A method and device for continuous annealing of galvannealed steel strip

Abstract

The application discloses a kind of hot galvanizing aluminium magnesium steel strip continuous annealing treatment method and device, by collecting component thickness state before furnace to identify the heat hysteresis demand of each section, in combination with the attenuation characteristics of furnace temperature disturbance when speed regulation, through disturbance response window attenuation fitting, identify heat conduction block position and adjacent candidate are merged, and the real thermal boundary of furnace section is parsed, so that segmented furnace temperature setting is established on the basis of measured state rather than preset section division;For segmented furnace temperature acquisition cross section transverse temperature difference, distinguish transient fluctuation and sustained threshold residence and convert into cross section heat damage amount that can be accumulated and transferred, avoid the influence of local uneven heating history discarded as soon as detection is completed;Accordingly, track the recrystallization evolution process of soaking zone, analyze the degree of grain refinement to evaluate the compliance of mechanical properties, extract cross section temperature gradient before plating for compliant steel strip, collect process parameter range and output control instruction, so that pre-plating determination and microstructure evolution tracking are connected rather than isolated detection.

Description

Technical Field

[0001] This invention relates to the field of metal heat treatment technology, and in particular to a method and apparatus for continuous annealing of hot-dip galvanized aluminum-magnesium steel strip. Background Technology

[0002] Hot-dip galvanized aluminum-magnesium steel strip undergoes multiple process stages in a continuous annealing furnace, including heating, homogenization, and cooling. The final mechanical properties of the steel strip and the quality of the coating bonding largely depend on the precision of furnace temperature control at each stage of the annealing process. Because the composition and thickness of the steel strip often vary between batches, and the production line belt speed is frequently adjusted, the actual heating state of different sections within the furnace is not constant. Traditional control methods that rely on a single set temperature and operate according to a fixed process curve are insufficient to adapt to these batch differences and dynamic disturbances. This can easily lead to uneven heating in the cross-section or width direction of the steel strip, thus affecting the uniformity and consistency of subsequent microstructure transformation.

[0003] Current control methods generally treat the incoming material condition before entering the furnace, temperature fluctuations at each stage within the furnace, the microstructure evolution process, and the final pre-plating judgment as relatively independent links. The detection and control of each stage are mostly completed locally, lacking continuity, making it difficult to effectively transmit the state information accumulated in the preceding stages and reflect it in the judgment criteria of the subsequent stages. This fragmentation means that local anomalies are often not discovered until the product is inspected after leaving the production line. It is impossible to intervene and adjust in a timely manner in the early stages of the process, and it is also difficult to trace the cause of the anomaly to see whether it stems from differences in incoming materials, furnace temperature fluctuations, or insufficient microstructure transformation. Summary of the Invention

[0004] This invention provides a method and apparatus for continuous annealing of hot-dip galvanized aluminum-magnesium steel strip. It aims to integrate multiple processing steps, including collecting the state of the steel strip before it enters the furnace, setting the furnace temperature in segments, monitoring the temperature state of the cross section, tracking the microstructure evolution, and determining compliance before galvanizing. This allows the key state information obtained in the preceding stages to be effectively utilized in subsequent stages, thereby improving the adaptability of the annealing process to differences in incoming materials and operational disturbances, improving the uniformity of the heating state of the steel strip cross section and width direction, and providing process assurance for the stability of the coating quality.

[0005] A first aspect of the present invention discloses a method for continuous annealing of hot-dip galvanized aluminum-magnesium steel strip, comprising:

[0006] The composition data and thickness deviation sequence of the steel strip before it enters the furnace are collected to generate a material specification set. Based on the material specification set, the heat requirements of each annealing section are calculated to obtain a process parameter set.

[0007] Based on the process parameter set, the furnace temperature disturbance amplitude of adjacent furnace sections is collected during unit speed regulation to obtain the furnace temperature disturbance sequence. The thermal boundary offset distribution of each furnace section is analyzed according to the furnace temperature disturbance sequence to form furnace area boundary parameters. The furnace area boundary parameters are used to integrate the furnace temperature setting range of each section to output the segmented furnace temperature setting.

[0008] For the segmented furnace temperature setting, the transverse temperature difference distribution of each furnace segment is collected to obtain the transverse temperature difference sequence. Based on the transverse temperature difference sequence, the residence time of the over-threshold temperature difference in each furnace segment is counted to determine the over-temperature residence time. Based on the over-temperature residence time, the cumulative amount of thermal damage in each furnace segment is calculated to establish the cross-sectional thermal history status.

[0009] Based on the cross-sectional thermal history status tracking, the recrystallization rate is generated at the recrystallization front of the homogenization zone. The grain refinement degree of each homogenization section is analyzed based on the recrystallization rate to obtain the grain size level. Based on the grain size level, mechanical performance indicators are estimated and the compliance degree is evaluated to obtain mechanical performance compliance results.

[0010] For steel strips whose mechanical properties meet the compliance requirements, the temperature gradient of the plating section is extracted to form a plating index. Based on the plating index, the range of pre-plating process parameters is collected to construct a plating parameter set. The pre-plating organization compliance characteristics are verified through the plating parameter set to determine the process control instructions.

[0011] A second aspect of the present invention discloses a continuous annealing apparatus for hot-dip galvanized aluminum-magnesium steel strip, comprising:

[0012] The specification acquisition module is used to collect the composition data and thickness deviation sequence of the steel strip before it enters the furnace to generate a material specification set, and calculate the heat requirements of each annealing section based on the material specification set to obtain a process parameter set.

[0013] The disturbance setting module is used to obtain the furnace temperature disturbance sequence based on the amplitude of the furnace temperature disturbance in adjacent sections when the unit speed is adjusted according to the process parameter set, analyze the thermal boundary offset distribution of each furnace section according to the furnace temperature disturbance sequence to form furnace area boundary parameters, and use the furnace area boundary parameters to integrate the furnace temperature setting range of each section to output the segmented furnace temperature setting.

[0014] The status monitoring module is used to collect the transverse temperature difference distribution of each furnace segment for the segmented furnace temperature setting to obtain the transverse temperature difference sequence, to count the residence time of the over-threshold temperature difference in each furnace segment based on the transverse temperature difference sequence to determine the over-temperature residence time, and to calculate the cumulative amount of thermal damage in each furnace segment based on the over-temperature residence time to establish the cross-sectional thermal history status.

[0015] The crystallization tracking module is used to track the recrystallization front of the homogenization zone based on the thermal history status of the cross section, generate the recrystallization rate, analyze the grain refinement degree of each homogenization section based on the recrystallization rate to obtain the grain size level, and estimate the mechanical performance indicators and evaluate the compliance degree based on the grain size level to obtain the mechanical performance compliance result.

[0016] The instruction output module is used to extract the temperature gradient of the plating section of the steel strip that meets the mechanical performance compliance results to form a plating index, collect the pre-plating process parameter range based on the plating index to construct a plating parameter set, and verify the pre-plating organization compliance characteristics through the plating parameter set to determine the process control instruction.

[0017] The beneficial effects of this invention are reflected in the following points: First, by collecting and analyzing the composition data and thickness deviation of the steel strip before it enters the furnace, the thermal hysteresis distribution characteristics of each annealing section are identified, and the heat demand is calculated accordingly. Then, combined with the furnace temperature disturbance response and attenuation law during the unit speed regulation process, the actual thermal boundary between furnace sections is analyzed, and a segmented furnace temperature setting that matches the actual heat conduction state of the furnace body is output. This allows the differences in incoming materials and the structural characteristics of the furnace body to be specifically reflected in the furnace temperature setting process, rather than relying on a uniform setting value for general coverage. Second, for the continuous monitoring of the transverse temperature difference of the steel strip under segmented furnace temperatures, instantaneous fluctuations and continuous over-threshold residence are distinguished, and the over-threshold residence state is converted into a quantitative accumulation of cross-sectional thermal damage. This allows the historical impact of local uneven heating to be preserved and transmitted to the subsequent microstructure determination process, avoiding the potential risks caused by continuous exposure that are ignored by judging solely based on instantaneous readings. Finally, by tracking the evolution trajectory of the recrystallization front in the soaking zone based on the thermal history of the cross section, the degree of grain refinement is analyzed and the mechanical properties are evaluated accordingly. For compliant steel strips, the temperature gradient state of the cross section before plating is further extracted, the range of process parameters is collected, and control instructions are output. This connects the microstructure evolution analysis with the pre-plating plating suitability determination, allowing the pre-plating control decision to be based on the actual tracking results of the previous furnace temperature and microstructure state, rather than relying solely on isolated judgments from single-point detections before plating. Attached Figure Description

[0018] Figure 1 This is a schematic flowchart of a continuous annealing treatment method for hot-dip galvanized aluminum-magnesium steel strip according to the present invention.

[0019] Figure 2 This is a schematic diagram of the production line equipment layout of the present invention.

[0020] Figure 3 This is a schematic diagram of furnace temperature disturbance attenuation and furnace zone boundary reconstruction according to the present invention.

[0021] Figure 4 This is a structural block diagram of a continuous annealing treatment device for hot-dip galvanized aluminum-magnesium steel strip according to the present invention.

[0022] Wherein: 1-Steel strip; 2-Pre-furnace measurement station; 3-Annealing furnace body; 4-Annealing section; 5-Furnace section boundary temperature sensor group; 6-Furnace section transverse temperature measurement point; 7-Soaking section; 8-Infrared thermal imager; 9-Pre-plating temperature measurement station group; 10-Platement station; 11-Original furnace section boundary; 12-Disturbance attenuation rate; 13-Candidate blocking position; 14-Blocking position after merging; 15-Furnace section boundary position set; 16-Thermal boundary section; 17-Boundary strength significant mark; 18-Segmented furnace temperature setting; 19-Transition margin; 20-Average value of segment disturbance attenuation rate. Detailed Implementation

[0023] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0024] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0025] The technical solutions of the embodiments of this application will be described below.

[0026] like Figure 1 As shown, this embodiment of the invention provides a continuous annealing treatment method for hot-dip galvanized aluminum-magnesium steel strip, including the following steps S101-S105:

[0027] Step S101: Collect the composition data and thickness deviation sequence of the steel strip before it enters the furnace to generate a material specification set, and calculate the heat requirements of each annealing section based on the material specification set to obtain a process parameter set.

[0028] Specifically, material specification sets are generated by collecting composition data and thickness deviation sequences of steel strip before it enters the furnace. For example... Figure 2As shown, the composition data of steel strip 1 before entering the furnace is obtained by scanning each steel coil cross-section point by point along the bandwidth at equal intervals using an X-ray fluorescence spectrometer at the pre-furnace measurement station 2. The interval between scanning points is one-twentieth of the bandwidth. The mass fraction readings of alloying elements such as carbon, silicon, and manganese are linearly corrected based on the comparison results of standard samples. The correction coefficients are recorded individually for each equipment according to the calibration certificate provided by the equipment manufacturer. The systematic deviation between different equipment is corrected separately instead of using the same set of coefficients. The thickness deviation sequence is continuously obtained by an online thickness gauge at a sampling rate of 500Hz. The deviation is the difference between the measured thickness and the nominal thickness in the order. This sampling rate ensures that each rolling position obtains at least three independent readings. The arithmetic mean of the three readings is used as the final value adopted at that position. Single skipped readings are smoothed by the mean and are not directly included in the thickness deviation sequence. Each record in the material specification set consists of three parts arranged sequentially: rolling position coordinates, composition data element distribution, and thickness deviation sequence deviation values. The composition and thickness deviations are aligned with the rolling position in 50mm increments and then merged into the record. This increment is taken from the minimum resolvable spacing of the mill roll gap adjustment; too fine a increment will introduce measurement repeatability errors, while too coarse a increment will miss local compositional abrupt changes. When the difference in element readings between adjacent scanning points at a certain location exceeds the allowable fluctuation range specified in the furnace front specification, a weighted interpolation based on the physical distance between the two effective points is used instead, prioritizing a simple arithmetic mean, as the actual distance from adjacent measuring points to that point is often unequal. The thermal response of each section of the steel strip during annealing is jointly determined by the cross-sectional composition deviation and thickness distribution before entering the furnace. The material specification set must accurately reflect the actual measured state of the batch; the nominal table for the same steel grade cannot replace the sampling of the current batch.

[0029] In some embodiments, the step of calculating the heat demand of each annealing section to obtain a process parameter set based on the material specification set includes: using the material specification set to identify the thickness-oriented thermal hysteresis distribution of each annealing section and obtaining a thickness-oriented thermal hysteresis spectrum; extracting the maximum hysteresis demand of each section based on the thickness-oriented thermal hysteresis spectrum to determine hysteresis compensation parameters; evaluating the degree of furnace heat flow redundancy in each section based on the hysteresis compensation parameters to obtain a heat flow redundancy rate; and calculating the heat demand range of the annealing section based on the heat flow redundancy rate to output the process parameter set.

[0030] Thick-axis thermal hysteresis distribution in each annealing section is identified using the material specification set to obtain the thick-axis thermal hysteresis spectrum. Locations in the material specification set where the absolute value of the thickness deviation exceeds 0.02 mm or the alloy content deviates from the nominal value by more than 5% are marked as high-risk points for thermal hysteresis and calculated separately. For other locations, the actual thermal hysteresis is calculated using the reference thermal hysteresis time τ_0 of annealing section 4 as the initial value, and the actual thermal hysteresis is calculated using the position-by-position formula τ_i = τ_0 × θ_t,i. τ_0 is the reference thermal hysteresis time (in ms) measured during the furnace temperature step experiment in the calibration stage, θ_t,i is the ratio of the measured thickness restored from the deviation to the nominal thickness at the i-th position in the material specification set, and τ_i is the time lag of the core relative to the surface temperature rise at that position. After substituting all positions from the material specification set, the τ_i of each section is arranged into a thermal hysteresis time sequence array according to the rolling position. The upper limit τ_max of τ_i is taken as the 95th percentile of the historical batch τ_i distribution of this section. It is re-statistically calculated every six months with the new batch, and the position of exceeding the limit is marked first. Once the number of exceeding the limit points exceeds 10% of the total number of positions in this section, a heat capacity abnormality alarm is triggered. The alarm section checks the original sampling of the material specification set. If it is confirmed to be an instrument failure, the field of this section is re-sampled and recalculated. Otherwise, it is filled with the average thermal hysteresis of the same section in history. The amplitude of each position of the thickness thermal hysteresis spectrum is normalized by the thermal hysteresis time series array according to the maximum thermal hysteresis of this section, falling between 0 and 1. The closer it is to 1, the closer the position is to the thermal hysteresis peak of this section. The amplitude sequence of each section is arranged in the order of furnace section. The aforementioned separately calculated high-risk thermal hysteresis points have their measured thermal hysteresis values ​​included in the normalization of the maximum thermal hysteresis value for this section. The amplitude is placed in the sequence as usual, with an additional high-risk marker added. This allows them to both compete for the main peak as candidates and be distinguished from regular points during the next sliding window peak selection process. The marker is present along with the amplitude value, allowing the peak selection process to determine whether the main peak falls at a high-risk position. The coarse-grained setting of the section-level furnace temperature cannot cover the thermal hysteresis dispersion caused by the cross-sectional composition and thickness. The thickness-oriented thermal hysteresis spectrum spreads it out position by position for subsequent section-specific compensation.

[0031] The hysteresis compensation parameters are determined by extracting the maximum hysteresis demand for each segment based on the thickness-oriented thermal hysteresis spectrum. The dispersion of amplitude values ​​in each segment of the thickness-oriented thermal hysteresis spectrum means that high and low hysteresis points coexist within the cross-section. If heat is injected according to the average value of the segment, the high hysteresis points will be underheated for a long time, and the low hysteresis points will be overheated for a long time. Therefore, the peaks of the amplitude sequence of each segment of the thickness-oriented thermal hysteresis spectrum are obtained by scanning with a sliding window. The window width is one-tenth of the length of the sequence of this segment and no less than five frames. If it is too narrow, a single noise peak will be mistaken for the main peak. If it is too wide, two adjacent real hysteresis peaks will be merged into the same window. The frame with the largest amplitude value in the window is recorded as the candidate peak. When there are multiple candidates, the one with the largest global value is taken as the main peak. The position of the main peak corresponds to the original thermal hysteresis amount τ_peak, which is the maximum hysteresis demand of this segment. τ_peak is converted to seconds and multiplied by the furnace segment heat flow reference value Q_ref (in watts, taken from the rated heating power calibration of this furnace segment) to obtain the maximum compensation heat Q_comp (in cokes). Candidate peaks with amplitudes lower than half the average amplitude of the thickness-oriented thermal hysteresis spectrum in this section are discarded as false peaks. False peaks are often caused by scaling at the temperature measurement point and an increased probability after continuous operation for more than two weeks. When all candidates are false peaks, the average maximum thermal hysteresis value of the historical batches is used to replace τ_peak for this section, and it is recorded as the estimation source in the hysteresis compensation parameters to distinguish it from the measured peaks. The hysteresis compensation parameters contain two types of fields: the heat compensation field is filled in Q_comp for each section according to the furnace section number, and the density correction field is filled in the ratio of the difference between τ_peak and the baseline thermal hysteresis time τ_0 divided by the furnace section length. The dimension of this ratio is time per meter, reflecting the hysteresis increment per unit furnace section length. A higher latter indicates that the thermal response of this section is more sensitive to changes in thickness, and the actual effectiveness of the same compensation amount is correspondingly discounted. The hysteresis compensation parameters thus reserve a larger margin for this type of section in the next step.

[0032] The heat flow redundancy rate is obtained by evaluating the degree of heat flow redundancy in each section of the furnace based on the lag compensation parameter. The difference between the lag compensation parameter Q_comp of each section and the upper limit of the output heat Q_max of the corresponding furnace section is the available margin. The margin is divided into three levels according to the absolute value of the margin relative to the nominal demand of the section: more than 40% is high redundancy, 20% to 40% is medium redundancy, and less than 20% is low redundancy. The weights of the three levels, 1.2, 1.0, and 0.8, are taken from the fitting of the furnace condition debugging records of the production line for three consecutive years and are refitted every three years with new data. The weight multiplied by the absolute value of the margin is the redundancy quantification value of each section. Multiplying by the weight before comparison is to amplify the margin of the high redundancy section by a larger coefficient, so that it will take priority in bearing the furnace condition fluctuations in subsequent calculations. For example, when the incoming material becomes thicker and the heat demand increases sharply, the high redundancy section will give up the margin first and the low redundancy section will not be used for the time being. Q_max is the product of the gas heating power of the corresponding furnace section and the transit time of the steel strip furnace section. Since the nozzle layout and rated flow rate are only changed during major overhauls, they are recalibrated after the overhaul and historical calibration values ​​are used during normal operation. If the gas flow sensor readings differ from each other by more than the upper limit of the factory accuracy specification for three consecutive times, it is judged as a fault. During the fault period, Q_max is reduced by 95% of the historical calibration value of this furnace section, and the redundancy quantization value of this section is marked with a sensor fault and the original reading at the trigger time is attached for reference. The heat flow redundancy rate field for each section consists of the redundancy quantization value and the gear number, arranged by furnace section. For sections with a large hysteresis compensation parameter density correction field, the redundancy quantization value is multiplied by the density correction field and divided by its historical maximum value, and the ratio is written into the heat flow redundancy rate. Because the thermal response of this type of section is nonlinear, the adjustable space corresponding to the same margin is not the same. Physical quantities with different dimensions are classified into the gear sequence in this way, and the redundancy level can be directly compared horizontally.

[0033] The process parameter set is output based on the heat flux redundancy rate to calculate the heat demand range of the annealing section. Each heat flux redundancy rate level corresponds to a baseline heat demand Q_req, which is obtained from the median actual heat supply of historical qualified batches for that level. The lower limit of the section heat demand is extended downward from Q_req according to the redundancy rate—the higher the redundancy, the larger the compressible space, and the lower the lower limit. The upper limit is extended upward from Q_req by multiplying the redundancy rate by a safety factor. This safety factor is taken as 1.20, 1.15, and 1.10 for high, medium, and low redundancy levels, respectively. The higher the redundancy, the larger the allowable upward floating space and the larger the corresponding safety factor. The value is derived from the regression of the historical overshoot amplitude of each level during the commissioning phase rather than a general empirical value. The leveling rather than taking a single coefficient is because even if the high redundancy section floats too high, there is still a margin to absorb it, while the low redundancy section must be tightened to prevent exceeding Q_max. The heat collection field of the process parameters is entered with the upper and lower limits of each section according to the furnace section number. The span between the two limits is the heat adjustment margin of that section. The arithmetic mean of the upper and lower limits is also stored with the field, serving as the nominal setpoint for the heat of that section. This is directly used for subsequent furnace temperature conversion without the need for temporary recalculation. The median is updated batch by batch with the upper and lower limits, and the two are always stored in pairs to prevent mismatch. For sections with heat flux redundancy carrying sensor fault markers, the upper limit of the tolerance in the process parameter set is expanded by 15%, with the expansion taken from the statistical upper bound of the actual deviation of Q_max in historical fault batches. After the process parameter set is summarized, cross-section consistency verification is performed. When the difference between the lower limits of adjacent sections exceeds the nominal requirement by 15% (this limit is taken from the historical maximum fluctuation under normal operating conditions of adjacent furnace sections), the average of the two lower limits is used for correction, and the original value is retained for traceability. Annealing heat should change batch by batch with cross-sectional state and furnace conditions. Single-point settings cannot cover the range of each section. The process parameter set allows cross-sectional differences to enter the parameter system, so they are no longer flattened by the section average.

[0034] Step S102: Based on the process parameter set, the furnace temperature disturbance amplitude of adjacent sections is collected during unit speed regulation to obtain the furnace temperature disturbance sequence. The thermal boundary offset distribution of each furnace section is analyzed according to the furnace temperature disturbance sequence to form furnace area boundary parameters. The furnace area boundary parameters are used to integrate the furnace temperature setting range of each section to output the segmented furnace temperature setting.

[0035] Specifically, the furnace temperature disturbance sequence is obtained by collecting the amplitude of furnace temperature disturbances in adjacent sections during unit speed regulation based on the process parameter set. At the moment of unit speed regulation, the actual furnace temperature of each section briefly shifts with the belt speed change. The shift amplitude is calculated using the formula ΔT_j = T_j - T_set, where T_j is the measured furnace temperature value of the j-th sampling, T_set is the reference furnace temperature for that section converted from the median of the heat range of the process parameter set, and ΔT_j is the disturbance amplitude of that sampling. This reference furnace temperature is the comparison benchmark for the shift calculation. During acquisition, each furnace section is independently compared with its T_set as the benchmark. For sections with sensor fault markers, the upper tolerance limit is maintained at 15% of the expanded process parameter set, and ΔT_j for that section is interpreted according to the corresponding relaxed threshold. The start and stop of the acquisition window are also triggered independently segment by segment based on the section boundaries defined by the process parameter set, ensuring that the disturbance amplitude of each segment is anchored to its own heat benchmark rather than a uniform target for the entire furnace. After each speed adjustment command is issued, the furnace temperature sensor continuously samples at 10Hz. This frequency allows for dozens of frames to be obtained within a complete speed adjustment transition. Five furnace section boundary temperature sensor groups are installed on both sides of the boundary between adjacent sections to record the measured values ​​on both sides. The installation spacing does not exceed 5% of the furnace section length to ensure that the readings closely reflect the true boundary conditions rather than being mixed with the temperature in the middle of the furnace section. The furnace temperature disturbance sequence records ΔT_j for each section moment by moment. Data from adjacent sections are paired according to section number and archived according to the sampling time. The sequence covers the entire speed adjustment until the furnace temperature falls back to near the reference furnace temperature. Falling back is defined as 20 consecutive samples of ΔT_j falling within the noise threshold. If five consecutive samples of a section differ from the previous mean by more than three times the standard deviation, it is treated as a sensor jump, and the batch is replaced by the previous mean and written into the furnace temperature disturbance sequence. If the jump accounts for more than 5% of the samples in this batch, the entire batch is discarded and resampled after speed adjustment.

[0036] In some embodiments, the step of analyzing the thermal boundary offset distribution of each furnace segment based on the furnace temperature disturbance sequence to form furnace area boundary parameters includes: screening the speed regulation response window of each furnace segment to obtain the disturbance response duration based on the furnace temperature disturbance sequence; fitting the disturbance amplitude attenuation law of each furnace segment based on the disturbance response duration to obtain the disturbance attenuation rate; identifying the heat conduction blocking positions between each furnace segment based on the disturbance attenuation rate to form a furnace segment boundary position set; and collecting the thermal boundary offset range of each furnace segment based on the furnace segment boundary position set to construct furnace area boundary parameters.

[0037] The speed regulation response window for each furnace section is screened to obtain the disturbance response duration based on the furnace temperature disturbance sequence. The first moment when the absolute value of ΔT_j in the furnace temperature disturbance sequence crosses the noise threshold of the section three consecutive times is recorded as the response start. The noise threshold is twice the standard deviation of ΔT_j in the steady-state phase of that section. Using twice, rather than one, is to exclude steady-state random fluctuations from the response judgment and avoid mistakenly identifying noise spikes as the start of the speed regulation response. This threshold is pre-calibrated from historical batches without speed regulation and reviewed quarterly. The first moment when ΔT_j falls back to within the threshold five consecutive times is recorded as the response end. The requirement of five consecutive falls excludes occasional rebounds during the fall. The sampling points between these two moments constitute the speed regulation response window. The disturbance response duration is the physical duration obtained by multiplying the number of sampling points in this window by the sampling period. Each section of the furnace temperature disturbance sequence is screened independently, and the start and end times may be asynchronous. The disturbance response duration of each section is archived by section number and accompanied by an absolute timestamp for cross-checking. The threshold for determining the initial hysteresis is set to 2 seconds after the speed adjustment command. This value is taken from the longest historical delay in a normal furnace section from the issuance of the command to the measurable deviation of the furnace temperature. If the initial hysteresis does not occur within this time, the sensor response in that section is recorded as hysteresis, and the disturbance response duration is changed to the value of the same section in the adjacent historical batch, while retaining the speed adjustment amplitude at the trigger time for attribution. If the end of the sampling range is not observed, the disturbance response duration is temporarily truncated at the end time and labeled with an extension attribute. At this point, the disturbance response duration of each segment is labeled in three categories: those with normal start and end are labeled with a regular attribute, those with truncated end are labeled with an extension attribute, and those with hysteresis substitution are labeled with a substitution attribute, along with the adjacent batch number used. The three types of attributes are archived together with the start and end sampling points. The next step of attenuation fitting determines the window selection method based on these attributes. Regular and extension segments are truncated according to their own start and end, substitution segments directly use the adjacent batch window, and segments with missing attributes are not included in the fitting.

[0038] The disturbance attenuation rate is obtained by fitting the attenuation law of the disturbance amplitude in each furnace section based on the disturbance response duration. Within the window defined by the disturbance response duration, the absolute value of ΔT_j decreases exponentially with the increase of the sampling sequence number. This window is segment by segment taken from the start and end sampling points of the disturbance response duration. The window labeled with the extension attribute still ends at the end cutoff point. For segments labeled with sensor response hysteresis and using the value of the adjacent batch, the original window boundary of the batch is directly used without re-cutting, so that the fitting object corresponds segment by segment with the source of the window without disconnection. Fitting is performed using the formula ΔT_n = ΔT_0 × e^(-λ × n), where ΔT_0 is the initial disturbance amplitude, n is the sampling sequence number within the window, and λ is the attenuation coefficient to be determined. After taking the natural logarithm of ΔT_n, the exponential relationship becomes linear with respect to n. The slope of ln|ΔT_n| versus n is calculated using least squares, and the negative value is the initial value of λ. Before fitting, tail points with amplitudes below the noise threshold are removed because the tail amplitudes are on the same order of magnitude as the noise, and taking the logarithm will amplify their relative error and significantly skew the slope. Figure 3As shown, the disturbance attenuation rate 12 is the fitted value of λ over the entire window, with the unit being the attenuation ratio per sampling period. When the disturbance response time is too short and there are fewer than eight effective points in the window, the least squares estimation of the coefficients becomes unstable, and the low freedom makes the residuals overly sensitive to individual points. The disturbance attenuation rate is then directly calculated from the ratio of the difference between the first and last points of the window and labeled with simplified algorithm attributes, without performing least squares. If the sum of squares of the fitted residuals exceeds 30% of the variance of ΔT_n, it is treated as distorted according to the attenuation law. The disturbance attenuation rate of the distorted section is replaced by the weighted average of the attenuation rates of the two adjacent furnace sections. The weights are distributed according to the reciprocal of the furnace section distance, with the closer the furnace section, the greater the weight, and the closer its characteristics are to the distorted section. The original residuals are also retained. The disturbance attenuation rates of each section are arranged by furnace section. The larger the value, the stronger the self-recovery of the section to speed regulation disturbances. If there is an order of magnitude jump between adjacent furnace sections, it is mostly caused by a sudden change in the furnace structure or the state of the insulation layer.

[0039] Based on the disturbance attenuation rate, the locations of heat conduction blockages between furnace sections are identified to form a set of furnace section boundary locations. When the absolute value of the difference between the disturbance attenuation rates of two adjacent furnace sections exceeds 40% of their average, the adjacent location is judged as a significant heat conduction blockage. The 40% threshold is taken from the upper limit of the batch-to-batch fluctuation of the attenuation rate under normal operating conditions within the same furnace section. If it is below this threshold, the difference falls within the range of measurement and operating condition fluctuations and is not judged as a blockage. If it exceeds this threshold, it means that there is a structural restriction on the heat transfer path on both sides. If it does not exceed this threshold, it is considered as continuous heat conduction and is not included in the candidate. The comparison covers all adjacent furnace sections one by one. After the above comparison is completed, the coordinates of the candidate blockage location 13 are registered according to the furnace section sequence. The coordinates are taken as the actual geometric position of the junction of the two furnace sections, rather than the furnace section number corresponding to the original furnace section boundary 11. The coordinates are cumulatively calibrated along the axis with the furnace inlet as the origin. For candidate furnace segments with an interval of less than three furnace segments, the local attenuation rate fluctuation is treated as a local attenuation rate fluctuation rather than a true blockage. This is because a true structural blockage corresponds to the entire insulation or furnace shell difference, and its impact range generally spans more than three furnace segments. If the interval is too close, it is more likely due to the fitting error of the attenuation rate at a single point. The two candidate coordinates are merged by taking the geometric midpoint and merged into a blockage position 14. The two coordinates before merging are kept on file for traceability. The furnace segment boundary position set 15 records the confirmed blockage coordinates one by one, arranged along the axis. The distance between adjacent coordinates is the coverage range of a complete thermal boundary segment. When the disturbance attenuation rate of a furnace segment is unreliable due to sensor failure, the comparison of the adjacent units on both sides is temporarily suspended. The attenuation rate of that segment is replaced by the historical average before being included. Before the comparison is completed, the corresponding interval of the furnace segment boundary position set uses the coordinates of the previous batch. The annealing furnace body 3 is spliced ​​together by multiple independent heating units along the axis. The structural differences at the splicing points make the heat conduction weaker, and these physical boundaries cannot be located by furnace segment numbers alone.

[0040] The furnace area boundary parameters are constructed by collecting the thermal boundary offset range of each furnace segment based on the set of furnace segment boundary locations. The interval between two adjacent blocking coordinates in the set of furnace segment boundary locations is defined as a thermal boundary segment (16). Each furnace segment within a segment shares the same set of thermal conduction characteristics. The difference between the coordinates at both ends and the thermal boundary values ​​determines the coverage area. Each segment is divided to cover the entire length of the furnace without gaps. The furnace area boundary parameters for each segment include the segment's start and end coordinates, a list of furnace segment numbers within the segment, and the average disturbance attenuation rate (20) for that segment. The average is the arithmetic mean of the attenuation rates of all furnace segments within the segment. The average is used instead of individual furnace segment values ​​because furnace segments within the same thermal boundary segment share thermal conduction characteristics, and their attenuation rates should be similar. Using the average can smooth out the fitting residuals of a single furnace segment and obtain more stable segment-level characteristics. When a thermal boundary section covers less than two furnace sections, it is treated as an overly dense boundary and merged with the adjacent shorter section. After merging, the boundary coordinates are recalculated as the endpoints of the union of the coordinates of the two sections. The intermediate coordinates corresponding to the furnace section boundary positions are then deleted, and the original values ​​are retained. When the average disturbance attenuation rate of a certain section of the furnace area boundary parameter (20) differs from that of the adjacent section by more than 50%, the boundary is marked with a significant boundary strength mark (17). The 50% threshold is higher than the aforementioned 40% blocking threshold for a single furnace section. Since the section-level average value has already suppressed noise, a difference of more than 50% is still required to be considered a strong boundary, indicating that the heat conduction characteristics change drastically here and the coordinate accuracy requirement is higher than that of ordinary sections. The next step setting will be based on this mark to reserve a wider transition. The design life and maintenance cycle of each heating unit along the furnace line are not consistent. After long-term operation, the actual thermal boundary often deviates from the original furnace section boundary. The furnace area boundary parameters are redefined based on the measured disturbance characteristics, so that the division of the temperature setting range fits the actual heat conduction state of the furnace body.

[0041] The furnace temperature setting range for each section is integrated using the furnace zone boundary parameters to output segmented furnace temperature settings. The start and end coordinates of each section of the furnace zone boundary parameters and the average disturbance attenuation rate are mapped to the furnace temperature setting fluctuation range for that section one by one. The larger the average attenuation rate, the stronger the thermal inertia and the faster the self-recovery from disturbances, thus allowing for a wider fluctuation without exceeding the limit, resulting in a wider fluctuation range. The upper and lower limits are symmetrically expanded around the reference temperature converted from the median of the original heat range of the process parameters. The conversion uses the heat-temperature conversion coefficient calibrated for that furnace section to convert the heat demand into an equivalent furnace temperature. The expansion ratio is determined by the ratio of the average attenuation rate of that section to the average attenuation rate of the entire furnace. When the ratio is less than one, the nominal fluctuation is used; when it is greater than one, the range is widened proportionally. The segmented furnace temperature setting 18 is output segment by segment according to the furnace zone boundary parameter division. Each furnace section in the same section shares a set of upper and lower limits. Step-like temperature differences are allowed between adjacent sections, and the upper limit of the step amplitude increases as the difference in the average attenuation rates between the two sections increases. At the boundary of the furnace zone, where the boundary strength is marked significantly (17), the furnace temperature is set in segments (18), with an additional 10% transition margin (19) on both sides of the steps. This margin is calculated proportionally to the difference in reference temperatures on both sides of the boundary to avoid localized thermal stress concentration at the steel strip boundary caused by excessively steep steps. When a section has too many merged and covered furnace sections, the furnace temperature is set in segments, and each furnace section within the segment is inversely weighted and fine-tuned according to its distance from the center of the segment. The closer to the boundary, the closer to the level of the adjacent segment. The fine-tuning range is locked within 3% of the original value to avoid introducing new steps through the fine-tuning itself. The steel strip needs to continuously pass through multiple heating units. If the furnace temperature is uniform throughout, it will not keep up with the differences in heat conduction between units. The furnace temperature is set in segments, gradually widening the range so that the temperature distribution conforms to the thermal characteristics of the furnace body.

[0042] Step S103: For the segmented furnace temperature setting, collect the transverse temperature difference distribution of each furnace segment to obtain the transverse temperature difference sequence, and determine the over-temperature residence time by statistically analyzing the over-temperature residence time of each furnace segment. Based on the over-temperature residence time, calculate the cumulative amount of thermal damage of each furnace segment to establish the cross-sectional thermal history status.

[0043] Specifically, a transverse temperature difference sequence is obtained by collecting the transverse temperature difference distribution of each furnace segment for segmented furnace temperature settings. The cross-section of the steel strip is affected by the difference in heat dissipation at the edges and heat transfer in the center within the furnace, resulting in inherent temperature inconsistencies between the edges and center of the same furnace segment. At least three transverse temperature measurement points (6) are distributed along the width of each furnace segment, covering both sides and the center, and are installed close to the strip surface to reduce furnace gas convection errors. The measured values ​​at each temperature measurement point are synchronously compared with the set values ​​of the corresponding segments of the segmented furnace temperature settings. The transverse temperature difference is calculated using the formula ΔT_lat,t = T_edge,t - T_center,t, where T_edge,t is the measured temperature at the edge at time t, T_center,t is the measured temperature at the center at the same time, and ΔT_lat,t is the transverse temperature difference at that time. The transverse temperature difference sequence is stored separately for each time ΔT_lat,t according to the furnace segment number. The sampling frequency is synchronized with the furnace temperature control cycle of the corresponding segment of the segmented furnace temperature settings, ensuring that the temperature difference readings and the set temperatures are within the same control cycle. The reference for setting the transverse temperature measurement points in each furnace section is taken from the median of a set of upper and lower limits of the segmented furnace temperature setting for that section. All furnace sections in the same section share the same reference. At the junction of the segmented furnace temperature setting and the boundary strength mark, the transverse temperature difference between the furnace sections on both sides is widened by a reserved transition margin of 10% to avoid the step transition zone being mistakenly counted as transverse unevenness. If the readings of a certain temperature measurement point differ from the adjacent points by more than 15℃ for ten consecutive times, 15℃ is taken from the physical upper limit of the normal width direction temperature difference. If it exceeds this, it is considered to be outside the range that can be explained by heat conduction. It is treated as scale blockage or probe offset and is removed. The readings of the adjacent points on the same side are used instead for calculation. When the cumulative number of removed events exceeds one-third of the temperature measurement points in a single furnace section, the transverse temperature difference sequence of the entire furnace section is suspended and manually checked. The check results are entered into the equipment maintenance record. The heat dissipation area of ​​the strip steel edge is large and the radiation loss is large. It is normal for the edge temperature to be lower than the middle. The segmented furnace temperature setting only gives a single target value. The transverse temperature difference sequence explicitly records this layer of unevenness in the width direction.

[0044] In some embodiments, determining the over-temperature residence time based on the statistical analysis of the over-threshold temperature difference residence period of each furnace segment according to the transverse temperature difference sequence includes: identifying the continuous over-threshold temperature difference period of each furnace segment based on the transverse temperature difference sequence to generate an over-threshold period record; extracting the duration of the over-threshold period of each furnace segment based on the over-threshold period record to establish an over-threshold duration sequence; evaluating the over-threshold duration sequence to determine the over-temperature judgment result for each furnace segment exceeding the residence time limit; and accumulating the effective over-threshold residence time of each furnace segment based on the over-temperature judgment result to determine the over-temperature residence time.

[0045] Based on the transverse temperature difference sequence, records of periods where the temperature difference in each furnace section continuously exceeds the threshold are generated. The first moment when the absolute value of ΔT_lat,t in the transverse temperature difference sequence crosses the allowable transverse temperature difference threshold for that furnace section is recorded as the start of the threshold exceedance. The allowable threshold is graded according to the steel grade and thickness specification; thinner specifications have smaller heat capacity and faster temperature changes under the same temperature difference, so a stricter value is taken. The threshold table is reviewed quarterly according to process requirements. The first moment when ΔT_lat,t continuously falls back to within the threshold and remains within three sampling cycles is recorded as the end of the threshold exceedance. Maintaining three cycles is to avoid repeatedly judging the end of the threshold exceedance and restarting when the temperature difference fluctuates near the threshold. The continuous sampling interval between two moments constitutes one threshold exceedance period. After scanning the transverse temperature difference sequence moment by moment, all threshold exceedance periods are registered in the threshold exceedance period record according to their start and end times; those that only last for one sampling cycle are treated as instantaneous fluctuations and are not included in the threshold exceedance period record. Each out-of-threshold period record stores the start and end times, corresponding furnace section number, and peak value of ΔT_lat,t for that period. The peak value is the single reading with the largest absolute value within the period. Each record also indicates the actual number of sampling points and the theoretically expected number within that period. If the two do not match, it indicates that there was a sampling interruption in the middle of the period, and the record is marked as continuously interrupted. When converting the duration downstream, this mark is used to determine whether to accumulate the data segment by segment according to the actual sampling interval or to directly subtract it, so as to avoid mistakenly including the interruption gap in the duration. If a furnace section accumulates more than eight out-of-threshold times in a single speed regulation batch, and the eight times are taken from the historical quartile of the number of out-of-threshold times in a single batch of normal furnace sections, it is treated as an abnormal response of the transverse temperature difference control system. The out-of-threshold period record uniformly marks all entries for that furnace section with a system abnormality mark to distinguish it from normal fluctuation entries. The temperature field in the width direction of the steel strip is inherently uneven, and the transverse temperature difference caused by the edge effect exists for a long time. Only recording the instantaneous out-of-threshold readings cannot show whether it has become a continuous risk. Therefore, the start and end boundaries of the out-of-threshold period become the basic unit for risk assessment.

[0046] Based on the records of out-of-threshold periods, the duration of each furnace segment's out-of-threshold period is extracted to establish an out-of-threshold duration sequence. The start and end times of each out-of-threshold period in the out-of-threshold period record are subtracted and then multiplied by the sampling period to convert it into the physical duration of the event, measured in seconds. For events spanning sampling breakpoints, the duration is accumulated segment by segment according to the actual sampling interval, rather than simply subtracting, because sampling may be briefly interrupted during equipment maintenance or data re-acquisition; directly subtracting the start and end times would include the interruption gap in the duration. The out-of-threshold duration sequence compiles the converted durations of all entries in the out-of-threshold period records segment by segment, and archives them in chronological order of event occurrence. Entries marked with system anomalies have their durations listed separately in the out-of-threshold duration sequence, with the source attribute listed separately from normal fluctuation entries, but not deleted, for tracing equipment failure history. The tracing record also includes the handling conclusion and signature confirmation for that particular anomaly. If the duration of a single event exceeds three times the reference duration corresponding to the allowable threshold for that furnace segment, the reference duration is the maximum allowable local overheating time for that steel grade. This is treated as an extreme event, and an additional copy of the original ΔT_lat,t waveform is made for its start and end times and checked separately. This copy is independent of the main over-threshold duration sequence and is not included in the accumulation. When there are zero over-threshold events within a furnace segment's statistical period, the over-threshold duration sequence outputs an empty sequence for that furnace segment instead of a default value, and this is noted in the furnace segment's periodic metadata. This clearly distinguishes the empty sequence from the uncollected state in the data structure, preventing empty values ​​from being misjudged as sensor failure. The duration of transverse temperature difference exceeding the threshold directly affects the depth of impact on the internal structure of the material at the locally heated area. The potential impact of instantaneous over-threshold and long-term over-threshold differs significantly; simply recording whether an over-threshold occurs without recording the duration misses this crucial dimension.

[0047] The over-threshold duration sequence is used to evaluate the dwell time of each furnace segment and generate over-temperature judgment results. The duration of each furnace segment in the over-threshold duration sequence is compared with the over-threshold threshold for that furnace segment. The threshold is determined based on the maximum allowable local overheating duration for the steel grade, which is obtained through regression analysis of experimental data on microstructure or surface defects caused by overheating of that steel grade. Different steel grades have different values: those reaching or exceeding the threshold are judged as valid over-temperature events and included in subsequent damage accumulation; those not reaching the threshold are judged as minor fluctuation events and are only stored and not included in the calculation. Entries in the over-threshold duration sequence marked with a system anomaly are not included in this round of judgment; their results are uniformly recorded as pending verification and will be re-judged after the system anomaly in that furnace segment is resolved and manually confirmed. Each over-temperature judgment result records the judgment category, corresponding furnace segment number, and original duration for each event. Valid over-temperature and minor fluctuation events are stored separately to distinguish statistical calibers. If the interval between two consecutive effective over-temperature events in the same furnace section is shorter than one complete cycle of that furnace section (the cycle being the travel time of the steel strip from the furnace section entrance to the exit), it indicates that the same steel strip has not yet left the section before over-temperature recurrence. These two over-temperature events are combined as continuous over-temperature events, and the over-temperature determination result is marked with a continuous attribute and the number of events before merging is recorded. If the original waveform of an extreme event shows no decline in ΔT_lat,t during the peak duration, the event category is further marked with a continuous, uninterrupted attribute to distinguish it from events with a brief decline but still exceeding the overall limit. Whether the lateral temperature difference exceeds the limit cannot be determined solely by whether the instantaneous value exceeds the limit; the same peak value but different durations result in significantly different degrees of thermal impact. Over-temperature determination results rely on classification to separate truly continuous events from brief fluctuations.

[0048] The over-temperature dwell time is determined by accumulating the effective over-threshold dwell time for each furnace segment based on the over-temperature judgment results. Items categorized as effective over-temperature in the over-temperature judgment results are grouped by furnace segment, and their original durations are accumulated one by one. For multiple events with continuous attributes, their individual durations are first restored to their original number of events before merging, and then summed to avoid duplicate accumulation of multiple records of continuous over-temperature in the same segment, thus artificially inflating the total dwell time. During restoration, the sum of each individual duration after restoration is simultaneously verified to ensure consistency with the original combined total duration of the over-temperature judgment results; if inconsistent, the merging process is reviewed. Minor fluctuations and items awaiting verification are not included in this round of accumulation. Items awaiting verification are continuously excluded until the system anomaly in the furnace segment is resolved. During the exclusion period, the accumulated result for that furnace segment is temporarily incomplete and includes the number of excluded items. The over-temperature dwell time is recorded as the total accumulated duration for each furnace segment, measured in seconds. Furnace segments with no effective over-temperature events within a statistical period are recorded as zero values ​​rather than empty, clearly distinguishing zero values ​​from unstatistical states. When accumulating events with the attribute of continuous and uninterrupted operation, the original duration is fully included without reduction, because continuous and uninterrupted operation means that the thermal effect is continuous during that period, and calculating the full duration will not underestimate its damage. If the accumulated overheating residence time of a certain furnace section exceeds 60% of the total duration of the statistical period, 60% is taken from the historical upper limit of the residence rate of normal furnace sections. Exceeding this limit indicates that the section is in overheating for most of the time, and is treated as almost continuous overheating. The overheating residence time is marked as continuous overheating for that furnace section, and the distribution and frequency of non-overheating intervals are recorded. The difference in the length of overheating residence time of each furnace section reflects the control level of the transverse temperature uniformity of that furnace section; the greater the difference, the more room for improvement in the furnace atmosphere distribution or nozzle layout.

[0049] The thermal history of each furnace section is established by calculating the cumulative thermal damage based on the duration of overheating. The duration of overheating is substituted into the formula H=Σ(Δt_i×ΔT_i) for each furnace section to calculate the cumulative thermal damage. Δt_i is the duration of the i-th effective overheating event in that furnace section (in seconds), and ΔT_i is the peak value of the lateral temperature difference corresponding to that event (in °C). The summation covers all effective overheating events within the statistical period of that furnace section. The dimension of H is °C·s, not thermodynamic energy, but an equivalent overheating exposure. Events with different durations and temperature differences are scaled down to the same scale, with longer durations or larger temperature differences being linearly superimposed for easier lateral sorting within the same steel grade. Different steel grades have different alloy compositions, and the actual microscopic damage corresponding to the same H is not equivalent. Therefore, before judgment, H is first normalized by dividing by the average H value of historically confirmed scrapped batches of that steel grade. The normalized relative value is then compared with a unified damage upper limit, with an upper limit of 1.0 corresponding to the average damage level of historical scrapped batches. The cross-sectional thermal history status stores the original H value and normalized relative value for each furnace segment. Furnace segments with normalized relative values ​​exceeding the upper limit are associated with a damage exceeding the limit indicator in the cross-sectional thermal history status, and downstream soaking section measurements are prioritized based on this indicator. The damage exceeding the limit indicator not only records whether the limit is exceeded, but also includes the specific multiple by which the normalized relative value exceeds the upper limit as the degree of exceeding the limit; the more the limit is exceeded, the greater the degree value. Downstream, the radiation interpretation threshold is relaxed according to this degree classification—the more severe the exceedance and the more significant the potential surface oxidation color difference, the greater the relaxation, rather than a uniform relaxation for all exceeding segments. If a furnace segment experiences some over-temperature events due to system anomalies that have not been verified and included in the accumulation for an extended period, the data for that furnace segment in the cross-sectional thermal history status will be incomplete, indicating that the actual damage may be higher than the currently recorded. The thermal effects of transverse temperature unevenness during annealing do not disappear instantaneously; the cross-sectional thermal history status fixes this accumulated amount, providing a unified comparative scale for the differences in thermal experience between different furnace segments of the same cross section.

[0050] Step S104: Based on the cross-sectional thermal history status tracking, the recrystallization rate is generated at the recrystallization front of the homogenization zone. The grain size level is obtained by analyzing the grain refinement degree of each homogenization section according to the recrystallization rate. Based on the grain size level, the mechanical performance indicators are estimated and the compliance degree is evaluated to obtain the mechanical performance compliance result.

[0051] In some embodiments, the step of tracking the recrystallization front of the homogenized zone based on the cross-sectional thermal history status to generate the recrystallization rate includes: determining the surface thermal radiation intensity distribution of each homogenized section based on the cross-sectional thermal history status to establish a thermal radiation distribution map; identifying the low-radiation zone boundary of each homogenized section from the thermal radiation distribution map to determine the recrystallization lag zone; tracking the migration dynamics of the low-radiation boundary of each homogenized section based on the recrystallization lag zone to form a front migration trajectory; and calculating the recrystallization area ratio of the homogenized section through the front migration trajectory to generate the recrystallization rate.

[0052] A thermal radiation distribution map is established by measuring the surface thermal radiation intensity distribution of each soaking section based on the cross-sectional thermal history. Soaking sections 7, which belong to furnace sections with damage exceeding limits in the cross-sectional thermal history, are prioritized for measurement. An infrared thermal imager 8 scans the surface of the soaking section point by point along the bandwidth, with each scan covering the entire bandwidth and recording the radiation intensity at each point with a spatial resolution of 0.5 mm. The intensity of each point obtained from each frame scan is indexed by both bandwidth coordinates and frame time. Readings of the same coordinate across frames are vertically aligned and horizontally cover the entire bandwidth, thus forming a two-dimensional radiation intensity array for subsequent retrieval by location or time. The thermal imager maintains a constant working distance from the strip surface. Because radiation readings decrease with the square of the distance, distance drift can be misread as changes in radiation intensity. The scanning frequency is synchronized with the furnace temperature control cycle of the soaking section, ensuring that each frame of thermal image corresponds to the instantaneous state of the same control cycle. In recrystallized areas, due to increased grain boundary density and more surface micro-undulations, the radiation intensity is slightly higher than in unreacted areas at the same temperature, exhibiting relatively high radiation. Unreacted areas, on the other hand, show relatively low radiation. The thermal radiation distribution map is arranged in a two-dimensional distribution using the scan coordinates as an index. If a scan point's readings are lower than the environmental background for three consecutive frames, it is treated as lens obstruction or localized scaling. The data is then replaced by interpolation of adjacent points in the same frame and written into the thermal radiation distribution map. If the interpolated points account for more than 8% of a single frame, the entire frame is discarded, triggering a lens cleaning prompt, while the original readings are retained for future reference. The damage exceeding the limit marker in the cross-sectional thermal history record also serves as a basis for relaxing the radiation anomaly threshold. The larger the multiple of the exceeding limit, the greater the relaxation of the interpretation threshold, to prevent surface oxidation color differences caused by localized thermal damage from being misjudged as scanning faults. The soaking zone maintains the steel strip at a high temperature, promoting microstructural transformation. Recrystallization continues here, and subtle differences in surface radiation provide a rare observable window for non-destructive monitoring of the internal transformation progress.

[0053] For example, the step of identifying the low-radiation zone boundary of each homogenized segment and determining the recrystallization lag zone from the thermal radiation distribution map includes: extracting the temporal stagnation degree of radiation intensity of each homogenized segment using the thermal radiation distribution map to form a temporal stagnation sequence; locating the radiation stagnation sites of each homogenized segment from the temporal stagnation sequence to obtain radiation stagnation boundaries; identifying the boundary positions of active and stagnation zones of each homogenized segment based on the radiation stagnation boundaries to obtain evolutionary boundary lines; and delineating the incomplete recrystallization zone of each homogenized segment according to the evolutionary boundary lines to form the recrystallization lag zone.

[0054] A time-series stagnation sequence is formed by extracting the temporal stagnation degree of radiation intensity in each homogenized heat segment using a thermal radiation distribution map. The thermal radiation distribution map is acquired frame-by-frame over time. Radiation readings at the same spatial location in consecutive frames constitute the time-series curve for that location. The absolute value of the difference between readings in two adjacent frames represents the instantaneous change in that frame. The smaller the change, the more stable the radiation characteristics. The stagnation degree is quantified as the reciprocal of the instantaneous change, ensuring that more stable locations have larger stagnation degrees, facilitating the use of a uniform threshold to determine stagnation. When the change approaches zero, the stagnation degree is capped by a preset upper limit to avoid division by zero and to prevent distortion caused by extremely small changes. The upper limit is determined by back-calculating the lower bound of the change distribution in the steady-state phase of historical batches. The time-series stagnation sequence is stored using a dual index of scan coordinates and frame number, storing the stagnation degree of each frame at each location. Frames at the same location are arranged chronologically. The index supports retrieval by coordinate or by frame number. When an entire frame of the thermal radiation distribution map is invalidated, the time-series stagnation sequence skips the stagnation degree at the location involved in that frame. Adjacent valid frames are bridged by weighted time intervals, and the source of the skipped frame is marked at the bridging point, along with the original timestamp of the skipped frame. The stagnation degree calculation window covers the most recent five frames, corresponding to the typical timescale of recrystallization stabilization. When three or more frames are skipped within the window, more than half of the valid frames are missing, the current window stagnation degree at that location is judged as unreliable, and the output is set to empty. It will be recalculated after subsequent frames fill the window. During the empty period, this location does not participate in the downstream boundary determination to avoid unreliable values ​​contaminating the determination. The strip steel is continuously heated in the soaking section, and the surface radiation characteristics change from rapid change to stabilization. The radiation intensity of a single frame cannot distinguish whether a certain point has stabilized or is still changing drastically. The time-series stagnation sequence uses the difference between consecutive frames to capture this convergence process.

[0055] Radiation stagnation boundaries are obtained by locating radiation stagnation points in each soaking section using a time-series stagnation sequence. A point in the time-series stagnation sequence is marked as stationary if its stagnation degree exceeds a preset stagnation threshold for three consecutive frames. Three consecutive frames, rather than a single frame, are used to exclude occasional false stagnation due to stabilization followed by rebound. The stagnation threshold is graded according to the recrystallization kinetics of the steel grade; steel grades with faster transformations stabilize earlier due to their radiation characteristics, hence a lower value is used. The threshold table is re-marked before each new steel grade is put into production. Points that do not exceed the threshold for three consecutive frames remain marked as active. The two states are distinguished by Boolean values ​​in the corresponding fields of the time-series stagnation sequence. The stationary status is checked point by point along the scanning coordinates of a soaking section. The coordinate point at the boundary between stationary and active points is registered as a candidate boundary point, with the registration content including the coordinate value and the frame number where the stationary condition is first met. Candidate boundary points for radiation stagnation boundaries are summarized for each homogenization segment. Candidate points with a spacing exceeding one-tenth of the total scan width of the homogenization segment are treated as sparse boundaries. This is because a continuous recrystallization front should not be without any stagnation points within one-tenth of the bandwidth; gaps are often caused by disturbances in the readings of that segment. Linear interpolation is used to fill in the gaps until the spacing meets the standard, and the interpolation source is indicated for each filler point. If a candidate boundary point repeatedly switches its stagnation state more than twice between two adjacent frames, it is treated as a critical point jitter, and its classification is temporarily suspended until it stabilizes for five consecutive frames before being included in the radiation stagnation boundary. During the suspension period, the coordinates of this point are undetermined and excluded from the summary; its boundary classification is determined after the state is confirmed. The recrystallization front advances asynchronously along the bandwidth; the transition speed often differs between the edges and the middle due to different heat dissipation and deformation histories. Radiation stagnation boundaries are judged point-by-point to capture the progress difference in the width direction of this layer, without assuming a single front.

[0056] The evolutionary boundary line is obtained by identifying the active and stagnant zones of each heat sink segment based on the radiation stagnation boundary. Candidate boundary points of the radiation stagnation boundary are connected by cubic splines in the order of scan coordinates to form a continuous curve. The smoothing window is set to five adjacent boundary points, sufficient to constrain the cubic polynomial without over-smoothing the actual curvature of the leading edge. Heat sink segments with fewer than five nodes are fitted using all available points. When the curvature of a segment exceeds a preset upper limit, it is treated as residual jitter, and the segment is replaced with a linear connection to avoid curve distortion, with the linear source marked. The evolutionary boundary line is recorded as a smoothed continuous coordinate sequence, extending to the physical boundary of the heat sink segment scan range to completely cover the bandwidth, ensuring the boundary line does not leave undefined sections at the bandwidth ends. The extension segment is linearly extrapolated from the nearest effective boundary point along the original trend. The extrapolation length does not exceed 5% of the total scan width, as the extrapolation lacks actual measurement support; exceeding this length will rapidly amplify the positional error with distance, rendering it unreliable. Points marked with critical point jitter within the radiation stagnation boundary but already stably included are additionally marked with low confidence on the evolutionary boundary line, distinguished from solid line segments by a dashed line attribute. This distinguishing attribute is written into the corresponding location metadata in the coordinate sequence for differentiated treatment based on reliability during interpretation. When all candidate points in a certain heat spread segment have low confidence, the evolutionary boundary line for that segment's output is delayed, temporarily replacing it with the boundary line from the previous cycle, until a more stable boundary point is added in the next acquisition cycle before regenerating. The active and stagnant regions are separated by a gradual transition band, not an instantaneous jump; the evolutionary boundary line uses a continuous curve to fit the center of the transition band, providing a traceable geometric benchmark for region division.

[0057] The recrystallization lag zone is formed by delineating the incomplete recrystallization areas in each soaking section according to the evolutionary boundary line. The evolutionary boundary line divides the soaking section scanning range into two sides. On the side facing the entrance, the radiation intensity continues to change, indicating incomplete recrystallization. On the side facing the exit, the radiation intensity has stabilized, indicating that recrystallization is basically complete. The classification is determined independently for each scan point and recorded in the classification field containing only two categories: active and completed. On the active side, consecutive scan points belonging to the same active category are combined into one segment. When multiple segments coexist within the same soaking section, the coordinate intervals of each segment are registered separately without cross-segment merging. The gaps between segments separated by completed points are left blank, so that the lag zone presents actual segmentation in bandwidth rather than a uniform area. The recrystallization lag zone records the coordinate range and corresponding area of ​​the active side. The area is estimated by multiplying the number of scan points within the range by the area represented by a single point. The area represented by a single point is the square of the scanning spatial resolution. Since each scan point represents a square area centered on it with a side length equal to the resolution, the number of points multiplied by this area is the actual coverage area of ​​the active side, archived in square millimeters. The soaking section, where the evolution boundary marker is generated with a delay, temporarily uses the previous cycle's range and marker data for the recrystallization lag zone. Once the boundary marker is regenerated, this marker is removed and replaced with the current measured range, recording the difference between the two ranges. If the recrystallization lag zone area in a soaking section accounts for more than 70% of the total scanned area of ​​that section, it is treated as a significant lag in recrystallization progress. The 70% threshold is taken from the allowable upper limit of the lag area when the microstructure should have basically completed its transformation at the current belt speed; exceeding this limit indicates that most areas are still incomplete. The recrystallization lag zone lags behind the progress of that section, and the lag ratio and the number of consecutive batches are recorded. If more than three consecutive batches occur, a process review prompt is triggered. The residence time of the steel strip in the soaking section is determined by the belt speed. If the recrystallization lag zone remains consistently large, it indicates that the residence time at the current belt speed is insufficient for the microstructure transformation to complete. This correlation can be directly used for belt speed adjustment.

[0058] The leading-edge migration trajectory is formed by tracking the low-radiation boundary migration dynamics of each soaking section based on the recrystallization lag zone. The coordinate range of the recrystallization lag zone is arranged by time for each acquisition cycle. The difference between the boundary coordinates of two adjacent cycles divided by the cycle interval gives the instantaneous migration velocity v_m at that location. The coordinate difference is in millimeters, the cycle interval is in minutes, and the division gives millimeters per minute. The sign is positive for the direction towards the exit and negative for the direction towards the entrance. This direction is uniformly applied to all coordinates of the soaking section, recorded in millimeters per minute with one decimal place for easy comparison with the production line belt speed. The leading-edge migration trajectory summarizes the boundary coordinates and corresponding instantaneous migration velocities of each cycle for each location, arranged and archived by time. The trajectory starting point is the first effective recrystallization lag zone boundary of the soaking section. For cycles with data delay markers in the recrystallization lag zone, the leading-edge migration trajectory skips the velocity of that cycle. During adjacent effective cycles, the total position of the two cycles is removed and the average of the total time is used to fill the gap. The source of the interpolation is marked, and the skipped cycle number is recorded. If the migration velocity at a certain position is negative for three consecutive cycles, it is treated as an abnormal retreat of the leading edge. A single negative value may originate from boundary identification noise, while a continuous retreat for three consecutive cycles indicates a true retreat of the leading edge, which is often related to a sudden drop in local furnace temperature. The leading edge migration trajectory is marked as retreating at this position, and the measured temperature of the corresponding furnace section at the time of retreat is recorded for attribution. A comparison of the set furnace temperature for this cycle is also provided to facilitate the determination of whether the retreat is related to the temperature drop. The advancement speed of the recrystallization leading edge is not constant. It is affected by local temperature fluctuations and differences in grain orientation, exhibiting alternating periods of fast and slow speed. The position and velocity of the leading edge migration trajectory are recorded cycle by cycle, completely recording this non-uniform advancement, without relying on averaging between the first and last points.

[0059] The recrystallization rate is generated by calculating the recrystallization area ratio of the soaking zone using the leading-edge migration trajectory. The boundary coordinates of the last sampling cycle of the soaking zone in the leading-edge migration trajectory are compared with the total scan width. The initial value of the recrystallization area ratio for that cycle is obtained by dividing the cumulative advance distance of the completed area boundary on the exit side by the total scan width. This is then applied to each scan coordinate point individually, and a width-weighted average is taken. The weights are allocated according to the actual width interval represented by each point and then normalized to 1. Because the spacing between edge measurement points may not be equal to that in the middle, an equal-weighted average would over-include densely sampled areas. Weighting by width interval ensures that each point's contribution to the total ratio is proportional to its represented area. For positions with a retreat mark in the leading-edge migration trajectory, the coordinates of the previous non-retreat cycle are used when calculating the ratio to avoid underestimating the ratio due to instantaneous retreat. The replaced original retreat coordinates are kept on file for verification. The recrystallization rate is obtained by taking the recrystallization area ratio of the last cycle of each soaking zone and archived as a percentage for each batch. The recrystallization rates of each sampling batch within the same soaking zone are stored separately by batch number for cross-sectional comparison. When the interpolated proportion of the leading edge migration trajectory exceeds 30% of the total recorded period for that soaking section, meaning that more than 30% of the period's position is obtained through interpolation rather than actual measurement, the reliability of the recrystallization rate data for that section decreases, indicating a bias in the proportion result due to data gaps. If the recrystallization rate of a certain soaking section is lower than the lower limit required by the process for that steel grade, it is treated as insufficient microstructure transformation. The recrystallization rate for that section shows insufficient transformation, and the specific difference from the lower limit is recorded. The recrystallization area proportion most directly measures the treatment effect of the soaking section. The period-by-period position given by the leading edge migration trajectory anchors the proportion to the latest measured leading edge, independent of theoretically calculated fixed transformation rates.

[0060] In some embodiments, the step of analyzing the grain refinement degree of each soaking section based on the recrystallization rate to obtain the grain size level includes: quantifying the migration hindrance differences of each soaking section based on the recrystallization rate to establish a migration hindrance distribution; identifying grain growth pinning force partitions of each soaking section based on the migration hindrance distribution to obtain a grain pinning region distribution; estimating the grain size range of each soaking section based on the grain pinning region distribution to obtain a grain size distribution; and assigning the grain size distribution to the corresponding level of grain refinement degree of each soaking section to obtain the grain size level.

[0061] A migration hindrance distribution was established to quantify the differences in migration hindrance across different soaking stages based on recrystallization rate. A higher recrystallization rate indicates more efficient grain boundary migration driven by deformation energy storage in that stage, resulting in a lower residual dislocation density hindering subsequent grain boundary movement. Therefore, the migration hindrance is calculated by linearly reducing the benchmark hindrance coefficient for that steel grade by the recrystallization rate. The recrystallization rate is decimal; its complement multiplied by the benchmark hindrance coefficient gives the quantified migration hindrance value at that location. Linear reduction is used because the residual dislocation density decreases approximately linearly with the transformation rate within the normal recrystallization range, eliminating the need for higher-order terms. The benchmark hindrance coefficient is chosen based on the steel grade to ensure the result falls between 0 and 1; a larger value indicates stronger residual hindrance. The migration hindrance distribution arranges the quantified migration hindrance values ​​at each location according to the soaking stage number. Data from different batches within the same soaking stage are stored separately to observe changes in hindrance levels over time. For soaking stages with a recrystallization rate marked as having reduced data reliability, the migration hindrance distribution only indicates reduced calculation reliability for the quantified value of that stage without numerical correction, providing a reference for downstream use. A migration retardation quantification value exceeding twice the historical batch average for that steel grade in a soaking section is treated as an anomaly. The migration retardation distribution marks this section as an anomaly and records the multiple of deviation from the historical average for traceability. Soaking sections with insufficient recrystallization rate markings naturally have higher migration retardation quantification values. The migration retardation distribution retains the original value for this type of entry without further correction; both the anomaly and insufficient transformation markings may coexist here. Grain boundary migration is followed by grain growth, a stage constrained by the pinning of residual dislocations and second-phase particles. The migration retardation distribution converts the recrystallization completion degree into the initial constraint of grain growth.

[0062] The grain pinning zone distribution is obtained by identifying the pinning force zones for grain growth in each soaking section based on the migration retardation distribution. The migration retardation quantification values ​​for each soaking section are divided into three levels according to the range: above 0.6 for strong pinning, 0.3 to 0.6 for moderate pinning, and below 0.3 for weak pinning. The thresholds of 0.6 and 0.3 are pre-calibrated based on the precipitation characteristics of the second phase particles in this steel grade, corresponding to the critical quantification values ​​for precipitates that are sufficient to fully pinnate and almost non-pinnate grain boundaries, respectively. These values ​​are reviewed annually with adjustments to the furnace charge composition. After each value of the migration retardation distribution is compared with the threshold, the corresponding pinning level, along with the soaking section number, is recorded in the grain pinning zone distribution. The level is stored as a number rather than the original quantification value. Since the next step of grain size estimation only requires matching the empirical model by level and no longer uses continuous values, storing the number reduces the amount of data and avoids downstream misuse of the original values. The original values ​​remain in the migration retardation distribution for future reference. Each of the three tiers is bound to a pre-calibrated Zener geometric correction coefficient α. The tier number serves as the index for retrieving α in the next step: the strong pinning tier is bound to the minimum α, and the weak pinning tier is bound to the maximum α. The binding table is reviewed concurrently with the two thresholds. The grain pinning region distribution stores the tier number position by position. Downstream, the corresponding α is directly retrieved by the tier number and substituted into the limit formula without needing to backtrack to the original quantization value. For the soaking section with the migration hindrance distribution marked with hindrance anomalies, the tier position of adjacent sections is additionally referenced during tier determination: if adjacent sections belong to the same tier but this section is classified into another tier due to an anomaly, the tier position of this section is corrected to be consistent with the adjacent section. The grain pinning region distribution records the tier position correction for this section and retains the tier number before correction. When the batch corresponding to the migration hindrance distribution of a certain soaking section is missing, the grain pinning region distribution temporarily fills the tier position of this section with the mode tier position of the historical batches of this section and marks the historical filling. The pinning of grain boundaries by second-phase particles is not a continuous gradual change, but rather a step-like transition within a specific precipitate density range; the distribution of grain pinning regions is characterized by discrete steps, which facilitates direct matching of empirical models for the corresponding ranges when estimating grain size.

[0063] The grain size distribution is obtained by estimating the grain size range of each soaking section based on the distribution of the grain pinning zone. The pinning level of each soaking section in the grain pinning zone distribution is then substituted into the corresponding Zener pinning limit formula D_lim = α × r / f, where r and f are the average radius (in μm) and volume fraction of the second phase particles obtained from metallographic statistics for that steel grade, and α is the geometric correction coefficient calibrated according to the pinning level, with the smallest value for the strong pinning level and the largest value for the weak pinning level. D_lim is the grain growth limit size corresponding to that level. When the pinning levels are inconsistent at different scanning positions within the same soaking section, D_lim is calculated for each position according to its respective level, and then the maximum value of D_lim at all positions in the entire section is taken as the upper limit of that section, because grain growth is controlled by the weakest pinning point, and the maximum limit size at that point determines the risk of coarsening throughout the entire section. The grain size distribution uses D_lim as the upper limit of the grain size range, because D_lim is the limit to which grains can grow under pinning action, and the actual size will not exceed it. This is archived for each soaking section. The lower limit is taken as the original grain size of that section as determined by metallographic analysis before entering the furnace. Since soaking only causes grain growth and does not reverse refinement, the original size is the lower limit. The upper and lower limits together constitute the grain size range of that section. For soaking sections with grade correction marks in the grain pinning zone distribution, the corresponding grain size distribution entries retain both the D_lim values ​​before and after correction for comparison during interpretation. If the D_lim of a certain soaking section is less than its original lower limit of grain size, it is treated as an anomaly in the grade parameter calibration. The grain size distribution for this section is anomaly in the calibration range and is temporarily replaced by the range of the adjacent soaking section. For soaking sections with historical fill marks in the grain pinning zone distribution, the grain size distribution results inherit these marks, indicating that the reliability of the range estimation for this section is lower than that of the measured grade entries. The range of grain size determines the uncertainty range of subsequent mechanical property predictions; the combination of pinning positions and limit size formulas allows size estimation to be grounded in materials science mechanisms, rather than just empirical interpolation.

[0064] Grain size distribution is assigned to the corresponding grain refinement level of each soaking section to obtain the grain size grade. The upper limit of the grain size range for each soaking section is determined by referring to a standard grain size grade conversion table. This table is established based on the correspondence between grain size grade and average grain diameter in national standards; smaller sizes have larger grade values ​​and higher refinement. The table lookup is performed independently for each soaking section, without cross-referencing. The grain size grade value is recorded for each soaking section after the table lookup and rounded. The critical values ​​of adjacent grade conversion intervals are uniformly assigned to the side with the larger value; that is, sizes falling on the boundary between two grades are counted as the finer grade, even if it means conservatively overestimating the refinement. The classification rules are consistent across all furnace sections and do not vary by location. For soaking sections with abnormal grain size distribution ranges, the grain size grade classification for that section also inherits the abnormal range mark, indicating that the grade value is uncertain due to upstream substitution calculations. If the grain size grade of a soaking section differs from that of an adjacent section by more than two levels, it is treated as a grade jump. If the temperature and microstructure of adjacent soaking sections should be continuous, a sudden jump of more than two levels is often introduced by abnormal upstream size estimations. The grain size grade for that section is recorded as a grade jump, and the specific grade difference with the adjacent section is recorded for verification. For soaking sections with historical filler marks in the grain size distribution, the grain size grade result inherits these marks. When calculating the overall grain size pass rate for that batch, its proportion is listed separately and not mixed with the measured items to avoid diluting the statistical significance of the measured results with filler data. Grain size grade is a common indicator in the steel industry for evaluating the degree of microstructure refinement. Converting continuous grain size estimates into grade numbers facilitates direct comparison with current national standards and customer acceptance specifications.

[0065] Mechanical performance compliance results are obtained by estimating mechanical performance indicators and assessing compliance levels based on grain size grades. The grain size grade values ​​for each soaking section are first converted to the average grain diameter *d* by referring to the standard grain size grade table. Then, they are substituted into the Hall-Page relation σ_y = σ_0 + k × d^(-1 / 2) to estimate the yield strength. σ_0 is the matrix strength constant for this steel grade, *k* is the grain boundary strengthening coefficient (unit: MPa·mm^0.5), and *d* is the converted average grain diameter. σ_y is the predicted yield strength of the corresponding steel strip section. The mechanical performance compliance results are summarized by soaking section, comparing σ_y with the standard yield range required for this steel grade, and categorized into compliant and non-compliant categories, written into the corresponding fields. For soaking sections with grain size grade jump markers, the corresponding σ_y in the mechanical performance compliance results inherits this marker, indicating that the strength prediction may deviate from reality due to grain size data jumps. σ_0 and k are obtained through regression testing of the standard tensile specimens of this steel grade using the Hol Page calibration test. If σ_y in a certain soaking stage is more than 5% lower than the lower limit of the standard range, it is treated as a significant failure to meet mechanical performance standards. 5% is a margin for judgment outside the error band of the prediction model itself. If the mechanical performance compliance result is insufficient for the standard performance of this stage, the percentage difference from the lower limit is recorded. When soaking stages with historical fill or grade jump markers in the grain size grade account for more than 20% of the batch, the reliability of the mechanical performance compliance result for the entire batch output needs to be verified, and a list of the marker types that triggered this state and the corresponding soaking stages is attached. The Hol Page relation connects grain refinement and material strength, directly converting the final result of the microstructure evolution during the soaking stage into a strength prediction that can be compared with the acceptance standard, providing a traceable quantitative basis for the microstructure control effect in the judgment process.

[0066] Step S105: Extract the temperature gradient of the plating section to form a plating index for steel strips whose mechanical properties meet the compliance requirements. Based on the plating index, collect the pre-plating process parameter range to construct a plating parameter set. Verify the pre-plating organization compliance characteristics through the plating parameter set to determine the process control instructions.

[0067] In some embodiments, the step of extracting the temperature gradient of the plating section to form a plating index for steel strips that meet the compliance requirements for mechanical properties includes: measuring the temperature difference between the surface and core of the steel strip before plating to obtain the cross-sectional temperature difference value; plotting the cross-sectional temperature gradient distribution at each time point based on the cross-sectional temperature difference value to generate a cross-sectional gradient distribution spectrum; using the cross-sectional gradient distribution spectrum to screen the process window corresponding to the gradient approaching zero time to obtain the gradient convergence interval; and evaluating the compliance of zinc-aluminum-magnesium alloy spreading and wetting in the gradient convergence interval to determine the plating index.

[0068] Based on the compliance results of mechanical properties, the cross-sectional temperature difference is obtained by measuring the surface and core temperature difference of the steel strip before plating. Only steel strips with compliant mechanical properties are allowed to enter this measurement stage; those that do not comply are diverted to the re-judgment process and do not participate in the temperature difference measurement. When the steel strip reaches the temperature measurement station group 9 before plating, the surface temperature is directly read by an infrared thermometer. Since the core temperature cannot be directly measured, a one-dimensional unsteady-state thermal conductivity model is used for inversion. One-dimensional model is chosen because the thickness of the steel strip is much smaller than its length and width, the temperature gradient is mainly along the thickness direction, and in-plane conduction can be neglected. The model input is the historical furnace temperature curve and thickness specification of the steel strip in the soaking and slow cooling sections. The measured surface temperature is used as the boundary condition to obtain the estimated core temperature T_core. The cross-sectional temperature difference is the absolute value of the difference between the measured surface temperature T_surf and the estimated core temperature T_core, recorded in degrees Celsius and attached to the belt speed reading at the time of measurement. The recording format is standardized and includes the steel coil batch number for traceability. When the core inversion model outputs three times consecutively with a difference exceeding 15℃ from the previous measurement cycle, it is treated as an abnormal missing segment of the furnace temperature curve input by the model. The current cross-sectional temperature difference value is replaced with the average value of the same steel grade and specification batch in history, indicating a model anomaly, while retaining the original model output value at the time of triggering the replacement. For steel strips with compliant mechanical properties and a status of pending credibility verification, the cross-sectional temperature difference value inherits this status, suggesting that the tolerance should be appropriately relaxed when determining plating suitability. Zinc-aluminum-magnesium hot-dip galvanizing is very sensitive to the cross-sectional temperature uniformity of the strip before plating. Excessive temperature difference between the surface and the core can cause sudden changes in local thermal stress at the moment of plating; the cross-sectional temperature difference value is the starting point for quantifying this risk.

[0069] A cross-sectional gradient distribution spectrum is generated by plotting the cross-sectional temperature gradient distribution at various times based on the cross-sectional temperature difference. The cross-sectional temperature difference is measured sequentially at several consecutive temperature measurement stations before the steel strip enters the plating stage. The distance between adjacent stations and the belt speed are converted into time intervals. The cross-sectional temperature difference at each time point is arranged chronologically to form a complete temperature difference evolution sequence for this batch from the last soaking station to the plating station 10. The first point is fixed as the measured temperature difference at the last soaking station, and the last point is the measured temperature difference at the last temperature measurement point before the plating station. The normalized temperature gradient value is recorded for each moment in the cross-sectional gradient distribution spectrum. The gradient value is the ratio of the cross-sectional temperature difference at that moment to the difference at the first temperature measurement station. The difference at the first station is used as the denominator for normalization, so that the decay curves of different batches with different initial temperature differences can be directly superimposed and compared. Theoretically, it decreases from 1 to close to 0. The horizontal axis of the spectrum is the travel time, and the vertical axis is the normalized gradient. The steel grade and thickness specification of the batch are also marked for archiving. When the cross-sectional temperature difference at a certain temperature measurement station is replaced by a substitute value due to an anomaly in the preceding model, the cross-sectional gradient distribution spectrum replaces the gradient source at that moment, and its weight is halved in subsequent gradient decay analysis. If the gradient values ​​of two adjacent stations show a non-monotonic increase exceeding 5% of the original gradient, it is treated as a local temperature measurement disturbance. The cross-sectional gradient distribution spectrum disturbs the gradient at that station and replaces the gradient at that point with a linear interpolation between the two adjacent stations. The interpolation does not change the original measured value and is archived. The original value is stored separately for later verification and equipment calibration comparison. When the strip steel enters the plating station from the soaking zone, the surface and core temperature difference gradually decays and does not disappear abruptly; the cross-sectional gradient distribution spectrum uses a continuous time series to preserve the complete shape of this decay.

[0070] For example, the step of using the cross-sectional gradient distribution spectrum to screen the process window corresponding to the gradient approaching zero time to obtain the gradient convergence interval includes: calculating the gradient decay degree at each time based on the cross-sectional gradient distribution spectrum to generate a gradient decay rate; identifying the time when the gradient continuously decays and approaches zero based on the gradient decay rate to generate the gradient approaching zero time; extracting the corresponding furnace temperature and belt speed matching range from the gradient approaching zero time to obtain a speed-temperature matching parameter; and using the speed-temperature matching parameter to define the process window corresponding to the gradient approaching zero to form a gradient convergence interval.

[0071] The gradient decay rate is generated by calculating the gradient decay degree at each time step based on the cross-sectional gradient distribution spectrum. The instantaneous decay amount for that interval is obtained by subtracting the normalized gradient values ​​of two adjacent time steps and dividing by the time interval. A positive value indicates a continuous decrease in the gradient, while a negative value indicates an abnormal rebound. After calculating the instantaneous decay amount for each adjacent interval of the cross-sectional gradient distribution spectrum, the instantaneous decay amount for each interval is arranged by time and archived along with the corresponding time interval endpoints. The archive structure supports retrieval by interval number and batch export. The gradient decay rate records the instantaneous decay amount for each interval, expressed as the normalized gradient decay per second, i.e., the decrease in the normalized gradient per unit time. The interval number corresponds one-to-one with the time step of the cross-sectional gradient distribution spectrum for traceability. For intervals in the cross-sectional gradient distribution spectrum marked with perturbation, the gradient decay rate is given lower weight and is not used as the primary basis for determining the decay trend; it is only used as a supplement when there are insufficient adjacent effective intervals. If the instantaneous attenuation in a certain interval is negative and its absolute value exceeds twice the average attenuation of that batch, it is treated as an abnormal attenuation rebound. The average attenuation of this batch itself is used as the benchmark instead of a fixed threshold. This is because the overall attenuation rate varies greatly among different thickness specifications, and a fixed threshold would frequently cause false alarms for thinner specifications. Using twice the average value allows the criterion to adapt to the actual attenuation rhythm of the batch. The gradient attenuation rate is used to attribute the abnormal rebound of the benchmark in this interval, and the corresponding measured furnace temperature before plating is provided for attribution. The measured belt speed for this period is also attached for comparison. The attenuation rate of the temperature difference between the steel strip surface and the core is not constant and is affected by both thickness specification and travel speed. The gradient attenuation rate is quantified interval by interval, and this speed difference is recorded without assuming a single attenuation constant.

[0072] The gradient decay rate is used to identify the moment when the gradient continuously decays and approaches zero, generating the gradient zeroing moment. The instantaneous decay amount of each interval of the gradient decay rate is substituted into the exponential decay model G(t) = G_0 × e^(-β × t) for fitting, where G_0 is the normalized gradient value at the initial time, β is the decay coefficient to be determined (in units of seconds), t is the travel time from the beginning, and G(t) is the predicted normalized gradient at time t. Before fitting, the instantaneous decay amount of each interval is cumulatively subtracted from G_0 to reconstruct the measured normalized gradient point sequence at each time as the fitting sample sequence, so that the fitting is anchored on the measured gradient points rather than just extrapolating from the decay rate. β is calculated using least squares on the measured gradient sequence. During the solution, intervals with low weights in the gradient decay rate are excluded. If the exclusion ratio exceeds one-third of the total intervals, the extra labeled samples for that batch of fitting are insufficient. The gradient approaches zero at the moment when the fitted curve G(t) first drops below the preset convergence threshold. The convergence threshold is 5% of the normalized gradient, meaning the temperature difference has decayed to one-twentieth of the initial value, and the residual temperature difference has negligible impact on immersion plating. At this point, the surface-core temperature difference is considered to have been substantially eliminated. For batches with an abnormal rebound in the gradient decay rate, the gradient approaches zero by extrapolating the fitted curve before the abnormal rebound interval separately. The extrapolation uses the same exponential model, only re-fitting the effective interval to avoid distorting the overall fit during the rebound segment. For batches whose fitted curves have not dropped below the convergence threshold within the measured time range of the cross-sectional gradient distribution spectrum, the gradient approaches zero at the moment when the batch outputs the non-converged value and displays the actual residual gradient value at the end of the range for process engineers to assess the risk. When the surface-core temperature difference is eliminated determines the state in which the strip enters the zinc bath; judging too early or too late will cause the subsequent process window to deviate from the true physical state. The precise positioning of the gradient approaches zero is the node connecting the front-end temperature measurement and the subsequent process decision-making.

[0073] Extract the furnace temperature and belt speed matching parameters from the gradient approaching zero moment. Multiply the travel time t_zero corresponding to the gradient approaching zero moment by the travel speed of the batch to calculate the physical travel distance required to eliminate the temperature difference. Since the strip travels at a constant belt speed, time and distance are one-to-one. t_zero multiplied by the belt speed is the distance traveled when the temperature difference drops to the convergence threshold. Compare this distance with the actual production line length from the soaking section to the plating station: if the actual length is sufficient, the current speed-temperature combination is feasible; if it is insufficient, the combination cannot guarantee that the temperature difference will be fully eliminated before plating. Record the specific values ​​of the excess or deficiency and the percentage of the total production line length, leaving one decimal place. The rapid temperature matching parameters record the belt speed, the corresponding set furnace temperature at the outlet of the soaking section, and the margin distance for each batch. These three factors together characterize whether the rapid temperature combination for that batch meets the convergence requirement. Each record is also linked to the steel grade and thickness specification of the batch as a grouping key. Because the distance required to eliminate temperature difference varies with the steel grade and thickness, and the margin distances of different specifications cannot be mixed, downstream processes use this key to group rapid temperature combinations by steel grade and specification before dividing them into intervals to avoid cross-specification samples interfering with each other. When the gradient approaches zero, batches that have not converged are output, and the rapid temperature matching parameters record the margin distance of that batch as a negative value. The larger the negative value, the further the current combination deviates from the feasible range, and the greater the adjustment required. If the margin distance of a certain rapid temperature combination is less than 5% of the production line length for five consecutive batches, it is treated as a long-term critical condition. A single batch being too small may be due to fluctuations in operating conditions; only when five consecutive batches are still close to the boundary does it indicate that the combination itself has insufficient margin and requires structural adjustments in the process settings. The rapid temperature matching parameters mark the combination as critical and record the number of consecutive batches and batch number list for verification. The belt speed and furnace temperature are not independently adjustable. Increasing the speed without simultaneously increasing the temperature will compress the temperature difference and eliminate the effective travel distance required. The speed-temperature matching parameter replaces this coupling with a verifiable distance margin, providing a quantitative basis for defining the process window.

[0074] The gradient convergence interval is formed by defining the process window corresponding to the gradient approaching zero using the speed-temperature matching parameters. Speed-temperature combinations with a margin greater than 10% of the production line length are classified into the margin interval; 0 to 10% is classified into the critical feasible interval; and less than 0 is classified into the infeasible interval. The two thresholds, 10% and 0%, are pre-calibrated based on the actual fluctuation margin of the production line. 10% corresponds to a safety margin that can absorb daily belt speed and furnace temperature fluctuations, while 0 is the critical line where the temperature difference is eliminated exactly at the end of the production line. These thresholds are reviewed during equipment overhaul cycles to reflect changes in effective length caused by roller wear and furnace aging. The initial boundary of the gradient convergence interval is taken from the range of belt speed and furnace temperature values ​​covered by the margin and critical feasible intervals after archiving all historical batches of the speed-temperature matching parameters. This is expressed as a closed region in the belt speed-furnace temperature two-dimensional plane and established separately according to steel type and specification combinations. For combinations of speed-temperature matching parameters marked with critical values, the gradient convergence interval should tighten its boundary range. Tightening should leave an additional safety margin on the boundary values ​​rather than directly taking the critical values. The margin ratio should be uniformly set at 5%, and should not be adjusted based on steel type. This margin addresses the uncertainty of the critical combination itself and is unrelated to the steel type. Setting different values ​​for each steel type would lead to inconsistent boundary standards and make horizontal comparisons difficult. If a speed-temperature combination is in an infeasible range but the distance from the critical feasible boundary is less than the set tolerance, the gradient convergence interval will not include it. Inclusion will be based on the allowable distance setting and will not be adjusted due to fluctuations in a single batch. The allowable range of belt speed and furnace temperature output by the gradient convergence interval are process constraint boundaries that can be directly referenced in plating feasibility assessments. These are updated and archived once per production cycle with the latest batch, while retaining the boundaries of the previous cycle for comparison of changes.

[0075] The plating applicability index is determined by assessing the compliance of zinc-aluminum-magnesium alloy spreading and wetting within the gradient convergence interval. The allowable ranges for belt speed and furnace temperature given by the gradient convergence interval are compared one by one with the actual operating parameters of the current batch: those falling within the excess range are judged as excellent; those falling within the critical feasible range are judged as acceptable but require attention; and those falling within the infeasible range are judged as non-compliant. This three-tier judgment is consistent for all steel grades and is performed independently for each batch without referring to historical data. Since plating applicability depends on whether the measured belt speed and temperature of this batch fall within the convergence boundary and is unrelated to historical batch results, independent judgment avoids historical acceptable records masking current batches exceeding the limits. The compliance judgment result of the plating applicability index is recorded for each batch and stored in three tiers: excellent, acceptable with attention, and non-compliant, along with the actual belt speed and furnace temperature at the time for post-event comparison and traceability. When the gradient convergence interval tightens its recommendations for a certain belt speed and temperature combination, the plating applicability index for batches falling within that range is re-evaluated with stricter judgment. Under stricter judgment, batches originally judged as excellent are automatically downgraded to acceptable with attention, ensuring that even if a critical combination falls within the excess range, it is not directly released. If a batch of steel coils is found to be non-compliant with the plating suitability index, and this is the only instance of non-compliance within that batch, then the plating suitability index alone will disqualify the steel coil from being scrapped. This indicates that a single instance of non-compliance should not lead to immediate scrapping; a comprehensive review of adjacent batches is necessary before further action, and this review should be conducted manually by process personnel. Zinc-aluminum-magnesium coatings are more sensitive to the substrate surface condition than pure zinc. Incomplete temperature differences can cause uneven alloy spreading, incomplete plating, or slag buildup. The plating suitability index directly translates the compliance conclusion for the specific alloy system based on the preceding temperature convergence analysis, rather than applying the general standards for pure zinc.

[0076] A plating parameter set is constructed based on the plating suitability index and the range of pre-plating process parameters. For batches judged as excellent in plating suitability, their actual belt speed and furnace temperature, along with the corresponding steel grade and specifications, are extracted as recommended process parameter candidates for that steel grade and specification combination. For batches judged as acceptable but not of concern, the candidate range is narrowed beyond the recommended values, retaining only the portion whose actual deviation from the center of the allowable range does not exceed the set tolerance. The tolerance is taken as 15% of the range span, using a relative proportion rather than a fixed value, allowing the tolerance to automatically scale with the width of the process window for each steel grade and specification. For each steel grade and specification combination, the plating parameter set records the recommended belt speed range, recommended furnace temperature range, and corresponding candidate batch number. More candidate batches indicate more thorough verification of the process stability for that range. These three are indexed according to the steel grade and specification combination number. Batches judged as non-compliant in plating suitability are not included in the candidate extraction, but their belt speed and furnace temperature values ​​are separately listed as exclusion areas in the plating parameter set for proactive avoidance in subsequent process settings. If there are fewer than five candidate batches rated as "excellent" for a certain steel grade and specification combination, insufficient statistical representativeness to support the recommended range, the sample size for this combination in the plating parameter set is insufficient. This suggests that the statistical confidence of the current recommended range is limited, and it is advisable to combine data from adjacent specification combinations for supplementary judgment and conduct periodic supplementary testing. Batches with a single-case rejection mark for plating index should not have their process parameters included in the candidate extraction to avoid contaminating the overall recommended range with a single abnormal judgment. The reasonable range of process parameters before plating cannot be determined solely by theoretical calculations; it requires reverse induction through plating verification of actual batches. The plating parameter set is the product of tracing back from the measured results to the process settings.

[0077] The process control instructions are determined by verifying the compliance characteristics of the pre-plating organization through the plating parameter set. The recommended belt speed range and recommended furnace temperature range for the current steel coil's corresponding steel grade and specification combination in the plating parameter set are compared one by one with the coil's current actual operating parameters. If the parameters fall within the recommended range, the verification is passed; otherwise, the verification fails, and the direction and specific magnitude of the deviation are recorded as a percentage of the recommended range boundary value. The process control instructions generate control suggestions for each steel coil. For those that pass verification, an instruction to maintain the current parameters is output; for those that fail, an adjustment instruction is output. The adjustment value is the difference between the actual parameter and the nearest boundary of the recommended range. For combinations in the plating parameter set marked with insufficient samples, the process control instructions have limited confidence in re-marking the verification results for that combination, suggesting that the adjustment magnitude should be manually verified rather than automatically issued. For belt speed and furnace temperature combinations marked with exclusion zones in the plating parameter set, if the current steel coil's actual parameters happen to fall into the exclusion zone, the process control instructions directly output a forced adjustment instruction for that steel coil. This instruction has higher priority than general verification failures and cannot be temporarily canceled by on-site personnel. If a steel coil receives three consecutive process control commands that are all in the same direction, and an occasional single adjustment is considered normal fluctuation, but the fact that three consecutive adjustments in the same direction point to the same steel grade and specification combination indicates a systematic deviation in the equipment reference, it should be treated as a systematic deviation. The process control commands will trigger an additional equipment reference verification prompt, not just a single coil parameter adjustment. In practice, this command is issued using structured fields: including the target furnace section number, the object to be adjusted (belt speed or furnace temperature), and the adjustment direction and magnitude, such as "Increase furnace temperature of No. 3 soaking section by 8℃" or "Decrease unit belt speed by 0.2m / min". Maintenance commands output the air conditioning adjustment amount and record the verification pass time. Forced adjustment commands are marked as non-cancellable and executed directly by the production line PLC.

[0078] To implement the continuous annealing method for hot-dip galvanized aluminum-magnesium steel strip corresponding to the above method embodiments, in order to achieve the corresponding functions and technical effects. See also Figure 4 , Figure 4 This application provides a structural block diagram of a continuous annealing treatment apparatus 400 for hot-dip galvanized aluminum-magnesium steel strip, comprising:

[0079] The specification acquisition module 401 is used to collect the composition data and thickness deviation sequence of the steel strip before it enters the furnace to generate a material specification set, and calculate the heat requirement of each annealing section based on the material specification set to obtain a process parameter set.

[0080] The disturbance setting module 402 is used to acquire the furnace temperature disturbance sequence based on the amplitude of the furnace temperature disturbance in adjacent sections when the unit speed is adjusted according to the process parameter set, analyze the thermal boundary offset distribution of each furnace section according to the furnace temperature disturbance sequence to form furnace area boundary parameters, and use the furnace area boundary parameters to integrate the furnace temperature setting range of each section to output the segmented furnace temperature setting.

[0081] The status monitoring module 403 is used to collect the transverse temperature difference distribution of each furnace segment for the segmented furnace temperature setting to obtain the transverse temperature difference sequence, to count the over-threshold temperature difference residence time of each furnace segment based on the transverse temperature difference sequence to determine the over-temperature residence time, and to calculate the cumulative amount of thermal damage of each furnace segment based on the over-temperature residence time to establish the cross-sectional thermal history status.

[0082] The crystallization tracking module 404 is used to track the recrystallization front of the homogenization zone based on the thermal history status of the cross section, generate a recrystallization rate, analyze the grain refinement degree of each homogenization section based on the recrystallization rate to obtain the grain size level, and estimate the mechanical performance indicators and evaluate the compliance degree based on the grain size level to obtain the mechanical performance compliance result.

[0083] The instruction output module 405 is used to extract the temperature gradient of the plating section of the steel strip that meets the mechanical performance compliance results to form a plating index, collect the pre-plating process parameter range based on the plating index to construct a plating parameter set, and verify the pre-plating organization compliance characteristics through the plating parameter set to determine the process control instruction.

[0084] The aforementioned continuous annealing apparatus 400 for hot-dip galvanized aluminum-magnesium steel strip can implement a continuous annealing method for hot-dip galvanized aluminum-magnesium steel strip as described in the above-described method embodiments. The options in the above method embodiments are also applicable to this embodiment and will not be detailed here. The remaining contents of this application's embodiments can be referred to the contents of the above method embodiments, and will not be repeated in this embodiment.

[0085] The purpose of the above embodiments is to reproduce and derive the technical solution of the present invention by way of example, and to fully describe the technical solution, purpose and effect of the present invention. The purpose is to enable the public to have a more thorough and comprehensive understanding of the disclosure of the present invention, and not to limit the scope of protection of the present invention.

Claims

1. A method for continuous annealing of hot-dip galvanized aluminum-magnesium steel strip, characterized in that, include: The composition data and thickness deviation sequence of the steel strip before it enters the furnace are collected to generate a material specification set. Based on the material specification set, the heat requirements of each annealing section are calculated to obtain a set of process parameters. Based on the process parameter set, the furnace temperature disturbance amplitude of adjacent furnace sections is collected during unit speed regulation to obtain the furnace temperature disturbance sequence. The thermal boundary offset distribution of each furnace section is analyzed according to the furnace temperature disturbance sequence to form furnace area boundary parameters. The furnace area boundary parameters are used to integrate the furnace temperature setting range of each section to output the segmented furnace temperature setting. For the segmented furnace temperature setting, the transverse temperature difference distribution of each furnace segment is collected to obtain the transverse temperature difference sequence. Based on the transverse temperature difference sequence, the residence time of the over-threshold temperature difference in each furnace segment is counted to determine the over-temperature residence time. Based on the over-temperature residence time, the cumulative amount of thermal damage in each furnace segment is calculated to establish the cross-sectional thermal history status. Based on the cross-sectional thermal history status tracking, the recrystallization rate is generated at the recrystallization front of the homogenization zone. The grain refinement degree of each homogenization section is analyzed based on the recrystallization rate to obtain the grain size level. Based on the grain size level, mechanical performance indicators are estimated and the compliance degree is evaluated to obtain mechanical performance compliance results. For steel strips whose mechanical properties meet the compliance requirements, the temperature gradient of the plating section is extracted to form a plating index. Based on the plating index, the range of pre-plating process parameters is collected to construct a plating parameter set. The pre-plating organization compliance characteristics are verified through the plating parameter set to determine the process control instructions.

2. The method according to claim 1, characterized in that, The process parameter set obtained by calculating the heat requirements of each annealing zone based on the material specification set includes: The thickness thermal hysteresis spectrum is obtained by identifying the thickness thermal hysteresis distribution of each annealing section using the material specification set. Based on the aforementioned thickness-oriented thermal hysteresis spectrum, the maximum hysteresis demand of each segment is extracted to determine the hysteresis compensation parameters; The heat flow redundancy rate is obtained by evaluating the degree of heat flow redundancy in each section of the furnace based on the hysteresis compensation parameters. Based on the heat flux redundancy rate, the heat demand range of the annealing section is calculated, and a set of process parameters is output.

3. The method according to claim 1, characterized in that, The process of analyzing the thermal boundary offset distribution of each furnace section based on the furnace temperature disturbance sequence to form furnace zone boundary parameters includes: The speed regulation response window of each furnace section is screened to obtain the disturbance response duration based on the furnace temperature disturbance sequence; The disturbance attenuation rate is obtained by fitting the disturbance amplitude attenuation law of each furnace section based on the disturbance response time; Based on the disturbance attenuation rate, the locations of heat conduction blockages between furnace sections are identified to form a set of furnace section boundary locations; The furnace zone boundary parameters are constructed by collecting the thermal boundary offset range of each furnace section based on the set of furnace section boundary locations.

4. The method according to claim 1, characterized in that, The determination of the over-temperature residence time based on the statistical analysis of the transverse temperature difference sequence for each furnace section includes: Based on the aforementioned transverse temperature difference sequence, the period of continuous temperature difference exceeding the threshold in each furnace section is identified, and a record of the exceeding threshold period is generated. Based on the recorded over-threshold periods, the duration of the over-threshold periods in each furnace segment is extracted to establish an over-threshold duration sequence; The above-threshold duration sequence is used to evaluate the conditions for exceeding the dwell time limit in each furnace section and generate over-temperature judgment results; The over-temperature residence time is determined by accumulating the effective over-threshold residence time of each furnace section based on the over-temperature determination results.

5. The method according to claim 1, characterized in that, The method of tracking the recrystallization front of the homogenization zone based on the thermal history status of the cross-section to generate the recrystallization rate includes: Based on the thermal history of the cross section, the surface thermal radiation intensity distribution of each homogenization section is determined and a thermal radiation distribution map is established. The recrystallization hysteresis zone is determined by identifying the boundaries of the low-radiation zones in each homogenization section from the aforementioned thermal radiation distribution map. Based on the recrystallization hysteresis zone, the dynamic migration of the low-radiation boundary of each homogenization section is tracked to form the front migration trajectory. The recrystallization rate is generated by calculating the recrystallization area ratio of the homogenization zone based on the aforementioned frontal migration trajectory.

6. The method according to claim 1, characterized in that, The process of determining the grain size grade by analyzing the grain refinement degree of each soaking section based on the recrystallization rate includes: A migration retardation distribution is established based on the quantification of migration retardation differences in each homogenization zone according to the recrystallization rate. Based on the migration retardation distribution, the grain pinning region distribution is obtained by identifying the grain growth pinning force partitions in each heat-soaking section. The grain size distribution is obtained by estimating the grain size range of each soaking zone based on the grain pinning zone distribution. The grain size distribution is assigned to the corresponding level of grain refinement in each soaking section to obtain the grain size level.

7. The method according to claim 1, characterized in that, The steel strip whose mechanical properties meet the compliance requirements is used to extract the temperature gradient of the plating section to form a plating index, including: Based on the compliance results of the mechanical properties, the temperature difference between the surface and the core of the steel strip before plating was measured to obtain the cross-sectional temperature difference value. Based on the temperature difference of the cross section, the temperature gradient distribution of the cross section at each time moment is plotted to generate the cross section gradient distribution spectrum; The gradient convergence interval is obtained by screening the process window corresponding to the gradient approaching zero time using the cross-sectional gradient distribution spectrum. The compliance of zinc-aluminum-magnesium alloy spreading and wetting with the gradient convergence interval is evaluated to determine the plating index.

8. The method according to claim 5, characterized in that, The step of identifying the low-radiation zone boundaries of each homogenization segment and determining the recrystallization lag zone from the thermal radiation distribution map includes: The temporal stagnation degree of radiation intensity in each homogenization zone is extracted using the aforementioned thermal radiation distribution map to form a temporal stagnation sequence. The radiation stagnation boundary is obtained by locating the radiation stagnation position of each homogenization segment from the time-series stagnation sequence. Based on the radiation stagnation boundary, the boundary positions of active and stagnation zones in each homogenization zone are identified to obtain the evolution boundary line; Based on the aforementioned evolutionary boundary line, recrystallization lag zones are formed by delineating recrystallization incomplete areas in each homogenization section.

9. The method according to claim 7, characterized in that, The step of using the cross-sectional gradient distribution spectrum to screen the process window corresponding to the gradient approaching zero time to obtain the gradient convergence interval includes: Based on the gradient distribution spectrum of the cross section, the gradient decay rate is generated by calculating the gradient decay degree at each time step. Based on the gradient decay rate, the moment when the gradient continuously decays and approaches zero is identified, and the moment when the gradient approaches zero is generated; Extract the furnace temperature and belt speed matching parameters from the corresponding furnace temperature and belt speed matching range at the moment when the gradient approaches zero. The gradient convergence interval is formed by defining the process window corresponding to the gradient approaching zero using the aforementioned rapid temperature matching parameters.

10. A continuous annealing treatment device for hot-dip galvanized aluminum-magnesium steel strip, characterized in that, include: The specification acquisition module is used to collect the composition data and thickness deviation sequence of the steel strip before it enters the furnace to generate a material specification set, and calculate the heat requirements of each annealing section based on the material specification set to obtain a process parameter set. The disturbance setting module is used to obtain the furnace temperature disturbance sequence based on the amplitude of the furnace temperature disturbance in adjacent sections when the unit speed is adjusted according to the process parameter set, analyze the thermal boundary offset distribution of each furnace section according to the furnace temperature disturbance sequence to form furnace area boundary parameters, and use the furnace area boundary parameters to integrate the furnace temperature setting range of each section to output the segmented furnace temperature setting. The status monitoring module is used to collect the transverse temperature difference distribution of each furnace segment for the segmented furnace temperature setting to obtain the transverse temperature difference sequence, to count the residence time of the over-threshold temperature difference in each furnace segment based on the transverse temperature difference sequence to determine the over-temperature residence time, and to calculate the cumulative amount of thermal damage in each furnace segment based on the over-temperature residence time to establish the cross-sectional thermal history status. The crystallization tracking module is used to track the recrystallization front of the homogenization zone based on the thermal history status of the cross section, generate the recrystallization rate, analyze the grain refinement degree of each homogenization section based on the recrystallization rate to obtain the grain size level, and estimate the mechanical performance indicators and evaluate the compliance degree based on the grain size level to obtain the mechanical performance compliance result. The instruction output module is used to extract the temperature gradient of the plating section of the steel strip that meets the mechanical performance compliance results to form a plating index, collect the pre-plating process parameter range based on the plating index to construct a plating parameter set, and verify the pre-plating organization compliance characteristics through the plating parameter set to determine the process control instruction.

Images