Shape measurement method and device for shape object, control method, manufacturing method and device and quality management method

Abstract

The shape measuring method of the strip-shaped object is a shape measuring method for measuring the shape of the strip-shaped object, and comprises the following steps: an imaging step, an image capturing unit that captures an image of thermal radiation light of the strip-shaped object such that an angle theta formed by a plane alpha serving as a reference surface of the surface of the strip-shaped object and an optical axis of the camera is not 90 degrees and an angle formed by an orthographic projection of the optical axis of the camera onto the plane alpha and a transport direction p of the strip-shaped object is not 0 degree; and an image processing step for calculating an index of the shape of the edge portion of the strip-shaped object by calculating the profile distribution of the strip-shaped object from the obtained image.

Description

Technical Field

[0001] This invention relates to a method for measuring the shape of a strip object, a method for controlling the shape of a strip object, a method for manufacturing a strip object, a method for quality management of a strip object, a device for measuring the shape of a strip object, and equipment for manufacturing a strip object. Background Technology

[0002] In the raw materials industry, shape management of strip-shaped raw materials is crucial, requiring the quantification of product shape. For example, in steelmaking processes, there is a high demand for steel shape measurement from the perspectives of operational stability and product quality assurance. In particular, shape measurement in rolling, used to shape products into the desired form, is important because the initial setting of rolling conditions and the feedback for rolling control during rolling are related to improved product quality and operational stability.

[0003] For example, in a hot-rolled steel sheet production line, a rectangular semi-finished product called a slab, which is taken out of the heating furnace at a high temperature, is processed into a sheet through finishing mill, rough rolling, and finish rolling, and then rolled into a coil. At this time, depending on the rolling state, the reduction may sometimes become uneven in the width direction of the steel sheet, resulting in local elongation and shape defects.

[0004] For example, compared to the central portion (central part in the width direction) of a steel sheet, if only the edge portion (end in the width direction) extends, the edge portion becomes an undulating shape. Conversely, if only the central portion extends, the central portion becomes an undulating shape. Such shape defects not only result in product defects, but also worsen the sheet's permeability in subsequent processes such as pickling and cold rolling, becoming a cause of failure. Therefore, it is strongly required to avoid shape defects.

[0005] To improve shape defects, it is necessary to appropriately set the load in the width direction during rolling. However, various interferences exist, such as roll wear, temperature distribution of the steel plate, and deviations in material property distribution, making it very difficult to determine the optimal rolling conditions solely through calculation. Therefore, to set the optimal rolling conditions, it is essential to derive the optimal initial settings by measuring the shape of the steel plate and interpolating with the rolling model, and then implement preset control, or to provide real-time feedback control as the rolling conditions. By implementing such control, shape defects can be suppressed first.

[0006] However, it is difficult to measure the shape of the steel sheet after it has become a roll, especially since measurement is required during rolling manufacturing in feedback control. Therefore, it is preferable to measure the shape immediately after the finishing mill exit, which is the final rolling process. For example, if the elongation of one edge portion relative to the center of the steel sheet is significant, leveling control can be performed to eliminate the deviation in elongation towards that edge portion by adjusting the load balance in the width direction of the rolling rolls. It should be noted that the shape described here mainly refers to the elongation in the length direction of the steel sheet that occurs locally in the width direction during rolling, specifically, "elongation of the central portion" and "elongation of the edge portion".

[0007] As a technique for determining the shape of a steel sheet immediately after finishing the rolling mill, various techniques have been proposed in the past, such as using a rod-shaped light source and a magnetic sensor. Among these, a particularly powerful method is disclosed in Patent Documents 1 to 3, which involves irradiating the surface of an object with a laser in the form of a point or line and measuring the reflected light to determine the shape.

[0008] Furthermore, Patent Document 4 discloses the following technique: for laser irradiation, three line lasers with their length direction orthogonal to the transport direction of the steel plate are used. In the technique disclosed in Patent Document 4, the line lasers are irradiated in parallel with the steel plate in a manner that they are three lines at equal intervals relative to the length direction of the steel plate, and their reflection images are obtained and the distribution of each laser is compared, thereby eliminating the influence of the vertical vibration of the steel plate.

[0009] In addition, patent documents 5 to 7 and non-patent document 1 disclose the following technology: by using a powerful LED light source to irradiate the surface of an object with a stripe pattern composed of multiple lines, the number of lines irradiated can be increased at a lower cost than laser, and the shape can be stably measured regardless of mirror properties or the tilt of the object.

[0010] It should be noted that the need for shape measurement of steel sheets in such manufacturing processes exists not only for hot-rolled steel sheets, but also for other strip-shaped raw materials, regardless of their state of being red-hot (e.g., above 600°C), warm (e.g., 300°C to 600°C), or cold (e.g., near room temperature). Furthermore, the term "strip-shaped material" here refers to long strips of raw material. Strip-shaped materials include not only products that are ultimately wound into rolls, such as non-ferrous metals like iron, paper, cloth, and aluminum, but also products formed into rectangular plates, such as thick steel sheets.

[0011] Prior art literature

[0012] Patent Literature

[0013] Patent Document 1: Japanese Patent Application Publication No. 56-124006

[0014] Patent Document 2: Japanese Patent Application Publication No. 55-40924

[0015] Patent Document 3: Japanese Patent Application Publication No. 58-11708

[0016] Patent Document 4: Japanese Patent Application Publication No. 61-40503

[0017] Patent Document 5: Japanese Patent Application Publication No. 2008-58036

[0018] Patent Document 6: Japanese Patent Application Publication No. 2011-99821

[0019] Patent Document 7: Japanese Patent Application Publication No. 2016-65863

[0020] Non-patent literature

[0021] Non-Patent Literature 1: Ise-kai, 3 others, “Development of a Flatness Tester for Hot-Rolled Steel Sheets Based on LED Dot Pattern Projection Method”, Iron and Steel, Japan Iron and Steel Association, 2019, Vol. 105, No. 1, pp. 20-29

[0022] Non-patent literature 2: Kim Jung, “Thermal measurement using a radiation thermometer”, Molding Processing, China Plastics Molding Processing Society, 2020, Vol. 32, No. 4, pp. 121-124. Summary of the Invention

[0023] Problems to be solved by the invention

[0024] The technologies disclosed in Patent Documents 1 through 7 all involve the following method: irradiating a hot-rolled steel sheet with light using a light source and capturing the reflected light using a camera, thereby determining the shape of the object. However, considering factors such as the influence of radiant heat on high temperatures, and the adhesion of steam, dust, oil, etc., to the steel sheet, a high level of technology is required to perform measurements stably and for a long period of time while keeping the light source and sensor close to the steel sheet being transported, which also increases maintenance costs.

[0025] To prevent this, one could consider keeping the light source away from the object, but this reduces the light intensity due to light diffusion, making it difficult to design an optical system that can stably focus the light to ensure sufficient intensity. Furthermore, while magnetic sensors could be used, this requires bringing the sensor itself close to the object being measured, which also makes setting and maintaining performance difficult.

[0026] The present invention was made in view of the above circumstances, and its object is to provide a method for measuring the shape of a strip object, a method for controlling the shape of a strip object, a method for manufacturing a strip object, a method for quality management of a strip object, a device for measuring the shape of a strip object, and a manufacturing equipment for a strip object, which can be used easily and stably with the edge of the strip object as the target, without the light source or sensor being close to the target object, and can also suppress maintenance costs.

[0027] Methods for solving problems

[0028] (1) The shape measurement method for a strip-shaped object of the present invention includes:

[0029] The shooting steps involve capturing an image of the thermal radiation of the strip object in a manner where the angle θ between the reference plane α of the strip object's surface and the camera's optical axis is not 90 degrees, and the angle φ between the orthographic projection of the camera's optical axis onto the plane α and the transport direction p of the strip object is not 0 degrees; and...

[0030] The image processing step involves calculating the contour distribution of the strip-shaped object based on the obtained image, thereby calculating an index of the shape of the edge of the strip-shaped object.

[0031] (2) In addition, based on the above-described (1) method for determining the shape of a strip object, the method for determining the shape of a strip object of the present invention extracts the region of the strip object from the obtained image in the image processing step, calculates the position of the edge portion in the extracted region, and thereby calculates the contour distribution of the strip object.

[0032] (3) In addition, the method for determining the shape of a strip object according to the present invention is based on the method for determining the shape of a strip object described in (1) or (2) above. In the image processing step, the steepness of the edge of the strip object, wave height, wave spacing, elongation, and elongation rate are calculated as any one or more of the indicators based on the obtained contour distribution of the strip object.

[0033] (4) In addition, in the method for determining the shape of a strip object according to any one of (1) to (3) above, in the image processing step, the resolution of the wave height direction and the wave distance direction is calculated according to the positional relationship between the camera and the strip object, and converted into the actual size.

[0034] (5) In addition, the shape control method of the strip object of the present invention measures the shape of the strip object by any one of the shape measurement methods of the strip object described in (1) to (4) above, and controls it based on the measurement results so that the shape of the strip object becomes the desired shape.

[0035] (6) In addition, the method for manufacturing the strip object of the present invention is to determine the shape of the strip object by any one of the strip object shape measurement methods described in (1) to (4) above, and to manufacture the strip object based on the measurement results.

[0036] (7) In addition, the quality management method for the strip object of the present invention measures the shape of the strip object by any one of the strip object shape measurement methods described in (1) to (4) above, and manages the quality of the strip object based on the measurement results.

[0037] (8) In addition, the shape measuring device for a strip-shaped object according to the present invention is a shape measuring device for measuring the shape of a strip-shaped object, comprising:

[0038] The imaging unit captures an image of the thermal radiation light of the strip object in a manner where the angle θ between the reference plane α of the strip object's surface and the optical axis of the camera is not 90 degrees, and the angle φ between the orthographic projection of the camera's optical axis onto the plane α and the transport direction p of the strip object is not 0 degrees; and

[0039] The image processing unit calculates an index of the shape of the edge of the strip-shaped object by calculating the contour distribution of the strip-shaped object based on the obtained image.

[0040] (9) In addition, the strip-shaped object manufacturing equipment of the present invention includes the strip-shaped object shape measuring device described in (8) above.

[0041] [Invention Effects]

[0042] The method for measuring the shape of a strip object, the method for controlling the shape of a strip object, the method for manufacturing a strip object, the method for managing the quality of a strip object, the device for measuring the shape of a strip object, and the equipment for manufacturing a strip object according to the present invention can be easily and stably used without bringing the light source or sensor close to the object being measured, and can also reduce maintenance costs. Attached Figure Description

[0043] Figure 1 This is a diagram showing a schematic structure of the shape measuring device for a strip-shaped object according to an embodiment of the present invention.

[0044] Figure 2 The diagrams are examples of the relationship between the direction of steel plate transport and the position of the camera. (a) is a diagram of the positional relationship viewed from an oblique angle, (b) is a diagram of the positional relationship viewed from above, and (c) is a diagram of the positional relationship viewed from the direction δ of (b).

[0045] Figure 3This is an example of the relationship between the direction of steel plate transport and the position of the camera. It is a diagram showing the case where the optical axis of the camera is configured perpendicular to the direction of steel plate transport.

[0046] Figure 4 This is an example of the relationship between the direction of steel plate transport and the position of the camera. It is a diagram showing the case where the optical axis of the camera is tilted relative to the direction of steel plate transport.

[0047] Figure 5 This is an example of an image showing a steel plate in its normal state and in a state of poor shape.

[0048] Figure 6 This is a flowchart illustrating the specific processing steps of the image processing apparatus of the shape measuring device for a strip-shaped object according to an embodiment of the present invention.

[0049] Figure 7 This is a diagram illustrating the binarization process in the image processing step of the method for determining the shape of a strip-shaped object according to an embodiment of the present invention.

[0050] Figure 8 This is a diagram illustrating the contour distribution calculation process in the image processing step of the method for determining the shape of a strip-shaped object according to an embodiment of the present invention.

[0051] Figure 9 This is a photographic example showing the fluctuations produced by the steel plate.

[0052] Figure 10 This shows an example of photographing a fixed wave generated on a steel plate.

[0053] Figure 11 This diagram illustrates an example of a situation where it is difficult to distinguish between the surface of a steel plate and a structure.

[0054] Figure 12 This figure illustrates an example of low-pass filter processing performed in the shape determination method for a strip-shaped object according to an embodiment of the present invention when it is difficult to distinguish between the surface of a steel plate and a structure.

[0055] Figure 13 This is an application example of using the shape measuring device for strip-shaped objects according to embodiments of the present invention to determine whether the shape is qualified or not.

[0056] Figure 14 This is an application example of applying the shape measuring device for strip objects according to the embodiments of the present invention to rolling feedback control.

[0057] Figure 15 This is an example of applying the shape measuring device for strip objects according to an embodiment of the present invention to a rolling control system that uses machine learning. Detailed Implementation

[0058] The method for measuring the shape of a strip object, the method for controlling the shape of a strip object, the method for manufacturing a strip object, the method for managing the quality of a strip object, the device for measuring the shape of a strip object, and the equipment for manufacturing a strip object, according to embodiments of the present invention, will be described with reference to the accompanying drawings. It should be noted that the constituent elements in the following embodiments include elements that can and are easily substituted by those skilled in the art, or elements that are substantially the same.

[0059] (Shape measuring device)

[0060] Reference Figures 1-12 The shape measuring apparatus for a strip object according to the embodiment will be described. The shape measuring apparatus is used to measure the shape of a strip object. Hereinafter, the application of the shape measuring apparatus to hot finishing rolling will be described. Furthermore, the case where the strip object to be measured is a steel plate will be described below. Additionally, the case where the shape measured by the shape measuring apparatus is the elongation of the edge of the steel plate will be described below.

[0061] like Figure 1 As shown, the shape measuring device of the embodiment includes a camera 2 and an image processing device 3. First, the details of the camera 2 will be explained.

[0062] As described below, camera 2 is configured such that the angle θ between the optical axis of camera 2 and the reference plane (plane α) of the strip-shaped object (steel plate S) is not 90 degrees, and the angle φ between the orthographic projection of the optical axis of camera 2 onto plane α and the transport direction p of the strip-shaped object (steel plate S) is not 0 degrees (refer to...). Figure 2 Then, using camera 2 configured in this way, an image of the thermal radiation of the steel plate S on the hot finishing output side after being rolled by roll 1 is captured.

[0063] When using a digital camera or similar device that is sold to the general public, problems such as camera shake or image thickening occur, making it impossible to clearly capture the shape of the steel plate S. Therefore, the following describes the technical points (1) to (4) for capturing the shape of the steel plate S with high precision using the camera 2.

[0064] (1) Ensure sufficient light during shooting

[0065] First, ensure sufficient light for obtaining a clear image. On the hot-rolled side, the steel sheet S sometimes passes at speeds exceeding 20 m / s, which is extremely high. To capture high-speed objects clearly without blurring, it is preferable to shorten the exposure time. For example, when shooting at 20 m / s with a resolution of approximately 2 mm and pixel jitter within 1 pixel (i.e., within 2 mm), the permissible exposure time is only 0.1 ms, and the amount of light received proportionally to the exposure time is also very small.

[0066] Furthermore, the temperature of the steel sheet on the hot-rolled end is approximately 900°C. (As stated in Non-Patent Document 2 above.) Figure 5 As shown, the peak wavelength of thermal radiation at approximately 900°C (1200K) is 2.5μm, resulting in low sensitivity in the visible region (0.4~0.7μm). Therefore, it is preferable to use a camera equipped with imaging elements such as InGaAs, PbS, or PbSe as camera 2.

[0067] On the other hand, cameras using imaging elements such as InGaAs, PbS, and PbSe are expensive, and the price increases further depending on the resolution. Therefore, if high-resolution imaging is desired, the implementation cost becomes high. Thus, for camera 2, it is preferable to use a camera with an inexpensive Si imaging element and capable of operating in the near-infrared sensitivity range of 0.8~1.0 μm. By using such a camera 2, sufficient thermal radiation light can be obtained inexpensively.

[0068] Furthermore, when applying this method to objects at even lower temperatures, it is preferable to use a camera 2 with an imaging element that is sensitive to longer wavelengths. For example, if the object being measured is around 400°C, and disregarding import costs, a camera 2 with an InGaAs imaging element that is sensitive to 1.2 to 1.7 μm can be used. Conversely, if the object being measured is around 200°C, a camera 2 with an imaging element such as PbS or PbSe that is sensitive to 3 to 5 μm can be used. This ensures sufficient light intensity. Thus, it is preferable to select a camera 2 with an imaging element sensitive to an appropriate wavelength range based on the object's transport speed, resolution, temperature, and depth of field (described later).

[0069] (2) Positional relationship between the steel plate and the camera

[0070] Second, in order to clearly capture the shape of the steel plate S, the positional relationship between the steel plate S and the camera 2 was studied. Here, focusing on the edge elongation of the steel plate S, the shape change of the edge was considered as the outline of the steel plate S. Figure 2 This illustrates an example of the relationship between the transport direction p of the steel plate S and the position of the camera 2. Figure 2 In the diagram, (a) is a view of the positional relationship from an oblique angle, (b) is a view of the positional relationship from above, and (c) is a view from the direction δ of (b). n Observe the diagram showing this positional relationship. Here, the direction δ n This refers to the direction of the optical axis of camera 2, which can be observed from the side. This direction δ is defined by... n This allows for accurate observation of the size of the light-receiving angle θ, which will be discussed later.

[0071] exist Figure 2In (a), (b), and (c), the plane parallel to the transport table of the steel plate S is designated as α, and the plane containing the transport direction p of the steel plate S and the normal n of the transport table is designated as β. Furthermore, the angle between the optical axis of the camera 2 and plane α is designated as θ (the angle of illumination θ), and the angle between the orthographic projection of the optical axis of the camera 2 onto plane α and the transport direction p is designated as φ. It should be noted that plane α includes the surface of the strip object (steel plate S) in a state where it is stably transported in a generally flat shape. Therefore, plane α is also called the reference plane of the surface of the strip object (steel plate S).

[0072] Furthermore, the direction of the normal n of the transport table is the same as the direction of the normal to the surface of the steel plate S when it is being transported stably in a generally flat shape. Therefore, plane α includes the transport direction p of the strip object (steel plate S) and is perpendicular to the normal direction of the surface of the strip object (steel plate S). Additionally, plane β is also a plane that includes both the transport direction p of the strip object (steel plate S) and the normal direction of the surface of the strip object (steel plate S). Furthermore, in Figure 2 In this diagram, the intersection of plane α and the optical axis is designated as point O. Additionally, the plane perpendicular to the transport direction of the steel plate is designated as plane γ. That is, the closer the angle φ is to 90 degrees, the closer plane γ is to being parallel to the optical axis.

[0073] exist Figure 2 In this context, when the resolution of camera 2 at the location of the steel plate S to be measured is set to "r (mm / pixel)," the resolution rn (mm / pixel) in the wave height direction and the resolution rp (mm / pixel) in the wave pitch direction can be expressed as shown in equation (1) and equation (2) below. Furthermore, a smaller resolution value indicates a higher resolution.

[0074] rn=r / cosθ···(1)

[0075] rp=r / sinφ···(2)

[0076] When the light-receiving angle θ is close to 90 degrees, that is, when the optical axis of camera 2 is nearly perpendicular to the plane α parallel to the transport table of steel plate S, the resolution rn in the wave height direction decreases, making it difficult to capture changes in the contour of the edge. Therefore, it is preferable for the light-receiving angle θ to be closer to 0 degrees. Similarly, when the angle φ is close to 0 degrees, that is, when the optical axis of camera 2 is nearly parallel to the transport direction p of steel plate S, the resolution rp in the wave pitch direction decreases. Therefore, it is preferable for the angle φ to be closer to 90 degrees. In particular, since the wave height to be measured is very small compared to the wave pitch, it is more preferable to minimize the light-receiving angle θ and photograph steel plate S from a low angle in order to capture the wave height with high accuracy.

[0077] Moreover, such as Figure 2 As shown, by observing the steel plate S at a low angle relative to the light-receiving angle θ and taking a picture, the distance between the two ends of the steel plate S in the image becomes smaller, allowing images of the two edges to be obtained with a smaller field of view. In addition, if the field of view of the object becomes smaller, the image size also becomes smaller. Therefore, the steel plate S that is being measured can be transported at a higher speed, which is advantageous when high-speed shooting and image processing are required.

[0078] Furthermore, in hot-rolled steel sheets, the wave pitch is sufficiently large compared to the wave height, thus allowing the resolution rp in the wave pitch direction to be smaller than the resolution rn in the wave height direction. Therefore, for example... Figure 3 As shown, it is preferable to set the angle φ to 90 degrees, that is, to position the camera 2 perpendicular to the transport direction of the steel plate S for lateral observation. However, if the angle φ is not extremely small, then for example, Figure 4 As shown, the camera 2 can also be set at an angle relative to the transport direction of the steel plate S.

[0079] (3) Distance between the steel plate and the camera

[0080] Third, increase the distance between the steel plate S being measured and the camera 2. The reason for this is that by keeping the camera 2 as far away from the production line as possible, a better environment can be achieved, and the differences in optical conditions such as the light-receiving angles θ and φ at the two edges of the steel plate S along its width can be reduced. Differences in optical conditions along the width of the steel plate S manifest as differences in appearance. Therefore, when judging by visual inspection, the degree of shape defect may be misjudged, requiring greater correction when quantifying the degree of shape defect through image processing. To increase the distance, it is preferable to use a telephoto lens during shooting. By using a telephoto lens, the differences in optical conditions can be reduced.

[0081] (4) Setting an appropriate aperture value

[0082] Fourth, to ensure depth of field, the aperture value of camera 2 should be appropriately determined. For example, assume that the entire surface of the steel plate S is photographed at the two edges near the front (lower) and the inner (upper) sides. In this case, if the width of the steel plate S is set as d (mm), the difference ΔL (mm) between the distance from camera 2 to the near-front edge and the distance from camera 2 to the inner edge can be expressed by the following formula (3).

[0083] ΔL=dcosθ / sinφ···(3)

[0084] To obtain an image focused on both edges of the steel plate S, a depth of field (focus range) of at least ΔLd (mm) is required. A larger aperture value results in a greater depth of field of the subject; therefore, it is preferable to set the aperture to ΔLd (mm) or larger. However, when the aperture value is large, insufficient light may occur, potentially causing camera shake. Therefore, it is preferable to determine the aperture value and focal length that balances camera shake and depth of field based on the transport speed of the steel plate S being measured and the required resolution. Next, the image processing apparatus 3 will be described.

[0085] The image processing device 3 is implemented, for example, by a workstation, a personal computer, or other general-purpose computer. This image processing device 3 can be located near the camera 2, or it can be located in the cloud if high speed is not required.

[0086] As described later, the image processing device 3 calculates the contour distribution of the steel plate S based on the image captured by the camera 2, thereby calculating the shape index of the edge portion of the steel plate S. Furthermore, as described later, the image processing device 3 extracts the region of the steel plate S from the image (…). Figure 6 In step S1), the position of the edge is calculated in the extracted area, thereby calculating the contour distribution of the strip-shaped object of the steel plate S. Figure 6 Step S4). Furthermore, as described later, the image processing device 3 appropriately selects one or more of the following from the contour distribution of the steel plate S: edge steepness, wave height, wave spacing, elongation, and elongation rate, and calculates them as the aforementioned indices. Hereinafter, the processing of calculating the indices of the shape of the edge of the steel plate S based on the image obtained from the camera 2 using the image processing device 3 will be explained.

[0087] Figure 5 (a) represents the image of steel plate S under normal conditions. Figure 5 (b) represents the image of the steel plate S when shape defects caused by plate elongation occur. For example... Figure 5 As shown, the elongation of the plate (specifically, the elongation of the edge of the strip-shaped object) can be visually determined. The image processing device 3, based on the acquired image of the steel plate S, for example, according to... Figure 6 The parameters of the edge portion of the steel plate S are calculated in the order shown. It should be noted that, for ease of explanation, the horizontal axis of the image will be approximately parallel to the transport direction of the steel plate S.

[0088] First, such as Figure 7 As shown, only the plate region of steel plate S is extracted through binarization processing. Figure 6 Step S1). Figure 7 (a) is the image before binarization. Figure 7 (b) is the image after binarization.

[0089] At this time, due to water droplets and cooling water spray on the plate surface, interference may occur, such as localized reduction in brightness on the plate surface, or the spray scattering into the space above the plate causing bright light emission due to thermal radiation from the steel plate S. These interferences are mostly high-frequency and small relative to the spacing of the shape to be calculated; therefore, it is preferable to perform connection / isolation point removal on the plate area extracted in the binarization process based on expansion / contraction processing, median filtering, etc. Figure 6 Step S2).

[0090] Additionally, after removing connections and isolated points, sometimes multiple candidate spots (blocks identified by connecting surrounding pixels during binarization) may appear as plate regions. In this case, clumps can also be extracted. Figure 6 Step S3), or determine whether it is a plate area based on the size, orientation, etc. of the steel plate S, thereby determining the plate area.

[0091] Next, based on the plate region of steel plate S obtained as described above, the contour distribution of the two edges is calculated. Figure 6 Step S4). Various methods can be considered for calculating the contour distribution of the edge region. As an example, in... Figure 8 The diagram illustrates a method for extracting the contours of edges through a search. Figure 8 (a) is a diagram showing the search for the contour of the edge. Figure 8 (b) is a diagram showing the outline of the extracted edge.

[0092] like Figure 8 As shown in (a), a search can be performed from outside the plate region in the image towards the plate region, and the coordinates of the plate region are recorded. Conversely, a search can also be performed from within the plate region towards outside the plate region. In this embodiment, at each point on the horizontal axis of the image, the contour distribution of the upper and lower edges is calculated as a one-dimensional vector by searching the contour in the vertical direction.

[0093] Here, the image obtained in the shooting step is taken at an angle relative to the transport direction of the steel plate S, resulting in different resolutions in the wave height direction and the wave spacing direction. Therefore, resolution correction is performed in both the wave height direction and the wave spacing direction based on the contour distribution on the obtained image. Figure 6 Step S5).

[0094] In step S5, specifically, the tilt of the optical axis of camera 2 is first corrected by rotation processing so that the length direction of steel plate S is aligned with the horizontal axis. Since the resolution of the wave height direction, i.e., the vertical axis, is "rn (mm / pixel)" and the resolution of the wave pitch direction, i.e., the horizontal axis, is "rp (mm / pixel)", it can be converted into the contour distribution of steel plate S.

[0095] Furthermore, the distance from camera 2 differs at the near-front edge and the inner edge, thus causing a change in resolution. Therefore, the resolution can be calculated and corrected separately at the near-front edge and the inner edge based on their distance from camera 2. That is, the resolution in the wave height direction and the wave pitch direction can also be calculated based on the positional relationship between camera 2 and steel plate S, and then converted into actual dimensions.

[0096] Alternatively, instead of performing the aforementioned processing on the obtained edge contour distribution, coordinate transformation can be performed using the pose parameters of camera 2 to obtain the coordinates of the plane β (refer to...). Figure 2 The orthographic projection of the plate S is used to calculate the profile distribution of the steel plate S. This allows for rigorous geometric correction throughout the entire field of view of camera 2.

[0097] In addition, the wave spacing and period of the plate elongation of the steel plate S are roughly determined. Therefore, noise that does not contribute to the shape can be removed from the contour distribution of the edge by applying a low-pass filter or a band-pass filter that removes frequency components other than these.

[0098] Next, based on the outline of the edge portion of the steel plate S obtained as described above, the shape index of the edge portion is calculated. Figure 6 Step S6). The shape of the edge is mostly discussed in terms of a parameter called steepness, which is the ratio of wave height to wave spacing. However, one or more parameters can be appropriately selected from wave height, wave spacing, elongation, elongation rate, etc. Various methods can be listed as the calculation methods for the shape of the edge of the steel plate S. For example, the following (i) to (iii) can be listed as representative methods.

[0099] (i) Calculate the maxima and minima, taking the vertical distance of the image as the wave height and the horizontal distance as the wave spacing. The kurtosis is calculated as the ratio of wave height to wave spacing.

[0100] (ii) Perform sine curve fitting and calculate wave height and inter-wave distance based on amplitude and period. Steepness is calculated as the ratio of wave height to inter-wave distance.

[0101] (iii) Calculate the elongation and elongation rate of the plate based on the outline length, and directly calculate the steepness using the method described in Non-Patent Document 2.

[0102] By using the shape parameters of the edge of the steel plate S obtained in this way, parameter control and feedback control during rolling, as well as the determination of the pass / fail status of the coil shape, can be implemented. It should be noted that the shape parameters are not only calculated based on the contour distribution of the steel plate S, but also, for example, wave height, wave period, steepness, elongation, elongation rate, etc., can be calculated based on the contour distribution on the image. Then, the resolution rn (mm / pixel) in the wave height direction and the resolution rp (mm / pixel) in the wave spacing direction are used for correction.

[0103] Furthermore, regarding the behavior during the handling of steel plate S, other curves may be detected even if plate elongation (specifically, elongation of the edge of the strip-shaped object) does not occur. For example, there may be phenomena such as steel plate S momentarily floating up and becoming a large wave (called wave undulation), or continuous handling of steel plate S with the wave shape remaining unchanged within the field of view (called stationary wave). These phenomena are different from plate elongation and are therefore preferably excluded from the detection results. Therefore, the following explains the method for distinguishing between plate elongation rate, wave undulation, and stationary wave.

[0104] Waves are a phenomenon that occurs when the transport speed of steel plate S increases during handling. Specifically, at a point on the upstream and downstream sides of a plate, if the downstream side has a higher speed than the upstream side, the plate, having nowhere else to go, bounces upwards, forming large waves. Because the plate edges are flat, this phenomenon can occur even without plate elongation, and therefore needs to be distinguished from the shape caused by plate elongation.

[0105] Characteristics of wave fluctuations include very large wave spacing, a sudden increase in wave height, and a sharp increase in wave spacing or wave height compared to preceding and following waves. Therefore, if the wave spacing and wave height can be directly calculated, a sharp increase in these values ​​can be identified as a wave fluctuation rather than plate elongation. Then, processing methods such as masking the data from the period identified as a wave fluctuation or filling in the gaps using preceding and following data can be performed. This allows wave fluctuations to be excluded from plate elongation detection results, reducing their impact on the measurement results.

[0106] Furthermore, indicators such as plate elongation calculated based on the profile of steel plate S, which do not directly calculate wave height and wave spacing, can also be distinguished using wave characteristics. For example, even without using the overall field of view within the image, such as... Figure 9 As shown, the outline position of a point along the length of the steel plate S within the field of view can also be monitored, and fluctuations are determined when the change in the outline position is extremely large.

[0107] Reference Figure 9 The method for determining fluctuations by utilizing changes in contour position is explained. Figure 9 (a) to (e) represent photographic examples of the undulation of steel plate S. Additionally, in Figure 9In the middle, arranged in chronological order (a), (b), (c), (d), (e). The determination of fluctuations can be carried out in the following steps (1) to (3).

[0108] (1) Determine the position Xt in the width direction of an image.

[0109] (2) For the image width direction position Xt determined by (1), calculate the lower contour position Yt of each image. The lower contour position Yt is the lower contour position of the horizontal center of the image.

[0110] (3) Monitor the changes in the calculated lower contour position Yt and detect areas with significant changes.

[0111] Figure 9 The lower part of the diagram (f) shows a schematic diagram of the lower contour position Yt plotted in chronological order over time. Using such a diagram, the presence or absence of fluctuations can be determined. Alternatively, as a method for determining fluctuations, one can monitor the number of maximum and minimum values ​​within a certain interval as the lower contour position Yt shifts; a sudden decrease in these values ​​indicates the presence of fluctuations.

[0112] In addition, Figure 9 In the example, the lower contour position Yt is used to determine the presence or absence of fluctuation, but the upper contour position can also be used. Furthermore, fluctuation may occur independently of the two edges of the steel plate S. Therefore, it is preferable to monitor the upper and lower contour positions separately to determine the presence or absence of fluctuation.

[0113] Next, the method for determining a fixed wave will be explained. A fixed wave is the phenomenon of continuously transporting a steel plate S while maintaining its wave shape. Within the field of view of camera 2, it is observed that the position and shape of the steel plate S's outline do not change within a certain time. Similar to a wave, this is a phenomenon where the plate speed ahead of the transport line steadily decreases, and the remaining steel plate S bends upwards, thus maintaining a curved outline. Like wave patterns, fixed waves can occur even without plate elongation, therefore they need to be distinguished from shapes caused by plate elongation.

[0114] Since a fixed wave, as the name suggests, produces a stable profile in the time direction (the profile remains unchanged), it can be detected by comparing the profile shape before and after a certain time and calculating its change. For example, the difference in absolute values ​​between the profile distributions can be taken, and if the sum and square of these differences are below a threshold, it can be identified as a fixed wave. Then, the data during the period identified as a fixed wave can be masked, or data from before and after can be used for hole filling, etc. This allows fixed waves to be excluded from the plate elongation detection results, reducing their influence on the measurement results.

[0115] Furthermore, similar to fluctuations, even without using the overall field of view within the image, such as... Figure 10 As shown, the outline position of a point along the length of the steel plate S within the field of view can also be monitored, and the change in the outline position is determined to be a fixed wave when there is no extreme change.

[0116] Reference Figure 10 The method for determining fixed waves that utilize changes in contour position is explained. Figure 10 (a) to (e) represent examples of photographs taken of a fixed wave on steel plate S. Additionally, in Figure 10 In the middle, arranged in chronological order (a), (b), (c), (d), (e). Specifically, the determination of a fixed wave can be carried out in the following order (1) to (3).

[0117] (1) Determine the position X2t in the width direction of an image.

[0118] (2) Calculate the upper contour position Y2t of each image based on the image width direction position X2t determined by (1). The upper contour position Y2t is the upper contour position of the horizontal center of the image.

[0119] (3) Monitor the changes in the calculated upper contour position Y2t and detect the parts that hardly change.

[0120] Figure 10 Figure (f) at the bottom shows a schematic diagram of the upper contour position Y2t drawn in chronological order over time. Using this schematic diagram, it is possible to determine whether a fixed wave exists. Alternatively, as a method for determining a fixed wave, one can monitor the standard deviation within a certain interval as the upper contour position Y2t shifts, identifying areas with a standard deviation below a threshold as fixed waves. Furthermore, one can monitor the difference between the maximum and minimum values ​​within a certain interval, identifying areas below a threshold as fixed waves.

[0121] In addition, Figure 10 In the example, the upper contour position Y2t is used to determine the presence or absence of a fixed wave, but the lower contour position can also be used. Furthermore, a fixed wave may occur independently of the two edges. Therefore, it is preferable to monitor the upper and lower contour positions separately to determine the presence or absence of a fixed wave.

[0122] On the other hand, even when the shape of the steel plate S is flat and does not produce waves, the profile shape does not change after a certain period of time. At this time, it is not a fixed wave, and the shape of the steel plate S and the state of the through-plate are good. Fixed waves are a type of poor through-plate condition caused by plate blockage. When the condition is severe, it comes into contact with the equipment and becomes the main cause of failure. Therefore, it is preferable to be able to detect it automatically.

[0123] Therefore, by incorporating information about the presence or absence of plate elongation into the evaluation of changes in profile shape, it is also possible to determine whether a fixed wave has occurred. For example, even if the change in profile shape is large, methods can be used such that a fixed wave is not identified when the plate elongation index is below a threshold, but only when it is above the threshold. Alternatively, methods can be developed that design evaluation functions and use thresholds for judgment, visually determine whether a wave is fixed and use this as correct data, or use machine learning to determine whether a wave is fixed.

[0124] Furthermore, during the binarization process in step S1 above, the brightness of the thermal radiation light varies depending on the temperature of the object being measured. Therefore, when binarization is performed with a fixed value, the following problem exists: when the steel plate S is bright, a portion of the background is detected as the plate area; conversely, when the steel plate S is dark, the background cannot be detected.

[0125] For this type of problem, brightness correction is preferred. One method for brightness correction is to multiply the overall brightness of the image by a fixed value, using representative values ​​such as the maximum, average, median, or percentile of the overall image brightness as the target value. Alternatively, dark current correction can be performed before multiplication.

[0126] In addition, for example, Figure 11 As shown, thermal radiation emitted from steel plate S illuminates structures such as conveying rollers and worktables. These structures are brightly captured in the image, making them difficult to distinguish from the plate surface. In this case, the plate surface directly receives the thermal radiation and appears bright in the image, while the structures are reflections of the thermal radiation emitted from the plate surface, thus appearing darker than the plate surface in many situations.

[0127] Therefore, a fixed value can be used to determine the binarization threshold for distinguishing structures and plates by implementing the aforementioned brightness correction. However, by automatically detecting valleys based on the shape of the overall image histogram, detection can be performed more stably. When detecting valleys, for example... Figure 12 As shown, by applying a low-pass filter to the brightness histogram itself, the minimum value can be stably calculated through differentiation and search processing. Figure 12 (a) is an example of a brightness histogram, which is a curve with brightness on the horizontal axis and the number of pixels N on the vertical axis. Figure 12 (b) is an example of applying a low-pass filter in the brightness direction of the brightness histogram. It should be noted that the valleys in the histogram exist at two locations: the boundary between the plate surface and the structure, and the boundary between the structure and the background. Therefore, it is preferable to select the larger of the two valleys. Figure 12 Within (b), the area divided by the dashed line is equivalent to the binarization threshold.

[0128] In addition, in this embodiment, an example is described in which the shape index of the edge of the steel plate S is calculated by processing the image of the camera 2 by the image processing device 3, but the obtained camera image can also be directly shown to the operator and fed back to the rolling control.

[0129] Alternatively, a machine learning-based discriminator can be used to infer the shape index of the edge of the steel plate S from an image. Specifically, first, for the obtained image of the steel plate S, learning data is created by establishing associations with the shape index of the edge determined visually or by some other method as the positive solution. Then, using the created learning data and the machine learning method, a discriminator is generated that takes the image as input and the shape index of the edge as output. Using this discriminator, the shape index of the edge of the steel plate S is calculated. Furthermore, the input data to the machine learning and discriminator can be features calculated from the image, in addition to images. Moreover, there are no restrictions on the machine learning method, but convolutional neural networks, etc., can be used as long as real-time performance is not required.

[0130] (Application example of determining whether the shape is qualified)

[0131] Reference Figure 13 An example of applying the shape measuring device of the embodiment to determine whether the shape is qualified will be described. Figure 13 The pass / fail determination system includes a camera 2, an image processing device 3, and a pass / fail determination device 6.

[0132] First, the image processing device 3 calculates the shape index (shape data) of the edge of the steel plate S based on the image captured by the camera 2, and sends it to the pass / fail determination device 6. In the pass / fail determination device 6, based on steel plate information obtained from the upper system (e.g., plate thickness, plate width, steel grade, temperature, etc.) and the shape index of the edge of the steel plate S, it determines whether the shape is qualified or not. The pass / fail determination result is sent to the upper system. In the upper system, based on the pass / fail determination result, it determines whether correction is necessary, whether the defective part should be cut off, and whether rolling in the next process is feasible. In this way, by using the shape index of the edge of the steel plate S for the actions of the next process, it is possible to help suppress defects and improve product quality.

[0133] (Example of application of rolling feedback control)

[0134] Reference Figure 14 The description provides an example of applying a device for determining the shape of a strip object, according to an embodiment, to rolling feedback control. Figure 14 The rolling control system includes a camera 2, an image processing device 3, and a rolling control device 7.

[0135] First, the image processing device 3 calculates the shape index of the edge of the steel plate S based on the image captured by the camera 2 and sends it to the rolling control device 7. In the rolling control device 7, based on steel plate information obtained from the upper system (e.g., plate thickness, plate width, steel grade, temperature, etc.), other measured data such as the plate position, and the shape index of the edge of the steel plate S, control parameters are calculated using the edge shape index. Furthermore, the rolling control device 7 implements leveling and feedback control by sending control signals to the rolls 1. By performing such feedback control, the shape of the product can be stabilized, reducing shape defects.

[0136] (An example of rolling control using machine learning)

[0137] Reference Figure 15 An example of applying the shape measuring device of the strip object described in the embodiment to rolling control using machine learning will be explained. Figure 15 The rolling control system includes a camera 2, an image processing device 3, a rolling control device 7, a data server 8, a machine learning device 9, and a control parameter estimation device 10.

[0138] First, the image processing device 3 calculates the shape index of the edge of the steel plate S based on the image captured by the camera 2, and sends it to the data server 8. In the data server 8, the shape index of the edge is accumulated in a state associated with steel plate information (such as plate thickness, plate width, steel grade, temperature, etc.) obtained from the upper system and the control parameters during rolling.

[0139] The accumulated data is sent to machine learning unit 9. In machine learning unit 9, a model is constructed using machine learning to estimate control parameters for rolling to suppress shape defects. This estimated model is sent to control parameter estimation unit 10, which estimates the control parameters based on steel plate information obtained from the host system. The control parameters are then sent to rolling control unit 7. Rolling control unit 7 implements leveling and preset control by sending control signals to roll 1. By performing such preset control, the shape of the product can be stabilized, reducing shape defects themselves.

[0140] Furthermore, while the control parameters in the preset control were described in this application example, machine learning can also be applied when calculating the control parameters for feedback control. Additionally, by combining preset control and feedback control, the shape quality of the product can be further improved.

[0141] In addition to the above-described application examples, the shape measuring device for the strip object in the embodiments can also be incorporated as part of the strip object manufacturing equipment.

[0142] Furthermore, the shape measurement method for a strip-shaped object according to the embodiment can be applied to a shape control method for a strip-shaped object. In this case, the shape of the strip-shaped object is measured using the above-described shape measurement method, and control is performed based on the measurement result to make the shape of the strip-shaped object the desired shape.

[0143] Furthermore, the method for determining the shape of a strip-shaped object according to the embodiment can be applied to a method for manufacturing a strip-shaped object. In this case, the shape of the strip-shaped object is determined by the above-described method for determining the shape of a strip-shaped object, and the strip-shaped object is manufactured based on the determination result.

[0144] Furthermore, the shape measurement method for a strip object according to the embodiment can be applied to a quality management method for strip objects. In this case, the shape of the strip object is measured using the above-described shape measurement method, and the quality of the strip object is managed based on the measurement results.

[0145] The shape measurement method, shape control method, manufacturing method, quality management method, shape measurement device, and manufacturing equipment of the strip object according to the above embodiments have the following effects.

[0146] First, by passively photographing the thermal radiation of the steel plate S being measured, it is possible to take pictures from a separate, well-designed environment without setting up light sources or sensors around the steel plate S in the through-plate, and to clearly capture the shape of the edge of the steel plate S.

[0147] Furthermore, by selecting an imaging element suitable for the temperature of the object as the imaging element of camera 2, thermal radiation light can be received efficiently. For example, for an object to be measured at around 900°C, the sensitivity can be improved by using the near-infrared component of the imaging element of Si. For a steel plate S being transported at high speed, a clear image without shaking can also be obtained by shortening the exposure time.

[0148] Furthermore, by taking pictures with the reference plane α of the surface of the strip object (steel plate S) forming an angle θ with the optical axis of camera 2 that is not 90 degrees, and with the angle φ between the orthographic projection of the optical axis of camera 2 onto plane α and the transport direction p of the strip object (steel plate S) that is not 0 degrees, the contours of the two edges of the steel plate S can be clearly captured, and the shape index of the edges can be quantitatively calculated, thus clearly capturing the plate elongation. Moreover, even better results can be obtained by shooting the strip object (steel plate S) from the width direction (i.e., the direction where angle φ is close to 90 degrees). Even better results can be obtained by shooting from a low angle (i.e., the direction where angle θ is close to 0 degrees). Of course, even better results can be obtained by shooting from the direction where angle φ is close to 90 degrees to the direction where angle θ is close to 0 degrees.

[0149] Furthermore, while this invention describes a steel plate S from the finishing exit side of a hot-rolling process, it is of course applicable to other high-temperature plate-shaped steel materials such as thick steel plates and slabs. Moreover, since this invention can obtain the thermal radiation of the object being measured, it can be applied not only to strip-shaped objects produced in steelmaking processes, but also to various strip-shaped objects made of different materials. Furthermore, this invention is preferred when the elongation of the edge of the strip-shaped object is used as the object of shape measurement, as it yields even greater results.

[0150] The above description details the method for measuring the shape of a strip object, the method for controlling the shape of a strip object, the method for manufacturing a strip object, the method for managing the quality of a strip object, the device for measuring the shape of a strip object, and the equipment for manufacturing a strip object, all of which are specific embodiments for carrying out the invention. However, the scope of the invention is not limited to these descriptions and must be interpreted broadly based on the claims. Furthermore, various modifications and alterations based on these descriptions are also included within the scope of the invention, which is self-evident.

[0151] [Explanation of reference numerals in the attached figures]

[0152] 1 roll

[0153] 2 cameras

[0154] 3 Image Processing Device

[0155] 6. Qualification / Failure Determination Device

[0156] 7 Rolling control device

[0157] 8 data servers

[0158] 9 machine learning devices

[0159] 10. Control Parameter Estimation Device

[0160] S-shaped steel plate.

Claims

1. A method for determining the shape of a strip-shaped object, wherein, include: The shooting steps involve capturing an image of the thermal radiation light of the strip object in such a way that the angle θ between the reference plane α of the surface of the strip object and the optical axis of the camera is not 90 degrees, and the angle φ between the orthographic projection of the optical axis of the camera onto the plane α and the transport direction p of the strip object is not 0 degrees. as well as The image processing step calculates the contour distribution of the strip-shaped object based on the obtained image, thereby calculating the shape index of the edge of the strip-shaped object.

2. The method for determining the shape of a strip-shaped object according to claim 1, wherein, In the image processing step, Extract the region of the strip-shaped object from the obtained image. The position of the edge portion is calculated in the extracted region, thereby calculating the contour distribution of the strip-shaped object.

3. The method for determining the shape of a strip-shaped object according to claim 1 or 2, wherein, In the image processing step, based on the obtained contour distribution of the strip-shaped object, one or more of the following parameters are calculated as indicators: the steepness of the edge of the strip-shaped object, the wave height, the wave spacing, the elongation, and the elongation rate.

4. The method for determining the shape of a strip-shaped object according to any one of claims 1 to 3, wherein, In the image processing step, the resolution in the wave height direction and the wave spacing direction is calculated based on the positional relationship between the camera and the strip-shaped object, and then converted into the actual size.

5. A method for controlling the shape of a strip-shaped object, comprising determining the shape of the strip-shaped object using the shape measurement method of any one of claims 1 to 4, and controlling the shape of the strip-shaped object based on the measurement results, so that the shape of the strip-shaped object becomes a desired shape.

6. A method for manufacturing a strip-shaped object, comprising determining the shape of the strip-shaped object using the shape determination method for any one of claims 1 to 4, and manufacturing the strip-shaped object based on the determination result.

7. A quality management method for a strip-shaped object, comprising determining the shape of the strip-shaped object using the shape determination method of any one of claims 1 to 4, and managing the quality of the strip-shaped object based on the determination results.

8. A shape measuring device for a strip-shaped object, wherein, have: The imaging unit captures an image of the thermal radiation light of the strip object in such a way that the angle θ between the reference plane α of the surface of the strip object and the optical axis of the camera is not 90 degrees, and the angle φ between the orthographic projection of the optical axis of the camera onto the plane α and the transport direction p of the strip object is not 0 degrees. and The image processing unit calculates the contour distribution of the strip-shaped object based on the obtained image, thereby calculating an index of the shape of the edge of the strip-shaped object.

9. A manufacturing apparatus for a strip-shaped object, comprising the shape measuring device for the strip-shaped object as described in claim 8.

Images