Industrial vision-based cold-rolled strip edge crack online early warning method and system

Abstract

The present application relates to the field of industrial vision technology, and proposes an online early warning method and system for edge cracks of cold-rolled strip steel based on industrial vision, which comprises the following steps: taking the connected boundary contour, geometric center position and maximum extension length along the running direction of the strip steel in the differentiated crack feature area of the edge surface of the strip steel as the crack labeling data; determining the crack evolution trajectory of the edge surface of the strip steel according to the geometric center position of the same crack object at different times and the maximum extension length at different times in the crack labeling data; and dynamically comparing the current crack length growth rate in the crack evolution with the preset crack propagation reference state to obtain the early warning information of the edge surface of the strip steel; and the present application can improve the efficiency of online early warning for edge cracks of cold-rolled strip steel.

Description

Technical Field

[0001] This invention relates to the field of industrial vision technology, and in particular to an online early warning method and system for edge cracks in cold-rolled strip steel based on industrial vision. Background Technology

[0002] Existing online detection technology for edge cracks in cold-rolled strip can only extract static crack features and cannot continuously track the dynamic evolution of cracks. It is difficult to accurately obtain information on the position changes and length extension of the same crack at different times. Existing technologies have not formed a standardized crack feature annotation system and cannot fully extract the crack connection boundary contour, geometric center position, and maximum extension length along the strip running direction. Crack feature data collection has gaps and biases.

[0003] Existing edge crack early warning technologies for cold-rolled strip steel only use a single static judgment rule and do not combine crack length growth rate for dynamic comparative analysis. They cannot distinguish between normal and accelerated crack propagation states. Existing early warning methods lack accurate judgment basis for crack propagation trends, and the generation of early warning information lacks accuracy and timeliness, making it difficult to meet the technical requirements for real-time prevention and control of edge cracks in the continuous production process of cold-rolled strip steel. Summary of the Invention

[0004] This invention provides an online early warning method and system for edge cracks in cold-rolled strip steel based on industrial vision, in order to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides an online early warning method for edge cracks in cold-rolled strip steel based on industrial vision, comprising: Pt.1, The connected boundary contour, geometric center position and maximum extension length along the strip running direction in the differential crack feature region on the edge surface of the strip are used as crack annotation data; Pt.2. Based on the geometric center position and maximum extension length of the same crack object at different times in the crack annotation data, determine the crack evolution trajectory on the surface of the strip edge. Pt.3. Dynamically compare the current crack length growth rate during crack evolution with the preset crack propagation baseline state to obtain early warning information for the strip edge surface.

[0006] In a preferred embodiment, the step of using the connected boundary contour, geometric center position, and maximum extension length along the strip running direction in the differentiated crack feature region on the strip edge surface as crack annotation data includes: A closed boundary trajectory is drawn pixel by pixel along the outer edge of the differential crack feature region on the edge surface of the strip to obtain the connected boundary contour of the edge surface of the strip. The geometric center is located at the intersection of the horizontal and vertical midlines of the differentiated crack feature region in the image plane. The span of the differential crack feature region is measured along the running direction of the strip edge surface to obtain the maximum extension length of the strip edge surface; By combining the connected boundary contour, geometric center position, and maximum extension length in a data association manner, crack annotation data of the strip edge surface is obtained.

[0007] In a preferred embodiment, the step of drawing a closed boundary trajectory pixel by pixel along the outer edge of the differential crack feature region on the strip edge surface to obtain the connected boundary contour of the strip edge surface includes: The pixel located at the top left corner of the differential crack feature region is retrieved as the starting point for tracking; Taking the starting point of tracking as the current observation point, explore adjacent pixels in the order of right, up, left and down directions, and determine the first adjacent pixel that belongs to the differential crack feature region as the boundary advancement point. Record the line connecting the current observation point and the boundary advancement point as a boundary line segment, and then update the boundary advancement point to the new current observation point; Each time, a boundary line segment is recorded and the current observation point is updated until the current observation point returns to the starting point of the tracking, thus obtaining the connected boundary profile of the strip edge surface.

[0008] In a preferred embodiment, the step of measuring the span of the differentiated crack feature region along the running direction of the strip edge surface to obtain the maximum extension length of the strip edge surface includes: In the image of the differential crack feature region, a reference scan line parallel to the strip running direction is established, and the reference scan line is placed outside the leading edge of the differential crack feature region. The reference scan line is translated sequentially along the running direction of the strip. After each translation, the intersection of the reference scan line and the differential crack feature area is checked, and the state of the intersection between the scan line and the area after each translation is recorded. When the reference scan line is shifted to a point where it no longer intersects with the differential crack feature region, the scan line is traced back to the last point where it intersected, thus obtaining the last intersection point on the strip edge surface. A straight line parallel to the strip's running direction is drawn from the leading edge of the differentiated crack feature region and extended to the final intersection point to obtain the maximum extension length of the strip's edge surface.

[0009] In a preferred embodiment, determining the crack evolution trajectory on the strip edge surface based on the geometric center position and maximum extension length of the same crack object at different times in the crack marking data includes: Retrieve all records belonging to the same crack object from the crack annotation data, extract the geometric center position of each record in the temporal order, and obtain the temporal sequence of the center position of the strip edge surface; For the time sequence of the center position, a direction mapping line segment is established from the geometric center position at the earlier time to the geometric center position at the later time to obtain the center migration vector of the strip edge surface; Retrieve all records belonging to the same crack object from the crack annotation data, extract the maximum extension length in each record according to the time sequence, and obtain the time sequence of the extension length of the strip edge surface; For the maximum extension length of each pair of adjacent moments in the extension length time sequence, an extension increment line segment is drawn from the end position of the maximum extension length at the earlier moment along the strip running direction to obtain the tip extension vector of the strip edge surface. Using the earliest geometric center position as the origin of the trajectory, the center migration vector and the tip expansion vector are sequentially connected end to end to obtain the crack evolution trajectory on the edge surface of the strip.

[0010] In a preferred embodiment, the step of establishing a direction mapping line segment from the geometric center position at an earlier time to the geometric center position at a later time for the center position time sequence, to obtain the center migration vector of the strip edge surface, includes: In the time sequence of the center position, the geometric center position with the earlier acquisition time is used as the mapping starting point, and the geometric center position with the later acquisition time is used as the mapping ending point. Starting from the mapping origin, draw a directed line segment along a straight path towards the mapping endpoint; The deflection angle of the directed line segment relative to the running direction of the strip is detected to obtain the center migration vector of the strip edge surface.

[0011] In a preferred embodiment, for each pair of adjacent times in the elongation length time sequence, drawing an extension increment line segment from the end position of the maximum elongation length at the earlier time along the strip running direction to obtain the tip extension vector of the strip edge surface includes: The maximum extension length at two adjacent moments is retrieved from the extension length time sequence. The crack tip position corresponding to the maximum extension length acquired earlier is taken as the extension reference point, and the maximum extension length acquired later is taken as the target extension length. Starting from the extension reference point, draw a ray along the running direction of the strip, and mark the target point on the ray whose distance from the starting point is equal to the target extension length. The line connecting the extension reference point and the target point is cut out to form a directed line segment with the starting point at the original crack tip and the ending point at the new crack tip, thus obtaining a single-step extension line segment on the edge surface of the strip. The length and direction of the single-step extension segment are attached as characteristic parameters to the single-step extension segment to obtain the tip extension vector of the strip edge surface.

[0012] In a preferred embodiment, determining the crack evolution trajectory on the edge surface of the strip further includes: The deflection angle of the directed line segment is extracted from the central migration vector and denoted as the first direction angle; The length of the directed line segment extracted from the tip extension vector is denoted as the extension step size; Using the natural constant as the base and the product of the expansion step size and the preset direction sensitivity coefficient as the exponent, an expansion trend factor is constructed. The calculation formula for the expansion trend factor is as follows: ; In the formula, To expand the trend factor, To extend the step size, The first direction angle, The preset reference length, The preset direction sensitivity coefficient, It is a natural constant; If the expansion trend factor exceeds the preset warning threshold, an accelerated crack expansion indicator is generated and attached to the crack evolution trajectory.

[0013] In a preferred embodiment, the step of dynamically comparing the current crack length growth rate during crack evolution with a preset crack propagation baseline state to obtain early warning information for the strip edge surface includes: Compare the length of the previous expansion step size with the current expansion step size in the crack evolution trajectory. If the current expansion step size is greater than the expansion step size at the previous time step, then an increment step size flag is generated. The number of consecutive occurrences of the step increment indicator is compared with the preset threshold value of the number of step increments in the crack propagation reference state. At the same time, the ratio of the difference between the current propagation step size and the previous propagation step size to the previous propagation step size is compared with the threshold value of the step size increase ratio in the crack propagation reference state. A Level 1 warning is generated when the number of consecutive occurrences exceeds the threshold for the number of steps incremented and the proportion exceeds the threshold for the proportion of steps incremented; otherwise, a Level 2 warning is generated.

[0014] To address the aforementioned problems, this invention also provides an online early warning system for edge cracks in cold-rolled strip steel based on industrial vision, the system comprising: The crack feature calibration module is used to take the connected boundary contour, geometric center position and maximum extension length along the strip running direction of the differential crack feature region on the edge surface of the strip as crack annotation data; The trajectory evolution construction module is used to determine the crack evolution trajectory of the strip edge surface based on the geometric center position and maximum extension length of the same crack object at different times in the crack annotation data. The extended early warning judgment module is used to dynamically compare the current crack length growth rate during crack evolution with the preset crack propagation benchmark state to obtain early warning information for the edge surface of the strip steel.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention obtains the outline of the connected boundary by tracing the closed boundary trajectory pixel by pixel along the outer edge of the crack feature area on the edge of the strip. It determines the geometric center position by locating the intersection of the horizontal and vertical central axes in the image plane of the crack area. It measures the maximum extension length of the crack by translating the scan line along the running direction of the strip. It associates the three types of core features to form standardized crack annotation data, which can completely, accurately and comprehensively depict the morphology, spatial location and extension scale of the crack. This makes crack feature acquisition more standardized, data more complete and information more accurate, and provides reliable data support for crack dynamic analysis.

[0016] 2. This invention constructs a crack evolution trajectory, including center migration and tip propagation, based on the geometric center position and maximum extension length of the same crack at different times. This clearly reconstructs the dynamic development process of the crack. Then, the real-time crack length growth rate is dynamically matched and compared with the preset extension benchmark state. Based on the extension change law, graded early warning information is generated. This invention can identify the crack extension state in real time, accurately and sensitively, greatly improving the response speed, judgment accuracy and early warning reliability of online early warning of edge cracks in cold-rolled strip steel, making online early warning more efficient and more in line with the actual needs of continuous industrial production. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating an embodiment of the present invention for an online early warning method for edge cracks in cold-rolled strip steel based on industrial vision. Figure 2 A functional module diagram of an online early warning system for edge cracks in cold-rolled strip steel based on industrial vision provided in an embodiment of the present invention; The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0018] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0019] This application provides an online early warning method for edge cracks in cold-rolled strip steel based on industrial vision. The executing entity of this method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this application: a server, a terminal, etc. In other words, the online early warning method for edge cracks in cold-rolled strip steel based on industrial vision can be executed by software or hardware installed on a terminal device or a server device. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cloud server cluster. The server can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.

[0020] Reference Figure 1 The diagram shown is a flowchart illustrating an online early warning method for edge cracks in cold-rolled strip steel based on industrial vision, according to an embodiment of the present invention. In this embodiment, the online early warning method for edge cracks in cold-rolled strip steel based on industrial vision includes: Pt.1, The connected boundary contour, geometric center position and maximum extension length along the strip running direction in the differential crack feature region on the edge surface of the strip are used as crack annotation data; In this embodiment of the invention, the step of using the connected boundary contour, geometric center position, and maximum extension length along the strip running direction in the differentiated crack feature region of the strip edge surface as crack annotation data includes: A closed boundary trajectory is drawn pixel by pixel along the outer edge of the differential crack feature region on the edge surface of the strip to obtain the connected boundary contour of the edge surface of the strip. The geometric center is located at the intersection of the horizontal and vertical midlines of the differentiated crack feature region in the image plane. The span of the differential crack feature region is measured along the running direction of the strip edge surface to obtain the maximum extension length of the strip edge surface; By combining the connected boundary contour, geometric center position, and maximum extension length in a data association manner, crack annotation data of the strip edge surface is obtained.

[0021] A closed boundary trajectory is drawn pixel by pixel along the outer edge of the differential crack feature region on the strip edge surface to obtain the connected boundary contour of the strip edge surface, including: The pixel located at the top left corner of the differential crack feature region is retrieved as the starting point for tracking; Taking the starting point of tracking as the current observation point, explore adjacent pixels in the order of right, up, left and down directions, and determine the first adjacent pixel that belongs to the differential crack feature region as the boundary advancement point. Record the line connecting the current observation point and the boundary advancement point as a boundary line segment, and then update the boundary advancement point to the new current observation point; Each time, a boundary line segment is recorded and the current observation point is updated until the current observation point returns to the starting point of the tracking, thus obtaining the connected boundary profile of the strip edge surface.

[0022] The step of measuring the span of the differentiated crack characteristic region along the running direction of the strip edge surface to obtain the maximum extension length of the strip edge surface includes: In the image of the differential crack feature region, a reference scan line parallel to the strip running direction is established, and the reference scan line is placed outside the leading edge of the differential crack feature region. The reference scan line is translated sequentially along the running direction of the strip. After each translation, the intersection of the reference scan line and the differential crack feature area is checked, and the state of the intersection between the scan line and the area after each translation is recorded. When the reference scan line is shifted to a point where it no longer intersects with the differential crack feature region, the scan line is traced back to the last point where it intersected, thus obtaining the last intersection point on the strip edge surface. A straight line parallel to the strip's running direction is drawn from the leading edge of the differentiated crack feature region and extended to the final intersection point to obtain the maximum extension length of the strip's edge surface.

[0023] In the image data acquired from the edge surface of the strip, the entire set of pixels covered by the differential crack feature region is located. The horizontal and vertical coordinate values ​​of each pixel in the set are traversed one by one. The horizontal coordinate values ​​of all pixels are sorted from smallest to largest, and the vertical coordinate values ​​of all pixels are sorted from smallest to largest. The unique pixel with the smallest horizontal and vertical coordinate values ​​is selected and fixed as the starting point for tracking the differential crack feature region.

[0024] Using the tracking starting point as the initial current observation point, strictly following the fixed priority order of adjacent pixels on the right, top, left, and bottom, check whether adjacent pixels in each direction belong to the differentiated crack feature region. The first adjacent pixel that is determined to belong to the region during the check is directly determined as the current boundary advancement point.

[0025] In the image plane, connect the pixel coordinates of the current inspection point with the pixel coordinates of the boundary advance point with a straight line. Record the line formed by this connection as a single boundary line segment. Then, overwrite the original pixel coordinates of the current inspection point with the pixel coordinates of the boundary advance point, making the boundary advance point the current inspection point for the new round of inspection.

[0026] Repeatedly execute the boundary segment recording and current observation point update operation. After each operation, compare the pixel coordinates of the current observation point with the pixel coordinates of the initial tracking starting point in real time. When the pixel coordinates of the current observation point and the pixel coordinates of the initial tracking starting point completely coincide, terminate all boundary tracking operations. The closed, gap-free trajectory formed by connecting all single boundary segments in sequence is the connected boundary contour of the strip edge surface.

[0027] Within the image plane of the strip edge surface, extract the lateral coordinate values ​​of all pixels covered by the differential crack feature region, select the leftmost pixel coordinate with the smallest value and the rightmost pixel coordinate with the largest value, add the values ​​of the leftmost and rightmost pixel coordinates together and take the average value, and determine the coordinates corresponding to the average value as the horizontal central axis of the differential crack feature region.

[0028] Extract the vertical coordinate values ​​of all pixels covered by the differential crack feature region, select the topmost pixel coordinate with the smallest value and the bottommost pixel coordinate with the largest value, add the values ​​of the topmost pixel coordinate and the bottommost pixel coordinate and take the average value, and determine the coordinate corresponding to the average value as the vertical central axis of the differential crack feature region.

[0029] Locate the unique coordinate point in the image plane where the horizontal and vertical central axes intersect. This coordinate point is the geometric center of the differentiated crack feature region and also serves as the geometric center of the strip edge surface.

[0030] In the image corresponding to the differential crack feature area, a straight reference scan line is drawn according to the actual running direction of the strip during the production process. The reference scan line is made to be completely parallel and without deviation from the running direction of the strip through direction calibration. The reference scan line is placed outside the leading edge of the differential crack feature area, and a distance of only one image pixel unit is retained between the reference scan line and the leading edge of the differential crack feature area.

[0031] Using a fixed single image pixel unit as the moving distance for each translation, the reference scan line is gradually translated along the strip running direction. After each translation operation is completed, the position of each pixel on the reference scan line is checked one by one to determine whether the pixel overlaps with the pixel in the differential crack feature area. If there is overlap, it is marked as an intersection state; if there is no overlap, it is marked as a non-intersection state. The corresponding state marking results after each translation are completely stored.

[0032] The baseline scan line translation and pixel position verification operations are continuously performed. When the verification result shows that the baseline scan line and the differential crack feature area are completely non-intersecting, all stored state mark results are retrieved, and the image coordinate position of the baseline scan line when it was last marked as intersecting is found in reverse. This coordinate position is the last intersection position of the strip edge surface.

[0033] Starting from the center pixel coordinates of the leading edge of the differential crack feature region, draw a straight line that is completely parallel to the strip running direction. Extend this line along the strip running direction to the last intersection point on the edge surface of the strip, and measure the actual physical length of the extended line in the image. This length value is the maximum extension length of the strip edge surface.

[0034] The three feature data of the strip edge surface—connected boundary contour, geometric center position, and maximum extension length—are classified and matched according to the unique identifier of the same differentiated crack feature region. This ensures that the three feature data form a fixed corresponding binding relationship with a single crack feature region, and that each set of connected boundary contour, geometric center position, and maximum extension length is accurately assigned to the corresponding differentiated crack feature region.

[0035] The three feature data that have been bound are integrated and stored according to a fixed data structure to form a complete data set containing crack morphology, spatial location and extension scale. This data set is the crack annotation data of the strip edge surface.

[0036] Pt.2. Based on the geometric center position and maximum extension length of the same crack object at different times in the crack annotation data, determine the crack evolution trajectory on the surface of the strip edge. In this embodiment of the invention, determining the crack evolution trajectory of the strip edge surface based on the geometric center position and maximum extension length of the same crack object at different times in the crack marking data includes: Retrieve all records belonging to the same crack object from the crack annotation data, extract the geometric center position of each record in the temporal order, and obtain the temporal sequence of the center position of the strip edge surface; For the time sequence of the center position, a direction mapping line segment is established from the geometric center position at the earlier time to the geometric center position at the later time to obtain the center migration vector of the strip edge surface; Retrieve all records belonging to the same crack object from the crack annotation data, extract the maximum extension length in each record according to the time sequence, and obtain the time sequence of the extension length of the strip edge surface; For the maximum extension length of each pair of adjacent moments in the extension length time sequence, an extension increment line segment is drawn from the end position of the maximum extension length at the earlier moment along the strip running direction to obtain the tip extension vector of the strip edge surface. Using the earliest geometric center position as the origin of the trajectory, the center migration vector and the tip expansion vector are sequentially connected end to end to obtain the crack evolution trajectory on the edge surface of the strip.

[0037] For the time sequence of the center position, a direction mapping line segment is established from the geometric center position at a earlier time to the geometric center position at a later time to obtain the center migration vector of the strip edge surface, including: In the time sequence of the center position, the geometric center position with the earlier acquisition time is used as the mapping starting point, and the geometric center position with the later acquisition time is used as the mapping ending point. Starting from the mapping origin, draw a directed line segment along a straight path towards the mapping endpoint; The deflection angle of the directed line segment relative to the running direction of the strip is detected to obtain the center migration vector of the strip edge surface.

[0038] For each pair of adjacent times in the elongation length time sequence, an extension increment line segment is drawn from the end position of the maximum elongation length at the earlier time along the strip running direction to obtain the tip extension vector of the strip edge surface, including: The maximum extension length at two adjacent moments is retrieved from the extension length time sequence. The crack tip position corresponding to the maximum extension length acquired earlier is taken as the extension reference point, and the maximum extension length acquired later is taken as the target extension length. Starting from the extension reference point, draw a ray along the running direction of the strip, and mark the target point on the ray whose distance from the starting point is equal to the target extension length. The line connecting the extension reference point and the target point is cut out to form a directed line segment with the starting point at the original crack tip and the ending point at the new crack tip, thus obtaining a single-step extension line segment on the edge surface of the strip. The length and direction of the single-step extension segment are attached as characteristic parameters to the single-step extension segment to obtain the tip extension vector of the strip edge surface.

[0039] The determination of the crack evolution trajectory on the edge surface of the strip also includes: The deflection angle of the directed line segment is extracted from the central migration vector and denoted as the first direction angle; The length of the directed line segment extracted from the tip extension vector is denoted as the extension step size; Using the natural constant as the base and the product of the expansion step size and the preset direction sensitivity coefficient as the exponent, an expansion trend factor is constructed. The calculation formula for the expansion trend factor is as follows: ; In the formula, To expand the trend factor, To extend the step size, The first direction angle, The preset reference length, The preset direction sensitivity coefficient, It is a natural constant; If the expansion trend factor exceeds the preset warning threshold, an accelerated crack expansion indicator is generated and attached to the crack evolution trajectory.

[0040] Examine each crack acquisition record stored in the crack annotation data. Based on the unique number of the crack object carried by each record, filter out all data entries with the same unique number to determine all acquisition data belonging to the same crack object. Sort all the selected data entries according to the value of the acquisition timestamp from smallest to largest to complete the chronological order. Accurately extract the horizontal and vertical coordinate information of the geometric center position from each sorted data entry. Piece together all the extracted geometric center position coordinates in the order of acquisition time from earliest to latest to finally form the time sequence of the center position of the strip edge surface.

[0041] According to the acquisition time sequence of the center position of the strip edge surface, the geometric center position coordinates of the earlier time in the sequence are selected as the starting endpoint of the directed line segment, and the geometric center position coordinates of the adjacent later time are selected as the ending endpoint of the directed line segment. The starting endpoint and the ending endpoint are connected by a straight line without curvature in the image plane. At the same time, the fixed direction from the starting endpoint to the ending endpoint is marked on the straight line to form a straight line segment with a clear direction. The coordinates, length and direction information of the straight line segment together constitute the center migration vector of the strip edge surface.

[0042] The crack annotation data is re-traversed and all data entries belonging to the same crack object are filtered out again based on the unique number of the crack object. These data entries are sorted by the collection timestamp value from smallest to largest completion time. The actual physical length value of the maximum extension length is extracted from each sorted data entry. All extracted maximum extension length values ​​are combined and arranged in order from earliest to latest collection time to form the extension length time sequence of the strip edge surface.

[0043] Two sets of maximum extension length data with adjacent acquisition times are selected from the extension length time sequence of the strip edge surface. The coordinate position of the crack tip end corresponding to the maximum extension length with earlier acquisition time is located by data association. This coordinate position is set as the extension start point. Starting from this extension start point, a directed straight line segment is drawn along the strip running direction pointing to the coordinate position of the crack tip end corresponding to the maximum extension length with later acquisition time. The length of this directed straight line segment and the direction information together constitute the tip extension vector of the strip edge surface.

[0044] The earliest geometric center coordinates in the time sequence of the location of the center position of the strip edge surface are used as the starting point of the crack evolution trajectory. All center migration vectors and tip propagation vectors generated in chronological order are sequentially connected end to end to ensure that the termination point of the previous vector coincides with the starting point of the next vector. The continuous and complete trajectory formed by connecting all vectors in sequence is the crack evolution trajectory of the strip edge surface.

[0045] The system fully reads the horizontal and vertical coordinates of all geometric center positions within the time sequence of the center position of the strip edge surface, as well as the acquisition timestamp value corresponding to each geometric center position. By comparing the magnitude of the acquisition timestamp values, the time sequence of each position is determined. The horizontal and vertical coordinates of the geometric center position with the smaller acquisition timestamp value are determined as the mapping start coordinates, and the horizontal and vertical coordinates of the geometric center position adjacent to the mapping start point and with the larger acquisition timestamp value are determined as the mapping end coordinates.

[0046] Using the coordinates of the starting point of the mapping as the initial position of the line segment, keeping the line segment straight and without bends, draw a straight line segment with a fixed direction to the coordinates of the ending point of the mapping, so that the starting end of the straight line segment completely coincides with the coordinates of the starting point of the mapping, and the ending end of the straight line segment completely coincides with the coordinates of the ending point of the mapping, forming a standard directed straight line segment.

[0047] The actual running direction of the strip is set as a fixed reference direction. An angle measurement method is used to detect the angle between the directed straight line segment and the reference direction. The coordinate position, length, direction and the measured deflection angle of the directed straight line segment are all integrated into one. The integrated complete feature information is the center migration vector of the strip edge surface.

[0048] Extract the data corresponding to the maximum extension length of two sets of data with adjacent acquisition times from the extension length time sequence of the strip edge surface. Read the acquisition timestamp values ​​of the two sets of data and compare their sizes. Determine the coordinate position of the crack tip end associated with the set of data with smaller acquisition timestamp values ​​as the extension reference point. Determine the actual physical length of the maximum extension corresponding to the set of data with larger acquisition timestamp values ​​as the target extension length.

[0049] Using the horizontal and vertical coordinates of the extension reference point as the starting position of the ray, draw a straight ray strictly in accordance with the running direction of the strip steel to ensure that the ray has no angular deviation from the running direction of the strip steel. Based on the image pixel distance corresponding to the target extension length, find the coordinate point on the ray whose pixel distance from the extension reference point is exactly the target extension length. This coordinate point is the target point.

[0050] In the image plane, the straight line segment between the extension reference point and the target point is extracted separately. The starting end of the extracted line segment coincides completely with the extension reference point, and the ending end of the line segment coincides completely with the target point. This line segment is the single-step extension line segment of the strip edge surface, with the starting point located at the original crack tip and the ending point located at the new crack tip.

[0051] The actual physical length value corresponding to the single-step extension segment and the directional information that is completely consistent with the running direction of the strip are taken as inherent features and additionally marked on the single-step extension segment. The complete directional segment information formed by integrating the length feature and the directional feature is the tip extension vector of the strip edge surface.

[0052] Extract the directed straight line segment corresponding to the center migration vector of the strip edge surface. Using the strip running reference direction as the measurement reference, detect the actual angle value between the directed straight line segment and the reference direction, and directly determine the angle value as the first direction angle.

[0053] Extract the directed straight line segment corresponding to the extension vector at the tip of the strip edge surface, measure the actual physical length of the directed straight line segment along the running direction of the strip, and directly determine the extension step size.

[0054] The expansion step size is multiplied by the preset direction sensitivity coefficient value in the cold rolling production process. The result of the multiplication is used as an exponent, and an exponential operation is performed on a fixed natural constant. At the same time, the expansion step size is multiplied by the sine value corresponding to the first direction angle. The result of the multiplication is then divided by the preset reference length value in the cold rolling production process. Finally, the result of the exponential operation is multiplied by the result of the division operation. The final result is the expansion trend factor.

[0055] The calculated expansion trend factor value is precisely compared with the pre-set crack expansion warning threshold value during the continuous production of cold-rolled strip steel to clarify the correspondence between the expansion trend factor value and the warning threshold value.

[0056] When the value of the expansion trend factor is greater than the preset crack expansion warning threshold, a unique crack acceleration expansion status identifier is generated. This crack acceleration expansion identifier is then associated and bound to the crack evolution trajectory on the edge surface of the strip steel through a unique crack number, so that the crack acceleration expansion identifier is stably attached to the corresponding crack evolution trajectory.

[0057] The value of the expansion step size is directly derived from the actual physical length value corresponding to the expansion vector of the edge surface of the strip. This value is obtained by extracting the crack tip position spacing at adjacent acquisition times of the same crack and then converting the image pixel spacing into the actual physical length.

[0058] The value of the first direction angle is directly derived from the deflection angle between the center migration vector of the strip edge surface and the strip running direction. This value is obtained by detecting the angle between the direction of the center migration vector and the reference running direction of the strip.

[0059] The reference length is a pre-set standard length value in the edge crack detection process of cold-rolled strip steel. This value is determined in advance based on the production specifications, material properties and crack judgment industry standards of the strip steel.

[0060] The orientation sensitivity coefficient is a pre-set weight value in the analysis of edge crack trends in cold-rolled strip steel. This value is calculated and calibrated in advance based on the degree of influence of the crack propagation direction on the overall propagation trend.

[0061] The natural constant uses a universally fixed mathematical constant value and directly participates in the quantitative calculation process of the crack propagation tendency factor.

[0062] The crack propagation tendency factor is calculated by integrating the numerical correlation between the propagation step size, the first direction angle, the reference length, the direction sensitivity coefficient, and the natural constant. It is used to accurately quantify the strength of the propagation tendency of cracks at the edge of the strip.

[0063] By comparing the calculated expansion trend factor with the preset warning threshold, it is possible to accurately determine whether the crack has entered a dangerous state of accelerated expansion. This generates an accelerated crack expansion indicator and adds it to the crack evolution trajectory, providing a clear status basis for real-time crack monitoring.

[0064] Pt.3. Dynamically compare the current crack length growth rate during crack evolution with the preset crack propagation baseline state to obtain early warning information for the strip edge surface.

[0065] In this embodiment of the invention, the step of dynamically comparing the current crack length growth rate during crack evolution with a preset crack propagation baseline state to obtain early warning information for the strip edge surface includes: Compare the length of the previous expansion step size with the current expansion step size in the crack evolution trajectory. If the current expansion step size is greater than the expansion step size at the previous time step, then an increment step size flag is generated. The number of consecutive occurrences of the step increment indicator is compared with the preset threshold value of the number of step increments in the crack propagation reference state. At the same time, the ratio of the difference between the current propagation step size and the previous propagation step size to the previous propagation step size is compared with the threshold value of the step size increase ratio in the crack propagation reference state. A Level 1 warning is generated when the number of consecutive occurrences exceeds the threshold for the number of steps incremented and the proportion exceeds the threshold for the proportion of steps incremented; otherwise, a Level 2 warning is generated.

[0066] From the time-series extended data stored in the completed strip edge surface crack evolution trajectory, the actual length value of the extended step corresponding to the previous acquisition time is accurately retrieved, and the actual length value of the extended step corresponding to the current acquisition time is retrieved. The two length values ​​are compared one by one according to their magnitude to clearly determine the size relationship between the current extended step length value and the extended step length value at the previous moment.

[0067] When the expansion step length value at the current acquisition time is greater than the expansion step length value at the previous acquisition time, a unique step increment flag is automatically generated to mark that the crack expansion step length is in an increasing state. This flag is bound to the crack expansion data at the current time and is used to permanently record the increasing state of the crack expansion step length at the current time.

[0068] The expansion step length changes in the crack evolution trajectory are traced back sequentially. The actual number of times the step length increment indicator appears continuously is counted. The actual number of consecutive occurrences is compared with the step length increment threshold set in the preset crack expansion reference state in the cold rolling production process. The current expansion step length is subtracted from the previous expansion step length to obtain the length difference between the two times. The length difference is then divided by the previous expansion step length to obtain the increase ratio of the current expansion step length compared to the previous time. The increase ratio is compared with the step length increase ratio threshold set in the preset crack expansion reference state. The two comparison operations are performed simultaneously.

[0069] If the actual number of consecutive occurrences of the step increment indicator is greater than the step increment number threshold, and the expansion step increment ratio is greater than the step increment ratio threshold, a first-level early warning message for the strip edge surface indicating that the crack is in a dangerous state of rapid propagation is generated. If the number of consecutive occurrences of the step increment indicator does not exceed the step increment number threshold, or the expansion step increment ratio does not exceed the step increment ratio threshold, i.e., if both conditions are not met simultaneously, a second-level early warning message for the strip edge surface indicating that the crack is in a normal propagation state is generated.

[0070] like Figure 2 The diagram shown is a functional block diagram of an online early warning system for edge cracks in cold-rolled strip steel based on industrial vision, provided in an embodiment of the present invention.

[0071] The online early warning system for edge cracks in cold-rolled strip steel based on industrial vision described in this invention can be installed in electronic devices. Depending on the functions implemented, the online early warning system for edge cracks in cold-rolled strip steel based on industrial vision may include a crack feature calibration module, a trajectory evolution construction module, and an extended early warning judgment module. The module described in this invention can also be called a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, and is stored in the memory of the electronic device.

[0072] In this embodiment, the functions of each module / unit are as follows: The crack feature calibration module is used to take the connected boundary contour, geometric center position and maximum extension length along the running direction of the strip steel as crack labeling data in the differentiated crack feature region on the edge surface of the strip steel. The trajectory evolution construction module is used to determine the crack evolution trajectory of the strip edge surface based on the geometric center position and maximum extension length of the same crack object at different times in the crack annotation data. The extended early warning determination module is used to dynamically compare the current crack length growth rate during crack evolution with the preset crack propagation benchmark state to obtain early warning information for the edge surface of the strip steel.

[0073] In the several embodiments provided by this invention, it should be understood that the disclosed methods and systems can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0074] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0075] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0076] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0077] This application embodiment can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0078] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. An online early warning method for edge cracks in cold-rolled strip steel based on industrial vision, characterized in that, The method includes: Pt.1, The connected boundary contour, geometric center position and maximum extension length along the strip running direction in the differential crack feature region on the edge surface of the strip are used as crack annotation data; Pt.

2. Based on the geometric center position and maximum extension length of the same crack object at different times in the crack annotation data, determine the crack evolution trajectory on the surface of the strip edge. Pt.

3. Dynamically compare the current crack length growth rate during crack evolution with the preset crack propagation baseline state to obtain early warning information for the strip edge surface.

2. The online early warning method for edge cracks in cold-rolled strip steel based on industrial vision as described in claim 1, characterized in that, The method of using the connected boundary contour, geometric center position, and maximum extension length along the strip running direction in the differentiated crack feature region on the edge surface of the strip as crack annotation data includes: A closed boundary trajectory is drawn pixel by pixel along the outer edge of the differential crack feature region on the edge surface of the strip to obtain the connected boundary contour of the edge surface of the strip. The geometric center is located at the intersection of the horizontal and vertical midlines of the differentiated crack feature region in the image plane. The span of the differential crack feature region is measured along the running direction of the strip edge surface to obtain the maximum extension length of the strip edge surface; By combining the connected boundary contour, geometric center position, and maximum extension length in a data association manner, crack annotation data of the strip edge surface is obtained.

3. The online early warning method for edge cracks in cold-rolled strip steel based on industrial vision as described in claim 2, characterized in that, A closed boundary trajectory is drawn pixel by pixel along the outer edge of the differential crack feature region on the strip edge surface to obtain the connected boundary contour of the strip edge surface, including: The pixel located at the top left corner of the differential crack feature region is retrieved as the starting point for tracking; Taking the starting point of tracking as the current observation point, explore adjacent pixels in the order of right, up, left and down directions, and determine the first adjacent pixel that belongs to the differential crack feature region as the boundary advancement point. Record the line connecting the current observation point and the boundary advancement point as a boundary line segment, and then update the boundary advancement point to the new current observation point; Each time, a boundary line segment is recorded and the current observation point is updated until the current observation point returns to the starting point of the tracking, thus obtaining the connected boundary profile of the strip edge surface.

4. The online early warning method for edge cracks in cold-rolled strip steel based on industrial vision as described in claim 2, characterized in that, The step of measuring the span of the differentiated crack characteristic region along the running direction of the strip edge surface to obtain the maximum extension length of the strip edge surface includes: In the image of the differential crack feature region, a reference scan line parallel to the strip running direction is established, and the reference scan line is placed outside the leading edge of the differential crack feature region. The reference scan line is translated sequentially along the running direction of the strip. After each translation, the intersection of the reference scan line and the differential crack feature area is checked, and the state of the intersection between the scan line and the area after each translation is recorded. When the reference scan line is shifted to a point where it no longer intersects with the differential crack feature region, the scan line is traced back to the last point where it intersected, thus obtaining the last intersection point on the strip edge surface. A straight line parallel to the strip's running direction is drawn from the leading edge of the differentiated crack feature region and extended to the final intersection point to obtain the maximum extension length of the strip's edge surface.

5. The online early warning method for edge cracks in cold-rolled strip steel based on industrial vision as described in claim 1, characterized in that, The method of determining the crack evolution trajectory on the strip edge surface based on the geometric center position and maximum extension length of the same crack object at different times in the crack annotation data includes: Retrieve all records belonging to the same crack object from the crack annotation data, extract the geometric center position of each record in the temporal order, and obtain the temporal sequence of the center position of the strip edge surface; For the time sequence of the center position, a direction mapping line segment is established from the geometric center position at the earlier time to the geometric center position at the later time to obtain the center migration vector of the strip edge surface; Retrieve all records belonging to the same crack object from the crack annotation data, extract the maximum extension length in each record according to the time sequence, and obtain the time sequence of the extension length of the strip edge surface; For the maximum extension length of each pair of adjacent moments in the extension length time sequence, an extension increment line segment is drawn from the end position of the maximum extension length at the earlier moment along the strip running direction to obtain the tip extension vector of the strip edge surface. Using the earliest geometric center position as the origin of the trajectory, the center migration vector and the tip expansion vector are sequentially connected end to end to obtain the crack evolution trajectory on the edge surface of the strip.

6. The online early warning method for edge cracks in cold-rolled strip steel based on industrial vision as described in claim 5, characterized in that, For the time sequence of the center position, a direction mapping line segment is established from the geometric center position at a earlier time to the geometric center position at a later time to obtain the center migration vector of the strip edge surface, including: In the time sequence of the center position, the geometric center position with the earlier acquisition time is used as the mapping starting point, and the geometric center position with the later acquisition time is used as the mapping ending point. Starting from the mapping origin, draw a directed line segment along a straight path towards the mapping endpoint; The deflection angle of the directed line segment relative to the running direction of the strip is detected to obtain the center migration vector of the strip edge surface.

7. The online early warning method for edge cracks in cold-rolled strip steel based on industrial vision as described in claim 5, characterized in that, For each pair of adjacent times in the elongation length time sequence, an extension increment line segment is drawn from the end position of the maximum elongation length at the earlier time along the strip running direction to obtain the tip extension vector of the strip edge surface, including: The maximum extension length at two adjacent moments is retrieved from the extension length time sequence. The crack tip position corresponding to the maximum extension length acquired earlier is taken as the extension reference point, and the maximum extension length acquired later is taken as the target extension length. Starting from the extension reference point, draw a ray along the running direction of the strip, and mark the target point on the ray whose distance from the starting point is equal to the target extension length. The line connecting the extension reference point and the target point is cut out to form a directed line segment with the starting point at the original crack tip and the ending point at the new crack tip, thus obtaining a single-step extension line segment on the edge surface of the strip. The length and direction of the single-step extension segment are attached as characteristic parameters to the single-step extension segment to obtain the tip extension vector of the strip edge surface.

8. The online early warning method for edge cracks in cold-rolled strip steel based on industrial vision as described in claim 5, characterized in that, The determination of the crack evolution trajectory on the edge surface of the strip also includes: The deflection angle of the directed line segment is extracted from the central migration vector and denoted as the first direction angle; The length of the directed line segment extracted from the tip extension vector is denoted as the extension step size; Using the natural constant as the base and the product of the expansion step size and the preset direction sensitivity coefficient as the exponent, an expansion trend factor is constructed. The calculation formula for the expansion trend factor is as follows: ； In the formula, To expand the trend factor, To extend the step size, The first direction angle, The preset reference length, The preset direction sensitivity coefficient, It is a natural constant; If the expansion trend factor exceeds the preset warning threshold, an accelerated crack expansion indicator is generated and attached to the crack evolution trajectory.

9. The online early warning method for edge cracks in cold-rolled strip steel based on industrial vision as described in claim 1, characterized in that, The step of dynamically comparing the current crack length growth rate during crack evolution with a preset crack propagation baseline state to obtain early warning information for the strip edge surface includes: Compare the length of the previous expansion step size with the current expansion step size in the crack evolution trajectory. If the current expansion step size is greater than the expansion step size at the previous time step, then an increment step size flag is generated. The number of consecutive occurrences of the step increment indicator is compared with the preset threshold value of the number of step increments in the crack propagation reference state. At the same time, the ratio of the difference between the current propagation step size and the previous propagation step size to the previous propagation step size is compared with the threshold value of the step size increase ratio in the crack propagation reference state. A Level 1 warning is generated when the number of consecutive occurrences exceeds the threshold for the number of steps incremented and the proportion exceeds the threshold for the proportion of steps incremented; otherwise, a Level 2 warning is generated.

10. An online early warning system for edge cracks in cold-rolled strip steel based on industrial vision, characterized in that, The system for implementing the online early warning method for edge cracks in cold-rolled strip steel based on industrial vision as described in claim 1 includes: The crack feature calibration module is used to take the connected boundary contour, geometric center position and maximum extension length along the strip running direction of the differential crack feature region on the edge surface of the strip as crack annotation data; The trajectory evolution construction module is used to determine the crack evolution trajectory of the strip edge surface based on the geometric center position and maximum extension length of the same crack object at different times in the crack annotation data. The extended early warning judgment module is used to dynamically compare the current crack length growth rate during crack evolution with the preset crack propagation benchmark state to obtain early warning information for the edge surface of the strip steel.

Images