process parameter optimization prediction model, using the prediction model to predict the optimized process parameters, and adjusting the process parameters in the silicon steel rolling process based on the predicted optimized process parameters includes: The process parameter optimization process is achieved through a process parameter optimization prediction model. Factors affecting the optimized process parameters include: the rate of change of thickness uniformity score, the rate of change of plate shape quality score, the rate of change of magnetic property trend score, the rate of change of process stability score, the comprehensive process score, the incoming material thickness, the incoming material temperature, the target thickness, the ambient temperature, and the ambient humidity. Acquire historical optimized process parameter data for production equipment and products. The historical optimized process parameter data includes the change rate of thickness uniformity score, change rate of plate shape quality score, change rate of magnetic property trend score, change rate of process stability score, comprehensive process score, incoming material thickness, incoming material temperature, target thickness, ambient temperature, ambient humidity, and historical optimized process parameters of production equipment. Based on the change rate of thickness uniformity score, change rate of plate shape quality score, change rate of magnetic property trend score, change rate of process stability score, comprehensive process score, incoming material thickness, incoming material temperature, target thickness, ambient temperature, ambient humidity and corresponding historical optimized process parameters in different historical optimized process parameter data, an optimized process parameter prediction set is generated and divided into the first training set and the first test set. A first convolutional neural network is constructed, and the following parameters from different historical optimized process parameters in the first training set are used as input data: thickness uniformity score change rate, plate shape quality score change rate, magnetic property trend score change rate, process stability score change rate, comprehensive process score, incoming material thickness, incoming material temperature, target thickness, ambient temperature and ambient humidity. The corresponding historical optimized process parameters in the first training set are used as output data of the first convolutional neural network. The first convolutional neural network is trained to obtain the first initial convolutional neural network. The first initial convolutional neural network is validated using the first test set. The first initial convolutional neural network, whose output is less than or equal to the preset first test error threshold, is used as the optimized process parameter prediction model. The real-time thickness uniformity score change rate, plate shape quality score change rate, magnetic property trend score change rate, process stability score change rate, comprehensive process score, incoming material thickness, incoming material temperature, target thickness, ambient temperature and ambient humidity of the production equipment and products are input into the optimized process parameter prediction model to obtain the predicted optimized process parameters. The predicted and optimized process parameters include the optimized rolling temperature, rolling speed, reduction rate, rolling force, and cooling water flow rate. The predicted and optimized process parameters are executed, and the change rate of each score and the corresponding comprehensive process score are monitored for y cycles after adjustment.

8. The intelligent optimization method for silicon steel production rolling process parameters according to claim 7, characterized in that, The process of acquiring historical energy consumption data and building an energy consumption prediction model, and then using that model to predict the energy consumption of production equipment, includes: The energy consumption refers to the energy consumed by the production equipment; Factors affecting the energy consumption of production equipment include: rolling temperature, rolling speed, reduction rate, rolling force, cooling water flow rate, strip width, incoming material thickness, incoming material temperature, and ambient temperature. Acquire historical energy consumption data of the production equipment in the production task. The historical energy consumption data includes the rolling temperature, rolling speed, reduction rate, rolling force, cooling water flow rate, strip width, incoming material thickness, incoming material temperature, ambient temperature and historical energy consumption of the production equipment in the production task. Based on the rolling temperature, rolling speed, reduction rate, rolling force, cooling water flow rate, strip width, incoming material thickness, incoming material temperature, ambient temperature and corresponding historical energy consumption of production equipment in different historical energy consumption data, an energy consumption prediction set is generated and divided into a second training set and a second test set. A second convolutional neural network is constructed. The rolling temperature, rolling speed, reduction rate, rolling force, cooling water flow rate, strip width, incoming material thickness, incoming material temperature and ambient temperature in different historical energy consumption data in the second training set are used as the input data of the second convolutional neural network, and the corresponding historical energy consumption in the second training set is used as the output data of the second convolutional neural network. The second convolutional neural network is trained to obtain the second initial convolutional neural network. The second initial convolutional neural network is validated using the second test set. The second initial convolutional neural network that outputs a second test error threshold less than or equal to the preset second test error threshold is used as the energy consumption prediction model. The real-time rolling temperature, rolling speed, reduction rate, rolling force, cooling water flow rate, strip width, incoming material thickness, incoming material temperature, and ambient temperature of the production equipment are input into the energy consumption prediction model to obtain the predicted energy consumption of the production equipment.

9. The intelligent optimization method for silicon steel production rolling process parameters according to claim 8, characterized in that, The process of calculating the energy consumption ratio before and after the predicted process parameters are optimized, setting a ratio threshold, and deciding whether to activate the energy-saving optimization strategy based on the energy consumption ratio and the ratio threshold includes: Both the optimized process parameters and the unoptimized process parameters are input into the energy consumption prediction model to obtain the predicted energy consumption after optimization and the predicted energy consumption before optimization. The energy consumption ratio is obtained by dividing the predicted energy consumption after optimization by the predicted energy consumption before optimization. Set an appropriate ratio threshold based on historical energy consumption ratio data. The historical energy consumption ratio data refers to the data set of previous energy consumption ratios. Compare the energy consumption ratio with the ratio threshold. If the energy consumption ratio is less than the ratio threshold, the energy-saving optimization strategy will not be triggered and the original optimized process parameters will be maintained. If the energy consumption ratio is greater than the ratio threshold, the energy-saving optimization strategy will be triggered. Energy-saving optimization strategies: Fine-tune each process parameter until the energy consumption ratio is lower than the ratio threshold, set the lower limit of each score, and after fine-tuning, ensure that each score is higher than the lower limit of each score, and that the overall process score is higher than the overall process score threshold. Energy consumption can also be reduced by the following operations: mass-producing similar specifications of products, optimizing the production rhythm, and avoiding peak electricity consumption periods.

Images