A method for detecting deviation of hot-rolled strip steel considering loop effect

Abstract

The application discloses a hot-rolled strip steel deviation detection method considering loop influence, and belongs to the technical field of plate and strip rolling. The method adopts a machine vision method to perform online detection on the width and deviation of the finished strip steel, and a detection device is arranged above a rolling mill to collect and detect the strip steel between two racks. First, a vision system is calibrated, then a strip steel image is collected, and a strip steel edge is extracted, then the rolling mill outlet strip steel thickness and the real-time rotation angle of the loop are obtained from the field first-level data through a communication system, and influence factors such as the outlet strip steel thickness and the real-time rotation angle of the loop are considered, and the real world coordinates of the strip steel edge are calculated according to the strip steel image coordinates, and finally the actual deviation of the strip steel is calculated according to the strip steel edge coordinates. The strip steel deviation detected by the method can improve the strip steel deviation detection precision to the millimeter level, and provides a reliable basis for subsequent strip steel deviation control.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of plate strip rolling, and particularly relates to a hot-rolled strip steel deviation detection method considering the influence of loop. BACKGROUND

[0002] As an indispensable raw material, plate strip products are applied in various fields such as automobile manufacturing, farm tool machinery, electronic products, food packaging and instrument equipment, providing great convenience for our work and life.

[0003] At present, the steel production of China ranks first in the world, but there is a big problem that the production capacity of ordinary steel is excessive, and high-end steel materials such as high-grade alloy steel, fatigue-resistant high-strength steel, high-temperature-resistant and radiation-resistant steel and special steel with high added value and high technical content are seriously dependent on imports. The reason is that the shape problem becomes one of the main problems that affect the production efficiency and product quality of these high-precision plate intermediate blanks. In recent years, with a large amount of research on the shape theory, detection and control of the plate in the finishing stage, the geometric size precision, processing performance and surface quality of the plate strip product have been greatly improved, and the shape quality of the hot-rolled plate has become one of the main indicators for evaluating the quality of the hot-rolled plate. Poor plate shape not only makes the product unqualified, but also affects the subsequent rolling process and increases the difficulty of plate shape control. Poor plate shape is mainly divided into symmetric plate shape and asymmetric plate shape. For hot continuous rolling, the control model and control system of symmetric plate shape have basically met the requirements, and the automation level and control precision have reached a high level. However, for asymmetric plate shape, there are still deficiencies in the research on the inducing causes and control model, resulting in that there is still no relatively mature online detection system, and the control process mainly depends on the subjective judgment and personal experience of the operators, which has great deviation and instability.

[0004] The running speed of the strip steel in the finishing rolling mill is fast, and when thin-gauge products are rolled, the most important factor that directly or indirectly affects the rolling stability is the deviation of the strip steel. The deviation of the strip steel between the stands is caused by the unevenness of the three-dimensional deformation of the metal along the width direction of the rolled piece, which causes the deviation defects of the size precision and internal stress, and not only causes the quality problems of the finished product, but also affects the continuity and stability of the whole rolling process.

[0005] The deviation of the strip steel causes the fluctuation of the wedge index of the strip steel, and in the thin-gauge rolling process, it is easy to produce wave shape between the stands, which affects the quality of the strip steel, and in severe cases, it will cause the tail throwing and steel stacking accidents between the stands, scratch the surface of the strip steel, produce edge damage and other defects, reduce the yield of the finished product and degrade the product, and at the same time, cause damage to the rollers, guides and other equipment. At present, there is no effective monitoring means for the deviation of the strip steel between the stands, and only the center line offset data can be obtained at the width measuring instrument at the outlet of the finishing rolling mill. Based on the current production status and demand, it is particularly important to design and develop a deviation detection system for the strip steel between the stands of the finishing rolling mill.

[0006] Adopt machine vision technology, establish a set of machine vision system including camera, processor and so on, use camera to carry out real-time shooting to moving strip, establish strip deviation detection method, eliminate the influence of loop swing on detection result, thereby real-time detect strip deviation. SUMMARY

[0007] The application provides a hot-rolled strip deviation detection method considering loop influence.

[0008] The method adopts machine vision method to detect the width and deviation of the finished strip on line. The method adopts the method of area array camera to detect the strip between the finishing stands, and the detection device is arranged above the rolling mill to collect and detect the strip between the two stands.

[0009] Due to the change of loop angle in the rolling process, the height of the strip changes constantly, which finally causes the conversion error between the world coordinate and the image coordinate, thereby causing detection error. The application can effectively eliminate the above error, thereby making the detection result more accurate.

[0010] To solve the above technical problem, the application provides the following technical scheme:

[0011] The method comprises the following steps:

[0012] S1: vision system calibration:

[0013] The loop is put down at the rolling gap, the camera is calibrated by Zhang Zhengyou calibration method, and the internal parameters and external parameters of the vision system at this time are obtained;

[0014] S2: collect the strip image and extract the strip edge:

[0015] When the strip enters the detection area, the camera is controlled to continuously collect the strip image at a fixed collection frequency (generally 10-30 Hz), the Canny operator and the sub-pixel edge detection operator are combined to detect the edge of the collected strip image, and the pixel coordinates of the left and right edges of the strip image at the collection time are obtained;

[0016] S3: obtain the thickness h1 of the strip at the outlet of the rolling mill and the real-time rotation angle θ of the loop from the on-site first-level data through the communication system;

[0017] S4: considering the thickness of the outlet strip and the real-time rotation angle of the loop, the real world coordinates of the strip edge are calculated according to the strip image coordinates;

[0018] S5: calculate the actual deviation of the strip according to the real world coordinates of the strip edge.

[0019] The vision system in S1 includes an area scan camera and camera lens, a water cooling device, a camera adjustment device, two photoelectric converters, an image processing server, a KVM extender, and a terminal display.

[0020] The area array camera is fixed between two finishing mills via a camera adjustment device, obliquely shooting the strip steel during the rolling process. The camera's fixed angle is adjusted according to specific conditions to center the strip steel in the image. The camera adjustment device consists of a pitch adjustment device and a rotation adjustment device, used to adjust the spatial pose of the area array camera. The entire adjustment device is welded and fixed above the mill archway on site. The area array camera is surrounded by a water-cooling device with high-pressure cooling water to prevent the camera from overheating. The water-cooling device is bolted and fixed below the camera adjustment device.

[0021] Two photoelectric converters are responsible for transmitting camera image information. One is located at the camera end and connected to the camera via a network cable, while the other is located at the image processing server end and connected to the image processing server via a network cable. The image information is transmitted to the photoelectric converter at the camera end via gigabit Ethernet, where the electrical signal is converted into an optical signal. This signal is then transmitted via optical fiber to the photoelectric converter at the image processing server end, and finally transmitted to the image processing server via gigabit Ethernet. The data information processed by the image processing server is then transmitted to the terminal display via a KVM extender to display the inspection results and strip information to the user.

[0022] The internal parameters in S1 include the scale factor s, the length of a unit pixel on the x-axis in the physical coordinate system dx, the length of a unit pixel on the y-axis in the physical coordinate system dy, the corresponding points u0 and v0 of the origin of the camera physical coordinate system in the camera pixel coordinate system, and the lens focal length f.

[0023] External parameters include the rotation matrix R and the translation matrix t.

[0024] The specific steps in S4 are as follows:

[0025] S41: Based on the geometric relationship between the strip, the looper, and the two finishing mills, calculate the real-time rotation angle α of the strip on the shooting side using the real-time rotation angle of the looper.

[0026]

[0027] In the formula:

[0028] θ — Real-time rotation angle of the looper, in rad;

[0029] H—The distance between the rotating shaft of the looper connecting rod and the horizontal running plane of the strip, in mm;

[0030] h-height of the loop center after the loop movement relative to the initial position of the loop, unit: mm;

[0031] L-distance between the loop center at the initial position of the loop and the previous rolling mill, unit: mm;

[0032] l-horizontal distance between the loop positions before and after the movement, unit: mm;

[0033] r1-radius of the loop, unit: mm;

[0034] r2-rotational radius of the loop, unit: mm;

[0035] β-angle between the loop connecting rod and the horizontal plane at the initial position of the loop, unit: rad;

[0036] Z-distance between the roll gaps of two adjacent finishing mills, unit: mm;

[0037] m-distance between the rotational axis of the loop connecting rod and the present rolling mill, unit: mm;

[0038] S42: according to the real-time rotation angle of the strip and the thickness of the strip at the outlet of the rolling mill, the coordinate system conversion model is optimized:

[0039] the calibration plane is plane I, and the plane where the strip is located after the loop rotates is plane II;

[0040] The rotation of the loop simultaneously drives the strip to rotate around the roll gap, and the real-time rotation angle of the strip is α;

[0041] In the calibration process, the distance from the roll gap to plane I is a, which is obtained by measurement;

[0042] The y-axis of the world coordinate system of the calibration plane is placed on the rolling center line;

[0043] In the process of changing the strip from plane I to plane II, the world coordinate system changes from coordinate system O-X w Y w Z w to coordinate system O1-X2Y2Z2;

[0044] The real world coordinates of the final strip edge are obtained as follows:

[0045]

[0046] In the formula:

[0047] u, v-pixel coordinates of the left and right edges of the strip image at the acquisition moment, unit: 1;

[0048] s-camera internal parameter, which is a scale factor and is obtained by camera calibration, unit: 1;

[0049] R - rotation matrix in camera extrinsic parameters, obtained through camera calibration, unit: 1;

[0050] t - translation matrix in camera extrinsic parameters, obtained through camera calibration, unit: 1;

[0051] dx, dy - camera intrinsic parameters, representing the length of unit pixel on the x, y axis in the physical coordinate system, obtained through camera calibration, unit: mm;

[0052] u0, v0 - camera intrinsic parameters, representing the corresponding point of the origin of the camera physical coordinate system in the camera pixel coordinate system, obtained through camera calibration, unit: 1;

[0053] f - camera intrinsic parameter, representing the lens focal length, obtained through camera calibration, unit: mm.

[0054] The plane in the calibration in the S42 is a loop real-time rotation angle 0 plane.

[0055] The S5 is specifically:

[0056] According to the S2, the pixel coordinates P1(u1, v1) and P2(u2, v2) of the left and right edges of the strip are obtained, and the world coordinate system coordinates of P1 and P2 are obtained through the S4, and are respectively denoted as P1(X 21 ,Y 21 ,Z 21 ), P2(X 22 ,Y 22 ,Z 22 );

[0057] Since the y axis of the world coordinate system coincides with the rolling center line, the width and the deviation of the strip detection are:

[0058] W=X 22 -X 21

[0059]

[0060] In the formula:

[0061] W - the width of the detected strip, unit: mm;

[0062] D - the deviation of the detected strip, unit: mm.

[0063] The hot-rolled strip deviation on-line detection method is used for real-time detection of the strip deviation between the hot-rolled production line finishing mill group, so as to provide an effective basis for the finishing strip deviation control.

[0064] Compared with the prior art, the technical scheme of the present application has the beneficial effects that:

[0065] In the scheme, the detection method adopts a non-contact detection method of machine vision, eliminates the subjective conjecture problem of strip steel deviation existing for a long time on site, has high detection precision, uses on-site loop swing angle data to establish a coordinate conversion model, thereby eliminating the influence of the loop in the rolling process, and effectively eliminates the detection error caused by the camera installation position through the detection model.

[0066] The application has the following characteristics: 1. The face array camera is used for detection, which ensures the detection precision while keeping the detection equipment simple; 2. The real-time angle of the strip steel is calculated by using the real-time angle of the on-site loop, which ensures the accuracy and real-time performance of the calculation of the rotation angle of the strip steel; 3. In the case that the strip steel has a rotation angle, a coordinate conversion model of the image coordinate to the world coordinate under the angle transformation of the strip steel is established, which can effectively eliminate the detection error caused by the swing of the loop angle; 4. The detection precision can reach ±2mm, which meets the accuracy requirement of the deviation detection on the rolling site. BRIEF DESCRIPTION OF DRAWINGS

[0067] In order to more clearly illustrate the technical solutions in the embodiments of the application, the drawings needed in the embodiment description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the application, and other drawings can be obtained by those skilled in the art without creative labor.

[0068] Figure 1 The figure shows the principle of the world coordinate system transformation for the hot-rolled strip steel deviation detection method considering the influence of the loop. DETAILED DESCRIPTION

[0069] In order to make the technical problems, technical solutions and advantages of the application more clear, the following will be described in detail in combination with the drawings and specific embodiments.

[0070] The application provides a hot-rolled strip steel deviation detection method considering the influence of the loop.

[0071] The method comprises the following steps:

[0072] S1: vision system calibration:

[0073] The loop is put down at the rolling gap, the camera is calibrated by using Zhang Zhengyou calibration method, and the internal parameters and external parameters of the vision system at this time are obtained;

[0074] S2: collect the strip steel image and extract the strip steel edge:

[0075] When the strip steel enters the detection area, the camera is controlled to continuously collect strip steel images at a collection frequency of 20Hz, the collected strip steel images are subjected to edge detection in a manner of combining a Canny operator and a sub-pixel edge detection operator, and pixel coordinates of left and right edges of the strip steel image at a collection time are obtained;

[0076] S3: Through a communication system, a strip steel thickness h1 at a mill outlet and a real-time rotation angle θ of a loop are obtained from field first-level data;

[0077] S4: Considering the strip steel thickness at the outlet and the real-time rotation angle of the loop, real-world coordinates of a strip steel edge are calculated according to strip steel image coordinates;

[0078] S5: An actual deviation amount of the strip steel is calculated according to the real-world coordinates of the strip steel edge.

[0079] In the specific implementation process, the following steps are included:

[0080] S1: A vision system is calibrated.

[0081] When the loop is put down at a rolling gap, a camera is calibrated by using Zhang Zhengyou calibration method, and an internal parameter matrix M of the vision system at this time is obtained in and an external parameter matrix M ex . The final calibration parameters are as follows:

[0082]

[0083]

[0084] S2: Strip steel images are collected, and a strip steel edge is extracted.

[0085] When the strip steel enters the detection area, the camera is controlled to continuously collect strip steel images, the collected strip steel images are subjected to edge detection in a manner of combining a Canny operator and a sub-pixel edge detection operator, and pixel coordinates of left and right edges of the strip steel image at a collection time are obtained.

[0086] S3: Through a communication system, a strip steel thickness h1 at a mill outlet and a real-time rotation angle θ of a loop are obtained from field first-level data;

[0087] S4: Considering the strip steel thickness at the outlet and the real-time rotation angle of the loop, real-world coordinates of a strip steel edge are calculated according to strip steel image coordinates, and the specific steps are as follows:

[0088] S41: According to a geometric relationship among the strip steel, the loop and two mills before and after, a real-time rotation angle α of the strip steel at a shooting side is calculated through the real-time rotation angle of the loop, and the following can be obtained:

[0089]

[0090] In the formula:

[0091] θ — Real-time rotation angle of the looper, in rad;

[0092] H—The distance between the rotating shaft of the looper connecting rod and the horizontal running plane of the strip, in mm;

[0093] h—the height of the center of the loop from the initial position of the loop after the loop has moved, in mm;

[0094] L—Initial position of the looper: Distance between the center of the looper and the previous rolling mill, in mm;

[0095] l — Horizontal distance between the positions of the loops before and after movement, in mm.

[0096] r1—Loop radius, in mm;

[0097] r2—Loop rotation radius, in mm;

[0098] β—Angle between the loop connecting rod and the horizontal plane at the initial position of the loop, in rad;

[0099] Z—The distance between two adjacent mill roll gaps, in mm;

[0100] m—the distance from the rotating shaft of the looper connecting rod to this rolling mill, in mm;

[0101] Finally, the value of α was obtained through calculation:

[0102]

[0103] S2: Optimize the coordinate system transformation model based on the real-time rotation angle of the strip and the strip thickness parameters at the mill exit.

[0104] The calibration plane, i.e., the plane where the looper rotates at an angle of 0, is denoted as Plane I. The plane where the strip lies after the looper rotates is denoted as Plane II. The rotation of the looper simultaneously causes the strip to rotate about the roll gap as an axis, and the real-time rotation angle of the strip is denoted as α. During the calibration process, the distance from the roll gap to Plane I is measured and denoted as a. Furthermore, for ease of calculation, the y-axis of the world coordinate system of the calibration plane is placed on the rolling center line.

[0105] like Figure 1 As the strip changes from plane I to plane II, the world coordinate system also changes accordingly, from coordinate system OX. w Y w Z w The coordinate system is transformed into O1-X2Y2Z2. Therefore, the final real-world coordinates of the strip edge are:

[0106]

[0107] In the formula:

[0108] u, v - pixel coordinates of the left and right edges of the strip image at the moment of acquisition, unit 1;

[0109] s - camera internal parameter, which is a scale factor, and can be obtained by camera calibration, unit 1;

[0110] R - rotation matrix in the camera external parameter, which can be obtained by camera calibration, unit 1;

[0111] t - translation matrix in the camera external parameter, which can be obtained by camera calibration, unit 1;

[0112] dx, dy - camera internal parameters, representing the length of unit pixel in the x, y axis of the physical coordinate system, which can be obtained by camera calibration, unit mm;

[0113] u0, v0 - camera internal parameters, representing the corresponding point of the origin of the camera physical coordinate system in the camera pixel coordinate system, which can be obtained by camera calibration, unit 1;

[0114] f - camera internal parameter, representing the focal length of the lens. It can be obtained by camera calibration, unit mm.

[0115] The real world coordinates of the strip edge are finally calculated as:

[0116]

[0117] S5: Calculate the actual run-out value of the strip according to the strip edge coordinates, the method is as follows:

[0118] According to the edge detection result, the left and right edge pixel coordinates of the strip are obtained, respectively denoted as P1(u1, v1), P2(u2, v2). Through formula (2), the world coordinate system coordinates of P1 and P2 can be obtained, respectively denoted as P1(X 21 ,Y 21 ,Z 21 ), P2(X 22 ,Y 22 ,Z 22 ).

[0119] Through the coordinate conversion relationship, the following is finally calculated:

[0120] X 21 = -723.70

[0121] X 22 = 756.31

[0122] Since the y axis of the world coordinate system coincides with the rolling center line, the width and run-out value of the strip detection can be obtained:

[0123] W = X 22 - X 21

[0124] = 756.31 - (-723.70)

[0125] = 1480.01 mm (3)

[0126]

[0127] In the formula:

[0128] W - detected strip width, in mm;

[0129] D - detected strip run-out amount, in mm.

[0130] Thus far, the final finishing strip width and strip run-out amount relative to the rolling center line detection is completed.

[0131] The above is the preferred embodiment of the present application, it should be noted that for those of ordinary skill in the art, without departing from the principles of the present application described, can also make a number of improvements and refinements, these improvements and refinements should also be considered within the scope of the present application.

Claims

1. A method for detecting deviation of hot-rolled strip steel considering the influence of looper, characterized in that, The steps include the following: S1: Vision system calibration: During the rolling gap, the looper is lowered, and the camera is calibrated using the Zhang Zhengyou calibration method to obtain the internal and external parameters of the vision system at this time. S2: Acquire images of the strip steel and extract its edges: When the strip enters the detection area, the camera is controlled to continuously acquire strip images at a fixed acquisition frequency. The edge detection of the acquired strip images is performed by combining the Canny operator and the subpixel edge detection operator to obtain the pixel coordinates of the left and right edges of the strip image at the acquisition time. S3: Obtain the strip thickness h1 at the mill exit and the real-time rotation angle θ of the looper from the primary data at the field via the communication system; S4: Considering the thickness of the exported strip and the real-time rotation angle of the looper, calculate the real-world coordinates of the strip edge based on the image coordinates of the strip. S5: Calculate the actual deviation of the strip based on the real-world coordinates of the strip edge; The specific steps in S4 are as follows: S41: Based on the geometric relationship between the strip, the looper, and the two finishing mills, calculate the real-time rotation angle α of the strip on the shooting side using the real-time rotation angle of the looper. In the formula: θ — Real-time rotation angle of the looper, in rad; H—The distance between the rotating shaft of the looper connecting rod and the horizontal running plane of the strip, in mm; h—the height of the center of the loop from the initial position of the loop after the loop has moved, in mm; L—Initial position of the looper: Distance between the center of the looper and the previous rolling mill, in mm; l—Horizontal distance between the positions of the looper before and after movement, in mm; r1—Loop radius, in mm; r2—Loop rotation radius, in mm; β—The angle between the loop connecting rod and the horizontal plane when the loop is initially in the position of the loop, in rad; Z—The distance between the roll gaps of two adjacent finishing mills, in mm; m—the distance from the rotating shaft of the looper connecting rod to this rolling mill, in mm; S42: Optimize the coordinate system transformation model based on the real-time rotation angle of the strip on the shooting side and the strip thickness at the mill exit: The plane for marking is plane I, and the plane where the strip lies after the looper rotates is plane II. The rotation of the looper simultaneously drives the strip to rotate around the roll gap as an axis, and the real-time rotation angle of the strip is α. During the calibration process, the distance from the roll gap to plane I was measured as a. Place the y-axis of the world coordinate system of the calibration plane on the rolling center line; During the process of the strip steel changing from plane I to plane II, the world coordinate system changes from coordinate system OX. w Y w Z w The coordinate system changes to O1-X2Y2Z2; The final real-world coordinates of the strip edge are: In the formula: u, v — the pixel coordinates of the left and right edges of the strip image at the time of acquisition, in units of 1; s—Camera intrinsic parameter, a scale factor obtained through camera calibration, with a unit of 1; R—The rotation matrix in the camera's extrinsic parameters, obtained through camera calibration, with a unit of 1; t—Translation matrix in the camera extrinsic parameters, obtained through camera calibration, with a unit of 1; dx, dy — camera internal parameters, representing the length of a unit pixel on the x and y axes in the physical coordinate system, obtained through camera calibration, and the unit is mm; u0, v0 — Camera internal parameters, representing the corresponding points of the camera's physical coordinate system origin in the camera's pixel coordinate system, obtained through camera calibration, with a unit of 1; f—Camera internal parameter, representing the lens focal length, obtained through camera calibration, in mm; Specifically, S5 refers to: Based on S2, the pixel coordinates P1(u1,v1) and P2(u2,v2) of the left and right edges of the strip are obtained. Then, the world coordinates of points P1 and P2 are obtained through S4, denoted as P1(X...). 21 ,Y 21 Z 21 ), P2(X 22 ,Y 22 Z 22 ); Since the y-axis of the world coordinate system coincides with the rolling centerline, the width and deviation of the strip steel inspection are as follows: W=X 22 -X 21 In the formula: W—The width of the strip being inspected, in mm; D—The amount of strip deviation detected, in mm.

2. The hot-rolled strip misalignment detection method considering the influence of looper according to claim 1, characterized in that, The vision system in S1 includes an area array camera and camera lens, a water cooling device, a camera adjustment device, two photoelectric converters, an image processing server, a KVM extender, and a terminal display. The area array camera is fixed between the two finishing mills by a camera adjustment device, and takes pictures of the strip steel during the rolling process at an angle downward to ensure that the strip steel is placed in the center of the image. The camera adjustment device consists of a pitch position adjustment device and a rotation position adjustment device, which are used to adjust the spatial position of the area array camera. The entire adjustment device is fixed above the mill archway on site by welding. The area array camera is surrounded by a water cooling device with high-pressure cooling water, and the water cooling device is fixed below the camera adjustment device by bolts. One of the two photoelectric converters is located at the camera end and connected to the camera via a network cable, while the other is located at the image processing server end and connected to the image processing server via a network cable. Image information is transmitted to the photoelectric converter at the camera end via gigabit Ethernet, converting electrical signals into optical signals, and then transmitted to the photoelectric converter at the image processing server end via optical fiber. Finally, it is transmitted to the image processing server via gigabit Ethernet. The data information processed by the image processing server is transmitted to the terminal display through a KVM extender to display the detection results and strip information to the user.

3. The hot-rolled strip misalignment detection method considering the influence of looper according to claim 1, characterized in that, The internal parameters in S1 include the scale factor s, the length dx of a unit pixel on the x-axis in the physical coordinate system, the length dy of a unit pixel on the y-axis in the physical coordinate system, the corresponding points u0 and v0 of the origin of the camera physical coordinate system in the camera pixel coordinate system, and the lens focal length f. External parameters include the rotation matrix R and the translation matrix t.

4. The hot-rolled strip misalignment detection method considering the influence of looper according to claim 1, characterized in that, The timing plane in S42 is the plane where the real-time rotation angle of the loop is 0.

Images