A cold rolling weld seam protection method suitable for high-grade non-oriented silicon steel

Abstract

The present application relates to the technical field of cold rolling weld joint protection, and discloses a cold rolling weld joint protection method suitable for high-grade non-oriented silicon steel, which comprises the following steps: performing surface inert gas pre-blowing during gradient ultrasonic impact, extracting dynamic characteristic parameters of a molten pool, continuously adjusting the flow of protective gas, the three-element mixed ratio of argon, helium and hydrogen, and the inclination angle of an annular gas nozzle to process the weld joint area; inputting a microstructure evolution and stress evolution coupling model, realizing continuous gradient normalization through an independent temperature control zone, applying a low-frequency pulse magnetic field in the holding stage, and adopting a segmented gradient cooling technology for cooling; dividing the weld joint area into continuous micro-sections along the length direction, calculating the maximum safe reduction amount of each micro-section, continuously and dynamically adjusting the reduction amount, rolling speed, front and rear tension, and roll gap crown of each rack for rolling, and correcting the influence of roll wear in real time; extracting the unique process fingerprint of each weld joint, establishing a machine learning fracture risk prediction model, and forming a full-process closed-loop feedback control mechanism.

Description

Technical Field

[0001] This invention relates to the field of cold-rolled weld protection technology, and specifically to a method for protecting cold-rolled welds applicable to high-grade non-oriented silicon steel. Background Technology

[0002] High-grade non-oriented silicon steel has high silicon and aluminum content and poor weldability. Cold-rolled welds are prone to oxidation embrittlement, uneven microstructure, and residual stress concentration, resulting in a high fracture rate during cold rolling. Existing technologies have many drawbacks and shortcomings that are difficult to overcome.

[0003] In existing technologies, the welding protection process often employs a single, fixed-parameter shielding gas supply method, which cannot respond in real time to dynamic changes in the molten pool temperature, flow rate, and area. This results in asymmetrical protection of the upper and lower surfaces, ineffective elimination of interference from smoke and arc light on molten pool detection data, failure to consider the response lag of the shielding gas actuator, and a lack of synergy between gradient ultrasonic impact pretreatment and surface pre-purging. Freshly impacted surfaces are prone to oxidation, leading to poor molten pool stability and defects such as porosity, inclusions, and oxide films. The normalizing process often uses an overall uniform temperature normalizing process, which cannot match the structural stress differences between the weld center, transition zone, and base metal. This results in uneven aluminum nitride precipitates with concentrated nucleation sites, insufficient residual stress release, and a lack of introduction of low-frequency pulsed magnetic fields to assist in precipitation control and stress release. Cooling methods are also problematic. The lack of a single, gradient-based design and the difference in thermal expansion coefficients between the weld and the base material lead to stress concentration and a tendency for microcracks to develop. Furthermore, the rolling control process often employs uniform rolling parameters without finely segmenting the weld area into micro-segments, making it impossible to accurately calculate the maximum safe reduction for each segment. Insufficient compensation for roll thermal expansion, wear, and tension system inertia results in uneven weld deformation, poor plate shape, and even breakage. The entire process control is largely based on independent control of single processes, lacking cross-process data correlation and closed-loop feedback mechanisms. The absence of a weld fracture risk prediction model based on the entire process fingerprint prevents early warnings and tiered intervention measures. The inability to effectively control weld quality fluctuations caused by batch-to-batch performance differences hinders the continuity of high-grade non-oriented silicon steel cold-rolled production and reduces product qualification rates.

[0004] Therefore, there is a need to provide a method for protecting cold-rolled welds of high-grade non-oriented silicon steel. Summary of the Invention

[0005] The purpose of this invention is to provide a method for protecting cold-rolled welds of high-grade non-oriented silicon steel. To solve the above-mentioned problems in the prior art, this invention achieves this through the following technical solution:

[0006] The first part, an embodiment of the present invention, provides a method for protecting cold-rolled welds of high-grade non-oriented silicon steel, specifically including the following steps:

[0007] Step 1: During the gradient ultrasonic impact process, surface inert gas pre-purging is performed. A four-section symmetrical gas supply technology is adopted to extract dynamic characteristic parameters of the molten pool. The shielding gas flow rate, argon-helium-hydrogen ternary mixing ratio and the tilt angle of the annular gas nozzle are continuously adjusted to treat the weld area.

[0008] Step 2: Combine the processed weld area, collect key weld parameters, input the coupled model of microstructure evolution and stress evolution, achieve continuous gradient normalization through independent temperature control zone, apply low frequency pulse magnetic field during the heat preservation stage, and use segmented gradient cooling technology for cooling.

[0009] Step 3: After the continuous gradient normalization is completed, the weld area is divided into continuous micro-segments along the length direction. The characteristic parameters of each micro-segment are collected, the maximum safe reduction of each micro-segment is calculated, and the reduction, rolling speed, front and rear tension and roll gap crown of each stand are continuously and dynamically adjusted for rolling. The influence of roll wear is corrected in real time.

[0010] Step 4: Collect full-process data for the three stages of welding, normalizing, and rolling for the entire rolling process, extract the unique process fingerprint of each weld, establish a machine learning fracture risk prediction model, and form a closed-loop feedback control mechanism for the entire process.

[0011] The second part, an embodiment of the present invention, provides a cold-rolled weld protection system suitable for high-grade non-oriented silicon steel, specifically including the following modules:

[0012] The protection module is used to pre-purge the surface with inert gas during gradient ultrasonic impact. It adopts a four-section symmetrical gas supply technology to extract dynamic characteristic parameters of the molten pool and continuously adjust the protective gas flow rate, argon-helium-hydrogen ternary mixing ratio and the tilt angle of the annular gas nozzle to treat the weld area.

[0013] Normalization control module: Combined with the processed weld area, key weld parameters are collected, and a coupled model of microstructure evolution and stress evolution is input. Continuous gradient normalization is achieved through an independent temperature control zone. A low-frequency pulsed magnetic field is applied during the heat preservation stage, and segmented gradient cooling technology is used for cooling.

[0014] Micro-segment control module: After continuous gradient normalization is completed, the weld area is divided into continuous micro-segments along the length direction. The micro-segment characteristic parameters of each micro-segment are collected, the maximum safe reduction of each micro-segment is calculated, and the reduction, rolling speed, front and rear tension and roll gap crown of each stand are continuously and dynamically adjusted for rolling. The effect of roll wear is corrected in real time.

[0015] Process control module: For the entire rolling process, data from the three stages of welding, normalizing and rolling are collected, a unique process fingerprint of each weld is extracted, a machine learning fracture risk prediction model is established, and a closed-loop feedback control mechanism for the entire process is formed.

[0016] The beneficial effects of this invention are:

[0017] 1. A welding protection technology combining gradient ultrasonic impact with symmetrical four-zone layered gas supply is employed. During gradient ultrasonic impact, continuous surface pre-blowing is performed. Four coaxial nested annular gas nozzles are used on both the upper and lower sides to correspond to the core reaction zone, solid-liquid interface zone, high-temperature zone, and low-temperature zone of the molten pool, respectively. By dynamically compensating for interference from smoke and arc light and the lag of the shielding gas actuator, dynamic characteristic parameters of the molten pool are extracted. The shielding gas flow rate, argon-helium-hydrogen ternary mixing ratio, and annular gas nozzle tilt angle are adjusted in real time to solve the problems of single fixed parameter shielding gas failing to respond to dynamic changes in the molten pool in real time, asymmetrical protection of the upper and lower surfaces, easy oxidation of fresh impact surfaces, and erroneous parameter adjustments due to distorted molten pool detection data. A coupled model of microstructure evolution and stress evolution is established to drive continuous gradient normalization in independent temperature control zones. Segmented gradient cooling technology is used to solve the problems of overall uniform temperature normalization failing to match the microstructure stress differences in different areas of the weld, aluminum nitride precipitate size exceeding the optimal range, and stress concentration during the cooling process easily initiating microcracks.

[0018] 2. The weld seam area is divided into continuous micro-segments. The maximum safe reduction is calculated based on the characteristic parameters of each micro-segment. The reduction, rolling speed, front and rear tension, and roll gap crown of each stand are dynamically adjusted in real time. At the same time, a roll thermal expansion and wear model and a tension fluctuation prediction model are established for early compensation. This solves the problems of uneven weld seam deformation, poor plate shape, roll condition changes, and rolling instability caused by tension inertia due to uniform rolling parameters. The fingerprint of the entire process of welding, normalizing, and rolling is extracted. A machine learning weld seam fracture risk prediction model is established and graded intervention is implemented. By synchronizing the time and location of data throughout the entire process and incremental learning to adapt to batch performance drift, a cross-process closed-loop feedback control mechanism is constructed to solve the problems of lack of data correlation in single-process independent control, inability to provide early warning of fracture risks, and difficulty in controlling quality fluctuations between batches. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart of the steps of a cold-rolled weld protection method for high-grade non-oriented silicon steel provided in Embodiment 1 of the present invention;

[0021] Figure 2 This is a schematic diagram of a cold-rolled weld protection system for high-grade non-oriented silicon steel provided in Embodiment 2 of the present invention;

[0022] Figure 3 This is a rolling schematic diagram of a cold-rolled weld protection method for high-grade non-oriented silicon steel provided in Embodiment 1 of the present invention. Detailed Implementation

[0023] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. 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 skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0024] Example 1: As Figure 1 and Figure 3 As shown in the figure, an embodiment of the present invention provides a method for protecting cold-rolled welds of high-grade non-oriented silicon steel, which specifically includes the following steps:

[0025] Step 1: During the gradient ultrasonic impact process, surface inert gas pre-purging is performed. A four-section symmetrical gas supply technology is adopted to extract dynamic characteristic parameters of the molten pool. The shielding gas flow rate, argon-helium-hydrogen ternary mixing ratio and the tilt angle of the annular gas nozzle are continuously adjusted to treat the weld area.

[0026] In a specific embodiment, a surface cleaning device and a gradient ultrasonic impact device are installed sequentially in front of the laser welding station; the surface cleaning device uses a rectangular nozzle with a nozzle width consistent with the strip width to continuously clean the entire strip surface.

[0027] The gradient ultrasonic impact device contains five independently controlled impact units, each with a width of 10 mm, covering an area of ​​25 mm on each side of the strip weld.

[0028] During the strip steel conveying process, when the weld seam marking point reaches the position of the surface purging device, the surface purging device is activated. High-purity argon gas of preset concentration is used to pre-purify the area 30 mm wide on both sides of the weld seam. The purging time is matched with the strip steel conveying speed to ensure that there is no oxide film or impurity residue on the surface of the weld seam area.

[0029] After the pre-purging is completed, the gradient ultrasonic impact device is started to perform gradient ultrasonic impact treatment on a 25 mm wide area on both sides of the weld. From the center line of the weld to both sides, the impact amplitude of the five impact units is set to 35 μm, 30 μm, 25 μm, 20 μm and 15 μm respectively, the vibration frequency is uniformly set to 20 kHz, and the processing speed is consistent with the strip conveying speed.

[0030] During the gradient ultrasonic impact treatment, the surface purging device works continuously to prevent the fresh surface generated by the impact from contacting the air and oxidizing.

[0031] Four coaxial nested and angle-adjustable annular air nozzles are installed above and below the laser welding gun head. The four upper annular air nozzles correspond to the core reaction zone, solid-liquid interface zone, heat-affected high temperature zone, and heat-affected low temperature zone on the upper surface of the molten pool, while the four lower annular air nozzles correspond to the same four areas on the lower surface of the molten pool, thus achieving symmetrical protection of the upper and lower surfaces of the molten pool.

[0032] During the welding process, welding current and welding voltage data are continuously collected by the welding power source, temperature distribution data of the upper and lower surfaces of the molten pool are continuously collected by a dual-color infrared thermometer arranged symmetrically above and below, flow velocity distribution data of the upper and lower surfaces of the molten pool are continuously collected by a high-speed camera combined with particle image velocimetry technology, and welding speed data are continuously collected by an encoder.

[0033] All collected data are time-synchronized to ensure that different parameter data at the same time can be accurately matched.

[0034] Dynamic compensation is performed for the interference of smoke and arc light in the molten pool detection data. A database of the correspondence between smoke concentration, arc light intensity and detection data error is established. The database of the correspondence is obtained by pre-calibration under different smoke concentration and arc light intensity conditions.

[0035] In the actual welding process, based on the real-time collected data on fume concentration and arc intensity, the corresponding error correction values ​​are retrieved from the database to dynamically correct the molten pool temperature distribution data and flow velocity distribution data.

[0036] The specific processing procedure is as follows: the original detection data is superimposed with the error correction value to obtain the true molten pool parameter data after eliminating interference, so as to avoid the incorrect adjustment of the protective gas parameters caused by smoke and arc interference;

[0037] To compensate for the lag of the protective gas parameter actuator, the response time characteristics of the protective gas flow regulating valve and the mixing proportion regulating valve are measured in advance, and an actuator lag model is established.

[0038] After the protective gas parameter adjustment command is calculated, the adjustment command is issued in advance according to the actuator lag model, so that the actual change time of the protective gas parameter is precisely synchronized with the change time of the molten pool state.

[0039] The specific process is as follows: predict the trend of the molten pool state in the future actuator response cycle, and adjust the protective gas parameters in advance according to the prediction results to eliminate the control deviation caused by the actuator lag.

[0040] Extract the compensated dynamic characteristic parameters of the molten pool, which include: the rate of change of temperature at the center of the molten pool, the rate of change of area of ​​the molten pool, the fluctuation amplitude of the surface velocity of the molten pool, and the fluctuation amplitude of the welding current.

[0041] The dynamic characteristic parameters of the molten pool are matched with the pre-established molten pool stability database, and the protective gas flow rate, argon-helium-hydrogen ternary mixing ratio and tilt angle of the upper and lower eight annular gas nozzles are continuously adjusted according to the matching results.

[0042] When the rate of temperature change at the center of the molten pool increases, the proportion of helium in the core reaction zone is increased to enhance heat dissipation; when the rate of change of the molten pool area increases, the flow rate of protective gas in all areas is increased to enhance the protective effect.

[0043] When the surface velocity fluctuation of the molten pool increases, adjust the tilt angle of the annular nozzle to counteract the effect of Marangoni convection; when the welding current fluctuation increases, increase the proportion of hydrogen in the shielding gas to stabilize the arc.

[0044] It should be noted that Marangoni convection is a fluid flow phenomenon driven by the surface tension gradient at the free surface of a liquid or at the interface between two fluids; it is also known as capillary convection.

[0045] The preheating temperature of the protective gas is continuously adjusted according to the average temperature of the molten pool. The preheating temperature is set to 0.3 times the average temperature of the molten pool, with a maximum of less than or equal to 600 degrees Celsius. This reduces the temperature difference between the protective gas and the molten pool and prevents surface shear forces from damaging the stability of the molten pool.

[0046] Step 2: Combine the processed weld area, collect key weld parameters, input the coupled model of microstructure evolution and stress evolution, achieve continuous gradient normalization through independent temperature control zone, apply low frequency pulse magnetic field during the heat preservation stage, and use segmented gradient cooling technology for cooling.

[0047] In a specific embodiment, after welding is completed, key parameters of the weld are collected. These key parameters include: initial nitrogen content data, silicon / aluminum atomic ratio data, initial microstructure grain size data, grain boundary orientation distribution data, residual stress distribution data, and temperature distribution data. The weld area is scanned point by point using a laser-induced breakdown spectrometer to collect initial nitrogen content data and silicon / aluminum atomic ratio data at different locations in the weld area.

[0048] The weld area was scanned point by point using an electron backscatter diffraction diffractometer to collect initial microstructure grain size data and grain boundary orientation distribution data at different locations in the weld area.

[0049] The weld area was scanned point by point using an X-ray stress analyzer to collect residual stress distribution data at different locations in the weld area.

[0050] By using an infrared temperature measurement array installed at the entrance of the normalizing furnace, the temperature distribution data of the weld area before entering the normalizing furnace is collected, and the welding cooling rate at different locations in the weld area is calculated.

[0051] All collected data are input into a pre-established coupled model of microstructure evolution and stress evolution. Based on the initial parameters at different locations in the weld area, the model simulates and calculates the aluminum nitride precipitation behavior, microstructure evolution process, and stress evolution process at the corresponding locations under different normalization temperatures, holding times, and cooling rates.

[0052] Through simulation calculations, the optimal combination of normalization parameters was determined for each location to ensure uniform precipitation of aluminum nitride in the [20, 50] nanometer size within the crystal, while minimizing residual stress.

[0053] The normalizing furnace is designed with twenty independent temperature control zones, each with a length of 0.25 meters, to achieve continuous temperature gradient changes. Based on the preset optimal normalizing parameter combination calculated by the model, the temperature of the twenty temperature control zones and the strip conveying speed are linearly adjusted so that each position in the weld area is processed according to the preset optimal temperature-time curve.

[0054] The spatial distribution of the temperature field in the normalizing furnace is compensated. The temperature field of the twenty temperature control zones of the normalizing furnace is calibrated in advance, and a database of the correspondence between the actual temperature and the set temperature at different locations in each temperature control zone is established.

[0055] In actual production, the set temperature of each temperature control zone is dynamically adjusted based on the real-time temperature distribution data of each temperature control zone to ensure that the actual temperature at each location in the weld area is consistent with the optimal normalization temperature calculated by the model.

[0056] The specific processing procedure is as follows: the optimal normalizing temperature calculated by the model is superimposed with the error value in the temperature field database to obtain the corrected set temperature for each temperature control zone, thereby eliminating the control deviation caused by the uneven temperature field of the normalizing furnace.

[0057] Dynamic compensation is made for the holding time to compensate for the fluctuation of strip running speed. Based on the real-time collected strip running speed data, the actual dwell time of each position in the weld area in each temperature control zone is calculated.

[0058] When the strip speed increases, the temperature of the corresponding temperature control zone is increased to compensate for the shortened holding time; when the strip speed decreases, the temperature of the corresponding temperature control zone is decreased to compensate for the extended holding time.

[0059] The specific processing procedure is as follows: establish an equivalent relationship model between heat preservation time and normalization temperature, and calculate the corresponding temperature adjustment amount based on the difference between the actual residence time and the target heat preservation time to ensure that each position obtains an equivalent normalization effect;

[0060] Temperature compensation is performed on the pulse magnetic field strength. The resistance of the electromagnetic coil changes with temperature in advance. A database of the correspondence between coil temperature and magnetic field strength is established. Based on the real-time collected electromagnetic coil operating temperature data, the power supply current of the coil is dynamically adjusted to keep the actual magnetic field strength at the set value.

[0061] The specific processing procedure is as follows: retrieve the current correction value at the corresponding temperature from the database, dynamically adjust the coil power supply current, and eliminate the influence of coil temperature changes on magnetic field strength;

[0062] During the heat preservation stage of normalization treatment, a low-frequency pulsed magnetic field is applied simultaneously above and below the weld area. The frequency of the pulsed magnetic field is set to 10Hz. The low-frequency pulsed magnetic field can promote the uniform nucleation of aluminum nitride precipitates, inhibit the growth of precipitates, and at the same time promote dislocation movement and accelerate the release of residual stress.

[0063] During the cooling stage, a segmented gradient cooling technology is adopted, gradually reducing the cooling rate from the weld center to the base material. The weld center area uses a cooling rate of 18 degrees Celsius per second, the weld transition section uses a cooling rate of 15 degrees Celsius per second, and the base material area uses a cooling rate of 12 degrees Celsius per second. The gradient cooling method matches the different thermal expansion coefficients of the weld and the base material, reduces the internal stress generated during the cooling process, and avoids the formation of microcracks.

[0064] Step 3: After the continuous gradient normalization is completed, the weld area is divided into continuous micro-segments along the length direction. The characteristic parameters of each micro-segment are collected, the maximum safe reduction of each micro-segment is calculated, and the reduction, rolling speed, front and rear tension and roll gap crown of each stand are continuously and dynamically adjusted for rolling. The influence of roll wear is corrected in real time.

[0065] In a specific embodiment, after the normalization process is completed, the weld area is tested point by point using a Vickers hardness tester to collect hardness distribution data of the weld area. The test step size is set to 5 mm and the test load is set to 1 kg.

[0066] Thickness distribution data of the weld area were collected using a laser thickness gauge with a sampling frequency of 200Hz; internal defect distribution data of the weld area were collected using a phased array ultrasonic flaw detector.

[0067] Data on the microstructure uniformity and grain boundary distribution of the weld region were collected using an electron backscatter diffraction instrument. The work hardening rate data at different locations in the weld region were determined by a single-pass compression test, and the relationship between the work hardening rate and the amount of deformation was established.

[0068] The weld area is divided into continuous 1 mm long micro-segments along its length. The characteristic parameters of each micro-segment are calculated, including: average hardness, average thickness, internal defect density, microstructure uniformity index, and work hardening rate. Based on the work hardening rate of each micro-segment, the hardness change of the corresponding micro-segment when passing through each frame is predicted.

[0069] A database of the correspondence between strip temperature and material yield strength is established. The database is obtained through tensile tests at different temperatures. Based on the real-time collected temperature distribution data of the weld area, the change in yield strength of each micro-segment is calculated, and then the adjustment of rolling force is calculated.

[0070] The specific processing procedure is as follows: multiply the maximum safe reduction calculated at room temperature by the temperature correction factor to obtain the actual maximum safe reduction at the corresponding temperature, thereby eliminating the influence of strip temperature changes on rolling force;

[0071] A roll thermal expansion model and a roll wear model were established. The roll thermal expansion model calculates the thermal expansion of the roll based on rolling force, rolling speed and cooling water flow data. The roll wear model calculates the wear of the roll based on cumulative rolling mileage, rolling force distribution and strip surface quality data.

[0072] The total roll gap change is obtained by superimposing the thermal expansion and wear. The specific process is as follows: adjust the roll gap setting value of each stand in real time according to the total roll gap change, and at the same time correct the roll gap crown parameter to eliminate the influence of roll thermal expansion and wear on the rolling process.

[0073] The inertial characteristics of the tension system are determined in advance, a tension fluctuation prediction model is established, and the tension fluctuation trend in the future is predicted based on the real-time collected rolling force change data and rolling speed change data.

[0074] The specific processing procedure is as follows: adjust the output of the tension actuator in advance according to the prediction results to ensure that the actual tension is kept within the set value range and eliminate the fluctuations caused by the inertia of the tension system;

[0075] Based on the initial hardness, predicted hardness change, internal defect density, and microstructure uniformity index of each microsegment, the maximum safe reduction that the corresponding microsegment can withstand in each frame is calculated.

[0076] The maximum safe reduction of all micro-segments is integrated to generate a reduction distribution curve for the entire weld area. Based on the reduction distribution curve, the reduction of each frame is continuously and dynamically adjusted to ensure that each micro-segment does not experience deformation greater than the preset maximum safe reduction when passing through each frame.

[0077] Based on the thickness distribution data of the weld area, the rolling speed of each stand is continuously and dynamically adjusted; for micro-segments with larger thickness, the rolling speed is reduced; for micro-segments with smaller thickness, the rolling speed is increased.

[0078] The adjustment range of the rolling speed is proportional to the thickness difference, ensuring that each micro-segment obtains stable deformation conditions when passing through the rolling mill;

[0079] The front and rear tensions of the rolling mill are continuously and dynamically adjusted based on the average hardness and internal defect density of the weld area.

[0080] When the average hardness of the weld area decreases or the density of internal defects increases, the front and back tensions are reduced to avoid excessive tension that could lead to weld breakage. The adjustment range of the front and back tensions is proportional to the changes in hardness and defect density.

[0081] Based on the hardness distribution and plate shape prediction results in the weld area, the roll gap convexity of each frame is continuously and dynamically adjusted; for the center section of the weld with lower hardness, the roll gap convexity is increased to compensate for the uneven deformation in the weld area and ensure the plate shape quality.

[0082] A real-time monitoring model for roll wear was established, and the wear distribution of the rolls was calculated in real time based on the cumulative rolling mileage, rolling force distribution, and strip surface quality data.

[0083] Based on the roll wear distribution, the reduction distribution and roll gap crown of each stand are adjusted in real time to ensure the accuracy of rolling parameters during long-term production.

[0084] Step 4: Collect full-process data for the three stages of welding, normalizing and rolling for the entire rolling process, extract the unique process fingerprint of each weld, establish a machine learning fracture risk prediction model, and form a closed-loop feedback control mechanism for the entire process.

[0085] In a specific embodiment, high-precision synchronous detection devices are installed at the welding station, the normalizing furnace exit, and the rolling mill exit of the production line, respectively. All detection devices use a unified clock signal for time synchronization.

[0086] A high-precision position encoder is installed on the edge of the strip to collect the position data of the strip in real time, ensuring that all detection data and processing parameters can be accurately matched to the specific position of the strip.

[0087] At the welding station, data from the entire process is collected and the fingerprint of the welding process is extracted. The data from the entire process includes: welding current, welding voltage, molten pool temperature distribution, molten pool flow rate distribution, shielding gas parameters, and weld appearance image.

[0088] At the outlet of the normalizing furnace, the hardness distribution, microstructure uniformity, nitrogen content and residual stress of the weld area were collected, and the fingerprint of the normalizing process was extracted.

[0089] At the mill exit, surface quality, plate shape, internal defects, and thickness accuracy of the weld area are collected, and the fingerprints of the rolling process are extracted.

[0090] Establish a full-process database for weld quality, and link and store the welding process fingerprint, normalization process fingerprint, rolling process fingerprint and final quality data of each weld.

[0091] By using big data analytics, a mapping relationship between process fingerprints and weld quality is established to identify key process characteristic parameters that affect the risk of weld fracture.

[0092] The transmission time of the weld from the welding station to the outlet of the normalizing furnace and from the outlet of the normalizing furnace to the inlet of the rolling mill was measured in advance, and a model of the correspondence between transmission time and strip running speed was established.

[0093] Based on the real-time collected strip speed data, the accurate time for the weld to reach the next process is predicted. The specific process is as follows: prepare the processing parameters for the next process in advance according to the predicted time, so that the best processing plan can be executed immediately when the weld arrives, eliminating the problem of untimely feedback caused by time delay between processes.

[0094] Based on the high-precision position encoder data of the strip edge, all detection data are aligned in position.

[0095] The specific processing procedure is as follows: the acquisition time of each detection data is matched with the time of the position encoder, and the position coordinates of the strip corresponding to the detection data are calculated;

[0096] Based on the position coordinates of the weld marking points, all detection data are aligned to the relative position of the weld to eliminate positional deviations caused by different installation positions of the detection devices and data transmission delays.

[0097] Establish a batch performance database to record the chemical composition, mechanical properties, and optimal processing parameters of each batch of strip steel;

[0098] When changing the strip steel batch, historical data of the corresponding batch is retrieved from the database to make initial corrections to the weld fracture risk prediction model and process parameter control model.

[0099] Based on the real-time production data of the corresponding batch, the incremental learning algorithm is used to continuously update the model. The specific process is as follows: compare the new production data with the model prediction results, calculate the model error, and adjust the model parameters according to the error value to ensure that the model adapts to the performance drift between batches.

[0100] A weld fracture risk prediction model was established to predict the fracture risk of the weld when it passes through the rolling mill, based on the welding process fingerprint and normalization process fingerprint of the weld.

[0101] The weld fracture risk prediction model uses machine learning algorithms to continuously optimize prediction accuracy by learning from new production data. When the predicted fracture risk exceeds the set safety level, an early warning is issued, and corresponding protective measures are taken according to the risk level.

[0102] For low-risk welds, reduce the rolling speed; for medium-risk welds, reduce the reduction in the weld area; for high-risk welds, automatically stop the machine and notify the operator for handling.

[0103] Establish a closed-loop feedback control mechanism for the entire process, and adjust the processing parameters of the preceding processes in real time based on the online detection results;

[0104] When welding quality inspection reveals a decrease in molten pool stability, the shielding gas parameters and welding current parameters of the welding station are adjusted in real time; when normalizing quality inspection reveals excessively large aluminum nitride precipitates, the temperature parameters of the normalizing furnace and the strip conveying speed parameters are adjusted in real time; when rolling quality inspection reveals poor plate shape, the roll gap crown parameters of the rolling mill are adjusted in real time.

[0105] Establish a performance database for different batches of strip steel, and automatically optimize processing parameters for different grades and batches of strip steel; when changing the batch of strip steel, call the preset optimal processing parameters for the corresponding batch of strip steel to avoid weld quality problems caused by batch performance differences.

[0106] Example 2: Figure 2 As shown in the figure, an embodiment of the present invention provides a cold-rolled weld protection system suitable for high-grade non-oriented silicon steel, which specifically includes the following modules:

[0107] The protection module is used to pre-purge the surface with inert gas during gradient ultrasonic impact. It adopts a four-section symmetrical gas supply technology to extract dynamic characteristic parameters of the molten pool and continuously adjust the protective gas flow rate, argon-helium-hydrogen ternary mixing ratio and the tilt angle of the annular gas nozzle to treat the weld area.

[0108] Normalization control module: Combined with the processed weld area, key weld parameters are collected, and a coupled model of microstructure evolution and stress evolution is input. Continuous gradient normalization is achieved through an independent temperature control zone. A low-frequency pulsed magnetic field is applied during the heat preservation stage, and segmented gradient cooling technology is used for cooling.

[0109] Micro-segment control module: After continuous gradient normalization is completed, the weld area is divided into continuous micro-segments along the length direction. The micro-segment characteristic parameters of each micro-segment are collected, the maximum safe reduction of each micro-segment is calculated, and the reduction, rolling speed, front and rear tension and roll gap crown of each stand are continuously and dynamically adjusted for rolling. The effect of roll wear is corrected in real time.

[0110] Process control module: For the entire rolling process, data from the three stages of welding, normalizing and rolling are collected, a unique process fingerprint of each weld is extracted, a machine learning fracture risk prediction model is established, and a closed-loop feedback control mechanism for the entire process is formed.

[0111] The above provides a detailed description of one embodiment of the present invention, but the content described is only a preferred embodiment of the present invention and should not be considered as limiting the scope of the present invention. The above formulas are all dimensionless numerical calculations, and the formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world situation. The preset parameters in the formulas are set by those skilled in the art based on actual conditions and historical experience, and can be adjusted according to actual conditions. The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. All equivalent changes and improvements made in accordance with the scope of the present invention should still fall within the patent coverage of the present invention.

Claims

1. A method for protecting cold-rolled welds of high-grade non-oriented silicon steel, characterized in that, Includes the following steps: During gradient ultrasonic impact, surface inert gas pre-purging is performed. A four-section gas supply technology with upper and lower symmetry is adopted to extract dynamic characteristic parameters of the molten pool. The shielding gas flow rate, argon-helium-hydrogen ternary mixing ratio and the tilt angle of the annular gas nozzle are continuously adjusted to treat the weld area. Combined with the processed weld area, key weld parameters are collected and input into the coupled model of microstructure evolution and stress evolution. Continuous gradient normalization is achieved through an independent temperature control zone. A low-frequency pulsed magnetic field is applied during the heat preservation stage, and segmented gradient cooling technology is used for cooling. After continuous gradient normalization is completed, the weld area is divided into continuous micro-segments along the length direction. The characteristic parameters of each micro-segment are collected, the maximum safe reduction of each micro-segment is calculated, and the reduction, rolling speed, front and rear tension and roll gap crown of each stand are continuously and dynamically adjusted for rolling. The influence of roll wear is corrected in real time. For the entire rolling process, data from the three stages of welding, normalizing, and rolling are collected, a unique process fingerprint of each weld is extracted, a machine learning fracture risk prediction model is established, and a closed-loop feedback control mechanism for the entire process is formed.

2. The method for protecting cold-rolled welds of high-grade non-oriented silicon steel according to claim 1, characterized in that, The method for surface inert gas pre-purging is as follows: A surface cleaning device is installed in front of the laser welding station. The surface cleaning device uses a rectangular air nozzle, and the width of the air nozzle is the same as the width of the strip. When the weld marking point reaches the position of the surface purging device, the surface purging device is activated, and high-purity argon gas of a preset concentration is used to pre-purify the areas on both sides of the weld. The purging time is matched with the strip conveying speed. During the gradient ultrasonic impact treatment, the surface purging device continues to work.

3. The method for protecting cold-rolled welds of high-grade non-oriented silicon steel according to claim 1, characterized in that, The method of symmetrical four-zone stratified gas supply is as follows: Four coaxial nested and angle-adjustable annular air nozzles are installed above and below the laser welding gun head. The four upper annular air nozzles correspond to the core reaction zone, solid-liquid interface zone, heat-affected high temperature zone, and heat-affected low temperature zone on the upper surface of the molten pool, while the four lower annular air nozzles correspond to the same four areas on the lower surface of the molten pool. Based on the dynamic characteristic parameters of the molten pool, the protective gas flow rate, argon-helium-hydrogen ternary mixing ratio, and tilt angle of the eight annular gas nozzles are continuously adjusted; the protective gas preheating temperature is continuously adjusted according to the average temperature of the molten pool.

4. The method for protecting cold-rolled welds of high-grade non-oriented silicon steel according to claim 1, characterized in that, The method for extracting dynamic characteristic parameters of the molten pool is as follows: Welding current and welding voltage data are continuously collected by the welding power source; temperature distribution data of the upper and lower surfaces of the molten pool are continuously collected by a dual-color infrared thermometer arranged symmetrically above and below; flow velocity distribution data of the upper and lower surfaces of the molten pool are continuously collected by a high-speed camera combined with particle image velocimetry technology; and welding speed data are continuously collected by an encoder. All collected data are processed for time synchronization, and dynamic compensation is performed for smoke and arc interference in the molten pool detection data. A database of the correspondence between smoke concentration, arc intensity and detection data error is established, and the corresponding error correction value is retrieved based on the real-time collected smoke concentration and arc intensity data. The hysteresis of the protective gas parameter actuator is compensated, the response time characteristics of the regulating valve are measured in advance, and the hysteresis model of the actuator is established; the temperature change rate of the molten pool center, the area change rate of the molten pool, the fluctuation amplitude of the molten pool surface velocity, and the fluctuation amplitude of the welding current are extracted after compensation.

5. The method for protecting cold-rolled welds of high-grade non-oriented silicon steel according to claim 1, characterized in that, The method for inputting the coupled model of tissue evolution and stress evolution is as follows: The weld area was scanned point by point using a laser-induced breakdown spectrometer to collect initial nitrogen content data and silicon / aluminum atomic ratio data at different locations in the weld area. The weld region was scanned point by point using an electron backscatter diffractometer to collect initial microstructure grain size data and grain boundary orientation distribution data at different locations in the weld region. All the collected data were input into a pre-established coupled model of microstructure evolution and stress evolution to simulate and calculate the aluminum nitride precipitation behavior, microstructure evolution process and stress evolution process at the corresponding locations under different normalization temperatures, holding times and cooling rates, and to determine the preset optimal combination of normalization parameters for each location.

6. The method for protecting cold-rolled welds of high-grade non-oriented silicon steel according to claim 1, characterized in that, The method for achieving continuous gradient normalization is as follows: The normalizing furnace is designed with twenty independent temperature control zones. Based on the preset optimal normalizing parameter combination obtained from the model calculation, the temperature of the twenty temperature control zones and the strip conveying speed are linearly adjusted. The spatial distribution of the temperature field in the normalizing furnace is compensated. The temperature field of the twenty temperature control zones of the normalizing furnace is calibrated in advance. A database of the correspondence between the actual temperature and the set temperature at different locations in each temperature control zone is established. The set temperature of each temperature control zone is dynamically adjusted based on the real-time temperature distribution data of each temperature control zone. Dynamic compensation is performed for the holding time due to fluctuations in the strip running speed. Based on the real-time collected strip running speed data, the actual dwell time of each position in the weld area within each temperature control zone is calculated. An equivalent relationship model between the holding time and the normalization temperature is established. Based on the difference between the actual dwell time and the target holding time, the corresponding temperature adjustment amount is calculated.

7. The method for protecting cold-rolled welds of high-grade non-oriented silicon steel according to claim 1, characterized in that, The method for applying the low-frequency pulsed magnetic field is as follows: During the heat preservation stage of normalization treatment, electromagnetic coils are installed simultaneously above and below the weld area; the resistance of the electromagnetic coils as a function of temperature is measured in advance, and a database of the correspondence between coil temperature and magnetic field strength is established. Based on the real-time collected electromagnetic coil operating temperature data, the current correction value at the corresponding temperature is retrieved from the database, and the coil power supply current is dynamically adjusted. To keep the actual magnetic field strength at the set value.

8. The method for protecting cold-rolled welds of high-grade non-oriented silicon steel according to claim 1, characterized in that, The segmented gradient cooling method is as follows: During the cooling stage of normalization treatment, the weld area is divided into the weld center zone, the weld transition zone, and the base metal zone; the weld center zone is cooled at a rate of 18 degrees Celsius per second; the weld transition zone is cooled at a rate of 15 degrees Celsius per second; and the base metal zone is cooled at a rate of 12 degrees Celsius per second.

9. A method for protecting cold-rolled welds of high-grade non-oriented silicon steel according to claim 1, characterized in that, The method for calculating the maximum safe reduction of each micro-segment is as follows: The weld area is divided into continuous micro-segments along its length; the hardness of the weld area is tested point by point using a Vickers hardness tester to collect hardness distribution data; the thickness distribution data of the weld area is collected using a laser thickness gauge; and the internal defect distribution data of the weld area is collected using a phased array ultrasonic flaw detector. Data on the microstructure uniformity and grain boundary distribution of the weld region were collected using an electron backscatter diffraction instrument; the work hardening rate data at different locations in the weld region were determined by a single-pass compression test, and the correlation between the work hardening rate and the amount of deformation was established. Calculate the average hardness, average thickness, internal defect density, microstructure uniformity index, and work hardening rate for each micro-segment. Based on the initial hardness, predicted hardness change, internal defect density, and microstructure uniformity index of each microsegment, the maximum safe reduction that the corresponding microsegment can withstand in each frame is calculated.

10. A method for protecting cold-rolled welds of high-grade non-oriented silicon steel according to claim 1, characterized in that, The method for forming a closed-loop feedback control mechanism throughout the entire process is as follows: A high-precision position encoder is installed on the edge of the strip to collect the position data of the strip in real time; at the welding station, welding current, welding voltage, molten pool temperature distribution, molten pool flow rate distribution, shielding gas parameters and weld appearance images are collected to extract the fingerprint of the welding process. At the normalizing furnace outlet, the hardness distribution, microstructure uniformity, nitrogen content, and residual stress of the weld area are collected to extract the normalizing process fingerprint; at the rolling mill outlet, the surface quality, plate shape, internal defects, and thickness accuracy of the weld area are collected to extract the rolling process fingerprint; the welding process fingerprint, normalizing process fingerprint, rolling process fingerprint, and final quality data of each weld are correlated and stored.

Images