Steel plate surface quality detection method and device, electronic equipment and storage medium

Abstract

The application provides a steel plate surface quality detection method and device, electronic equipment and storage medium, and belongs to the field of surface quality detection. The method comprises the following steps: determining an original defect area and binding the original defect area with a unique identifier of a material; acquiring physical process parameters and material attribute parameters of the material in a cross-process flow process; calculating a target longitudinal coordinate; triggering local data collection; obtaining an area evolution coefficient, generating a deterioration tolerance and a flatness tolerance matched with the material attribute parameters, and classifying and determining the evolved quality grade according to the comparison result of the area evolution coefficient and the deterioration tolerance and the flatness tolerance. The steel plate surface quality detection method and device, electronic equipment and storage medium provided by the application solve the problems of cross-process defect coordinate mapping drift and evolution state confusion, and realize accurate addressing and quantitative determination of the same defect.

Description

Technical Field

[0001] This application belongs to the field of surface quality testing technology, and more specifically, relates to methods and apparatus for testing the surface quality of steel plates, electronic equipment, and storage media. Background Technology

[0002] Modern continuous steel sheet manufacturing processes encompass multiple transfer steps, including hot rolling, shearing, cold rolling, and galvanizing. The strip steel flows between processes in coil form, with each process physically modifying the strip surface and simultaneously deploying independent surface inspection equipment to collect data on the surface condition at that location. The inspection equipment in each process segment is relatively isolated, and defect data is transmitted from top to bottom along the material flow direction.

[0003] Chinese Patent Publication No. CN121502200A discloses an intelligent annotation and analysis method for steel production quality based on multi-process spatiotemporal mapping. This method acquires the time-series curves of characteristic parameters of the target material at each process, material tracking data, and quality data. Based on the production start time, end time, and material length, it constructs a linear conversion relationship between time and processing position within a process. At cross-process stages, it establishes a position mapping model based on material flipping information and changes in the upstream and downstream length ratios. Based on the statistical and frequency domain features extracted from the time-series curves, it compares the target material with a reference material at the same processing position and outputs diagnostic results.

[0004] During the physical process of strip steel moving from upstream to downstream, the head and tail sections are cut off, the side edges are trimmed, and the middle section is rolled and elongated along the rolling direction. The physical reference boundary of the material shifts forward due to the head trimming, causing a sudden change in the coordinate origin; the side reference bases shrink inward due to the trimming; the dimensions are enlarged along the length direction due to the rolling extension; some processes also involve plate flipping, causing a mirror reversal of the original coordinate system. The defect coordinates recorded upstream no longer point to the same physical location on the downstream physical entity. Linear conversion based on the total length ratio of the material still uses the length ratio before trimming for position allocation on the remaining section after the head and tail are cut off, causing the calculated mapped coordinate points to deviate from the actual physical location of the defect. The correlation method based on time-linear interpolation cannot accommodate the inward shift of the lateral reference caused by the trimming, nor can it accommodate the mirror reversal caused by the strip steel flipping. After the mapped coordinates deviate from the physical location, the grayscale data collected by the downstream detection nodes at the deviation points is not the result of the evolution of the original defect. The difference index obtained at this time reflects process fluctuations at irrelevant locations rather than the morphological changes of the defect itself. Some existing technologies that rely on time-series feature comparisons based on misaligned data (such as mean and variance analysis of time-series curves) can only reflect the overall macroscopic fluctuations of process parameters. They fail to separate the geometric effects of the passive elongation and amplification of defect area as the strip rolls under pressure from the actual morphological evolution of defects caused by physical tearing or pressing. The drift of coordinate references and the resulting confusion in evaluation dimensions make it impossible to accurately locate the same surface defect across processes, and its evolutionary state cannot be quantitatively determined. Summary of the Invention

[0005] The purpose of this application is to provide a method and device for detecting the surface quality of steel plates, electronic equipment, and storage medium to solve the problem of cross-process defect coordinate mapping drift and evolution state confusion, and to achieve accurate addressing and quantitative determination of the same defect.

[0006] A first aspect of this application provides a method for detecting the surface quality of a steel plate, comprising: When the local grayscale gradient value on the material surface is greater than the preset first gradient threshold, the original defect region is determined, and the initial horizontal coordinate, initial vertical coordinate and initial defect area of ​​the original defect region are extracted and bound to the unique identifier of the material. In response to the received material subsequent processing in place trigger signal corresponding to the unique identifier, the physical process parameters and material attribute parameters of the material in the cross-process flow process are obtained. The physical process parameters include absolute cut length, single-sided cut width, comprehensive pressing elongation and plate flipping status position. The target lateral coordinate is calculated based on the plate flipping state position, the initial lateral coordinate, and the single-sided cutting edge width. The difference between the initial longitudinal coordinate and the absolute cutting length is calculated, and the difference is multiplied by the comprehensive compression elongation rate to obtain the target longitudinal coordinate. The target expected window is determined based on the target horizontal coordinate and the target vertical coordinate. When the material enters the vertical interval corresponding to the target expected window, the local data acquisition is triggered to extract the evolved defect area after the original defect area has evolved, based on the length measurement data of the material acquired in real time during the subsequent processing. The ratio of the evolved defect area to the initial defect area is normalized by the comprehensive reduction elongation rate to obtain the area evolution coefficient. Based on the material property parameters, a preset metallurgical experience database is queried to dynamically generate a deterioration tolerance and a recovery tolerance that match the material property parameters. The evolved quality level is classified and determined according to the comparison results of the area evolution coefficient, the deterioration tolerance, and the recovery tolerance.

[0007] In one possible implementation, calculating the target lateral coordinate based on the board flip state position, the initial lateral coordinate, and the single-sided cut edge width includes: Obtain the real-time nominal width of the material during the subsequent processing; When the flip state position indicates that a flip has occurred, the target horizontal coordinate is obtained by subtracting the difference between the initial horizontal coordinate and the single-sided cut edge width from the real-time nominal width. When the board flip state position indicates that no flip has occurred, the difference between the initial lateral coordinate and the width of the single-sided cut edge is used as the target lateral coordinate.

[0008] In another possible implementation, determining the target expected viewport based on the target's horizontal and vertical coordinates includes: Obtain a preset process drift tolerance that matches the subsequent processing steps; The target expectation window is generated by superimposing the preset process drift tolerance along the horizontal and vertical axes with the target horizontal and vertical axes as the origin. After generating the target expectation window, the process also includes: The target expectation window is encapsulated into a control command carrying a reference timestamp and issued to trigger the caching of feature data acquired by the local data acquisition when the communication link is determined to be disconnected, and to perform asynchronous offline alignment of the cached feature data based on the reference timestamp after the communication is restored.

[0009] In another possible implementation, when the material is determined to have entered the longitudinal section corresponding to the target expected window based on the length measurement data of the material acquired in real time during subsequent processing, local data acquisition is triggered to extract the evolved defect area of ​​the original defect region, including: Obtain the pulse frequency of the master control node used to collect the length measurement data; When the pulse frequency is lower than the preset drop threshold, supplementary pulse data of the auxiliary guide node fixed adjacent to the main control node is extracted; The supplementary pulse data is mapped to a distance increment and accumulated with the collected length measurement data to determine whether the material has entered the longitudinal interval based on the accumulation result; When it is determined that the material has entered the longitudinal section, local data acquisition is triggered to extract the evolved defect area after the original defect area has evolved.

[0010] In another possible implementation, classifying the evolved quality level based on the comparison results of the area evolution coefficient, the deterioration tolerance, and the smoothing tolerance includes: When the area evolution coefficient is greater than the sum of the preset evolution benchmark value and the deterioration tolerance, the quality level is determined to be a fatal degradation level and a degradation blocking signal is output. When the area evolution coefficient is less than the difference between the preset evolution benchmark value and the smoothing tolerance, the quality level is determined to be a qualified smoothing level; When the area evolution coefficient is between or equal to the difference between the preset evolution benchmark value and the smoothing tolerance, and between the preset evolution benchmark value and the deterioration tolerance, the area of ​​the evolution defect is overwritten to the underlying map library, and the benchmark quality level corresponding to the original defect area is maintained.

[0011] In another possible implementation, after outputting the downgrade blocking signaling, the following is also included: The target's horizontal coordinates, vertical coordinates, and fatal degradation level are synchronized to the preset downstream finishing terminal; A virtual coordinate system is constructed, and the longitudinal coordinates of the target are updated in the virtual coordinate system using the real-time travel length of the material in the downstream finishing terminal, thereby generating a dynamic tracking trajectory of the original defect area. When the relative distance between the dynamic tracking trajectory and the preset physical cutting station in the downstream finishing terminal is less than a preset distance threshold, a fixed-point cutting instruction is issued to the downstream finishing terminal.

[0012] In another possible implementation, the preset metallurgical experience base is configured with custom weight rules that are bound to the physical material characteristics in the material attribute parameters; The method of dynamically generating deterioration tolerance and recovery tolerance that match the material attribute parameters by querying a preset metallurgical experience base based on the material attribute parameters includes: Retrieve the actual two-dimensional grayscale image obtained during the extraction of the original defect region; The multidimensional morphological features of the original defect region are extracted from the real two-dimensional grayscale image. The multidimensional morphological features include the defect aspect ratio, hole density and grayscale dispersion. Extract the physical material characteristics from the material attribute parameters to determine the corresponding custom weight rules in the preset metallurgical experience library; The multidimensional morphological features are input into the preset metallurgical experience base as query conditions. Fuzzy mapping query is performed on the multidimensional morphological features using the determined custom weight rules. A confidence scalar representing the evolution tendency is output. The deterioration tolerance and the recovery tolerance are generated based on the confidence scalar.

[0013] A second aspect of this application provides a steel plate surface quality inspection device, comprising: The defect extraction module is used to determine the original defect region when the local grayscale gradient value on the material surface is greater than a preset first gradient threshold, extract the initial horizontal coordinate, initial vertical coordinate and initial defect area of ​​the original defect region, and bind them with the unique identifier of the material. The parameter retrieval module is used to respond to the received material subsequent processing placement trigger signal corresponding to the unique identifier, and to obtain the physical process parameters and material attribute parameters of the material in the cross-process flow process. The physical process parameters include absolute cutting length, single-sided cutting edge width, comprehensive pressing elongation and plate flipping status position. The coordinate calculation module is used to calculate the target lateral coordinate based on the plate flipping state position, the initial lateral coordinate and the single-sided cutting edge width, calculate the difference between the initial longitudinal coordinate and the absolute cutting head length, and multiply the difference by the comprehensive compression elongation rate to obtain the target longitudinal coordinate. The target acquisition module is used to determine the target expected window based on the target's horizontal coordinate and the target's vertical coordinate, and to determine the longitudinal interval corresponding to the target expected window when the material enters the longitudinal interval corresponding to the target expected window based on the length measurement data of the material acquired in real time during the subsequent processing. This triggers local data acquisition to extract the evolved defect area after the original defect area has evolved. The quality assessment module is used to normalize the ratio of the evolved defect area to the initial defect area using the comprehensive reduction elongation rate to obtain the area evolution coefficient, query a preset metallurgical experience library based on the material attribute parameters to dynamically generate a deterioration tolerance and a recovery tolerance that match the material attribute parameters, and classify and determine the evolved quality level based on the comparison results of the area evolution coefficient, the deterioration tolerance, and the recovery tolerance.

[0014] A third aspect of this application provides an electronic device, including a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the processor executes the computer program to implement the steps of the above-described steel plate surface quality detection method.

[0015] In a fourth aspect of this application, a computer-readable storage medium is provided, the computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-described steel plate surface quality detection method.

[0016] The beneficial effects of the steel plate surface quality inspection method and apparatus, electronic equipment, and storage medium provided in this application are as follows: 1. By introducing absolute cut length, single-sided cut width, comprehensive pressing elongation, and board flip state position as physical process parameters, cut removal and elongation product calculations are performed on the longitudinal coordinates of the defect, and flip compensation and cut inward calculations are performed on the transverse coordinates. This maps the upstream defect coordinates to the real position of the downstream physical entity, avoiding coordinate drift caused by abrupt changes in the reference and proportional conversion, and achieving accurate addressing of the same defect across processes.

[0017] 2. A target expectation window is generated by overlaying preset process drift tolerances with the target coordinates as the center. When the pulse frequency of the main control node drops, the pulse data of the auxiliary guide node is switched to accumulate the length measurement. This ensures the reliability of triggering local acquisition when the material enters the longitudinal section. At the same time, the window command carries a reference timestamp to asynchronously align cached data offline after disconnection, which improves the robustness of data acquisition in complex industrial environments.

[0018] 3. The ratio of the evolved defect area to the initial defect area is normalized by the comprehensive reduction elongation rate to obtain the area evolution coefficient. Based on the material property parameters, the deterioration tolerance and recovery tolerance are dynamically matched from the metallurgical experience database to make a three-way classification judgment. The degree of active evolution of the defect is separated from the geometric amplification effect, realizing the quantitative differentiation of quality level and avoiding misjudging process fluctuations as defect deterioration. Attached Figure Description

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

[0020] Figure 1 A schematic flowchart of a steel plate surface quality inspection method provided in an embodiment of this application; Figure 2This is a structural block diagram of a steel plate surface quality inspection device provided in an embodiment of this application; Figure 3 This is a schematic block diagram of an electronic device provided in an embodiment of this application.

[0021] Reference numerals: 300, electronic device; 301, processor; 302, input device; 303, output device; 304, memory; 305, communication bus. Detailed Implementation

[0022] 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.

[0023] It is understood that in the embodiments of this application, data such as user information are involved. When the embodiments of this application are applied to specific products or technologies, user permission or consent is required, and the collection, use and processing of related data must comply with relevant laws, regulations and standards.

[0024] It should be noted that the terms "first," "second," etc., used in the specification, claims, and drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such use of data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in sequences other than those illustrated or described herein.

[0025] In existing technologies, during the physical process of strip steel moving from upstream to downstream, the head and tail sections are cut off, the side edges are trimmed, and the middle section is rolled and elongated along the rolling direction. The physical reference boundary of the material shifts forward due to the head trimming, causing a sudden change in the coordinate origin; the side reference bases shrink inward due to the trimming; and the dimensions are enlarged along the length direction due to the rolling extension. Some processes also involve plate flipping, causing a mirror reversal of the original coordinate system. The defect coordinates recorded upstream no longer point to the same physical location on the downstream physical entity. Linear conversion based on the total length ratio of the material still uses the pre-trimming length ratio for position allocation on the remaining section after the head and tail are cut off, causing the calculated mapped coordinate points to deviate from the actual physical location of the defect. The correlation method based on time-linear interpolation cannot accommodate the inward shift of the lateral reference caused by the trimming, nor can it accommodate the mirror reversal caused by the strip steel flipping. After the mapped coordinates deviate from the physical location, the grayscale data collected by the downstream detection nodes at the deviation points is not the result of the evolution of the original defect. The difference index obtained at this time reflects process fluctuations at irrelevant locations rather than the morphological changes of the defect itself. Some existing technologies that rely on time-series feature comparisons based on misaligned data (such as mean and variance analysis of time-series curves) can only reflect the overall macroscopic fluctuations of process parameters. They fail to separate the geometric effects of the passive elongation and amplification of defect area as the strip rolls under pressure from the actual morphological evolution of defects caused by physical tearing or pressing. The drift of coordinate references and the resulting confusion in evaluation dimensions make it impossible to accurately locate the same surface defect across processes, and its evolutionary state cannot be quantitatively determined.

[0026] To address the aforementioned technical problems, this application provides a method for inspecting the surface quality of steel plates, resolving the issue of cross-process defect coordinate mapping drift and evolution state confusion, and achieving accurate addressing and quantitative determination of the same defect.

[0027] To make the objectives, technical solutions, and advantages of this application clearer, the following description will be provided in conjunction with the accompanying drawings and specific embodiments.

[0028] Please refer to Figure 1 , Figure 1This is a schematic flowchart of a steel plate surface quality inspection method provided in an embodiment of this application. The steel plate surface quality inspection method provided in this embodiment can be executed by a steel plate surface quality inspection device. The device deploys a first edge inspection node directly above the roller conveyor at the exit of the upstream process and a second edge inspection node directly above the roller conveyor at the entrance of the downstream process. A process data central control gateway is connected to a data platform centralized control node. A main control metering node is fixed to the downstream entrance roller conveyor, and an auxiliary guide metering node is fixed to the adjacent roller conveyor of the main control metering node. A downstream finishing terminal, as the execution end of the downstream finishing section, is connected in series after the second edge inspection node. Data interaction between the first edge inspection node, the second edge inspection node, the main control metering node, the auxiliary guide metering node, and the data platform centralized control node is achieved through the process data central control gateway. The process data central control gateway acts as the clock master node, and each edge node synchronizes its time with the gateway via the PTP protocol, with cross-node timestamps aligned to the same time base.

[0029] S1, the coordinates and area of ​​the original defect are fixed.

[0030] The first edge inspection node includes a linear camera array, a surface light source, and an image processing board. The scanning direction is perpendicular to the material's travel direction. The camera array can also be implemented using a surface camera array. The surface camera acquires two-dimensional sub-images with a fixed exposure period and then stitches them into a complete grayscale matrix along the material's length direction within the image processing board. The camera array uses encoder pulses from the same source as the upstream main mill stand as the shutter trigger source. Each valid pulse received acquires one row of pixel data. The row of pixels covers the entire plate surface along the material's width direction. Continuous rows of pixels are stitched into a two-dimensional grayscale matrix along the material's travel direction. The horizontal pixel index of the matrix corresponds to the physical position in the material's width direction, and the vertical pixel index corresponds to the physical position in the material's length direction.

[0031] The image processing board calculates the grayscale difference between each pixel in the grayscale matrix and its eight neighboring pixels, and takes the maximum absolute value as the local grayscale gradient value of that pixel. Pixels with local grayscale gradient values ​​greater than a preset first gradient threshold are marked as candidate defect pixels. The first gradient threshold is set based on the upper limit of the background noise amplitude of the upstream production line under no-load conditions, and typically takes a value corresponding to 5% to 8% of the grayscale quantization level. This value preserves the abrupt change of the defect boundary while suppressing the spurious response caused by iron oxide scale and slag.

[0032] Candidate defect pixels are aggregated into the original defect region through eight-neighbor connected components. The connected component aggregation adopts a two-pass scanning method. In the first pass, local labels are assigned and equivalence relations are recorded by scanning rows. In the second pass, the labels are merged according to the equivalence relations. The aggregation result removes isolated points whose total number of pixels is less than the preset minimum number of connected pixels.

[0033] The image processing board calculates the minimum bounding rectangle of the original defect area. The initial horizontal coordinate is obtained by multiplying the index of the center pixel of the rectangle in the width direction by the physical width of a single pixel. The initial vertical coordinate is obtained by multiplying the index in the length direction by the physical length of a single pixel. The physical references are the left edge of the material and the cut at the head of the material. The initial defect area is obtained by multiplying the total number of pixels in the connected region by the physical area of ​​a single pixel. The physical area of ​​a single pixel is determined by the field of view parameters calibrated by the camera array during the production line debugging phase. The unique identifier ID of the material is written into the cache of the data platform control node. The ID is converted from the steel coil number assigned by the L1 system during the upstream uncoiling stage.

[0034] S2, Feedforward retrieval of physical process parameters and material property parameters.

[0035] The material is wound into finished steel coils by the upstream coiler and then transferred to the downstream uncoiling section by a crane or ground transport device. After the downstream uncoiling machine bites the head of the steel coil and completes the centering, the uncoiling position flag bit of the downstream PLC flips from 0 to 1. The flipping event carries the unique identifier ID and timestamp of the material coil and is sent to the data platform centralized control node through the process data central control gateway.

[0036] The data platform control node initiates parameter retrieval requests to the data interfaces of upstream and downstream L2 systems using the ID as the query key. The upstream L2 returns the cumulative cutting length performed by the cutting shear on the current roll of material as the absolute cutting length. The unit is millimeters; the single-sided trimming width of the return trimming shear on both sides of the material in this roll is used as the single-sided trimming width. The unit is millimeters; the ratio of the cumulative elongation of the returned strip after passing through the downstream mill stand to the original length is used as the comprehensive reduction elongation. Dimensionless; returns the execution status of the rewinding table on the current roll of material as the plate flipping status position. If the operation is reversed, it returns 1; otherwise, it returns 0. Downstream L2 returns the strip grade, export specifications (including nominal export width and nominal export thickness), and target thickness, which together form the material attribute parameters.

[0037] When a timeout window is set for parameter retrieval requests, if the L2 system does not return data within the timeout window, the device uses the historical average of the three most recent rolls of the same steel type and specification as the fallback parameter and writes a parameter degradation flag in the independent label field of the mapping result table. The subsequent quality judgment process will then perform wide-range expansion processing on tolerance α and β based on this flag. , , , After the material attribute parameters reach the data platform control node, they are cached in stage S1. By concatenating the values ​​under the same ID key name, the complete input required for cross-process coordinate mapping is formed.

[0038] S3, cross-process composite branch coordinate mapping.

[0039] The data platform's centralized control node reads the real-time nominal width of the material during subsequent processing from the downstream L2 layer. , The width measurement value is taken from the online measurement of the width meter in the downstream production line. The sampling frequency is the same as that of the meter counter. The physical reference is the distance between the actual edges on both sides of the strip after it is trimmed. This distance is also the actual coverage width of the pixel row of the downstream inspection node. The flipping action is characterized by up-and-down flipping along the length axis of the strip. The flipping action does not change the orientation of the head and tail of the material, and the longitudinal zero point does not change.

[0040] Horizontal mapping basis Branch execution. When the value is 1, the strip plate surface undergoes vertical flipping, and the coordinates are calculated from the left edge of the material in the original coordinate system. The defect point, after being flipped, is located starting from the right edge of the material. Position, superimposed single-sided cut edge width After the inward contraction, the lateral position measured from the downstream left edge is: minus and The difference is used as the target's lateral coordinate. . When the value is 0, the board surface orientation remains consistent, and the position of the defect point relative to the left edge is reduced only due to the inward shrinkage of the cut edge. ,and The difference is . When the value exceeds the lower bound (less than zero), the value is reset to zero and an edge defect is marked. Beyond the upper limit (greater than) )time setting And mark the edge defects.

[0041] Vertical mapping does not branch. minus The local ordinate with the new head end as the zero point is obtained. The local ordinate is scaled up proportionally by λ, and the product of the local ordinate and λ is used as the target ordinate. When the local ordinate is less than zero, the defect has been removed by the cut-off action. The mapping result is discarded and the defect label is sent back to the data platform control node. The mapping result table of the data platform central control node is written along with the ID, and the parameter degradation label of the S2 stage is transmitted along with it.

[0042] S4, Targeted window delivery and local trigger acquisition.

[0043] Data platform control node Centered on the window, a rectangular window is generated by superimposing a preset lateral drift tolerance ±ΔX along the horizontal direction and a preset longitudinal drift tolerance ±ΔY along the vertical direction. The lateral half-width ΔX is set based on the upper limit of the shearing position fluctuation of the downstream cutting edge shear, and the longitudinal half-width ΔY is set based on the upper limit of the measured dispersion of the comprehensive reduction elongation λ. Typical values ​​for ΔX are 10 mm to 30 mm and ΔY are 50 mm to 150 mm. The window center coordinates, lateral half-width, longitudinal half-width, reference timestamp, and the unique identifier ID of this roll of material are encapsulated as the targeted expected window instruction. The data is sent to the second edge inspection node via the process data central control gateway.

[0044] The second edge detection node receives Then, the window parameters are registered in the local mapping table using the ID as the key, and polling of the pulse accumulation value of the main control metering node is initiated. The main control metering node is an incremental encoder installed on the downstream inlet roller conveyor, which can also be replaced by a rotary transformer or an absolute encoder. It outputs a fixed number of pulses per revolution. The physical length corresponding to a single pulse is calibrated by the encoder resolution and the roller conveyor diameter. The pulse accumulation value multiplied by the physical length corresponding to a single pulse gives the real-time length of the material that has entered the downstream. The camera array of the second edge inspection node uses... The calibration input is the width of the starting point of the pixel row. The physical coordinates are used to calculate the corresponding pixel index based on the physical width of a single pixel.

[0045] The pulse frequency of the main control metering node is calculated in real time. When the pulse frequency falls below the preset drop threshold, the device determines that the main control node is slipping, losing pulses, or has an encoder malfunction. It then uses supplementary pulse data from the auxiliary guide metering node. The auxiliary guide metering node is a backup incremental encoder installed on the adjacent roller conveyor of the main control node, maintaining a fixed roller gap with the main control node. The supplementary pulse data is mapped to a distance increment and then added to the already collected real-time length. The preset drop threshold is set based on the downward displacement of the nominal pulse frequency of the main control node at normal production line speed, typically 60% to 70% of the nominal frequency.

[0046] Real-time length falls into Subtracting ΔY gives the lower limit. When the vertical threshold range with ΔY as the upper limit is reached, the second edge detection node triggers a local high-frequency exposure to the camera array, and the exposure only covers... Subtract ΔX to Adding the horizontal pixel range of ΔX, a grayscale sub-image within the viewport is acquired. This sub-image is then processed on the local image processing board at the edge nodes using the first gradient threshold and eight-neighbor connected region aggregation process to extract the evolutionary defect region and calculate its total number of pixels. The total number of pixels is multiplied by the physical area of ​​a single pixel to obtain the evolutionary defect area. .

[0047] If the communication link is lost after the command is issued or during partial data acquisition, the second edge detection node will... The corresponding reference timestamp and ID are cached in local non-volatile storage. The communication link can also be implemented at the underlying level using Industrial Ethernet, PROFINET fieldbus, or EtherCAT bus. After communication is restored, the second edge detection node transmits the cached feature data back based on the reference timestamp and ID key. The data center control node receives the data and uses it to map the data to the corresponding reference timestamp and ID key in the mapping result table. Timestamps perform asynchronous offline alignment, The field representing the evolution result of returning to the original defect area.

[0048] S5, Area Normalization and Three-Branch Adaptive Hierarchy.

[0049] The data platform control node will Divide by The original area magnification factor is obtained. Dividing the magnification factor by the overall compression elongation λ yields the area evolution coefficient K. This division will affect the area along the length direction. The passive geometric amplification is separated from the total amplification effect. K = 1 indicates that the defect area only increases proportionally with rolling and does not undergo active evolution. K > 1 indicates active deterioration. K < 1 indicates active recovery.

[0050] The data platform's centralized control node queries the metallurgical experience database using the strip steel grade and export specification from the material attribute parameters as composite index keys. The metallurgical experience database is a pre-built lookup table, with the strip steel grade as the primary index and the export specification range as the secondary index. Each index node registers the deterioration tolerance α and recovery tolerance β obtained based on historical sample statistics. α and β represent the maximum allowable active deterioration and maximum active recovery range for that grade and specification during the rolling process, respectively. The metallurgical experience database returns α and β that match the current index key; if no results are returned, it reverts to the default α and β values ​​for all specifications of the same steel grade.

[0051] The data platform control node performs a three-branch comparison on K: when K is greater than the sum of the preset evolution benchmark value 1 and α, the quality level is determined to be a fatal degradation level, and the process data control gateway sends a degradation blocking signal to the downstream finishing section. The signal carries ID, When K is less than the difference between the preset evolution baseline value 1 and β, the quality level is judged as a satisfactory recovery level; when K is between or equal to 1-β and 1+α, The data is overwritten to the underlying atlas library, maintaining the baseline quality level corresponding to the original defect areas. For materials with parameter downgrade labels set in the mapping results table, α and β are amplified by a preset expansion factor before comparison.

[0052] After the downgrade blocking signaling is issued, the data platform control node will... Synchronizing with the fatal downgrade level, the downstream finishing terminal constructs a virtual coordinate system with its own entrance as the origin, utilizing the real-time travel length of the material in this roll within the finishing terminal to... Synchronous updates are performed in the virtual coordinate system to generate a dynamic tracking trajectory of the original defect area within the finishing terminal. When the relative distance between the dynamic tracking trajectory and the preset physical resection station within the finishing terminal is less than a preset distance threshold, the data platform control node sends a fixed-point resection command to the finishing terminal, carrying the resection start position and resection length.

[0053] As another implementation method for querying the metallurgical experience database, the data platform control node extracts physical material features from material attribute parameters to determine the corresponding custom weight rules in the preset metallurgical experience database; retrieves the real two-dimensional grayscale image obtained when extracting the original defect area; extracts the multidimensional morphological features of the original defect area from the real two-dimensional grayscale image, the multidimensional morphological features including defect aspect ratio, pore density, and grayscale dispersion; performs fuzzy mapping query on the multidimensional morphological features using the determined custom weight rules, the output of the fuzzy mapping is a confidence scalar characterizing the evolution tendency, and generates α and β corresponding to this query based on the confidence scalar.

[0054] Based on the same inventive concept, this application also provides a steel plate surface quality inspection device for implementing the steel plate surface quality inspection method described above. The solution provided by this system is similar to the solution described in the above method; therefore, the specific limitations in the embodiments of the steel plate surface quality inspection device provided below can be found in the limitations of the steel plate surface quality inspection method described above, and will not be repeated here.

[0055] This application provides a steel plate surface quality inspection device, such as... Figure 2 As shown, the steel plate surface quality inspection device mainly includes the following core functional modules: defect extraction module, parameter extraction module, coordinate calculation module, targeted acquisition module, and quality judgment module. Each module is coupled to the data platform centralized control node through the process data central control gateway.

[0056] The defect extraction module is carried on the image processing board of the first edge inspection node. It is configured to receive continuous rows of pixels output by a linear array camera array or an area array camera array and stitch them into a two-dimensional grayscale matrix according to the material travel direction. For each pixel in the matrix, the grayscale difference between it and its eight neighboring pixels is calculated and the maximum absolute value is taken to obtain the local grayscale gradient value. When the local grayscale gradient value is greater than the preset first gradient threshold, the pixel is marked as a candidate defect pixel and aggregated through the eight neighboring connected domains to form the original defect region. The initial horizontal coordinate, initial vertical coordinate, and initial defect area are extracted from the minimum bounding rectangle of the original defect region. The initial horizontal coordinate, initial vertical coordinate, initial defect area are bound to the unique identifier of the material and written into the cache of the data platform central control node.

[0057] The parameter retrieval module resides on the data platform's centralized control node and is communicatively coupled to the process data central control gateway. It is configured to listen for the uncoiling and positioning flag flip-over event from the downstream PLC. In response to the received material subsequent processing positioning trigger signal corresponding to the unique identifier, it initiates a parameter retrieval request to the upstream and downstream L2 systems using the unique identifier as the query key. This retrieves physical process parameters, including absolute cut length, single-sided cut edge width, overall reduction elongation, and plate flip-over status, as well as material attribute parameters, including strip grade, export specifications, and target thickness. When a parameter retrieval request times out, the module activates a historical average fallback branch, setting the parameter degradation in the independent label field of the mapping result table.

[0058] The coordinate calculation module is hosted on the data platform control node and logically coupled to the output of the parameter retrieval module. It is configured to select between two horizontal branches based on the value of the board flip state bit: the flip branch uses the difference between the real-time nominal width and the initial horizontal coordinate and the single-sided cut edge width as the target horizontal coordinate; the non-flip branch uses the difference between the initial horizontal coordinate and the single-sided cut edge width as the target horizontal coordinate. Vertically, the coordinate calculation module calculates the difference between the initial vertical coordinate and the absolute cut length, and multiplies this difference by the overall reduction elongation rate to obtain the target vertical coordinate. The coordinate calculation module performs edge defect marking on out-of-bounds target horizontal coordinates, and marks defects with a local vertical coordinate less than zero after removing the absolute cut length as removed and discarded from this mapping.

[0059] The targeted acquisition module is distributed between the data platform control node and the second edge inspection node. The data platform control node is configured to generate a target expectation window instruction by overlaying preset process drift tolerances along the horizontal and vertical axes, using the target's horizontal and vertical coordinates as the window center. The instruction carries the window center coordinates, horizontal half-width, vertical half-width, reference timestamp, and unique identifier, and is sent through the process data control gateway. The second edge inspection node is configured to receive the target expectation window instruction and initiate polling of the pulse accumulation value of the main control metering node. The main control metering node is an incremental encoder fixed to the downstream inlet roller conveyor, and the auxiliary guide metering node is a backup incremental encoder fixed to the roller conveyor adjacent to the main control metering node. When the pulse frequency of the main control metering node is lower than a preset drop threshold, the second edge inspection node switches to the supplementary pulse data from the auxiliary guide metering node. When the real-time length falls within the vertical threshold range determined by the target's vertical coordinates and vertical half-width, it triggers local high-frequency exposure of the camera array to extract the evolved defect area. The second edge inspection node is equipped with non-volatile storage and is configured to cache feature data when the communication link is lost. After communication is restored, the data platform control node performs asynchronous offline alignment based on the reference timestamp.

[0060] The quality assessment module is hosted on the data platform control node and logically coupled to the output of the targeted acquisition module and the query interface of the metallurgical experience database. It is configured to obtain the area evolution coefficient by normalizing the ratio of the evolved defect area to the initial defect area using the comprehensive reduction elongation rate. The module queries the metallurgical experience database using the strip steel grade and export specification as composite index keys to dynamically generate the deterioration tolerance and the recovery tolerance. When the area evolution coefficient is greater than the sum of the preset evolution benchmark value and the deterioration tolerance, it is judged as a fatal downgrade level and a downgrade blocking signal is issued through the process data control gateway. When the area evolution coefficient is less than the difference between the preset evolution benchmark value and the recovery tolerance, it is judged as a qualified recovery level. When the area evolution coefficient is between or equal to the difference between the preset evolution benchmark value and the recovery tolerance, and between the preset evolution benchmark value and the deterioration tolerance, the evolved defect area is overwritten to the underlying map library and the benchmark quality level corresponding to the original defect area is maintained. When the parameter degradation label is set in the mapping result table, the quality judgment module amplifies the deterioration tolerance and the recovery tolerance by a preset expansion coefficient. If no result is found in the metallurgical experience database, it reverts to the default tolerance for all specifications of the same steel type.

[0061] The quality judgment module is logically coupled to the downstream finishing terminal. The downstream finishing terminal is configured to construct a virtual coordinate system with the local entry point as the origin. Based on the real-time travel length of the material in the finishing terminal, the longitudinal coordinates of the target are updated in the virtual coordinate system to generate a dynamic tracking trajectory of the original defect area. When the relative distance between the dynamic tracking trajectory and the preset physical cutting station is less than the preset distance threshold, the module receives a fixed-point cutting instruction issued by the data center control node.

[0062] The process data central control gateway acts as the clock master node, synchronizing with each edge node via the PTP protocol, ensuring that cross-node timestamps are aligned to the same time base. The camera array in the device is implemented using a linear or area array, the main control metering node is implemented using an incremental encoder, absolute encoder, or rotary transformer, and the communication link is implemented using Industrial Ethernet, PROFINET fieldbus, or EtherCAT bus.

[0063] See Figure 3 , Figure 3 This is a schematic block diagram of an electronic device provided according to an embodiment of this application. Figure 3 The electronic device 300 in this embodiment may include one or more processors 301, one or more input devices 302, one or more output devices 303, and one or more memories 304. The processors 301, input devices 302, output devices 303, and memories 304 communicate with each other via a communication bus 305. The memories 304 store computer programs, including program instructions. The processors 301 execute the program instructions stored in the memories 304. Specifically, the processors 301 are configured to invoke the program instructions to perform the functions of each module / unit in the above-described device embodiments, for example... Figure 2 The functions of each module are shown.

[0064] It should be understood that, in the embodiments of this application, the processor 301 may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor.

[0065] Input device 302 may include a touchpad, a fingerprint sensor (for collecting the user's fingerprint information and fingerprint orientation information), a microphone, etc., and output device 303 may include a display (LCD, etc.), a speaker, etc.

[0066] The memory 304 may include read-only memory and random access memory, and provides instructions and data to the processor 301. A portion of the memory 304 may also include non-volatile random access memory. For example, the memory 304 may also store device type information.

[0067] In specific implementations, the processor 301, input device 302, and output device 303 described in the embodiments of this application can execute the implementation methods described in the embodiments of this application, or they can execute the implementation methods of the electronic devices described in the embodiments of this application, which will not be repeated here.

[0068] In another embodiment of this application, a computer-readable storage medium is provided. This computer-readable storage medium stores a computer program, which includes program instructions. When executed by a processor, the program instructions implement all or part of the processes in the methods described above. Alternatively, the computer program can instruct related hardware to implement these processes. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or certain intermediate forms. The computer-readable medium can include any entity or device capable of carrying computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc.

[0069] The computer-readable storage medium can be an internal storage unit of the electronic device in any of the foregoing embodiments, such as a hard disk or memory of the electronic device. The computer-readable storage medium can also be an external storage device of the electronic device, such as a plug-in hard disk, SmartMediaCard (SMC), Secure Digital (SD) card, FlashCard, etc., equipped on the electronic device. Furthermore, the computer-readable storage medium can include both internal and external storage units of the electronic device. The computer-readable storage medium is used to store computer programs and other programs and data required by the electronic device. The computer-readable storage medium can also be used to temporarily store data that has been output or will be output.

[0070] Those skilled in the art will recognize that the modules / units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this application.

[0071] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the electronic devices and units described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0072] In the several embodiments provided in this application, it should be understood that the disclosed electronic devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For instance, the division of modules / units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules, units, or components may be combined or integrated into another system, or some features may be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces or modules / units, or it may be an electrical, mechanical, or other form of connection.

[0073] The modules / units described as separate components may or may not be physically separate. Similarly, the components shown as modules / units may or may not be physical modules / units; they may be located in one place or distributed across multiple network modules / units. Some or all of the modules / units can be selected to achieve the purpose of the embodiments of this application, depending on actual needs.

[0074] Furthermore, the functional modules / units in the various embodiments of this application can be integrated into one processing module / unit, or each module / unit can exist physically separately, or two or more modules / units can be integrated into one module / unit. The integrated modules / units described above can be implemented in hardware or in the form of software functional modules / units.

[0075] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in this application, and these modifications or substitutions should all be covered within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for inspecting the surface quality of steel plates, characterized in that, include: When the local grayscale gradient value on the material surface is greater than the preset first gradient threshold, the original defect region is determined, and the initial horizontal coordinate, initial vertical coordinate and initial defect area of ​​the original defect region are extracted and bound to the unique identifier of the material. In response to the received material subsequent processing in place trigger signal corresponding to the unique identifier, the physical process parameters and material attribute parameters of the material in the cross-process flow process are obtained. The physical process parameters include absolute cut length, single-sided cut width, comprehensive pressing elongation and plate flipping status position. The target lateral coordinate is calculated based on the plate flipping state position, the initial lateral coordinate, and the single-sided cutting edge width. The difference between the initial longitudinal coordinate and the absolute cutting length is calculated, and the difference is multiplied by the comprehensive compression elongation rate to obtain the target longitudinal coordinate. The target expected window is determined based on the target horizontal coordinate and the target vertical coordinate. When the material enters the vertical interval corresponding to the target expected window, the local data acquisition is triggered to extract the evolved defect area after the original defect area has evolved, based on the length measurement data of the material acquired in real time during the subsequent processing. The ratio of the evolved defect area to the initial defect area is normalized by the comprehensive reduction elongation rate to obtain the area evolution coefficient. Based on the material property parameters, a preset metallurgical experience library is queried to dynamically generate a deterioration tolerance and a recovery tolerance that match the material property parameters. The evolved quality level is classified and determined according to the comparison results of the area evolution coefficient, the deterioration tolerance, and the recovery tolerance.

2. The method for detecting the surface quality of steel plates according to claim 1, characterized in that, The step of calculating the target lateral coordinate based on the board flip state position, the initial lateral coordinate, and the single-sided cut edge width includes: Obtain the real-time nominal width of the material during the subsequent processing; When the flip state position indicates that a flip has occurred, the target horizontal coordinate is obtained by subtracting the difference between the initial horizontal coordinate and the single-sided cut edge width from the real-time nominal width. When the board flip state position indicates that no flip has occurred, the difference between the initial lateral coordinate and the width of the single-sided cut edge is used as the target lateral coordinate.

3. The method for detecting the surface quality of steel plates according to claim 2, characterized in that, The step of determining the target expected view window based on the target's horizontal and vertical coordinates includes: Obtain a preset process drift tolerance that matches the subsequent processing steps; The target expectation window is generated by superimposing the preset process drift tolerance along the horizontal and vertical axes with the target horizontal and vertical axes as the origin. After generating the target expectation window, the process also includes: The target expectation window is encapsulated into a control command carrying a reference timestamp and issued to trigger the caching of feature data acquired by the local data acquisition when the communication link is determined to be disconnected, and to perform asynchronous offline alignment of the cached feature data based on the reference timestamp after the communication is restored.

4. The method for detecting the surface quality of steel plates according to claim 3, characterized in that, When the material enters the longitudinal section corresponding to the target expected window based on the length measurement data of the material acquired in real time during subsequent processing, local data acquisition is triggered to extract the evolved defect area of ​​the original defect region, including: Obtain the pulse frequency of the master control node used to collect the length measurement data; When the pulse frequency is lower than the preset drop threshold, supplementary pulse data of the auxiliary guide node fixed adjacent to the main control node is extracted; The supplementary pulse data is mapped to a distance increment and accumulated with the collected length measurement data to determine whether the material has entered the longitudinal interval based on the accumulation result; When it is determined that the material has entered the longitudinal section, local data acquisition is triggered to extract the evolved defect area after the original defect area has evolved.

5. The method for detecting the surface quality of steel plates according to claim 4, characterized in that, The classification and determination of the evolved quality level based on the comparison results of the area evolution coefficient, the deterioration tolerance, and the smoothing tolerance includes: When the area evolution coefficient is greater than the sum of the preset evolution benchmark value and the deterioration tolerance, the quality level is determined to be a fatal degradation level and a degradation blocking signal is output. When the area evolution coefficient is less than the difference between the preset evolution benchmark value and the smoothing tolerance, the quality level is determined to be a qualified smoothing level; When the area evolution coefficient is between or equal to the difference between the preset evolution benchmark value and the smoothing tolerance, and between the preset evolution benchmark value and the deterioration tolerance, the area of ​​the evolution defect is overwritten to the underlying map library, and the benchmark quality level corresponding to the original defect area is maintained.

6. The method for detecting the surface quality of steel plates according to claim 5, characterized in that, After outputting the downgrade blocking signaling, the following is also included: The target's horizontal coordinates, vertical coordinates, and fatal degradation level are synchronized to the preset downstream finishing terminal; A virtual coordinate system is constructed, and the longitudinal coordinates of the target are updated in the virtual coordinate system using the real-time travel length of the material in the downstream finishing terminal, thereby generating a dynamic tracking trajectory of the original defect area. When the relative distance between the dynamic tracking trajectory and the preset physical cutting station in the downstream finishing terminal is less than a preset distance threshold, a fixed-point cutting instruction is issued to the downstream finishing terminal.

7. The method for detecting the surface quality of steel plates according to claim 1, characterized in that, The preset metallurgical experience base is configured with custom weight rules that are bound to the physical material characteristics in the material attribute parameters; The method of dynamically generating deterioration tolerance and recovery tolerance that match the material attribute parameters by querying a preset metallurgical experience base based on the material attribute parameters includes: Retrieve the actual two-dimensional grayscale image obtained during the extraction of the original defect region; The multidimensional morphological features of the original defect region are extracted from the real two-dimensional grayscale image. The multidimensional morphological features include the defect aspect ratio, hole density and grayscale dispersion. Extract the physical material characteristics from the material attribute parameters to determine the corresponding custom weight rules in the preset metallurgical experience library; The multidimensional morphological features are input into the preset metallurgical experience base as query conditions. Fuzzy mapping query is performed on the multidimensional morphological features using the determined custom weight rules. A confidence scalar representing the evolution tendency is output. The deterioration tolerance and the recovery tolerance are generated based on the confidence scalar.

8. A steel plate surface quality inspection device, used to perform the steel plate surface quality inspection method as described in any one of claims 1 to 7, characterized in that, include: The defect extraction module is used to determine the original defect region when the local grayscale gradient value on the material surface is greater than a preset first gradient threshold, extract the initial horizontal coordinate, initial vertical coordinate and initial defect area of ​​the original defect region, and bind them with the unique identifier of the material. The parameter retrieval module is used to respond to the received material subsequent processing placement trigger signal corresponding to the unique identifier, and to obtain the physical process parameters and material attribute parameters of the material in the cross-process flow process. The physical process parameters include absolute cutting length, single-sided cutting edge width, comprehensive pressing elongation and plate flipping status position. The coordinate calculation module is used to calculate the target lateral coordinate based on the plate flipping state position, the initial lateral coordinate and the single-sided cutting edge width, calculate the difference between the initial longitudinal coordinate and the absolute cutting head length, and multiply the difference by the comprehensive compression elongation rate to obtain the target longitudinal coordinate. The target acquisition module is used to determine the target expected window based on the target's horizontal coordinate and the target's vertical coordinate, and to determine the longitudinal interval corresponding to the target expected window when the material enters the longitudinal interval corresponding to the target expected window based on the length measurement data of the material acquired in real time during the subsequent processing. This triggers local data acquisition to extract the evolved defect area after the original defect area has evolved. The quality assessment module is used to normalize the ratio of the evolved defect area to the initial defect area using the comprehensive reduction elongation rate to obtain the area evolution coefficient, query a preset metallurgical experience library based on the material attribute parameters to dynamically generate a deterioration tolerance and a recovery tolerance that match the material attribute parameters, and classify and determine the evolved quality level based on the comparison results of the area evolution coefficient, the deterioration tolerance, and the recovery tolerance.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 7.

Images