Measuring method and measuring apparatus of bend of steel material, and manufacturing method of steel material

Abstract

To provide a technique capable of measuring bend of steel material such as an H-section steel in relatively small space and in short time.SOLUTION: A measuring method of bend of steel material for measuring bend of steel material includes the steps of: photographing the whole area in a longitudinal direction of the steel material placed stationary by a camera; calculating a shape of the steel material from an image photographed by the camera by using image recognition; and measuring the bend of the steel material by using a shape calculation result by the image recognition.SELECTED DRAWING: Figure 4

Description

[Technical field]

[0001] The present invention relates to a method and device for measuring the curvature of a steel material such as an H-beam, and a method for manufacturing the steel material. [Background technology]

[0002] After rolling, the H-section steel is air-cooled on a cooling bed, and then goes through a straightening process to remove the "belly bulge" on the web surface and the inclination of the flange surface, before going through a material inspection process and a shipping label attachment process.

[0003] In the material inspection process, product dimensions such as web warpage (vertical warpage when the flange faces are in an upright H-shape, hereafter simply referred to as "warpage") and flange bending (lateral bending when the flange faces are in an upright H-shape, hereafter simply referred to as "bending") are inspected by visual inspection. If the visual inspection reveals significant warping or bending of the H-shaped steel, the steel is taken offline and bending is measured and judged by hand.

[0004] However, visual inspection is a sensory inspection, and manual measurement results vary depending on the person doing the measurement, making it difficult to quantitatively grasp the bending. Also, when measuring the maximum bending position and bending amount of H-beam by manual measurement, it is necessary to tie the front and rear ends of the H-beam with piano wire or to attach a long member over the entire length of the H-beam, which takes several tens of minutes to measure per H-beam, resulting in poor inspection efficiency.

[0005] Patent Document 1 discloses a technology capable of solving such problems in which an H-shaped steel beam is placed on a support table, a distance sensor is installed so as to be able to move along the longitudinal direction of the placed H-shaped steel beam, and the distance sensor measures the facing distance to the flange surface of the H-shaped steel beam at multiple points from one end to the other end of the H-shaped steel beam, and the bending state of the flange surface of the H-shaped steel beam is determined based on the information on the facing distance measured at multiple points. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] JP 2013-228325 A Summary of the Invention [Problem to be solved by the invention]

[0007] However, in the technology of Patent Document 1, it is necessary to run a distance sensor along the longitudinal direction of the H-shaped steel, which requires a large space to install a device including a distance sensor for determining the bending state. Also, in the technology of Patent Document 1, it is necessary to leave the H-shaped steel stationary while the distance sensor is running, which means that the time for leaving the H-shaped steel stationary is relatively long, which affects product manufacturing in continuous operation.

[0008] Therefore, the present invention provides a technique capable of measuring the bending of steel materials such as H-beams in a relatively small space and in a short period of time. [Means for solving the problem]

[0009] In order to solve the above problems, the present invention provides the following [1] to

[11] .

[0010] [1] A method for measuring the bending of a steel material, comprising: A step of photographing the entire longitudinal area of ​​the stationary steel material with a camera; A step of calculating a shape of the steel material from the image captured by the camera using image recognition; Measuring the bending of the steel material using a shape calculation result by the image recognition; A method for measuring the bending of steel material.

[0011] [2] The method for measuring the bend of steel material described in [1], wherein the shape calculating step outputs the shape of the steel material obtained by the image recognition as coordinates on the photographed image.

[0012] [3] The method for measuring the curvature of steel material described in [2], wherein the step of measuring the curvature of the steel material uses coordinates on the image to measure the curvature of the steel material.

[0013] [4] The method for measuring the curvature of steel material described in [3], wherein the step of measuring the curvature includes converting coordinates on the image into real coordinates and measuring the curvature based on the real coordinates. [5] A method for measuring the curvature of steel material described in any one of [1] to [4], wherein the step of calculating the shape of the steel material uses segmentation by a neural network for the image recognition.

[0014] [6] A method for measuring the bend of a steel material described in any of [1] to [4], further comprising a step of determining that the steel material is placed at a predetermined location, and the photographing step is carried out after it is determined that the steel material is placed at the predetermined location.

[0015] [7] The method for measuring the bend of steel material described in [6], in which the process of determining whether the steel material is placed at a predetermined location determines whether the steel material is placed at a predetermined location by utilizing the fact that the distance measured by a distance sensor is different when the steel material is placed at a predetermined location from when it is not.

[0016] [8] A steel bending measurement device for measuring the bending of steel, comprising: A camera that captures the entire longitudinal area of ​​the stationary steel material; an image recognition means for calculating a shape of the steel material from an image captured by the camera using image recognition; a bending measurement means for measuring a bending of the steel material using a shape calculation result by the image recognition means; A steel bend measuring device having the above structure.

[0017] [9] The steel bend measuring device described in [8], wherein the image recognition means uses segmentation by a neural network for the image recognition.

[0018]

[10] rolling the steel material to a predetermined shape; A step of cooling the rolled steel material; Measuring the bending of the cooled steel material; having The step of measuring the bending of the steel material includes: A step of photographing the entire longitudinal area of ​​the stationary steel material with a camera; A step of calculating a shape of the steel material from the image captured by the camera using image recognition; Measuring the bending of the steel material using a shape calculation result by the image recognition; The method for producing a steel material comprising the steps of:

[0019]

[11] A method for manufacturing steel material as described in

[10] , wherein the step of calculating the shape of the steel material uses image recognition as the image processing, and the image recognition is segmentation by a neural network. Effect of the Invention

[0020] According to the present invention, there is provided a technique capable of measuring the bending of steel materials such as H-beams in a relatively small space and in a short period of time. [Brief description of the drawings]

[0021] [Figure 1] 1 is a block diagram showing an example of a bend measuring device for carrying out a method for measuring bend of a steel material according to an embodiment; [Diagram 2] FIG. 1 is a diagram showing an example of equipment having a predetermined location where an H-shaped steel is placed. [Diagram 3] FIG. 13 is a diagram for explaining detection of the presence or absence of an H-shaped steel by a distance sensor. [Figure 4] 1 is a flowchart showing a method for measuring a bend of a steel material according to one embodiment. [Diagram 5] 5 is a flowchart specifically showing step ST1 in FIG. 4. [Figure 6] 5 is a flowchart specifically showing step ST4 in FIG. 4. [Figure 7] FIG. 13 is a diagram showing an example of converting coordinates on an image into real coordinates. [Figure 8] FIG. 13 is a diagram showing an example of bending measurement. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0022] Hereinafter, an embodiment of the present invention will be described with reference to the accompanying drawings. FIG. 1 is a block diagram showing an example of a bend measuring device for carrying out a method for measuring bend of a steel material according to an embodiment.

[0023] As shown in Fig. 1, the steel material bend measuring device 100 measures the bend of an H-shaped steel 200 as a steel material placed at rest in a predetermined location 300, and has a camera 1, a distance sensor 2, a main PC 3, and an image recognition PC 4. The main PC 3 has an image capture unit 5, a stationary state determination unit 6, and a bend measurement unit 7. The image recognition PC 4 has an image recognition unit 8.

[0024] The predetermined location 300 where the H-shaped steel is placed is not particularly limited as long as the entire area in the longitudinal direction of the H-shaped steel 200 can be measured by the camera 1 as described below and the H-shaped steel 200 is stationary for about one second. For example, in a facility in which the H-shaped steel 200 is transported while intermittently operating on a chain conveyor 400 as shown in Fig. 2, the H-shaped steel 200 is stationary for a short time at a predetermined location, so that the bending of the H-shaped steel can be measured at that location. In addition, the H-shaped steel is placed at a location where air cooling is performed on an air-cooled floor, so that the bending of the H-shaped steel can be measured at that location.

[0025] The camera 1 photographs the entire longitudinal area of ​​the H-shaped steel 200 placed stationary in a predetermined location 300. The camera 1 is required to photograph the H-shaped steel while the steel is stationary for about one second. The image taken by the camera 1 is captured by the image capture unit 5 of the main PC 3.

[0026] The distance sensor 2 constantly measures the predetermined place 300 where the H-shaped steel 200 is placed, and detects whether the H-shaped steel 200 is present at the predetermined place 300. That is, the distance measured by the distance sensor 2 differs between when the H-shaped steel 200 is present (placed stationary) at the predetermined place 300 as shown in FIG. 3(a) and when the H-shaped steel 200 is not present at the predetermined place 300 as shown in FIG. 3(b), and is shorter when the H-shaped steel 200 is present. This can be used to detect whether the H-shaped steel 200 is present. Also, when the H-shaped steel 200 is moving with respect to the predetermined place 300, the measured value of the distance sensor 2 differs from when the H-shaped steel 200 is placed stationary, so that the presence or absence of the H-shaped steel 200 can be detected. The distance sensor 2 is not particularly limited as long as it can perform such detection.

[0027] The stationary state determination unit 6 in the main PC 3 determines whether the H-shaped steel 200 has been placed stationary for a specified time. That is, the stationary state determination unit 6 takes in the measurement result of the distance sensor 2, determines whether the H-shaped steel 200 is placed stationary in the specified place 300, and instructs the camera 1 to capture an image if it determines that the H-shaped steel 200 is placed stationary. As described above, the distance sensor 2 produces different measured values ​​depending on whether the H-shaped steel 200 is present in the specified place 300 or not. Therefore, the stationary state determination unit 6 determines that the H-shaped steel 200 is in a stationary state if the measured value of the distance sensor 2 remains smaller than the threshold value for a specified time or more.

[0028] The image recognition unit 8 in the image recognition PC 4 calculates the shape of the H-shaped steel 200 captured by the camera 1 using image recognition from the image captured by the image capture unit 5, and outputs it as coordinates representing the shape of the H-shaped steel 200. Image recognition is a type of image processing, and is a technology that grasps features from an image and identifies an object. For example, segmentation by a general neural network, so-called AI segmentation, can be used as image recognition. In segmentation by a neural network (AI segmentation), it is possible to determine where a detection target is located in an input image from an input image, and output the shape of the detection target as coordinates in pixel units. As AI segmentation, Mask R-CNN and Fast R-CNN using deep learning are well-known methods. Since the image recognition unit 8 only needs to recognize the contour of the object (in this case, the flange top surface of the H-shaped steel) in pixel units on the image by image recognition, a color clustering method may be used to extract an area that is statistically similar to the color of the flange top surface of the H-shaped steel.

[0029] The bend measuring unit 7 in the main PC 3 measures the bend of the H-shaped steel 200 based on the shape calculation result by image recognition by the image recognition unit 8. Specifically, the bend measuring unit 7 converts the coordinates representing the shape of the H-shaped steel 200 output from the image recognition unit 8 into real coordinates (coordinates in real space), and measures the bend of the H-shaped steel 200 based on the real coordinates.

[0030] Next, a method for measuring the curvature of a steel material carried out by the steel material curvature measuring device configured as above will be described.

[0031] FIG. 4 is a flowchart showing a method for measuring the bend of a steel material according to one embodiment. First, it is determined whether or not the H-shaped steel 200 is placed at rest in a predetermined location 300 (step ST1).

[0032] When it is determined that the H-shaped steel 200 is placed still, the entire area in the longitudinal direction of the H-shaped steel 200 is photographed by the camera 1 (step ST2).

[0033] Next, the image recognition unit 8 calculates the shape of the H-shaped steel 200 from the image captured by the camera 1 using image recognition (step ST3).

[0034] Next, the bending of the H-shaped steel 200 is measured by the bending measurement unit 7 using the shape calculation result by image recognition (step ST4).

[0035] Specifically, in step ST1, as shown in the flowchart of FIG. 5, the distance sensor 2 measures the distance to the predetermined location 300 (step ST1-1), and based on the result, the stationary state determination unit 6 determines whether the H-shaped steel 200 is stationary at the predetermined location 300 (step ST1-2). If it is determined that the H-shaped steel 200 is stationary, an instruction to shoot is given to the camera 1 (step ST1-3). Next, to avoid making a determination on the same H-shaped steel, a certain specified time is awaited (step ST1-4), and the initial measurement state is restored, and the same process is carried out on the next H-shaped steel. This makes it possible to automatically and continuously perform stationary determination on multiple H-shaped steels.

[0036] In step ST1-2, as shown in Figure 3 above, the distance measured by the distance sensor 2 is shorter when the H-shaped steel 200 is present at the specified location 300 than when it is not present, so if the measurement value of the distance sensor 2 remains smaller than the threshold value for a specified time or longer, it is determined that the H-shaped steel 200 is in a stationary state.

[0037] In step ST2, in a state in which it is determined in step ST1 that the steel is left stationary and imaging by the camera 1 is possible, the camera 1 images the entire area in the longitudinal direction of the H-shaped steel 200. The captured image is captured by the image capture unit 5.

[0038] In step ST3, the image recognition unit 8 uses image recognition to calculate the shape of the H-shaped steel 200 from the image captured by the camera 1 captured in the image capture unit 5, and outputs the calculated shape as coordinates representing the shape of the H-shaped steel. As described above, the image recognition can be performed using segmentation using a general neural network, so-called AI segmentation. As AI segmentation, Mask R-CNN and Fast R-CNN using deep learning are well-known methods. As described above, it is sufficient to recognize the contour of the target object, i.e., the flange top surface of the H-shaped steel, in pixel units on the image, so that a color clustering method can be used for image recognition to extract an area that is statistically similar to the color of the flange top surface of the H-shaped steel.

[0039] In step ST4, the bending of the H-shaped steel 200 is measured by the bending measurement unit 7 based on the coordinates representing the shape of the H-shaped steel 200 output from the image recognition unit 8 as a result of shape calculation by image recognition. Specifically, as shown in Fig. 6, the bending measurement unit 7 converts the coordinates representing the shape of the H-shaped steel 200 on the image output from the image recognition unit 8 into real coordinates (coordinates in real space) by coordinate conversion (step ST4-1), and measures the bending of the H-shaped steel 200 based on the real coordinates (step ST4-2).

[0040] In step ST4-1, the coordinates after conversion can be calculated from the relational expressions between the camera coordinate system, the image coordinate system, and the world coordinate system, and the condition where the shape of the H-shaped steel (the upper surface of the flange) exists (the horizontal plane obtained by adding the height of the H-shaped steel to the height of the bottom surface of the conveyor table). In practice, the coordinates on the image are converted to real coordinates using a coordinate conversion expression created by taking these into account. The coordinate conversion expression at this time is premised on the upper surface of the H-shaped steel 200.

[0041] In step ST4-1, for example, if one point in the coordinates on the output image is assumed to be A(X1, Y1) as shown in Fig. 7, it can be converted from a reference point O' (this reference point can be anywhere) to, for example, X1'm to the east and Y1'm to the north. All points required to calculate the curvature can be converted from the reference point of the real coordinates in this way.

[0042] In step ST4-2, in order to measure the bending, for example, as shown in FIG. 8, A(X1',Y1'), B(X2',Y2'), C(X3',Y3'), D(X4',Y4'), and E(X5',Y5') are obtained as the actual coordinates constituting the outline of the H-shaped steel. Here, A(X1',Y1') and D(X4',Y4') are points constituting one short side of the H-shaped steel, the point with the smallest X' coordinate is A(X1',Y1'), and the point with the largest X' coordinate is D(X4',Y4'). Similarly, B(X2',Y2') and E(X5',Y5') are points constituting the other short side of the H-shaped steel, the point with the smallest X' coordinate is B(X2',Y2'), and the point with the largest X' coordinate is E(X5',Y5'). Regarding the Y' coordinate, Y1'>Y2' and Y4'>Y5'. C (X3', Y3') is a point on one of the long sides connecting points A and B. In the example of FIG. 8, since the bending direction of the H-shaped steel 200 is the + direction in the figure, the X' coordinate X3' is not less than X1' and not more than X4', and the point with the largest X' coordinate on one of the long sides connecting points A and B is C. The Y' coordinate Y3' is not less than Y2' and not more than Y1'. The bending of the H-shaped steel 200 is measured using these points. The amount of bending is measured as the length of a perpendicular line dropped from C (X3', Y3') to a straight line connecting A (X1', Y1') and B (X2', Y2'). When the bending direction is the - direction in the figure, the point with the smallest X' coordinate is C (X3', Y3').

[0043] In this embodiment, the entire longitudinal area of ​​the H-shaped steel 200 as the steel material is photographed by the camera 1, the shape of the H-shaped steel 200 is calculated from the photographed image using image recognition, and the calculation result is used to measure the bending of the H-shaped steel 200 by the bending measurement unit 7. Therefore, the only equipment added to the product manufacturing equipment is the camera 1, which requires a smaller space than Patent Document 1, and since the time that the H-shaped steel 200 is left stationary is the time that it is photographed by the camera 1, the time to measure the bending can be shorter than that of Patent Document 1.

[0044] Furthermore, by using the distance sensor 2 and the stationary state determination unit 6 to confirm that the H-shaped steel 200 is placed at the specified location 300 before photographing with the camera 1, it is possible to reliably photograph the H-shaped steel 200 in a stationary state, and to automatically and continuously measure the bending of a plurality of H-shaped steels 200. Furthermore, the distance sensor 2 has little effect on the space, making it possible to maintain a small space.

[0045] Furthermore, since the shape of the H-shaped steel 200 is calculated using image recognition and the bending is measured based on that, the bending can be measured easily and with high accuracy.

[0046] When manufacturing H-beam steel, the steel is rolled to the specified shape, cooled, and then subjected to post-processing processes, and the above-mentioned bending measurements are carried out as part of material inspection.

[0047] Although the embodiments of the present invention have been described above, these are merely examples and should not be considered as limiting. The above embodiments may be omitted, substituted, or modified in various ways without departing from the gist of the present invention.

[0048] For example, in the above embodiment, an H-shaped steel is used as the steel material, but the present invention is not limited to this and can be applied to other steel materials such as shaped steel other than H-shaped steel. The image recognition method and bending measurement method are also not limited to the above embodiment and various methods can be applied. Furthermore, in the above embodiment, an example was shown in which image recognition is performed by an image recognition PC separate from the main PC, but it may be performed by a single PC. [Explanation of symbols]

[0049] 1 Camera 2 Distance Sensor 3. Main PC 4. Image Recognition PC 5 Image capture section 6. Stationary state determination unit 7 Bend measurement section 8 Image Recognition Unit 200 H-shaped steel (steel material) 300 designated location 400 Chain Conveyor

Claims

1. A method for measuring the bending of a steel material, comprising: A step of photographing the entire longitudinal area of ​​the stationary steel material with a camera; A step of calculating a shape of the steel material from the image captured by the camera using image recognition; Measuring the bending of the steel material using a shape calculation result by the image recognition; A method for measuring the bending of steel material.

2. 2. The method for measuring a bend of a steel material according to claim 1, wherein the step of calculating the shape includes outputting the shape of the steel material obtained by the image recognition as coordinates on the captured image.

3. The method for measuring a bend in a steel material according to claim 2 , wherein the step of measuring the bend comprises measuring the bend in the steel material using coordinates on the image.

4. 4. The method for measuring a bend in a steel material according to claim 3, wherein the step of measuring the bend comprises converting coordinates on the image into real coordinates, and measuring the bend based on the real coordinates.

5. The method for measuring the curvature of a steel material according to claim 1 , wherein the step of calculating the shape of the steel material uses segmentation by a neural network as the image recognition.

6. A method for measuring the bend of a steel material described in any one of claims 1 to 4, further comprising a step of determining that the steel material is placed at a predetermined location, and the photographing step is performed after it is determined that the steel material is placed at the predetermined location.

7. The method for measuring the bend of steel material described in claim 6, wherein the process of determining whether the steel material is placed at a predetermined location determines whether the steel material is placed at a predetermined location by utilizing the fact that the distance measured by a distance sensor is different when the steel material is placed at a predetermined location from when the steel material is otherwise placed at a predetermined location.

8. A steel bending measurement device for measuring the bending of a steel material, A camera that captures the entire longitudinal area of ​​the stationary steel material; an image recognition means for calculating a shape of the steel material from an image captured by the camera using image recognition; a bending measurement means for measuring a bending of the steel material using a shape calculation result by the image recognition means; A steel bend measuring device having the above structure.

9. The steel material bend measuring device according to claim 8 , wherein the image recognition means uses segmentation by a neural network for the image recognition.

10. rolling the steel material to a predetermined shape; A step of cooling the rolled steel material; Measuring the bending of the cooled steel material; having The step of measuring the bending of the steel material includes: A step of photographing the entire longitudinal area of ​​the stationary steel material with a camera; Calculating the shape of the steel material from the image captured by the camera using image recognition; Measuring the bending of the steel material using a shape calculation result by the image recognition; The method for producing a steel material comprising the steps of:

11. The method for manufacturing a steel material according to claim 10 , wherein the step of calculating the shape of the steel material uses segmentation by a neural network as the image recognition.

Images