Method for manufacturing steel pipe, method for predicting roundness of steel pipe, control method, method for generating prediction model, and prediction device

Abstract

The method for generating a true roundness prediction model of a steel pipe according to the present application is characterized by generating a plurality of sets of data of true roundness information of a steel pipe after an expanding process corresponding to a set of operation condition data as learning data, by performing a numerical calculation including the set of operation condition data as input data and the true roundness information of the steel pipe after the expanding process as output data a plurality of times while changing the set of operation condition data each time, the set of operation condition data including one or more operation parameters selected from operation parameters of a U-punching process and one or more operation parameters selected from operation parameters of an O-punching process, generating a true roundness prediction model including the set of operation condition data as input data and the true roundness information of the steel pipe after the expanding process as output data by machine learning using the plurality of sets of learning data, and generating a true roundness prediction model including the set of operation condition data as input data and the true roundness information of the steel pipe after the expanding process as output data by machine learning using the plurality of sets of learning data.

Description

Technical Field

[0001] This invention relates to a method for generating a roundness prediction model for steel pipes after the pipe expansion process in the manufacturing process of UOE steel pipes, a roundness prediction method, a roundness control method, a manufacturing method, and a roundness prediction device. Background Technology

[0002] The manufacturing technology for large-diameter, thick-walled steel pipes used as pipelines, etc., involves stamping steel plates with specified length, width, and thickness into a U-shape, then stamping them into an O-shape and welding the butt joints to form a steel pipe. Further enlarging the diameter (so-called pipe expansion) improves the roundness of the steel pipe (so-called UOE steel pipe). In recent years, the requirements for UOE steel pipes, which utilize thicker-walled or high-strength materials compared to the past, have increased. Simultaneously, the required precision of the steel pipe's roundness has also increased.

[0003] In response, Patent Document 1 describes a method for manufacturing UOE steel pipes that includes C-shaped stamping (end bending), U-shaped stamping (U-bending), and O-shaped stamping (O-bending). By appropriately selecting the end bending width (C-bending length) in the C-shaped stamping process, the U-bending width in the U-shaped stamping process, and the radius of curvature of the U-bending shoulder of the U-shaped stamping tool, incomplete forming in the O-shaped stamping process can be suppressed. According to the method described in Patent Document 1, it is believed that if the steel pipe is formed under conditions within a specified range regardless of the thickness and material of the raw material, a suitable steel pipe shape can be obtained.

[0004] On the other hand, Patent Document 2 describes a method that improves the roundness of a steel pipe by setting the ratio of the outer diameter of the die used in the pipe-expanding process of the manufacturing process of the UOE steel pipe before expansion to the inner diameter of the manufactured steel pipe within a specified range, thereby reducing the amount of sharp edges in the steel pipe, which is called the peak value. Furthermore, Patent Document 3 describes a method where, for steel pipes of specified strength and dimensions, the width of the U-shaped stamping tool used in the U-shaped stamping process is 70% or less of the outer diameter of the product. According to the method described in Patent Document 3, the contact state between the O-shaped stamping die and the formed body in the O-shaped stamping process is optimized, and the roundness of the open pipe after the O-shaped stamping process is improved.

[0005] Existing technical documents

[0006] Patent documents

[0007] Patent Document 1: Japanese Patent Application Publication No. 55-50916

[0008] Patent Document 2: Japanese Patent Application Publication No. 4-71737

[0009] Patent Document 3: Japanese Patent Application Publication No. 2004-141936 Summary of the Invention

[0010] The problem that the invention aims to solve

[0011] However, the method described in Patent Document 1 is insufficient to meet the current requirements for the roundness of UOE steel pipes and cannot produce UOE steel pipes with good roundness. Furthermore, it is a method that provides appropriate operating conditions for the C-forming and U-forming stamping processes, not a method for predicting the roundness of the steel pipe after the expansion process. On the other hand, in the method described in Patent Document 2, it is difficult to achieve good roundness for thick-walled and high-strength UOE steel pipes from the perspective of the strength of the expansion equipment. In addition, the manufacturing process of UOE steel pipes includes at least several processes besides the expansion process, such as the U-forming stamping process and the O-forming stamping process. However, the method described in Patent Document 2 does not consider the influence of the operating conditions of processes other than the expansion process on the roundness of the steel pipe after the expansion process. Therefore, it may not always be able to improve the roundness of the steel pipe after the expansion process.

[0012] Furthermore, Patent Document 3 describes how adjusting the width of the U-shaped stamping tool to a specified range improves the roundness of the open tube after the O-shaped stamping process, implying that the roundness after the O-shaped stamping process varies depending on the relationship between the U-shaped and O-shaped stamping processes. However, the roundness of UOE steel pipe products is affected by the operating conditions of multiple manufacturing processes, including the pipe expansion process; therefore, there is room for improvement in enhancing the roundness of the steel pipe. Additionally, the method described in Patent Document 3 is not a method for predicting the roundness of the steel pipe after the pipe expansion process.

[0013] This invention addresses the aforementioned issues and aims to provide a method for generating a roundness prediction model for steel pipes. This method can generate a high-precision roundness prediction model that accurately predicts the roundness of steel pipes after the expansion process in the manufacturing process of UOE steel pipes, which includes multiple processes. Furthermore, this invention aims to provide a method and apparatus for predicting the roundness of steel pipes, capable of accurately predicting the roundness of steel pipes after the expansion process in the manufacturing process of UOE steel pipes, which includes multiple processes. Additionally, this invention aims to provide a method for controlling the roundness of steel pipes and a manufacturing method that can produce UOE steel pipes with excellent roundness.

[0014] Methods for solving problems

[0015] The method for generating a steel pipe roundness prediction model according to the present invention predicts the roundness of the steel pipe after the pipe expansion process in a steel pipe manufacturing process, which includes a U-shaped stamping process that forms a U-shaped cross-section of a steel plate using a U-shaped stamping tool, an O-shaped stamping process that forms the U-shaped cross-section of the steel plate into an open pipe, and a pipe expansion process that performs a pipe expansion forming process on the steel pipe formed by joining the width-direction ends of the open pipe together. The method includes: a basic data acquisition step, which involves numerical calculation for a steel pipe after the pipe expansion process, using an operating condition dataset as input data and the roundness information of the steel pipe as output data, while changing the operating condition dataset. The condition dataset is executed multiple times, thereby generating multiple sets of data on the roundness information of the steel pipe after the expansion process corresponding to the operation condition dataset as learning data. The operation condition dataset includes one or more operation parameters selected from the operation parameters of the U-shaped stamping process and one or more operation parameters selected from the operation parameters of the O-shaped stamping process. The roundness prediction model generation step uses the multiple learning data generated in the basic data acquisition step to generate a roundness prediction model through machine learning, which takes the operation condition dataset as input data and the roundness information of the steel pipe after the expansion process as output data.

[0016] Preferably, the basic data acquisition step includes the step of using the finite element method to calculate the roundness information of the steel pipe after the pipe expansion process based on the operating condition dataset.

[0017] Preferably, the roundness prediction model, as part of the operating condition dataset, includes one or more parameters selected from the attribute information of the steel plate.

[0018] Preferably, the roundness prediction model, as part of the operating condition dataset, includes one or more parameters selected from the operating parameters of the tube expansion process.

[0019] Preferably, the manufacturing process of the steel pipe includes a C-shaped stamping process, which is formed by bending the end of the steel plate in the width direction prior to the U-shaped stamping process, and the roundness prediction model, as the operating condition dataset, includes one or more operating parameters selected from the operating parameters of the C-shaped stamping process.

[0020] Preferably, the operating parameters of the U-shaped stamping process include at least one of the following operating parameters: the shape information of the U-shaped stamping tool, the U-shaped stamping reduction amount, the initial interval of the U-shaped stamping support, and the final interval of the U-shaped stamping support.

[0021] Ideally, the machine learning method used is selected from neural networks, decision tree learning, random forests, Gaussian process regression, and support vector regression.

[0022] The method for predicting the roundness of steel pipes according to the present invention includes: an operation parameter acquisition step, which uses the method for generating the roundness prediction model of steel pipes according to the present invention as input to obtain an operation condition dataset set as the operation conditions of the manufacturing process of the steel pipes online; and a roundness prediction step, which inputs the operation condition dataset obtained in the operation parameter acquisition step into the roundness prediction model to predict the roundness information of the steel pipes after the pipe expansion process.

[0023] The roundness control method for steel pipes according to the present invention includes the following steps: using the roundness prediction method for steel pipes according to the present invention, before the start of a resetting target process selected from a plurality of forming processing processes constituting the manufacturing process of the steel pipe, predicting the roundness information of the steel pipe after the pipe expansion process; and based on the predicted roundness information of the steel pipe, resetting at least one or more operating parameters selected from the operating parameters of the resetting target process or from the operating parameters of the forming processing process downstream of the resetting target process.

[0024] The steel pipe manufacturing method of the present invention includes the step of manufacturing the steel pipe using the steel pipe roundness control method of the present invention.

[0025] The steel pipe roundness prediction device of the present invention predicts the roundness of the steel pipe after the pipe expansion process in a steel pipe manufacturing process, including a U-shaped stamping process that processes a steel plate into a U-shaped cross-section using a U-shaped stamping tool, an O-shaped stamping process that processes the U-shaped cross-section into an open pipe, and a pipe expansion process that performs a pipe expansion forming process on the steel pipe formed by joining the width-direction ends of the open pipe together. The device includes: a basic data acquisition unit that performs numerical calculations multiple times while changing the operating condition dataset, including an operating condition dataset as input data and the roundness information of the steel pipe after the pipe expansion process as output data, thereby generating multiple sets of data corresponding to the operating condition dataset for the roundness information of the steel pipe after the pipe expansion process as learning data. The data set includes one or more operating parameters selected from the operating parameters of the U-shaped stamping process and one or more operating parameters selected from the operating parameters of the O-shaped stamping process; a roundness prediction model generation unit, which uses multiple learning data generated in the basic data acquisition unit to generate a roundness prediction model by machine learning, taking the operating condition dataset as input data and the roundness information of the steel pipe after the tube expansion process as output data; an operating parameter acquisition unit, which acquires the operating condition dataset set as the operating conditions of the steel pipe manufacturing process in an online manner; and a roundness prediction unit, which uses the roundness prediction model generated in the roundness prediction model generation unit to predict the roundness information of the steel pipe after the tube expansion process corresponding to the operating condition dataset acquired by the operating parameter acquisition unit in an online manner.

[0026] Preferably, a terminal device is provided, which has an input unit for acquiring input information based on user operation and a display unit for displaying the roundness information. The operation parameter acquisition unit updates part or all of the operation condition dataset in the manufacturing process of the steel pipe based on the input information acquired by the input unit, and the display unit displays the roundness information of the steel pipe predicted by the roundness prediction unit using the updated operation condition dataset.

[0027] Invention Effects

[0028] According to the method for generating a roundness prediction model for steel pipes disclosed in this invention, a roundness prediction model can be generated with high accuracy to predict the roundness of steel pipes after the pipe expansion process in the manufacturing process of UOE steel pipes, which includes multiple processes. Furthermore, according to the roundness prediction method and apparatus for steel pipes disclosed in this invention, the roundness of steel pipes after the pipe expansion process in the manufacturing process of UOE steel pipes, which includes multiple processes, can be predicted with high accuracy. Additionally, according to the roundness control method and manufacturing method for steel pipes disclosed in this invention, UOE steel pipes with good roundness can be manufactured. Attached Figure Description

[0029] Figure 1 This is a diagram illustrating the manufacturing process of a steel pipe as an embodiment of the present invention.

[0030] Figure 2 This is a perspective view showing the overall structure of the C-shaped stamping device.

[0031] Figure 3 It is shown Figure 2 The diagram shows the structure of the stamping mechanism.

[0032] Figure 4 This is a schematic diagram showing the overall structure of the U-shaped stamping device.

[0033] Figure 5 This is a schematic diagram used to illustrate the operation of the U-shaped stamping device.

[0034] Figure 6 This is a schematic diagram showing a modified example of a U-shaped stamping device.

[0035] Figure 7 This is a schematic diagram used to illustrate the O-ring stamping process.

[0036] Figure 8 This is a schematic diagram showing the structure of the tube expansion device.

[0037] Figure 9 This is a diagram illustrating the method for determining the outer diameter shape of a steel pipe.

[0038] Figure 10 This is a block diagram showing the structure of a device for generating a roundness prediction model for a steel pipe, as an embodiment of the present invention.

[0039] Figure 11 It is shown Figure 10 The diagram shows the structure of the offline roundness calculation unit.

[0040] Figure 12 This is a diagram illustrating an example of the relationship between pipe thickness and expansion capacity under various yield stresses of a steel pipe.

[0041] Figure 13 This is a diagram illustrating an example of the relationship between pipe thickness and expansion capacity for various outer diameters of a steel pipe.

[0042] Figure 14 This is a diagram showing an example of measuring the cross-sectional shape of a steel pipe before the expansion process.

[0043] Figure 15 It is shown Figure 14 The graph shows the relationship between the expansion rate and the peak value at the three protrusions.

[0044] Figure 16This is a diagram used to illustrate the definition of a peak value.

[0045] Figure 17 This is a diagram used to illustrate the shape of a U-shaped stamping tool.

[0046] Figure 18 This is a diagram illustrating a method for controlling the roundness of a steel pipe as an embodiment of the present invention.

[0047] Figure 19 This is a diagram showing the structure of a steel pipe roundness prediction device as an embodiment of the present invention.

[0048] Figure 20 This is a diagram illustrating an example of a finite element model in the O-ring stamping process. Detailed Implementation

[0049] Hereinafter, with reference to the accompanying drawings, a method for generating a roundness prediction model for a steel pipe, a method for predicting roundness, a method for controlling roundness, a manufacturing method, and a device for predicting roundness, as embodiments of the present invention, will be described.

[0050] [Steel pipe manufacturing process]

[0051] First, refer to Figure 1 The manufacturing process of the steel pipe, which is one embodiment of the present invention, will be described.

[0052] Figure 1 This diagram illustrates the manufacturing process of a steel pipe according to one embodiment of the present invention. In the manufacturing process of the steel pipe according to one embodiment of the present invention, the steel plate used as the raw material is a thick steel plate manufactured by a thick plate rolling process, which is a preceding process in the manufacturing process of the steel pipe. Representative thick steel plates have a yield stress of 245–1050 MPa, a tensile strength of 415–1145 MPa, a thickness of 6.4–50.8 mm, a width of 1200–4500 mm, and a length of 10–18 m.

[0053] like Figure 1 As shown in (a), the steel plate used as raw material is first machined in a pre-processing step. Specifically, in order to adjust the width of the steel plate to a specified range, the width-direction ends of the steel plate are cut or diced by a plate width processing device. As a result, the outer diameter of the steel pipe after being formed into a UOE steel pipe is within the range required as a product. In addition, the width-direction ends of the steel plate are pre-cut or melted into a chamfered shape called a bevel by a beveling processing device. This is so that in the subsequent welding process, molten metal can easily flow in the thickness direction of the weld, ensuring the strength of the weld.

[0054] Next, as Figure 1As shown in (b), as part of the C-forming stamping process, the ends of the steel sheet using a C-forming stamping device are sometimes bent (also known as edge rolling). By pre-applying bending deformation to the ends of the steel sheet in the width direction using a die, the roundness of the steel pipe after the expansion process can be improved. Moreover, as... Figure 1 As shown in (c), as a U-shaped stamping process, a steel sheet using a U-shaped stamping device is formed into a U-shaped shaped body (a shaped body with a U-shaped cross-section). Next, as... Figure 1 As shown in (d), the U-shaped forming body has its seam gap reduced by using the O-shaped stamping process of the O-shaped stamping device, and is formed into an open tube with a roughly tubular cross section.

[0055] in addition, Figure 1 The subsequent welding process shown in (e) involves binding the seam gap formed at the width-direction ends of the open pipe in such a way that the width-direction ends come into contact with each other, and joining the width-direction ends together using a welding device. Thus, the open pipe becomes a steel pipe with its width-direction ends joined together. Figure 1 The subsequent pipe-expanding process shown in (f) involves using a pipe-expanding device equipped with multiple pipe-expanding tools that divide the arc into multiple curved surfaces. The process expands the steel pipe by bringing the curved surfaces of the pipe-expanding tools against the inner surface of the steel pipe. Regarding steel pipes manufactured in this way, Figure 1 In the inspection process shown in (g), it is determined whether the material, appearance, dimensions, and other qualities meet the specified specifications, and then the product is shipped as a UOE steel pipe. It should be noted that in this embodiment, the inspection process includes a roundness measurement process that uses a roundness measuring machine to measure the roundness of the steel pipe.

[0056] In this embodiment, in the series of manufacturing processes involving forming a steel sheet into an open tube and then expanding it after welding, the C-shaped stamping process, the U-shaped stamping process, the O-shaped stamping process, and the tube expansion process are referred to as "forming processes." These processes are common to each other as processes that impart plastic deformation to the steel sheet to control the size / shape of the steel tube. Hereinafter, refer to... Figures 2-9 The manufacturing process of steel pipes is explained in detail for each step.

[0057] (C-shaped stamping process)

[0058] First, refer to Figure 2 , Figure 3 This section explains the case of bending the ends of steel plates using a C-shaped stamping device as part of a C-shaped stamping process.

[0059] Figure 2 This is a perspective view showing the overall structure of the C-shaped stamping device. (For example...) Figure 2As shown, the C-shaped stamping device 10 includes a conveying mechanism 11 that conveys a steel plate S in such a way that its length direction is the conveying direction; a stamping mechanism 12A that bends one of the width-direction ends Sc of the steel plate S to a predetermined curvature with the downstream side of the conveying direction as the front; a stamping mechanism 12B that bends the other width-direction end Sd to a predetermined curvature; and an interval adjustment mechanism (not shown) that adjusts the interval between the left and right stamping mechanisms 12A and 12B according to the width of the steel plate S in which the end bending is performed. The conveying mechanism 11 includes a plurality of rotary-driven conveying rollers 11a arranged in front of and behind the stamping mechanisms 12A and 12B, respectively. It should be noted that the reference numeral Sa in the figure indicates the top end (front end in the length direction) of the steel plate S.

[0060] exist Figure 3 (a) shows a cross-section in the width direction of a stamping mechanism 12A that bends one end Sc of the steel plate S when viewed from the upstream side of the transport direction of the steel plate S to the downstream side of the transport direction. It should be noted that stamping mechanisms 12A and 12B are symmetrical and have the same structure. Stamping mechanisms 12A and 12B include an upper die 13 and a lower die 14, which are a pair of dies arranged opposite each other in the vertical direction, and a hydraulic cylinder 16, which is a die-moving unit that lifts the lower die 14 together with a tool holder 15 (moving it towards the upper die 13) and closes the die with a predetermined stamping force. It should be noted that stamping mechanisms 12A and 12B sometimes include a clamping mechanism 17 that clamps the steel plate S in a release manner inside the width direction of the upper die 13 and the lower die 14. The length of the steel plate S in the longitudinal direction of the upper die 13 and the lower die 14 is usually shorter than the length of the steel plate S. In this case, while utilizing the transport mechanism 11 (see reference...) Figure 2 The steel plate S is intermittently conveyed along its length while undergoing multiple end bending processes.

[0061] In the C-shaped stamping process, the lower die 14, which contacts the outer surfaces of the width-direction ends Sc and Sd of the steel sheet S to be bent, has a pressing surface 14a facing the upper die 13. The upper die 13 has a convex curved forming surface 13a facing the pressing surface 14a and having a radius of curvature corresponding to the inner diameter of the manufactured steel tube. The pressing surface 14a has a concave curved surface that approaches the upper die 13 as it moves outward in the width direction. However, although the pressing surface 14a of the lower die 14 is concave, it can be an inclined plane as long as it approaches the upper die 13 as it moves outward in the width direction. As for the curved surface shapes of the upper die 13 and the lower die 14, sometimes a suitable shape is designed according to the thickness of the steel sheet S, the outer diameter of the steel tube, etc., and the appropriate shape is selected according to the material being processed.

[0062] Figure 3(b) is with Figure 3 (a) A cross-section of the stamping mechanism 12A at the same location, but showing the state after the lower die 14 is lifted by the hydraulic cylinder 16 and the die is closed. The lower die 14 is lifted by the hydraulic cylinder 16, and the width end Sc of the steel plate S is bent into a shape along the arc forming surface 13a of the upper die 13. The width of the end bending forming (end bending processing width) varies depending on the width of the steel plate S, but is generally about 100 to 400 mm.

[0063] (U-shaped stamping process)

[0064] Next, refer to Figures 4-6 The following is an explanation of the U-shaped stamping process.

[0065] exist Figure 4 The diagram shows the overall structure of a U-shaped stamping apparatus for performing the U-shaped stamping process. Various structures exist for U-shaped stamping apparatuses, but a typical example is shown. Figure 4 In the U-shaped stamping apparatus shown, a lifting cylinder 21 is installed on the upper part of the frame 20 such that the upper rod is positioned downwards. The upper end of a suspension member 23 supporting the U-shaped stamping tool (U-shaped punch) 22 is mounted on the upper rod of the lifting cylinder 21. Additionally, a sliding cylinder 26 is provided at the center of the lower floor surface 24 of the frame 20 such that the rod 25 extends into the frame 20. A pair of sliding blocks 27 are provided on its side. A base (seat) 28 is mounted on the head of the rod 25 of the sliding cylinder 26. Furthermore, the rod 25 and the sliding blocks 27 are connected by a connecting rod 29. This connecting rod 29 has a fixed rotation center 30 at the sliding block 27, and a brake roller (U-shaped bending support) 32 for bending the steel plate S is mounted at the top of an arm 31 extending from this point. The steel plate S, which is the raw material for the U-shaped stamping process, is a steel plate whose ends have been bent in the aforementioned C-shaped stamping process. However, steel sheets that have not undergone the C-shaped stamping process and have not been end-bending can also be used. Steel sheet S to make... Figure 4 The U-shaped stamping device shown is positioned on the left and right brake rollers 32 in a manner that aligns with the width direction of the steel plate S. At this time, the steel plate S is approximately symmetrically positioned with respect to the centers of the left and right brake rollers 32. Subsequently, if the U-shaped stamping tool 22 is lowered using the lifting cylinder 21, a three-point bending deformation is applied between the U-shaped stamping tool 22 and the left and right brake rollers 32, resulting in a downwardly convex bending deformation of the steel plate S.

[0066] Figure 5This illustrates the state where the U-shaped stamping tool 22 is lowered to a preset maximum lowering position using the lifting cylinder 21. As the U-shaped stamping tool 22 is gradually lowered using the lifting cylinder 21, the steel plate S abuts against the base 28, and the rod 25 gradually descends via the base 28. At this time, the connecting rod 29 closes, the sliding block 27 moves towards the center of the machine frame 20, and the arm 31 stands up. The left and right brake rollers 32 move in a direction that closes their intervals. Thus, the brake rollers 32 approach from the side of the steel plate S, which is being formed into a U-shape, and the steel plate S is processed into a U-shaped formed body.

[0067] During the U-shaped stamping process, to ensure the U-shaped formed body achieves the desired shape, a suitable shape is selected as the top shape of the U-shaped stamping tool 22 (the shape of the area abutting the steel plate S) based on the thickness of the steel plate S, the steel grade, and the target outer diameter of the steel pipe. Furthermore, the greater the U-shaped stamping reduction amount of the U-shaped stamping tool 22 (the amount pressed in from the position where the U-shaped stamping tool 22 abuts the upper surface of the steel plate S to the lowest point), the larger the winding angle of the steel plate S relative to the U-shaped stamping tool 22, and the closer the shape of the area of ​​the U-shaped formed body in contact with the U-shaped stamping tool 22 is to the top shape of the U-shaped stamping tool 22. On the other hand, by changing the position of the sliding block 27 before the start of the U-shaped stamping process, the interval between the left and right brake rollers 32 (the initial interval of the U-shaped stamping support) can be set. This changes the distance between the support points during the three-point bending process of the steel plate S, thus altering the curvature imparted to the steel plate S. Furthermore, by adjusting the height of the support 28 or the length of the rod 25, the height at which the steel plate S abuts against the support 28 changes during processing by the U-shaped stamping tool 22. Consequently, the opening and closing position of the connecting rod 29 is altered, and the position of the brake roller 32 during forming and the interval between the left and right brake rollers 32 (final interval of the U-shaped stamping support) when the U-shaped stamping tool 22 reaches its lowest position change. As a result, the opening amount of the U-shaped formed body changes. Therefore, when performing the U-shaped stamping process, these operating parameters should be appropriately set according to the thickness of the steel plate S, the steel grade, and the target outer diameter of the steel pipe.

[0068] Figure 4 , Figure 5 The U-shaped stamping device shown is called a Kaiser-type U-shaped stamping device because the brake roller 32 moves in a direction that closes the mutual gap via the connecting rod 29. On the other hand, in the U-shaped stamping device, there are also... Figure 6 The device shown is called a Burson-type U-shaped stamping device. Among them, Figure 4 , Figure 5 The functions of the bracket 28 and brake roller 32 in the Caesar-type U-shaped stamping device shown are respectively determined by... Figure 6The support portion 33 and the U-shaped bending support portion 34 are supported by a component that integrates the two into one unit, namely the lower die (rocker arm die) 35. In the Bosen-type U-shaped stamping device, the left and right rocker arm dies 35 rotate around the pivot point 36. Furthermore, if the U-shaped stamping tool 37 presses down on the steel plate S, the support portion 33 is pressed down, thereby causing the left and right rocker arm dies 35 to rotate, and the left and right U-shaped bending support portions 34 to move in a mutually closing interval. As a result, the steel plate S is formed into a U-shaped cross section. It should be noted that... Figure 6 Reference numeral 38 in the figure refers to a component called a buffer, which is used to suppress the sudden drop of the steel plate S during forming or to allow the U-shaped formed body to rise after forming.

[0069] After using Figure 6 In the U-shaped stamping process of the Bosen-type U-shaped stamping apparatus shown, the top shape of the U-shaped stamping tool 37 is selected and the U-shaped stamping reduction amount is set according to the thickness of the steel plate S, the steel grade, and the target outer diameter of the steel pipe. On the other hand, the initial interval of the U-shaped stamping support can be set by changing the setting position of the left and right rocker arm dies 35. Furthermore, the final interval of the U-shaped stamping support can be changed by setting the initial height of the support portion 33 (or the initial angle of the rocker arm die 35). Therefore, in the U-shaped stamping process, regardless of which U-shaped stamping apparatus is used, the operating conditions of the U-shaped stamping process can be determined using the same operating parameters. It should be noted that... Figure 6 The U-shaped stamping device shown is a symmetrical structure. The figure on the left side of the symmetry plane F shows the state in which the steel plate S is placed in the U-shaped stamping device, and the figure on the right side of the symmetry plane F shows the state in which the U-shaped stamping tool 37 descends to the preset lowest position.

[0070] (O-ring stamping process)

[0071] Next, refer to Figure 7 The O-ring stamping process will be explained.

[0072] The O-forming process is performed by assembling a semi-circular die into an O-forming device. Figure 7 The diagram schematically illustrates the gradual deformation of a U-shaped formed body during an O-forming process. In the O-forming process, the U-shaped formed body is first loaded onto the lower die 40 of the O-forming device. Then, if the upper die 41, with its lower side open in a semi-circular shape, is lowered, it becomes... Figure 7 The initial state before the O-shaped stamping process is shown in (a). Then, if the upper die 41 is gradually lowered using a die lifting device (not shown), then as... Figure 7 As shown in (b), the U-shaped formed body becomes a cylindrical shape with a roughly circular cross-section along the upper and lower dies. Then, if the upper die 41 is raised to release the constraint of the die on the U-shaped formed body, the U-shaped formed body springs back from... Figure 7As shown in (b), the roughly circular shape elastically recovers to a slightly U-shaped form. As a result, the U-shaped formed body after the O-forming stamping process becomes as follows: Figure 7 The shape shown in (c) is a slightly elongated oval. This state is called an open tube.

[0073] The operating parameters set during the O-forming stamping process can be used. Figure 7 The distance between the uppermost part of the inner surface of the upper die 41 at its lowest point, as shown in (b), and the lowermost part of the inner surface of the lower die 40 at this point (called the O-ring reduction) is used to determine this distance. Alternatively, it can be determined by... Figure 7 The gap between the lowest part of the upper die 41 and the uppermost part of the lower die 40 when the upper die 41 is at its lowest point, as shown in (b), is used to determine the O-shaped stamping pressing position. On the other hand, when the width of the steel plate S, which will become the raw material before the U-shaped stamping process, is set to W, and the circumferential length of a portion equivalent to half the thickness of the steel plate S when the upper die 41 is locked (the portion obtained by adding the part where the die contacts the steel plate S and the gap between the upper die 41 and the lower die 40) is set to L, the compression ratio in the O-shaped stamping process is defined by (WL) / W (×100%). Alternatively, the radius of curvature of the curved surfaces of the upper die 41 and the lower die 40 that contact the U-shaped forming body, i.e., the radius of the O-shaped stamping die, can also be used as the operating parameter of the O-shaped stamping process. It should be noted that when placing the U-shaped molded body on the lower die 40, it is usually placed symmetrically from left to right so that the lowest point of the U-shaped cross section is consistent with the lowest part of the inner surface of the lower die 40. However, there are also cases where the position of the U-shaped molded body may be misaligned due to the shape of the U-shaped molded body and the change in the shape of the die caused by the wear of the lower die 40.

[0074] (Welding process)

[0075] Next, the welding process will be explained.

[0076] For open tubes formed by O-forming, the end faces of the open portions are then joined together and welded using a welding machine (joining unit) to form a steel pipe. The welding machine (joining unit) is typically composed of three types of welding mechanisms: a positioning welding machine, an inner surface welding machine, and an outer surface welding machine. In these machines, the positioning welding machine uses rollers to continuously and tightly press the joined end faces together in a suitable positional relationship, performing positioning welding along the entire length of the tube axis. The steel pipe after positioning welding is then welded from the inner surface of the joined portion by the inner surface welding machine (submerged arc welding), and from the outer surface welding machine by the outer surface welding machine (submerged arc welding).

[0077] (Pipe expansion process)

[0078] Next, refer to Figure 8 The process of expanding the pipe is explained below.

[0079] Regarding the steel pipe after the joint gap is welded, a pipe expander is inserted into the inside of the steel pipe to increase the diameter of the steel pipe (so-called pipe expansion). Figure 8 Figures (a) to (c) are structural examples of the tube expander device. Figure 8 As shown in (a), the pipe expanding device has multiple expanding dies 51 along the circumferential direction of the outer peripheral surface 52 of the cone, each having a curved surface obtained by dividing the arc into multiple parts. When expanding a steel pipe using the pipe expanding device, as... Figure 8 As shown in (b) and (c), firstly, the expansion die 51 is aligned with the expansion start position by moving the steel pipe P using the steel pipe moving device, and the first expansion process is performed by retracting the pull rod 53 from the expansion start position.

[0080] As a result, the expanding die 51, which slides in contact with the outer circumferential surface 52 of the cone through wedge action, is displaced radially, and the steel pipe P is expanded. Furthermore, the unevenness of the cross-sectional shape of the steel pipe P decreases, and the cross-sectional shape of the steel pipe P approaches a true circle. Next, the pull rod 53 is advanced to the expanding start position, and the expanding die 51 is reset to its axially perpendicular inward position using the release mechanism. The steel pipe P is then moved further by an amount corresponding to the spacing (axial length) of the expanding die 51. Then, the expanding die 51 is aligned with the new expanding position, and the aforementioned operation is repeated. Thus, the expanding process can be performed along the entire length of the steel pipe P each time, with the spacing of the expanding dies 51.

[0081] At this point, operational parameters that determine the operating conditions of the pipe expansion process include the expansion ratio, the number of expansion die pieces, and the expansion die radius. The expansion ratio refers to the ratio of the difference between the outer diameter of the expanded steel pipe P and the outer diameter of the steel pipe P before expansion, relative to the outer diameter of the steel pipe P before expansion. The outer diameter of the steel pipe P before and after expansion can be calculated by measuring the circumference of the steel pipe P. The expansion ratio can be adjusted by the stroke of the expansion die 51 when expanding in the radial direction. The number of expansion die pieces refers to the number of expansion die pieces that abut against the circumferentially arranged steel pipe P during pipe expansion. The expansion die radius refers to the circumferential radius of curvature of the portion of each expansion die that abuts against the steel pipe P.

[0082] Among the parameters that allow for easy adjustment of the roundness after the tube-expanding process when the yield stress, thickness, and other properties of the steel plate used as raw material change, the tube-expanding ratio is the operational parameter that facilitates this adjustment. Increasing the tube-expanding ratio ensures that the curvature of the area in contact with the tube-expanding die is imparted evenly across the entire circumference, based on the die radius, thereby improving roundness. Furthermore, a greater number of tube-expanding die pieces effectively suppresses localized circumferential curvature variations in the steel pipe, resulting in better roundness after the tube-expanding process.

[0083] However, to ensure the diameter of the steel pipe produced is within the specified dimensional tolerance, the upper limit of the expansion ratio is constrained. Furthermore, if the expansion ratio is too large, the gap between the circumferential expansion dies increases during expansion, sometimes compromising the roundness of the steel pipe. Moreover, for sections softened by heat during welding, localized deformation concentrations occur, leading to greater thinning in these areas and sometimes causing the pipe thickness to deviate from the specified tolerance range. Additionally, the compressive yield strength of the steel pipe product sometimes decreases due to the Bauschinger effect. In cases of use under high compressive stress in the circumferential direction (e.g., deep-sea pipelines), an upper limit for the expansion ratio needs to be considered based on material constraints. Therefore, in practical operation, the expansion ratio is set such that the roundness of the steel pipe remains within the specified value at an expansion ratio lower than the pre-set upper limit.

[0084] (Inspection process)

[0085] Finally, refer to Figure 9 The inspection process will be explained.

[0086] In the final inspection step of the steel pipe manufacturing process, a quality inspection is performed on the steel pipe, and its roundness is measured. The roundness measured in the roundness measurement step is an index of the degree of deviation of the outer diameter shape of the steel pipe from a perfectly circular shape. Generally, the closer the roundness is to zero, the closer the cross-sectional shape of the steel pipe is to a perfect circle. Roundness is calculated based on the outer diameter information of the steel pipe measured by a roundness measuring machine. For example, if the steel pipe is divided into equal parts circumferentially at any position along its length, and the outer diameters at opposite positions are selected, with the maximum and minimum diameters set as Dmax and Dmin respectively, the roundness can be defined as Dmax-Dmin. In this case, the more equal parts there are, the more accurately it reflects the small irregularities in the steel pipe after the expansion process, which is preferable. Specifically, it is best to use information of 4 to 36,000 equal parts. More preferably, it is 360 or more equal parts. However, the roundness of a steel pipe can also be determined not by its outer diameter, but by its inner diameter shape, using the difference between the inner diameter shape and the inner diameter of the pipe to define roundness.

[0087] Furthermore, the position along the length of the steel pipe for measuring roundness can be arbitrarily chosen. Roundness can be measured near the ends of the steel pipe along its length, or it can be measured at the center of the steel pipe along its length. Additionally, even if multiple roundness measurement positions are selected along the length of the steel pipe and the roundness at each position is measured, the average of the roundness measured at multiple positions along the length can be calculated. However, roundness does not necessarily have to be based on the difference between the maximum and minimum diameters. An equivalent hypothetical circle (diameter) with the same area as the inner side of the curve can be calculated from a graph representing the outer diameter shape of the steel pipe using a continuous line graph. This hypothetical circle is then used as a reference to define the area deviating from the outer diameter shape of the steel pipe as the content of the image. This is because image information can be used as output in the machine learning described later. For example, the following method can be used as a means of measuring the outer diameter shape of the steel pipe.

[0088] (a) such as Figure 9 As shown in (a), a device is used that has an arm 60 capable of rotating 360 degrees about the approximate central axis of the steel pipe P, displacement gauges 61a and 61b mounted on the top of the arm 60, and a rotation angle detector 62 that detects the rotation angle of the rotation axis of the arm 60. The distance between the rotation center of the arm 60 and the measuring point on the outer periphery of the steel pipe P is measured using displacement gauges 61a and 61b for each minute angular unit of rotation of the arm 60, and the outer diameter shape of the steel pipe P is determined based on the measured value.

[0089] (b) such as Figure 9 As shown in (b), a device is used that includes a rotating arm 63 that rotates about the central axis of the steel pipe P, a frame (not shown) disposed at the end of the rotating arm 63 so as to be movable in the radial direction of the steel pipe P, a pair of pressing rollers 64a and 64b that abut against the outer and inner surfaces of the end of the steel pipe P and rotate with the rotation of the rotating arm 63, and a pair of pressing cylinders (not shown) fixed relative to the frame to press the pressing rollers 64a and 64b against the outer and inner surfaces of the steel pipe P. The outer diameter shape of the steel pipe is determined based on the amount of radial movement of the frame and the pressing position of each pressing cylinder against the pressing rollers 64a and 64b.

[0090] In this embodiment, the prediction accuracy of the roundness prediction model described later can be verified by comparing it with the measured roundness value obtained in the aforementioned inspection process. Therefore, the prediction accuracy of the roundness prediction model can be further improved by adding the actual value of its prediction error to the prediction result of the roundness prediction model, relative to the prediction result described later.

[0091] [A device for generating a prediction model for the roundness of steel pipes]

[0092] Next, refer to Figures 10-17 The apparatus for generating a roundness prediction model for steel pipes, which is one embodiment of the present invention, will be described.

[0093] Figure 10 This is a block diagram showing the structure of a device for generating a roundness prediction model for a steel pipe, as an embodiment of the present invention. Figure 11 It is shown Figure 10 A block diagram showing the structure of the offline roundness calculation unit 112. (See diagram below.) Figure 10 As shown, the steel pipe roundness prediction model generation device 100, which is an embodiment of the present invention, is composed of an information processing device such as a workstation, and includes a basic data acquisition unit 110, a database 120 and a roundness prediction model generation unit 130.

[0094] The basic data acquisition unit 110 includes an operation condition dataset 111 that quantifies the factors affecting the roundness of the steel pipe after the U-shaped stamping process, O-shaped stamping process, welding process and pipe expansion process, and an offline roundness calculation unit 112 that outputs the roundness information after the pipe expansion process using the operation condition dataset 111 as input conditions.

[0095] In this embodiment, the operating condition dataset 111 includes at least the operating parameters for the U-shaped stamping process and the O-shaped stamping process. This is because these are factors that significantly affect the roundness of the steel pipe after the tube expansion process and influence the deviation in roundness. However, in addition to these, it may also include the property information of the steel plate used as raw material, the operating parameters for the C-shaped stamping process, the operating parameters for the welding process, and the operating parameters for the tube expansion process. The data used in the operating condition dataset 111 will be described later.

[0096] The basic data acquisition unit 110 calculates the roundness information of the steel pipe after the pipe expansion process corresponding to multiple operating condition datasets 111 by changing various parameters contained in the operating condition dataset 111 and performing numerical calculations by the roundness offline calculation unit 112. The range of parameters to be changed in the operating condition dataset 111 can be determined based on the size of the manufactured steel pipe, the specifications of the equipment for each process, etc., based on the range that can be changed as normal operating conditions.

[0097] The roundness offline calculation unit 112 calculates the shape of the steel pipe after the expansion process through numerical analysis corresponding to a series of manufacturing processes up to the pipe expansion process, and determines the roundness information of the steel pipe based on the shape of the steel pipe after the expansion process. Here, the series of manufacturing processes includes a U-shaped stamping process, an O-shaped stamping process, and a pipe expansion process. For example... Figure 11As shown, the roundness offline calculation unit 112 includes finite element model generation units 112a to 112c corresponding to each process and a finite element analysis solver 112d. It should be noted that the roundness offline calculation unit 112 may also include a finite element model generation unit corresponding to the C-shaped stamping process.

[0098] As the finite element analysis solver 112d, many commercially available general-purpose analysis software programs exist, and can be utilized by appropriately selecting and introducing them. This is because if a finite element model corresponding to each process is generated, numerical analysis can be performed using a single finite element analysis solver. Alternatively, the finite element analysis solver 112d can be mounted on a computer independent of the roundness offline calculation unit 112, and input data including the finite element model and output data as calculation results can be exchanged between the computer and the computer equipped with the finite element analysis solver 112d. That is, it can also be configured such that the finite element model generation units 112a to 112c of the roundness offline calculation unit 112 are mounted on a client computer, and the finite element analysis solver 112d is mounted on a server computer, and input data including the finite element model and calculation results related to the shape of the steel pipe after the pipe expansion process are exchanged.

[0099] The finite element method (FEM) is an approximate solution method that divides a continuum into a finite number of elements. Although it is an approximate method, the FEM finds solutions that satisfy the equilibrium of forces and the continuity of displacements at the nodes of the elements, and can obtain highly accurate solutions even under non-uniform deformation conditions. In the FEM, stress, strain, and displacement within each element are defined independently and correlated with the displacements (velocities) at the nodes, thus formulating a problem for solving simultaneous equations. Therefore, the method of treating the displacements (velocities) at the nodes of the elements as unknowns and evaluating strain (increment) and stress accordingly is widely used.

[0100] Furthermore, the finite element method (FEM) is characterized by its calculation based on the principle of virtual work expressed in integral form, relative to the stress equilibrium conditions within elements. The accuracy of the analysis results varies depending on conditions such as element segmentation, and computation time is also required. However, the FEM is also characterized by its ability to provide solutions to problems that are difficult to solve using other methods, as it serves as a solution to fundamental plasticity equations within nodes or elements. Therefore, even for the complex processing history in steel pipe manufacturing, solutions to the displacement, stress, and strain fields of the processed material that closely approximate actual phenomena can be obtained.

[0101] It should be noted that a portion of the finite element analysis solver 112d can also be replaced with various numerical analysis methods or approximate solution methods such as the slip line field method and the energy method. This can shorten the overall calculation time. Furthermore, the finite element analysis used in this embodiment performs elastoplastic analysis and does not include temperature field analysis such as heat conduction analysis. However, in cases where the processing speed is high and the temperature rise of the steel plate is large due to processing heat, an analysis combining heat conduction analysis and elastoplastic analysis can be performed. Additionally, the elastoplastic analysis in this embodiment is a two-dimensional cross-sectional analysis for the U-shaped stamping process, the O-shaped stamping process, and the tube expansion process; numerical analysis of the cross-section of the stable portion along the length direction when the steel plate is formed into a U-shaped cross-section, an open tube, or a steel pipe is sufficient. However, in cases where the shape of unstable portions such as the top and bottom ends of the steel pipe needs to be predicted with high accuracy, a finite element model generation unit capable of performing three-dimensional analysis including the top and bottom ends is sufficient.

[0102] Regarding the steel sheet used in the U-shaped stamping process, its property information is provided as input data. In the case where the preceding process to the U-shaped stamping process includes a C-shaped stamping process, the shape and stress / strain distribution of the steel sheet obtained from the finite element analysis of the C-shaped stamping process become the initial conditions for the workpiece in the U-shaped stamping process. Here, the finite element model generation unit 112a of the U-shaped stamping process performs element segmentation within the steel sheet based on the dimensions and shape of the steel sheet before the U-shaped stamping process. Element segmentation is automatic based on pre-set element segmentation conditions. Alternatively, the distribution of residual internal stress and strain can be allocated to each element based on the manufacturing history assigned to the steel sheet in the preceding process. This is because, in the U-shaped stamping process, where bending is the main process, the initial residual stress sometimes affects the shape of the U-shaped formed body of the processed steel sheet.

[0103] Along with the finite element model of the U-shaped stamping process generated in this way, the calculation conditions of the U-shaped stamping process are sent as input data to the finite element analysis solver 112d. At this time, the calculation conditions of the U-shaped stamping process include the operating parameters of the U-shaped stamping process, as well as all the information required to perform the finite element analysis, including the physical property values ​​of the steel plate, tools, etc., geometric boundary conditions, mechanical boundary conditions, etc.

[0104] The finite element analysis solver 112d performs numerical analysis under the given calculation conditions to determine the shape of the U-shaped formed body after the U-shaped stamping process and the distribution of residual stress and strain within it. The results of this calculation are used as input data for the subsequent O-shaped stamping process in the roundness offline calculation unit 112. In the finite element model generation unit 112b for the O-shaped stamping process, element segmentation of the interior of the U-shaped formed body is performed based on the calculated shape after the U-shaped stamping process. Element segmentation is performed automatically based on pre-set element segmentation conditions. It is preferable to allocate the stress and strain distribution calculated in the previous process to each element.

[0105] Along with the finite element model of the O-forming stamping process generated in this way, the calculation conditions of the O-forming stamping process are sent as input data to the finite element analysis solver 112d. At this time, the calculation conditions of the O-forming stamping process include the operating parameters of the O-forming stamping process, as well as all the information required to perform the finite element analysis, including the physical property values ​​of the steel plate, tools, etc., geometric boundary conditions, mechanical boundary conditions, and all other boundary conditions.

[0106] The finite element analysis solver 112d performs numerical analysis under the given calculation conditions to determine the shape of the open tube after the O-ring stamping process and the distribution of residual stress and strain inside. The results of this calculation are used as input data in the finite element model generation unit 112c for the subsequent tube expansion process. Furthermore, the welding process, which involves welding the joint gap of the open tube, can also be analyzed numerically to determine the residual stress and strain generated in the steel tube after welding.

[0107] However, due to factors such as the heat conduction behavior associated with the melting of the steel plate during welding and the influence of the heat-affected zone on mechanical properties, rigorous numerical analysis of the welding process is often difficult. Furthermore, since heat input is reduced to suppress material changes during actual welding, and the heat-affected zone of the weld is limited to a localized area relative to the overall shape of the steel pipe, the impact of such an area on the roundness of the steel pipe after the expansion process can be ignored.

[0108] Furthermore, during the welding process, the open pipe is bound from the outside while welding is performed in a way that reduces the joint gap of the open pipe. Therefore, the behavior of binding from the outside in a way that makes the joint gap of the open pipe zero can be numerically analyzed using the finite element method using the finite element analysis solver 112d, and the results can be used as the stress / strain state after the welding process.

[0109] On the other hand, when the reduction of the joint gap in such a welding process is an elastic deformation, the stress and strain analysis solution relative to the bending beam based on beam theory can be superimposed with the stress and strain distribution inside the open pipe calculated by finite element analysis to obtain the stress / strain distribution after the welding process. This can shorten the calculation time.

[0110] The finite element model generation unit 112c for the pipe expansion process performs element segmentation of the interior of the steel pipe based on the shape of the steel pipe after the welding process obtained as described above. The element segmentation is automatic based on pre-set element segmentation conditions. At this time, it is preferable to distribute the stress and strain distribution calculated as described above to each element. The generated finite element model of the pipe expansion process, together with the calculation conditions in the pipe expansion process, is sent to the finite element analysis solver 112d. The calculation conditions in the pipe expansion process include the operating parameters of the pipe expansion process in this embodiment, as well as all the information required to perform the finite element analysis, including the physical property values ​​of the steel plate, tools, etc., geometric boundary conditions, mechanical boundary conditions, etc.

[0111] The finite element analysis solver 112d performs numerical analysis under the calculation conditions given above, determining the shape of the steel pipe after the expansion process and the distribution of internal stress and strain. The calculated shape of the steel pipe has a non-uniform curvature distribution in the circumferential direction, and the roundness of the steel pipe is determined corresponding to the definition of roundness in the roundness measurement process. It should be noted that in the numerical analysis using the finite element method in the roundness offline calculation unit 112, the calculation time for one operating condition dataset (1 example) sometimes requires about 1 to 10 hours. However, since the processing is performed offline, there is no limitation on the calculation time. However, in order to shorten the calculation time for many operating condition datasets, multiple computers can be used to perform numerical calculations corresponding to multiple operating condition datasets in parallel. As a result, a database for generating roundness prediction models in a short period of time can be constructed. Moreover, in recent years, with the use of GPGPU calculations, the calculation time per example has been reduced to about 1 / 2 to 1 / 10 compared to the past, and such computer tools can also be used.

[0112] return Figure 10Database 120 stores operating condition dataset 111 and data related to the roundness of the steel pipe after the corresponding pipe expansion process. The data stored in database 120 can be obtained offline. Unlike databases that accumulate actual operational data, database 120 allows for arbitrary setting of operating condition datasets, thus reducing the likelihood of statistical bias in operating conditions set by operating condition dataset 111, making it suitable for machine learning. Furthermore, since it accumulates rigorous numerical analysis results rather than time-varying learning data, the more data accumulated, the more beneficial the database becomes. Additionally, database 120, generated through offline calculations, can calculate roundness under conditions different from actual manufacturing specifications, thus enabling prediction of roundness within a range where no manufacturing records exist.

[0113] The roundness prediction model generation unit 130 generates a roundness prediction model M, learned through machine learning, based on the relationship between multiple sets of operating condition datasets 111 stored in the database 120 and the roundness information of the steel pipe. This model calculates the roundness information of the steel pipe after the pipe expansion process relative to the input operating condition datasets 111. It should be noted that the relationship between the operating conditions in each process and the roundness information of the steel pipe after the pipe expansion process sometimes exhibits complex nonlinearity. Modeling using influence coefficients that assume linearity results in low accuracy. High-precision prediction can be achieved through machine learning techniques such as neural networks that utilize nonlinear functions. Here, modeling means replacing the input-output relationship in numerical calculations with an equivalent functional form.

[0114] The number of data points required to generate the roundness prediction model M varies depending on the manufacturing range of the steel pipe's dimensions, but 200 or more data points are sufficient. Preferably, 500 or more data points are used, and more preferably, 2000 or more data points are used. Known learning methods can be used for machine learning. For example, known machine learning techniques based on deep learning, including convolutional neural networks (CNNs) and recurrent neural networks (RNNs), are acceptable. Other methods include decision tree learning, random forests, Gaussian process regression, support vector regression, and k-nearest neighbors. Additionally, ensemble models combining multiple models can be used. It should be noted that while the roundness prediction model M is generated offline, the roundness prediction model generation unit 130 can also be integrated into an online control system, using a database that is continuously computed and accumulated offline to periodically update the roundness prediction model.

[0115] The roundness prediction model M of the steel pipe generated after the expansion process, as described above, has the following characteristics: The U-shaped stamping process involves the U-shaped stamping tool contacting the steel plate near the center of its width direction, causing the steel plate S to be wound relative to the tip of the U-shaped stamping tool. In this case, the bending moment applied to the steel plate varies depending on the contact position with the U-shaped stamping tool, resulting in a bending deformation with a curvature distribution. Furthermore, the tip of the U-shaped stamping tool sometimes has a shape formed by connecting curves with multiple curvatures; in this case, the curvature of the steel plate also varies along the surface of the U-shaped stamping tool. Here, the information used to determine the shape of the tip of the U-shaped stamping tool is referred to as the shape information of the U-shaped stamping tool.

[0116] On the other hand, in the O-forming process, if a combined deformation of compression and bending is applied to the U-shaped form, the bending moment applied in the O-forming process is distributed according to the local curvature distribution of the steel plate applied in the U-forming process, similar to how the bending moment acting on a so-called "bending beam" varies with the curvature of the beam. Furthermore, in the O-forming process, deformations where bending strain is locally concentrated, known as "plastic hinges," sometimes occur in areas with large bending moments. Additionally, the compressive force and bending moment applied to the U-shaped form in the O-forming process vary depending on the amount of opening at the width end of the U-shaped form. Therefore, the deformation state of the U-shaped form differs in the O-forming process, and the circumferential curvature distribution in the formed open tube also differs, thus affecting the roundness of the steel tube after the tube expansion process.

[0117] That is, the curvature distribution of the U-shaped formed body, which varies due to the operating conditions of the U-shaped stamping process, changes the deformation state imparted to the steel sheet in the O-shaped stamping process, thus affecting the circumferential curvature distribution of the steel pipe after the expansion process. Both the operating parameters of the U-shaped and O-shaped stamping processes are used as input parameters for the roundness prediction model M because the operating conditions of both processes collectively affect the roundness of the steel pipe after the expansion process.

[0118] It is known that the load of the pipe expansion process is proportional to the pipe thickness multiplied by the yield stress of the steel pipe. In thicker steel pipes and steel plates with high yield stress, the load (expansion load) of the pipe expansion process increases. On the other hand, the strength of the pipe expansion equipment tends to decrease inversely proportional to the outer diameter of the expansion tool. Therefore, if the outer diameter of the steel pipe decreases, the outer diameter of the expansion tool inserted into the steel pipe also decreases, and the equipment strength also tends to decrease. Furthermore, the closer the load of the pipe expansion process is to the equipment strength, the more the ability to improve the roundness of the steel pipe during the expansion process tends to decrease. Therefore, sometimes the pipe expansion process alone is insufficient to achieve sufficiently good roundness of the steel pipe. Therefore, it is preferable to appropriately optimize the operating parameters of both the U-shaped stamping process and the O-shaped stamping process to improve the roundness of the steel pipe after the expansion process. From this perspective, it is also necessary to include the operating parameters of both processes as input to the roundness prediction model M.

[0119] Here, the ability to improve the roundness of steel pipes during the pipe-expanding process (referred to as pipe-expanding capacity) is evaluated using the ratio of the strength of the pipe-expanding equipment to the pipe-expanding load (pipe-expanding load). This is an indicator of the equipment's margin relative to the load required for pipe expansion; a larger value indicates a higher pipe-expanding capacity. The strength of the pipe-expanding equipment is roughly inversely proportional to the outer diameter of the steel pipe, while the pipe-expanding load is proportional to the pipe thickness multiplied by the yield stress of the steel pipe. Therefore, pipe-expanding capacity decreases when manufacturing small-diameter, thick-walled, high-strength steel pipes. As a specific example of pipe-expanding capacity, we take the equipment with a steel pipe thickness of 50.8 mm, an outer diameter of 914.4 mm, and a yield stress of 300 MPa as the upper limit. Figure 12 , 13 The example illustrates the relationship between outer diameter, pipe thickness, and yield stress. Figure 12 This graph illustrates the relationship between pipe thickness and expansion capacity for a steel pipe with an outer diameter of 914.4 mm at various yield stresses. As the pipe thickness increases, the expansion capacity decreases; for the same pipe thickness, a higher yield stress results in a lower expansion capacity. From this perspective, in this embodiment, a steel pipe with a yield stress of 400–800 MPa and a pipe thickness of 19–55 mm is preferred. A yield stress of 500–800 MPa and a pipe thickness of 25–55 mm are more preferred. Furthermore, Figure 13 This graph shows the relationship between pipe thickness and expansion capacity for steel pipes with a yield stress of 300 MPa, for various outer diameters. The thicker the pipe, the lower the expansion capacity; conversely, for the same pipe thickness, the smaller the outer diameter, the lower the expansion capacity. From this perspective, in this embodiment, steel pipes with an outer diameter of 16–48 inches and a thickness of 12–55 mm are preferred. More preferably, steel pipes with an outer diameter of 16–36 inches and a thickness of 19–55 mm are preferred.

[0120] Furthermore, the roundness prediction model M preferably includes one or more parameters selected from the steel sheet's property information. As steel sheet properties, such as yield stress and thickness, certain deviations occur during the manufacturing of the raw steel sheet. During the bending process when the U-shaped stamping tool presses in, these parameters affect the curvature applied to the steel sheet and the curvature after the load is released. That is, by including parameters that affect the deformation state of the steel sheet during bending as steel sheet property information, the influence of the yield stress and thickness of each raw material on roundness can be considered individually. In addition, the O-shaped stamping process also uses a die to apply bending and compressive forces. Due to the yield stress and thickness, the curvature of the steel sheet after the load is released changes. Therefore, it is preferable to use the steel sheet's property information as input parameters for the roundness prediction model M.

[0121] On the other hand, as input to the roundness prediction model M, it is preferable to include one or more operating parameters selected from the operating parameters of the tube expansion process. More preferably, the tube expansion rate is used as the operating parameter of the tube expansion process. This is because, in the manufacturing process of UOE steel pipes made from high-strength steel plates, the tube expansion rate in the tube expansion process also has a significant impact on the final roundness of the product. However, if the tube expansion rate can only be set within a narrow range due to the tube expansion capacity of the tube expansion equipment, the range that can be changed as an operating parameter is narrow, so it may not be included in the input of the roundness prediction model M.

[0122] Furthermore, the input to the roundness prediction model M preferably includes one or more operating parameters selected from the operating parameters of the C-forming stamping process. This is because the area where the steel sheet is subjected to bending processing in the C-forming stamping process is limited to the vicinity of the end of the steel sheet in the width direction, which may not coincide with the area where bending deformation is applied in the U-forming stamping and O-forming stamping processes. Therefore, by using operating parameters from these multiple forming processes, the roundness prediction accuracy of the steel pipe after the tube expansion process is improved.

[0123] exist Figures 14-16 The diagram illustrates the effect of operating conditions for the U-shaped stamping and O-shaped stamping processes on the roundness of the steel pipe after the expansion process. The steel pipe in question is a product standard X65 steel pipe with dimensions of 911.8 mm outer diameter × 28.9 mm thickness. Figure 14 This is a diagram showing the measurement results of the cross-sectional shape of the steel pipe before the expansion process. It should be noted that the weld seam is overlaid with weld, therefore the measurement results for this part are excluded from the diagram. As for the operating conditions of the U-shaped stamping process, the horizontal r ( ) of the punch of the U-shaped stamping tool is selected. Figure 17The U-shaped stamping tool 22 shown is formed using a tool with a side radius of curvature of 178 mm. In this case, the side radius of curvature of the U-shaped stamping tool is smaller than the inner radius of the steel tube that becomes the product, therefore, the positions A1 and A2 of the steel plate that contact the side of the U-shaped stamping tool are... Figure 14 The cross-section of the steel pipe is convex in the vicinity of the 4 o'clock and 8 o'clock directions. Furthermore, if an O-forming process is performed on such a steel plate during a U-forming stamping process, localized bending deformation will concentrate in a portion of the steel plate, sometimes... Figure 14 The steel pipe section shown has a localized convex shape caused by a plastic hinge at two points (position B).

[0124] Such bulges formed on a portion of the circumference of the steel pipe sometimes become a cause of poor roundness, known as peaks, after the pipe expansion process. Peaks are bulges in the cross-section of the steel pipe after the expansion process. Figure 16 As shown, the index is defined by the distance between the outer circumference of the steel pipe P at the center of the chord and the arc P1, which corresponds to the outer diameter of the steel pipe at both ends of the chord, within a specified distance (150 mm in this case). The peak value is defined as positive if it is on the convex side relative to the arc corresponding to the outer diameter of the steel pipe, and negative if it is on the concave side. That is, if the peak value is 0, it means that the point is on the arc corresponding to the outer diameter of the steel pipe. In other words, the smaller the absolute value of the peak value along the entire outer circumference of the steel pipe, the better the roundness.

[0125] Figure 15 This shows a focus on Figure 14 The results of investigating the peak value after expanding the steel pipe were obtained by examining the three protrusions A1, A2, and B shown. Figure 15 The horizontal axis represents the tube expansion rate, an operational parameter of the tube expansion process, while the vertical axis represents the peak value (Body PK). For example... Figure 15 As shown, regarding the protrusions A1 and A2 generated in the U-shaped stamping process, increasing the expansion ratio slightly reduces their peak value, but the decrease is slow. On the other hand, the peak value of the protrusion B generated in the O-shaped stamping process tends to decrease with increasing expansion ratio. Thus, the peak value decreases with expansion ratio in the expansion process differs between the cross-sectional shapes formed in the U-shaped stamping process and the O-shaped stamping process. Therefore, appropriately setting the operating conditions in each forming process is necessary to improve roundness.

[0126] The roundness prediction model of this embodiment can take into account the influence of the operating parameters of multiple manufacturing processes on the roundness of the steel pipe after the expansion process, and can perform high-precision roundness prediction. Furthermore, since the roundness prediction model is generated through machine learning, even if the variables used as input conditions are changed, the roundness of the output can be calculated immediately. Therefore, it has the feature that operating conditions can be set and corrected immediately even when used online. The parameters used as inputs to the roundness prediction model will be explained below.

[0127] (Property information of the steel plate)

[0128] The steel plate properties used as input data for the roundness prediction model M can include any parameters that affect the roundness of the steel pipe after the tube expansion process, such as yield stress, tensile strength, longitudinal elastic modulus, thickness, thickness distribution within the plate surface, yield stress distribution in the thickness direction, magnitude of the Bauschinger effect, and surface roughness. In particular, it is preferable to use the deformation state and springback factors of the bending process in the U-shaped stamping process, and the deformation state and springback factors of the steel plate caused by compression / bending processes in the O-shaped stamping process as indicators.

[0129] The yield stress, thickness distribution, and thickness of the steel plate directly affect the stress and strain state during bending. Tensile strength, as a parameter reflecting the work hardening state during bending, influences the stress state during bending deformation. The Bauschinger effect affects the yield stress and subsequent work hardening behavior under load reversal caused by bending deformation, thus impacting the stress state during bending deformation. Furthermore, the longitudinal elastic modulus of the steel plate affects the springback behavior after bending. Moreover, the thickness distribution within the plate surface alters the distribution of bending curvature in the U-shaped stamping process, and surface roughness affects the roundness of the steel pipe after the expansion process by influencing the friction state between the die and the steel plate in the O-shaped stamping process.

[0130] Of these attributes, yield stress, representative plate thickness, thickness distribution information, and representative plate width are particularly preferred. This is because they correspond to information measured during the quality inspection process of the steel plate manufacturing process, namely the thick plate rolling process, and influence the deformation behavior in the U-shaped stamping and O-shaped stamping processes, thus affecting the roundness of the steel pipe after the tube expansion process. Furthermore, these are attributes representing the deviations of each steel plate becoming the raw material.

[0131] The yield stress corresponds to the information obtained from tensile tests on small test pieces used for quality verification of the thick steel plate used as raw material, and can be used as a representative value for the in-plane thickness of the steel plate used as raw material. Furthermore, the representative plate thickness is the thickness representing the in-plane thickness of the steel plate used as raw material, and can be the average of the thickness at the center of the steel plate in the width direction at any position along the length direction and the thickness in the length direction. Alternatively, the average thickness of the entire in-plane thickness of the steel plate can be used as the representative plate thickness. Thickness distribution information refers to information representing the thickness distribution in the in-plane thickness of the steel plate. As representative information, the thickness distribution in the width direction of the steel plate, i.e., convexity, can be cited. Convexity represents the difference between the thickness at the center of the steel plate in the width direction and the thickness at a position located a specified distance (e.g., 100 mm, 150 mm, etc.) from the end of the steel plate in the width direction. However, the thickness distribution information is not limited to this; the coefficients of an approximation obtained by approximating the thickness distribution in the width direction using a function of quadratic or higher can also be used as thickness distribution information. Alternatively, the thickness distribution in the length direction can be used instead of the thickness distribution in the width direction of the steel plate. This information represents the plate thickness and thickness distribution, and corresponds to the data measured by the plate thickness gauge during the rolling process of the thick plate and the data measured during the inspection process of the thick steel plate.

[0132] Furthermore, the representative plate width is a representative value relating to the width of the steel plate used as raw material. Deviations in the width of the thick steel plate used as raw material will affect the deviation in the outer diameter accuracy of the steel pipe produced. The representative width value can be the width at any position along the length of the steel plate, or it can be the average width along the length.

[0133] On the other hand, in the O-forming process, the steel sheet is subjected to bending and bending recovery deformation, so it is preferable to include property information representing the Bauschinger effect of the steel sheet. As property information representing the Bauschinger effect, a structural formula representing the stress-strain relationship of the steel sheet exhibiting the Bauschinger effect and the values ​​of the parameters used to determine this structural formula can be used. This is because, thereby, the mechanical properties of the steel sheet, such as movement hardening and isotropic hardening, and the anisotropy of yield stress, can also be reflected in the finite element analysis.

[0134] (Operating parameters for C-shaped stamping process)

[0135] When the operating parameters of the C-forming stamping process are used as inputs to the roundness prediction model M, parameters that determine the shape of the forming surface 13a of the upper die 13 and the pressing surface 14a of the lower die 14 used in the C-forming stamping apparatus can be used as operating parameters. Additionally, the end bending processing width (width for end bending forming), the feed rate of the steel sheet, the feed direction and number of feeds, the lifting force of the hydraulic cylinder 16 (C-forming stamping force), and the clamping force in the C-forming stamping process can also be used as operating parameters. This is because these are factors that can affect the deformation at the width-direction end of the steel sheet in the C-forming stamping process.

[0136] Regarding the shape of the forming surface 13a of the upper die 13, there are cases where it is given by a continuous arc shape with multiple radii of curvature, or by an involute curve, etc., and parameters for determining the geometric cross-sectional shape can be used. For example, when the cross-sectional shape is formed by a parabolic shape, the cross-sectional shape can be determined by using the first term and the coefficient of the quadratic expression representing the parabola passing through the origin, and therefore such coefficients can be used as operating parameters for the C-shaped stamping process.

[0137] On the other hand, if multiple molds are kept and used to determine the shape of the forming surface 13a of the upper mold 13 according to the outer diameter / wall thickness / steel grade of the manufactured steel pipe, the mold management number used to determine the mold used in the C-shaped stamping process can also be used as the operating parameter of the C-shaped stamping process.

[0138] (Operating parameters for the U-shaped stamping process)

[0139] In this embodiment, the operating parameters of the U-shaped stamping process are used as inputs to the roundness prediction model M. The operating parameters for the U-shaped stamping process can include the shape information of the U-shaped stamping tool (information used to determine the tip shape of the U-shaped stamping tool), the U-shaped stamping reduction amount, the initial spacing of the U-shaped stamping support, and the final spacing of the U-shaped stamping support. This is because these operating parameters have a significant impact on the deformation behavior of the steel sheet in the U-shaped stamping process.

[0140] As mentioned earlier, the U-shaped stamping reduction, the initial interval of the U-shaped stamping support, and the final interval of the U-shaped stamping support are common operating parameters that can be defined as parameters representing the deformation mode imparted to the steel sheet in both the Kaiser-type and Boson-type U-shaped stamping devices. However, for each device, parameters that indirectly affect these parameters can also be used. For example, the opening angle of the connecting rod 29 and the position information of the sliding block 27 in the Kaiser-type U-shaped stamping device can also be used. This is because these are operating parameters that can indirectly establish a correspondence with any of the U-shaped stamping reduction, the initial interval of the U-shaped stamping support, and the final interval of the U-shaped stamping support.

[0141] On the other hand, as a U-shaped stamping tool, sometimes it is used for example Figure 17 The tool shown is of the shape. Figure 17 In the shape of the U-shaped stamping tool 22 shown, the area that the U-shaped stamping tool 22 contacts when pressing down the steel plate is angled from a predetermined center point. The range gives rise to an arc shape of radius R, which, as the area where the steel plate contacts the brake roller, gives rise to a side surface shape of radius r that smoothly connects from the top end to the arc shape of radius R. In this case, the angle is utilized. The shape of the U-shaped stamping tool 22 is determined by three parameters: the top radius R (bottom R) and the side radius r (horizontal r). The parameters used to determine the top shape of the U-shaped stamping tool are called the shape information of the U-shaped stamping tool. However, if multiple U-shaped stamping tools with different top shapes exist depending on the outer diameter, wall thickness, and steel grade of the manufactured steel pipe, the U-shaped stamping tool management number used to determine the U-shaped stamping tool to be used can also be used as an operating parameter for the U-shaped stamping process.

[0142] (O-ring stamping process operating parameters)

[0143] In this embodiment, the operating parameters of the O-bending process are used as inputs to the roundness prediction model M. The operating parameters for the O-bending process include the O-bending reduction amount, the O-bending reduction position, and the O-bending die R. The O-bending reduction amount is particularly suitable because increasing the O-bending reduction amount results in a concentration of bending and compression deformation in the region between the point where the upper die applies restraint / pressure and the point where the lower die applies restraint, primarily in the 3-point and 9-point directions of the steel pipe, where there is no die restraint. This increases the curvature in this region, affecting the final roundness. At this time, the O-bending reduction amount, the O-bending reduction position, and the O-bending die R are information needed to control the O-bending device and correspond to the set values ​​set by the host computer.

[0144] (Operating parameters for the tube expansion process)

[0145] When the operating parameters of the tube expansion process are used as inputs to the roundness prediction model M, the tube expansion ratio can also be used as an operating parameter for the tube expansion process. A higher tube expansion ratio results in improved roundness of the steel pipe after the expansion process, but values ​​below a pre-set upper limit are used. In this case, the tube expansion ratio is information needed to control the tube expansion device and corresponds to the set value set by the host computer. It should be noted that, in addition to the tube expansion ratio, the number of tube expansion die pieces and the diameter of the tube expansion die can also be used as operating parameters for the tube expansion process.

[0146] [Method for predicting the roundness of steel pipes]

[0147] Next, a method for predicting the roundness of a steel pipe, as one embodiment of the present invention, will be described.

[0148] In the steel pipe roundness prediction method according to one embodiment of the present invention, the roundness prediction model M generated as described above is used for online processing in the steel pipe manufacturing process to predict the roundness of the steel pipe after the expansion process. During the roundness prediction of the steel pipe after the expansion process, an operation condition dataset (operation parameter acquisition step) is obtained online, which sets the operation conditions for the steel pipe manufacturing process. This is a step of obtaining the necessary data from a host computer overseeing the steel pipe manufacturing process or a control computer for each forming process, as the operation condition dataset that becomes the input to the roundness prediction model generated as described above. Here, "online" means the period from the start of the steel pipe manufacturing process to the completion of the expansion process, encompassing a series of manufacturing processes. Therefore, processing may not be performed in any single forming process. Even the period of waiting between forming processes to transport the steel plate to the next process is included in the "online" definition of this embodiment. Furthermore, even the period before the start of the steel pipe manufacturing process and after the completion of the thick plate rolling process that produces the raw material steel plate can be included in the "online" definition. This is because once the thick plate rolling process, which produces the steel plate as raw material, is completed, the state becomes one in which the operating condition dataset, which serves as the input to the roundness prediction model of this embodiment, can be obtained.

[0149] As described above, by using the operating condition dataset obtained from the operating parameter acquisition step as input and employing the roundness prediction model generated in the roundness prediction model generation step, the roundness of the steel pipe after the tube expansion process corresponding to the input operating condition dataset can be predicted (roundness prediction step). Therefore, in the steel pipe manufacturing process, including the U-shaped stamping process (forming a U-shaped body by pressing with a U-shaped stamping tool) and the tube expansion process (forming an open tube by reducing the joint gap of the U-shaped body and then joining the ends of the open tubes together to expand the inner diameter of the tube), the appropriateness of the manufacturing conditions in each process can be verified. The operating conditions of the U-shaped stamping process and the O-shaped stamping process significantly affect the roundness of the steel pipe after the tube expansion process, and the influence of these factors on the roundness of the product can be quantitatively evaluated. Furthermore, based on the roundness prediction model M using the property information of the steel plate used as the raw material, even if deviations occur in the property information of the steel plate in the upstream process, the influence of these factors on the roundness of the product can be quantitatively evaluated. Therefore, it is possible to predict the roundness deviation of steel pipe products based on the actual state of the deviation of the property information of the steel plate that becomes the raw material, and to change the operating conditions of the U-shaped stamping process and the O-shaped stamping process that take into account such raw material deviation.

[0150] [Methods for controlling the roundness of steel pipes]

[0151] Next, refer to Figure 18 The method for controlling the roundness of a steel pipe, as one embodiment of the present invention, will be described.

[0152] In this embodiment, the roundness prediction method for the steel pipe after the expansion process, which utilizes the roundness prediction model M for the steel pipe, is used to control the roundness of the steel pipe as follows: First, a resetting target process is selected from among the multiple forming processes constituting the steel pipe manufacturing process. Then, before the start of the resetting target process, the roundness prediction model M is used to predict the roundness of the steel pipe after the expansion process. Next, in order to reduce the roundness of the steel pipe after the expansion process, at least one or more operating parameters selected from the operating parameters of the resetting target process or at least one operating parameter selected from the operating parameters of the forming process downstream of the resetting target process are reset.

[0153] Here, the multiple forming processes constituting the steel pipe manufacturing process refer to the C-shaped stamping process, U-shaped stamping process, O-shaped stamping process, and pipe expansion process, which plastically deform the steel plate to form a steel pipe into a specified shape. The target process is then selected from these forming processes. Before performing the forming process in the selected target process, the roundness prediction model M of the steel pipe is used to predict the roundness of the steel pipe after the pipe expansion process. At this time, for the forming processes upstream of the target process, the forming of the steel plate has already been completed; therefore, when using the operating parameters of the upstream forming processes, their performance data can be used as input to the roundness prediction model M. On the other hand, for the forming processes downstream of the target process, since operational performance data cannot be collected, pre-set values ​​in a host computer or similar device are used as input to the roundness prediction model M of the steel pipe. In this way, the roundness of the steel pipe after the pipe expansion process of the target material can be predicted.

[0154] Then, it is determined whether the predicted roundness of the steel pipe after the expansion process is within the acceptable roundness for the product. Therefore, if the roundness of the steel pipe after the expansion process is smaller than the predicted value, the operating conditions in the resetting target process and the forming process downstream of the resetting target process can be reset. Here, the reset operating parameters can be either the operating parameters in the resetting target process or the operating parameters in the forming process downstream of the resetting target process. Based on the difference between the predicted roundness and the acceptable roundness for the product, the operating parameters of the forming process suitable for changing the roundness of the steel pipe after the expansion process can be selected. Alternatively, the operating parameters of both the resetting target process and any forming process downstream of the resetting target process can be reset. This is because when the difference between the predicted roundness and the acceptable roundness for the product is large, the roundness of the steel pipe after the expansion process can be effectively changed.

[0155] Table 1 shows specific examples of forming processes selected as resetting targets and corresponding forming processes whose operating parameters can be reset. Example 1 is a steel pipe manufacturing process that includes a C-shaped stamping process, where the C-shaped stamping process is selected as the resetting target process. In this case, before the start of the C-shaped stamping process, the roundness of the steel pipe after the tube expansion process is predicted using the set values ​​of the operating parameters in the forming processes including the U-shaped stamping and O-shaped stamping processes. If the predicted roundness is high, any operating parameter in each forming process of the C-shaped stamping, U-shaped stamping, O-shaped stamping, and tube expansion processes can be reset. The operating parameters to be reset are not limited to the operating parameters of the C-shaped stamping process; they can also be the operating parameters of other forming processes. It should be noted that when the steel plate's property information is included as input to the roundness prediction model M, actual data, including measured values ​​related to the steel plate's property information, can be used as input before the start of the resetting target process, i.e., the C-shaped stamping process.

[0156] Examples 2 and 3 can also be approached using the same logic as Example 1 to select the target process for resetting and to reset the operating parameters. On the other hand, Example 4 involves using the tube expansion process as the target process for resetting. In this case, a roundness prediction model M is used before the tube expansion process begins to predict the roundness of the steel pipe after the expansion process. As input to the roundness prediction model M, at least the operational data from the U-shaped stamping and O-shaped stamping processes can be used. Alternatively, operational data from the steel plate's property information and the operational data from the C-shaped stamping process can also be used. By comparing the predicted roundness of the steel pipe after the expansion process with the roundness allowed for the product, and if a reduction in roundness is desired, the operating parameters in the tube expansion process are reset. The expansion rate is preferably used as the reset operating parameter for the tube expansion process. It should be noted that the change in the reset expansion rate relative to the initial setting value can be reset based on operational experience data. However, if the roundness prediction model M includes the expansion rate of the expansion process in its input, the value of the re-set expansion rate can also be used as the input of the roundness prediction model M to re-predict the roundness of the steel pipe after the expansion process and determine whether the re-set conditions are appropriate.

[0157] [Table 1]

[0158] (Table 1)

[0159]

[0160] ○: Forming process with reconfigurable operating parameters

[0161] Here, refer to Figure 18The method for controlling the roundness of a steel pipe, as one embodiment of the present invention, will be described. Figure 18 The example shown illustrates an instance where an O-forming stamping process is selected as the resetting target process, and the U-shaped forming body is transferred to the O-forming stamping device after the U-forming stamping process is completed. At this time, the operational performance data from the U-forming stamping process is sent to the operation condition resetting unit 140. Ideally, the operational performance data is sent via a network from the control computer installed in each forming processing equipment (the equipment performing the forming process). However, it is also possible to temporarily send the data from the control computer of each forming processing equipment to the host computer 150, which oversees the steel pipe manufacturing process, and then send it from the host computer 150 to the operation condition resetting unit 140. Furthermore, the operation condition resetting unit 140 sends performance data regarding the steel plate's properties from the host computer 150 as needed. In the case of a pre-processing process, the operation performance data from the pre-processing process is used to correct a portion of the steel plate's property information, and this correction is sent to the operation condition resetting unit 140. For example, if the width of the steel plate that becomes the raw material changes during the pre-processing process, the width of the steel plate's property information is corrected to the width after the pre-processing process. Furthermore, operational performance data from the C-forming process can be sent as needed. Additionally, the setting values ​​for the operation parameters of the resetting target process and the downstream forming processes, namely the O-forming and tube-expanding processes, are sent from the control computer of each forming equipment to the operation condition resetting unit 140. However, if the setting values ​​for the operation parameters of the O-forming and tube-expanding processes are stored in the host computer 150, they can also be sent from the host computer 150 to the operation condition resetting unit 140. It should be noted that the roundness target value, determined by the specifications of the steel pipe becoming the product, is sent from the host computer 150 to the operation condition resetting unit 140.

[0162] The operating condition resetting unit 140 uses the roundness prediction model M online to predict the roundness of the steel pipe after the tube expansion process based on this information, and compares the predicted roundness (predicted roundness value) with the target roundness (target roundness value). Furthermore, if the predicted roundness is smaller than the target roundness value, the operating condition resetting unit 140 determines the operating conditions for the remaining forming processes without changing the set values ​​of the operating conditions for the U-shaped stamping process, the O-shaped stamping process, and the tube expansion process, and manufactures the steel pipe. On the other hand, if the predicted roundness is larger than the target roundness value, the operating condition resetting unit 140 resets at least the operating conditions for the O-shaped stamping process or the tube expansion process. Specifically, the O-shaped stamping reduction amount of the O-shaped stamping process can be reset. Additionally, the tube expansion rate of the tube expansion process can be reset. Moreover, both the O-shaped stamping reduction amount and the tube expansion rate can be reset.

[0163] It should be noted that the operation condition resetting unit 140 can also use the reset values ​​of the reset operation parameters as input data for the roundness prediction model M to perform roundness prediction again, confirm whether the predicted roundness has become smaller than the roundness target value, and determine the reset values ​​of the operation conditions for the O-forming stamping process and the tube expansion process. The reset operation conditions for the O-forming stamping process and the tube expansion process are sent to each control computer to determine the operation conditions for the O-forming stamping process and the tube expansion process. By repeatedly performing the roundness determination in the operation condition resetting unit 140, even if the roundness target value is set small, appropriate operation conditions for the O-forming stamping process and the tube expansion process can be set, thus enabling the manufacture of steel pipes with better roundness. Moreover, after performing roundness control on the steel pipe after the tube expansion process, in which the O-forming stamping process is the resetting target process, the roundness control on the steel pipe after the tube expansion process, which has been formed into an open tube and welded, can be performed again. This is because obtaining operational data on the O-ring stamping process improves the accuracy of predicting the roundness of the steel pipe.

[0164] As described above, according to the steel pipe roundness control method of one embodiment of the present invention, since a roundness prediction model M that considers the influence of the interaction between the U-shaped stamping process and the O-shaped stamping process on roundness is used, suitable operating conditions for ensuring good roundness of the steel pipe after the expansion process can be set, and steel pipes with high roundness can be manufactured. Furthermore, high-precision roundness control reflecting deviations in the property information of the steel sheet used as raw material can be achieved.

[0165] [Steel pipe roundness prediction device]

[0166] Next, refer to Figure 19 The roundness prediction device for steel pipes, which is one embodiment of the present invention, will be described.

[0167] Figure 19 This is a diagram showing the structure of a steel pipe roundness prediction device as an embodiment of the present invention. Figure 19 As shown, the roundness prediction device 160 for steel pipes according to one embodiment of the present invention includes an operation parameter acquisition unit 161, a storage unit 162, a roundness prediction unit 163, and an output unit 164.

[0168] The operation parameter acquisition unit 161 has an interface capable of acquiring any roundness prediction model M generated by the machine learning unit, for example, from the roundness prediction model generation device 100 of the steel pipe. For example, the operation parameter acquisition unit 161 preferably has a communication interface for acquiring the roundness prediction model M from the roundness prediction model generation device 100 of the steel pipe. In this case, the operation parameter acquisition unit 161 can also receive the roundness prediction model M from the roundness prediction model generation device 100 of the steel pipe using a prescribed communication protocol. Furthermore, the operation parameter acquisition unit 161 acquires the operating conditions of the forming equipment (the equipment performing the forming process) from, for example, the control computer or host computer of the equipment used in each forming process. For example, the operation parameter acquisition unit 161 preferably has a communication interface for acquiring the operating conditions. Additionally, the operation parameter acquisition unit 161 can acquire input information based on user operations. In this case, the roundness prediction device 160 of the steel pipe also has an input unit including one or more input interfaces that detect user input and acquire input information based on user operations. Examples of input units include physical buttons, electrostatic capacitive buttons, touchscreens integrated with the display of the output unit, and microphones that accept voice input, but are not limited to these. For example, the input unit accepts input relative to the operating conditions of the roundness prediction model M obtained by the operation parameter acquisition unit 161 from the roundness prediction model generation device 100 of the steel pipe.

[0169] The storage unit 162 includes at least one semiconductor memory, at least one magnetic memory, at least one optical memory, or a combination of at least two of these. The storage unit 162 functions as, for example, a main storage device, an auxiliary storage device, or a flash memory. The storage unit 162 stores any information used in the operation of the roundness prediction device 160 for steel pipes. For example, the storage unit 162 stores the roundness prediction model M obtained by the operation parameter acquisition unit 161 from the roundness prediction model generation device 100 for steel pipes, the operating conditions obtained by the operation parameter acquisition unit 161 from the host computer, and the roundness information predicted by the roundness prediction device 160 for steel pipes. The storage unit 162 may also store system programs and application programs, etc.

[0170] The roundness prediction unit 163 includes one or more processors. In this embodiment, the processor can be a general-purpose processor or a dedicated processor for a specific process, but is not limited to either. The roundness prediction unit 163 is communicatively connected to each component of the roundness prediction device 160 constituting the steel pipe, and controls the overall operation of the roundness prediction device 160. The roundness prediction unit 163 can be any general-purpose electronic device, such as a PC (Personal Computer) or a smartphone. The roundness prediction unit 163 is not limited to these; it can also be one or multiple server devices that can communicate with each other, or other electronic devices dedicated to the roundness prediction device 160. The roundness prediction unit 163 uses the operating conditions obtained through the operating parameter acquisition unit 161 and the roundness prediction model M obtained from the roundness prediction model generation device 100 of the steel pipe to calculate the predicted value of the roundness information of the steel pipe.

[0171] The output unit 164 outputs the predicted value of the roundness information of the steel pipe calculated by the roundness prediction unit 163 to a device for setting the operating conditions of the forming processing equipment. The output unit 164 may include one or more output interfaces for outputting information and notifying the user. The output interface is, for example, a display. The display is, for example, an LCD or an organic EL display. The output unit 164 outputs the data obtained by the operation of the steel pipe roundness prediction device 160. The output unit 164 may also replace the roundness prediction device 160 installed on the steel pipe and be connected to the steel pipe roundness prediction device 160 as an external output device. As a connection method, any method such as USB, HDMI (registered trademark), or Bluetooth (registered trademark) can be used. For example, as the output unit 164, examples include a display that outputs information in the form of an image, a speaker that outputs information in the form of voice, etc., but it is not limited to them. For example, the output unit 164 prompts the user with the predicted value of the roundness information calculated by the roundness prediction unit 163. Users can appropriately set the operating conditions of the forming equipment based on the predicted roundness value provided by the output unit 164.

[0172] A more preferred embodiment of the roundness prediction device 160 for steel pipes after the expansion process described above is a terminal device such as a tablet terminal, which has an input unit 165 that acquires input information based on user operation and a display unit 166 that displays the predicted value of the roundness information calculated by the roundness prediction unit 163. It acquires the input information based on user operation from the input unit 165 and uses this information to update part or all of the operating parameters of the forming process that have been input to the roundness prediction device 160 for steel pipes. That is, regarding the steel plate being processed in the forming equipment, if the roundness information of the steel pipe has been predicted by the roundness prediction unit 163, the operator uses the terminal device to accept the operation of correcting a portion of the operating parameters of the forming process that have been input to the operating parameter acquisition unit 161. At this time, the operating parameter acquisition unit 161 maintains the original input data for operating parameters that are not corrected from the terminal device in the operating parameters of the forming process, and only changes the operating parameters that have been corrected. Therefore, in the operation parameter acquisition unit 161, new input data for the roundness prediction model M is generated, and the roundness prediction unit 163 calculates the predicted value of the roundness information based on the input data. Furthermore, the calculated predicted value of the roundness information is displayed on the display unit 166 of the terminal device via the output unit 164. Thus, the operator or factory manager of the forming equipment can immediately confirm the predicted value of the roundness information when the operation parameters of the forming process are changed, and quickly adjust to suitable operating conditions.

[0173] Example

[0174] [Example 1]

[0175] In this embodiment, Figure 10 In the basic data acquisition unit 110 shown, an operation condition dataset including the following operation parameters is set. The steel pipe that becomes the product after the pipe expansion process is an API grade X56 pipe with a thickness of 31.8 mm and an outer diameter of 914.4 mm. First, the steel plate's property information is set as a thickness of 31.8 mm and a width of 2751 mm. The yield stress is included in the operation condition dataset as a property information of the steel plate, corresponding to a tensile strength of 480–600 MPa. It should be noted that the width is set after the pre-processing step. Regarding the operation parameters for the C-shaped stamping process, the radius of curvature of the forming surface of the upper die is set to R310 mm. Calculation conditions for end bending processing are set for a 180 mm range from the end in the width direction as the end bending processing width, but this is not included in the input of the roundness prediction model.

[0176] The operating parameters of the U-shaped stamping process are based on the U-shaped stamping process of the Kaiser-type U-shaped stamping device, and a corresponding finite element model of the U-shaped stamping process is generated. Regarding the shape information of the U-shaped stamping tool, the base radius (R) is 362mm, and the base radius angle... The angle is 120 degrees. Calculations are performed for the horizontal radius (r) at two levels: 178 mm and 191 mm. These are included in the operating condition dataset. Additionally, as operating parameters for the U-shaped stamping process, the final spacing of the U-shaped stamping support is varied within a range of 564 ± 30 mm, and the U-shaped stamping reduction is varied within a range of 782.6 ± 12.7 mm. These are also included in the operating condition dataset for the U-shaped stamping process.

[0177] On the other hand, regarding the operating parameters for the O-buck stamping process, the upper and lower dies are set with an O-buck stamping die radius R451mm, an arc depth of 451mm for the upper die, and an arc depth of 438mm for the lower die. Furthermore, the calculation conditions are modified to make the distance between the uppermost point of the upper die and the lowermost point of the lower die (O-buck stamping reduction) 903±3mm, and this is included in the operating condition dataset for the O-buck stamping process. Figure 20 An example of a finite element model in the O-ring stamping process is shown. Furthermore, in this embodiment, as an operating parameter for the tube expansion process, the tube expansion die is assumed to have a shape with a tube expansion die radius of 390 mm after being divided into 12 sections in the circumferential direction. The tube expansion rate is varied in the range of 0.4 to 1.6%, and this is included in the operating condition dataset of the tube expansion process.

[0178] Regarding the welding process, the behavior of binding the open tube from the surroundings to achieve a zero joint gap is numerically analyzed using the finite element method, and the results are used as the stress / strain state after the welding process. Specifically, for the open tube after the O-ring stamping process, finite element analysis was performed on the behavior of pressing the rollers into a circle with a diameter of 905.5 mm by positioning nine rollers at the joint gap and at distances of 24, 65, 105, and 148 degrees from the joint gap. However, the heat effect during welding was not considered. The stress / strain state of the open tube with the joint gap at zero after pressing in the rollers was used as the stress / strain state of the steel pipe after the welding process and sent to the finite element model generation unit for the subsequent tube expansion process. It should be noted that the roundness of the steel pipe after the tube expansion process, which is the output of the roundness prediction model, is obtained by dividing the outer diameter shape of the steel pipe after the tube expansion process obtained by the roundness offline calculation unit 112 into 1080 points in the circumferential direction, and using the difference between the maximum diameter Dmax and the minimum diameter Dmin.

[0179] As described above, the roundness prediction model for the steel pipe after the expansion process in this embodiment includes the final spacing of the U-shaped stamping support and the U-shaped stamping reduction as operating parameters for the U-shaped stamping process, and the O-shaped stamping reduction as operating parameters for the O-shaped stamping process. Furthermore, the yield stress is included in the operating condition dataset as property information of the steel plate, and the expansion rate is included in the operating condition dataset as an operating parameter for the expansion process. The data required for the finite element analysis of the forming process, including such an operating condition dataset, is provided to the roundness offline calculation unit 112, and the roundness of the steel pipe after the expansion process is calculated. The learning dataset obtained through calculation is accumulated in the database 120, and a roundness prediction model M is generated using the roundness prediction model generation unit 130.

[0180] In the roundness offline calculation unit 112, a two-dimensional plane strain element was obtained by dividing the steel plate, which would become the raw material before the C-shaped stamping process, into 720 sections in the width direction and 18 sections in the thickness direction. Finite element analysis was performed in the order of C-shaped stamping, U-shaped stamping, O-shaped stamping, welding, and tube expansion processes. In this embodiment, the finite element analysis solver used was Abaqus 2019, and the calculation time for each case was approximately 3 hours. The number of datasets stored in the database 120 was 300, and Gaussian process regression with radial basis functions was used as the machine learning model. The roundness prediction model M generated in this way was mounted as an online model. Figure 18 The system shown sets the roundness target value to 8mm and obtains the yield stress value of the steel plate from the material inspection results in the previous process, namely the thick plate rolling process, by using the actual performance data of the property information of the steel plate as the raw material from the host computer.

[0181] In the first embodiment, the U-shaped stamping process is first designated as the re-setting process. After the C-shaped stamping process is completed and before the transition to the U-shaped stamping process, the actual value of the yield stress, which is a property information of the steel plate, is used as the input to the roundness prediction model M, and the operating parameters for the U-shaped stamping process and the O-shaped stamping process are re-set. However, the setting value of the operating conditions for the tube expansion process is fixed at 1.0% and is not included in the re-set operating parameters. On the other hand, the re-set operating parameters for the U-shaped stamping process are the final interval of the U-shaped stamping support, and the re-set operating parameters for the O-shaped stamping process are the O-shaped stamping reduction amount. The re-set operating parameters are all subject to constraints within the range of the operating parameters set by the roundness offline calculation unit 112. By executing this first embodiment of roundness control, 100 steel tubes were manufactured. As a result, without using the roundness prediction model M, the average roundness was 7.9 mm and the pass rate was 40%, while in the first embodiment, the average roundness was improved to 6.2 mm and the pass rate was increased to 75%.

[0182] As a second embodiment, similar to the first embodiment described above, the U-shaped stamping process was used as the re-setting process. After the C-shaped stamping process and before transitioning to the U-shaped stamping process, the actual value of the yield stress, which is the property information of the steel plate, was used as the input to the roundness prediction model M, and the operating parameters of the U-shaped stamping process, the O-shaped stamping process, and the tube expansion process were re-set. It should be noted that the re-set operating parameter for the tube expansion process is the tube expansion rate, which was re-set within the range of 0.6% to 1.3%. The other re-set operating parameters are the same as in the first embodiment. By implementing this second embodiment of roundness control, 100 steel pipes were manufactured. As a result, the average roundness was further improved to 5.1 mm, and the pass rate was further improved to 90%.

[0183] [Example 2]

[0184] In this embodiment, Figure 10 In the basic data acquisition unit 110 shown, an operation condition dataset including the following operation parameters is set. The steel pipe that becomes the product after the pipe expansion process is an API grade X80 pipe with a thickness of 25.4 mm and an outer diameter of 558.8 mm. First, the steel plate's property information is set as a thickness of 25.0–27.0 mm and a width of 1662 mm. The steel plate thickness and yield stress are included in the operation condition dataset as property information, corresponding to steel plates with a tensile strength of 600–780 MPa. It should be noted that the width is set after the pre-processing step. Regarding the operation parameters for the C-shaped stamping process, the radius of curvature of the forming surface of the upper die is set to R170 mm. Calculation conditions for end bending processing are set for a 135 mm range from the end in the width direction as the end bending processing width, but this is not included in the input of the roundness prediction model.

[0185] The operating parameters of the U-shaped stamping process are based on the U-shaped stamping process of the Kaiser-type U-shaped stamping device, and a corresponding finite element model of the U-shaped stamping process is generated. Regarding the shape information of the U-shaped stamping tool, the base radius (R) is 225mm, and the base radius angle... The angle is 120 degrees, and the horizontal radius (r) is calculated with a condition of 110 mm. This is included in the operating condition dataset. In addition, as operating parameters in the U-shaped stamping process, the final spacing of the U-shaped stamping support is varied within the range of 314 ± 30 mm, and the U-shaped stamping reduction is varied within the range of 706.4 ± 25.4 mm. These are also included in the operating condition dataset for the U-shaped stamping process.

[0186] On the other hand, regarding the operating parameters for the O-ring stamping process, the upper and lower dies are set to an O-ring stamping die radius R276mm, the arc depth of the upper die is 276mm, and the arc depth of the lower die is 264mm. Furthermore, the calculation conditions are modified to make the distance between the uppermost point of the upper die and the lowermost point of the lower die (O-ring stamping reduction) 578±3mm, and this is included in the operating condition dataset for the O-ring stamping process. Moreover, in this embodiment, as operating parameters for the tube expansion process, the tube expansion die is based on a die with a shape having a circumferentially divided 10-fold radius of 240mm, and the tube expansion rate is varied within a range of 0.4% to 1.6%, and this is included in the operating condition dataset for the tube expansion process.

[0187] Regarding the welding process, the behavior of binding the open tube from the surroundings to achieve zero joint gap is numerically analyzed using the finite element method, and the results are used as the stress / strain state after the welding process. Specifically, for the open tube after the O-ring stamping process, finite element analysis was performed on the behavior of pressing the rollers into a circle with a diameter of 553.3 mm by positioning seven rollers at the joint gap and at distances of 38, 84, and 130 degrees from the joint gap. However, the heat effect during welding was not considered. The stress / strain state of the open tube with zero joint gap after pressing in the rollers was used as the stress / strain state of the steel pipe after the welding process and sent to the finite element model generation unit for the subsequent tube expansion process. It should be noted that the roundness of the steel pipe after the tube expansion process, which is the output of the roundness prediction model, is obtained by dividing the outer diameter shape of the steel pipe after the tube expansion process obtained by the roundness offline calculation unit 112 into 1080 points in the circumferential direction, and using the difference between the maximum diameter Dmax and the minimum diameter Dmin.

[0188] As described above, the roundness prediction model for the steel pipe after the expansion process in this embodiment includes the final spacing of the U-shaped stamping support and the U-shaped stamping reduction as operating parameters for the U-shaped stamping process, and the O-shaped stamping reduction as operating parameters for the O-shaped stamping process. Furthermore, the plate thickness and yield stress are included in the operating condition dataset as property information of the steel plate, and the expansion rate is included in the operating condition dataset as an operating parameter for the expansion process. The data required for the finite element analysis of the forming process, including such an operating condition dataset, is provided to the roundness offline calculation unit 112, and the roundness of the steel pipe after the expansion process is calculated. The learning dataset obtained through calculation is accumulated in the database 120, and a roundness prediction model M is generated using the roundness prediction model generation unit 130.

[0189] In the roundness offline calculation unit 112, a two-dimensional plane strain element was obtained by dividing the steel plate, which would become the raw material before the C-shaped stamping process, into 720 sections in the width direction and 18 sections in the thickness direction. Finite element analysis was performed in the order of C-shaped stamping, U-shaped stamping, O-shaped stamping, welding, and tube expansion processes. In this embodiment, the finite element analysis solver used was Abaqus 2019, and the computation time for each case was approximately 3 hours. The dataset stored in database 120 contained 500 data points. For the machine learning model, an ensemble model combining a neural network and a decision tree was used. The neural network had one intermediate layer with 5 nodes in each layer. The ReLU function was used as the activation function. The decision tree had a maximum depth of 3 levels, and the maximum number of leaves in the generated decision tree was 160. The roundness prediction model M generated in this way was then used as an online model. Figure 18 The system shown sets the roundness target value to 5mm and obtains the yield stress value of the steel plate from the material inspection results in the previous process, namely the thick plate rolling process, by using the actual performance data of the property information of the steel plate as the raw material from the host computer.

[0190] As a third embodiment, firstly, the U-shaped stamping process is treated as a re-setting process. After the C-shaped stamping process is completed and before the transition to the U-shaped stamping process, the actual values ​​of the plate thickness and yield stress, which are attribute information of the steel plate, are used as inputs to the roundness prediction model M, and the operating parameters for the U-shaped stamping process and the O-shaped stamping process are re-set. However, the setting value of the operating conditions for the tube expansion process is fixed at 1.0% and is not included in the re-set operating parameters. On the other hand, the re-set operating parameters for the U-shaped stamping process are the final interval of the U-shaped stamping support, and the re-set operating parameters for the O-shaped stamping process are the O-shaped stamping reduction amount. The re-set operating parameters are all subject to constraints within the range of the operating parameters set by the roundness offline calculation unit 112. By executing this third embodiment of roundness control, 100 steel pipes were manufactured. As a result, without using the roundness prediction model M, the average roundness was 5.0 mm and the pass rate was 60%. In the third embodiment, the average roundness was improved to 4.1 mm and the pass rate was increased to 81%.

[0191] As a fourth embodiment, similar to the third embodiment described above, the U-shaped stamping process is used as the re-setting process. After the C-shaped stamping process and before transitioning to the U-shaped stamping process, the actual values ​​of the plate thickness and yield stress, which are attribute information of the steel plate, are used as inputs to the roundness prediction model M, and the operating parameters of the U-shaped stamping process, O-shaped stamping process, and tube expansion process are re-set. It should be noted that the re-set operating parameter for the tube expansion process is the tube expansion rate, which is re-set within the range of 0.6% to 1.3%. The other re-set operating parameters are the same as in the third embodiment. By implementing this fourth embodiment of roundness control, 100 steel pipes were manufactured. As a result, the average roundness was further improved to 2.6 mm, and the pass rate was further improved to 95%.

[0192] [Example 3]

[0193] In this embodiment, Figure 10 In the basic data acquisition unit 110 shown, an operation condition dataset including the following operation parameters is set. The steel pipe that becomes the product after the pipe expansion process is an API grade X100 pipe with a thickness of 12.7 mm and an outer diameter of 1219.2 mm. First, the steel plate's property information is set as a thickness of 12.7–14.3 mm and a width of 3760 mm. The steel plate thickness and yield stress are included in the operation condition dataset as property information, corresponding to steel plates with a tensile strength of 720–900 MPa. It should be noted that the width is set after the pre-processing step. Regarding the operation parameters for the C-shaped stamping process, the radius of curvature of the forming surface of the upper die is set to R310 mm. Calculation conditions for end-bending processing are set for a 180 mm range from the end in the width direction as the end-bending processing width, but this is not included in the input of the roundness prediction model.

[0194] The operating parameters of the U-shaped stamping process are based on the U-shaped stamping process of the Kaiser-type U-shaped stamping device, and a finite element model of the corresponding U-shaped stamping process is generated. Regarding the shape information of the U-shaped stamping tool, the base radius (R) is 1300mm, and the base radius angle... The angle is 27 degrees, and the horizontal radius (r) is calculated with a condition of 120 mm. This is included in the operating condition dataset. In addition, as operating parameters in the U-shaped stamping process, the final spacing of the U-shaped stamping support is varied within the range of 394 ± 40 mm, and the U-shaped stamping reduction is varied within the range of 858.8 ± 25.4 mm. These are also included in the operating condition dataset for the U-shaped stamping process.

[0195] On the other hand, regarding the operating parameters for the O-ring stamping process, the upper and lower dies are set with an O-ring stamping die radius R602mm, an arc depth of 602mm for the upper die, and an arc depth of 590mm for the lower die. Furthermore, the calculation conditions are modified to make the distance between the uppermost point of the upper die and the lowermost point of the lower die (O-ring stamping reduction) 1200±5mm, and this is included in the operating condition dataset for the O-ring stamping process. Moreover, in this embodiment, as operating parameters for the tube expansion process, the tube expansion die is based on a die with a shape having a radius of 545mm after being divided into 12 sections in the circumferential direction, and the tube expansion rate is varied within the range of 0.5% to 1.6%, and this is included in the operating condition dataset for the tube expansion process.

[0196] Regarding the welding process, the behavior of binding the open tube from the surroundings to achieve a zero joint gap is numerically analyzed using the finite element method, and the results are used as the stress / strain state after the welding process. Specifically, for the open tube after the O-ring stamping process, finite element analysis was performed on the behavior of pressing the rollers into a circle with a diameter of 1270.1 mm by positioning nine rollers at the joint gap and at distances of 22, 70, 105, and 150 degrees from the joint gap. However, the heat effect during welding was not considered. The stress / strain state of the open tube with the joint gap at zero after pressing in the rollers was used as the stress / strain state of the steel pipe after the welding process and sent to the finite element model generation unit for the subsequent tube expansion process. It should be noted that the roundness of the steel pipe after the tube expansion process, which is the output of the roundness prediction model, is obtained by dividing the outer diameter shape of the steel pipe after the tube expansion process obtained by the roundness offline calculation unit 112 into 1080 points in the circumferential direction, and the difference between the maximum diameter Dmax and the minimum diameter Dmin is used.

[0197] As described above, the roundness prediction model for the steel pipe after the expansion process in this embodiment includes the final spacing of the U-shaped stamping support and the U-shaped stamping reduction as operating parameters for the U-shaped stamping process, and the O-shaped stamping reduction as operating parameters for the O-shaped stamping process. Furthermore, the plate thickness and yield stress are included in the operating condition dataset as property information of the steel plate, and the expansion rate is included in the operating condition dataset as an operating parameter for the expansion process. The data required for the finite element analysis of the forming process, including such an operating condition dataset, is provided to the roundness offline calculation unit 112, and the roundness of the steel pipe after the expansion process is calculated. The learning dataset obtained through calculation is accumulated in the database 120, and a roundness prediction model M is generated using the roundness prediction model generation unit 130.

[0198] In the roundness offline calculation unit 112, a two-dimensional plane strain element was obtained by dividing the steel plate, which would become the raw material, into sections 720 in the width direction and 18 in the thickness direction before the C-shaped stamping process. Finite element analysis was performed in the order of C-shaped stamping, U-shaped stamping, O-shaped stamping, welding, and tube expansion processes. In this embodiment, the finite element analysis solver used is Abaqus 2019, and the calculation time for each case is approximately 3 hours. The number of datasets stored in the database 120 is 400. For the machine learning model, gradient boosting, a method of ensemble learning using decision trees, was used. The number of decision trees constituting the gradient boosting decision tree was set to 10, the maximum depth of its levels was set to 5, and the maximum number of leaves in the generated decision tree was set to 180. The roundness prediction model M generated in this way was then used as an online model. Figure 18 The system shown sets the roundness target value to 10mm and obtains the yield stress value of the steel plate from the material inspection results in the previous process, namely the thick plate rolling process, by using the actual performance data of the property information of the steel plate as the raw material from the host computer.

[0199] As a fifth embodiment, firstly, the U-shaped stamping process is treated as a re-setting process. After the C-shaped stamping process is completed and before the transition to the U-shaped stamping process, the actual values ​​of the plate thickness and yield stress, which are attribute information of the steel plate, are used as inputs to the roundness prediction model M, and the operating parameters for the U-shaped stamping process and the O-shaped stamping process are re-set. However, the setting value of the operating conditions for the tube expansion process is fixed at 1.2% and is not included in the re-set operating parameters. On the other hand, the re-set operating parameters for the U-shaped stamping process are the final interval of the U-shaped stamping support, and the re-set operating parameters for the O-shaped stamping process are the O-shaped stamping reduction amount. The re-set operating parameters are all subject to constraints within the range of the operating parameters set by the roundness offline calculation unit 112. By executing this fifth embodiment of roundness control, 100 steel pipes were manufactured. As a result, without using the roundness prediction model M, the average roundness was 10.5 mm and the pass rate was 20%. In the fifth embodiment, the average roundness was improved to 6.3 mm and the pass rate was increased to 69%.

[0200] As a sixth embodiment, similar to the fifth embodiment described above, the U-shaped stamping process is used as the re-setting process. After the C-shaped stamping process and before transitioning to the U-shaped stamping process, the actual values ​​of the plate thickness and yield stress, which are attribute information of the steel plate, are used as inputs to the roundness prediction model M, and the operating parameters of the U-shaped stamping process, O-shaped stamping process, and tube expansion process are re-set. It should be noted that the re-set operating parameter for the tube expansion process is the tube expansion rate, which is re-set within the range of 0.9% to 1.5%. The other re-set operating parameters are the same as in the fifth embodiment. By implementing this sixth embodiment of roundness control, 100 steel pipes were manufactured. As a result, the average roundness was further improved to 5.1 mm, and the pass rate was further improved to 92%.

[0201] [Example 4]

[0202] In this embodiment, Figure 10 In the basic data acquisition unit 110 shown, an operation condition dataset including the following operation parameters is set. The steel pipe that becomes the product after the pipe expansion process is an API grade X42 pipe with a thickness of 44.5 mm and an outer diameter of 1422.4 mm. First, the steel plate's property information is set as a thickness of 45.6 mm and a width of 4295 mm. The yield stress is included as a property information of the steel plate in the operation condition dataset, corresponding to a steel plate with a tensile strength of 500 MPa. It should be noted that the width of the plate is also set after the pre-processing process. Regarding the operation parameters for the C-shaped stamping process, the radius of curvature of the forming surface of the upper die is set to R420 mm. Calculation conditions for end bending processing are set for a 300 mm range from the end in the width direction as the end bending processing width, but this is not included in the input of the roundness prediction model.

[0203] The operating parameters of the U-shaped stamping process are based on the U-shaped stamping process of the Caesar-type U-shaped stamping device, and a finite element model of the corresponding U-shaped stamping process is generated. Regarding the shape information of the U-shaped stamping tool, the base radius (R) is 501–552 mm, and the base radius angle... For a 120-degree angle, calculations corresponding to multiple shape information are performed within the range of 200–300 mm for the horizontal radius r, and this is included in the operating condition dataset. Additionally, as operating parameters for the U-shaped stamping process, the final interval of the U-shaped stamping support is varied within the range of 914 ± 80 mm, and the U-shaped stamping reduction is varied within the range of 858.8 ± 25.4 mm; these are also included in the operating condition dataset for the U-shaped stamping process.

[0204] On the other hand, regarding the operating parameters of the O-ring stamping process, the upper and lower dies are set with an O-ring stamping die radius R702mm, an arc depth of 702mm for the upper die, and an arc depth of 683mm for the lower die. Furthermore, the calculation conditions are modified so that the distance between the uppermost point of the upper die and the lowermost point of the lower die (O-ring stamping reduction) is 1415±5mm, and this is included in the operating condition dataset for the O-ring stamping process. It should be noted that in this embodiment, regarding the tube expansion process, the tube expansion die is assumed to have a shape with a tube expansion die radius of 620mm after being divided into 12 sections in the circumferential direction, and the tube expansion rate is set to 0.9%.

[0205] Regarding the welding process, the behavior of binding the open tube from the surroundings to achieve a zero joint gap is numerically analyzed using the finite element method, and the results are used as the stress / strain state after the welding process. Specifically, for the open tube after the O-ring stamping process, finite element analysis was performed on the behavior of pressing the rollers into a circle with a diameter of 1408.3 mm by positioning nine rollers at the joint gap and at distances of 19, 70, 105, and 154 degrees from the joint gap. However, the heat effect during welding was not considered. The stress / strain state of the open tube with the joint gap at zero after pressing in the rollers was used as the stress / strain state of the steel pipe after the welding process and sent to the finite element model generation unit for the subsequent tube expansion process. It should be noted that the roundness of the steel pipe after the tube expansion process, which is the output of the roundness prediction model, is obtained by dividing the outer diameter shape of the steel pipe after the tube expansion process obtained by the roundness offline calculation unit 112 into 1080 points in the circumferential direction, and using the difference between the maximum diameter Dmax and the minimum diameter Dmin.

[0206] As described above, the roundness prediction model of the steel pipe after the expansion process in this embodiment includes the shape information of the U-shaped stamping tool, the final interval of the U-shaped stamping support, and the U-shaped stamping reduction amount as operating parameters for the U-shaped stamping process, and the O-shaped stamping reduction amount as operating parameters for the O-shaped stamping process. The data required for the finite element analysis of the forming process, including such an operating condition dataset, is provided to the roundness offline calculation unit 112, and the roundness of the steel pipe after the expansion process is calculated. The learning dataset obtained through calculation is accumulated in the database 120, and the roundness prediction model M is generated using the roundness prediction model generation unit 130.

[0207] In the roundness offline calculation unit 112, a two-dimensional plane strain element was obtained by dividing the steel plate, which would become the raw material before the C-shaped stamping process, into 720 sections in the width direction and 18 sections in the thickness direction. Finite element analysis was performed in the order of C-shaped stamping, U-shaped stamping, O-shaped stamping, welding, and tube expansion processes. In this embodiment, the finite element analysis solver used was Abaqus 2019, and the calculation time for each case was approximately 3 hours. During the stage when 300 data points were accumulated in the database 120, a roundness prediction model M was generated. The roundness prediction model M is a machine learning model that includes the operating parameters of the U-shaped stamping process and the operating parameters of the O-shaped stamping process as inputs. For the machine learning model, Gaussian process regression was used, and for the kernel function, the RBF kernel (Radial Basis Function kernel) for evaluating the similarity between variables and the White kernel for functionalizing the influence of noise on the objective variable were used.

[0208] The roundness prediction model M, generated as described above, was used for online processing in the steel pipe manufacturing process. After the O-ring stamping process, the roundness of the steel pipe after the expansion process was predicted. When predicting the roundness of the steel pipe after the expansion process, steel pipes meeting the aforementioned manufacturing conditions were used as the target. Operational data from the steel pipe manufacturing process were used to obtain the shape information of the U-ring stamping tool, the final spacing of the U-ring stamping support, the U-ring stamping reduction, and the O-ring stamping reduction, generating an operational condition dataset that became the input to the roundness prediction model M. Furthermore, as the output of the roundness prediction model M, the predicted value of the roundness of the steel pipe after the expansion process was calculated and compared with the actual value of the roundness of the steel pipe after the expansion process (the actual roundness value). The results showed that the difference between the predicted roundness value output by the roundness prediction model M and the actual roundness value was an average error of 0.3%, with a standard deviation of 4.3%, confirming that the roundness prediction model M accurately predicts the roundness of the steel pipe after the expansion process.

[0209] [Example 5]

[0210] In this embodiment, Figure 10The basic data acquisition unit 110 shown includes an operation condition dataset containing the following operation parameters. The steel pipe that becomes the product after the pipe expansion process is defined as having an API grade of X52, a thickness of 6.4 mm, and an outer diameter of 508.0 mm. First, the steel plate's property information is set as a thickness of 6.4–7.4 mm and a width of 1564 mm. The thickness and yield stress of the steel plate are included in the operation condition dataset as property information, corresponding to steel plates with a tensile strength of 440–640 MPa. It should be noted that the width is set after the pre-processing step. In this embodiment, the process of manufacturing a steel pipe by performing a U-shaped stamping process, an O-shaped stamping process, a welding process, and a pipe expansion process on the steel plate that becomes the raw material is considered. That is, the steel pipe is manufactured without the end bending process of the C-shaped stamping process.

[0211] The operating parameters of the U-shaped stamping process are based on the U-shaped stamping process of the Caesar-type U-shaped stamping device, and a finite element model of the corresponding U-shaped stamping process is generated. Regarding the shape information of the U-shaped stamping tool, the base radius (R) is 210mm, and the base radius angle... The angle is 120 degrees, and the horizontal radius (r) is calculated with a condition of 131 mm. Furthermore, as operating parameters in the U-shaped stamping process, the final interval of the U-shaped stamping support is varied within a range of 154 ± 30 mm, and the U-shaped stamping reduction is varied within a range of 656.6 ± 25.4 mm; these are included in the operating condition dataset for the U-shaped stamping process.

[0212] On the other hand, regarding the operating parameters for the O-ring stamping process, the upper and lower dies are set with an O-ring stamping die radius R251mm, an arc depth of 251mm for the upper die, and an arc depth of 239mm for the lower die. Furthermore, the calculation conditions are modified so that the distance between the uppermost point of the upper die and the lowermost point of the lower die (O-ring stamping reduction) is 501±3mm, and this is included in the operating condition dataset for the O-ring stamping process. It should be noted that in this embodiment, as operating parameters for the tube expansion process, the tube expansion die is based on a die with a shape having a circumferentially divided 10-section tube expansion die radius of 226mm, and the tube expansion rate is set to 1.1%.

[0213] Regarding the welding process, the behavior of binding the open tube from the surroundings to achieve zero joint gap is numerically analyzed using the finite element method, and the results are used as the stress / strain state after the welding process. Specifically, for the open tube after the O-ring stamping process, finite element analysis was performed on the behavior of pressing the rollers into a circle with a diameter of 503.0 mm by positioning seven rollers at the joint gap and at distances of 42, 84, and 126 degrees from the joint gap. However, the heat effect during welding was not considered. The stress / strain state of the open tube with zero joint gap after pressing in the rollers was used as the stress / strain state of the steel pipe after the welding process and sent to the finite element model generation unit for the subsequent tube expansion process. It should be noted that the roundness of the steel pipe after the tube expansion process, which is the output of the roundness prediction model, is obtained by dividing the outer diameter shape of the steel pipe after the tube expansion process obtained by the roundness offline calculation unit 112 into 1080 points in the circumferential direction, and the difference between the maximum diameter Dmax and the minimum diameter Dmin is used.

[0214] As described above, the roundness prediction model for the steel pipe after the expansion process in this embodiment includes the final spacing of the U-shaped stamping support and the U-shaped stamping reduction as operating parameters for the U-shaped stamping process, and the O-shaped stamping reduction as operating parameters for the O-shaped stamping process. Furthermore, the plate thickness and yield stress are included in the operating condition dataset as property information of the steel plate. The data required for the finite element analysis of the forming process, including such an operating condition dataset, is provided to the roundness offline calculation unit 112, and the roundness of the steel pipe after the expansion process is calculated. The learning dataset obtained through calculation is accumulated in the database 120, and a roundness prediction model M is generated using the roundness prediction model generation unit 130.

[0215] In the roundness offline calculation unit 112, a two-dimensional plane strain element was obtained by dividing the steel plate, which would become the raw material, into 720 sections in the width direction and 18 sections in the thickness direction. Finite element analysis was performed in the order of U-shaped stamping, O-shaped stamping, welding, and tube expansion processes. In this embodiment, the finite element analysis solver used was Abaqus 2019, and the calculation time for each case was approximately 3 hours. During the stage of accumulating 250 data points in the database 120, a roundness prediction model M was generated. The roundness prediction model M is a machine learning model that includes the property information of the steel plate, the operating parameters of the U-shaped stamping process, and the operating parameters of the O-shaped stamping process as inputs. Support vector regression was used for the machine learning model, and a Sigmoid kernel was used for the kernel function.

[0216] The roundness prediction model M, generated as described above, was used for online processing in the steel pipe manufacturing process. After the O-ring stamping process, the roundness of the steel pipe after the expansion process was predicted. In predicting the roundness of the steel pipe after the expansion process, steel pipes meeting the aforementioned manufacturing conditions were used as the object. Operational performance data for the steel pipe manufacturing process were obtained, including the steel plate thickness, yield stress, final spacing of the U-ring stamping support, U-ring stamping reduction, and O-ring stamping reduction, generating an operational condition dataset that became the input to the roundness prediction model M. Furthermore, the predicted roundness value of the steel pipe after the expansion process was calculated as the output of the roundness prediction model M and compared with the actual roundness value (actual roundness value). The results showed that the difference between the predicted roundness value output by the roundness prediction model M and the actual roundness value was an average error of 0.2% and a standard deviation of 4.6%, confirming that the roundness prediction model M accurately predicts the roundness after the expansion process.

[0217] [Example 6]

[0218] In this embodiment, Figure 10 In the basic data acquisition unit 110 shown, an operation condition dataset including the following operation parameters is set. The steel pipe that becomes the product after the pipe expansion process is an API grade X65 pipe with a thickness of 44.5 mm and an outer diameter of 914.4 mm. First, the steel plate's property information is set as a thickness of 44.5 mm and a width of 2711 mm. The yield stress is included in the operation condition dataset as a property information of the steel plate, corresponding to a tensile strength of 520–690 MPa. It should be noted that the width is set after the pre-processing step. Regarding the operation parameters for the C-shaped stamping process, the radius of curvature of the forming surface of the upper die is set to R310 mm. Calculation conditions for end bending processing are set for a 195 mm range from the end in the width direction as the end bending processing width, but this is not included in the input of the roundness prediction model.

[0219] The operating parameters of the U-shaped stamping process are based on the U-shaped stamping process of the Kaiser-type U-shaped stamping device, and a finite element model of the corresponding U-shaped stamping process is generated. Regarding the shape information of the U-shaped stamping tool, the base radius (R) is 310mm, and the base radius angle... The angle is 120 degrees, and the horizontal radius (r) is calculated with a condition of 178 mm. Furthermore, as operating parameters in the U-shaped stamping process, the final interval of the U-shaped stamping support is varied within a range of 634 ± 40 mm, and the U-shaped stamping reduction is varied within a range of 782.6 ± 25.4 mm; these are included in the operating condition dataset for the U-shaped stamping process.

[0220] On the other hand, regarding the operating parameters for the O-ring stamping process, the upper and lower dies are set with an O-ring stamping die radius R451mm, an arc depth of 451mm for the upper die, and an arc depth of 438mm for the lower die. Furthermore, the calculation conditions are modified to make the distance between the uppermost point of the upper die and the lowermost point of the lower die (O-ring stamping reduction) 903±4mm, and this is included in the operating condition dataset for the O-ring stamping process. Moreover, in this embodiment, as operating parameters for the tube expansion process, two types of dies with radii of 390mm and 410mm after 12 circumferential divisions are used as the tube expansion die, and the tube expansion rate is varied within the range of 0.5% to 1.1%, and this is included in the operating condition dataset for the tube expansion process.

[0221] Regarding the welding process, the behavior of binding the open tube from the surroundings to achieve zero joint gap is numerically analyzed using the finite element method, and the results are used as the stress / strain state after the welding process. Specifically, for the open tube after the O-ring stamping process, finite element analysis was performed on the behavior of pressing the rollers into a circle with a diameter of 905.3 mm by positioning nine rollers at the joint gap and at distances of 24, 65, 105, and 148 degrees from the joint gap. However, the heat effect during welding was not considered. The stress / strain state of the open tube with zero joint gap after pressing in the rollers was used as the stress / strain state of the steel pipe after the welding process and sent to the finite element model generation unit for the subsequent tube expansion process. It should be noted that the roundness of the steel pipe after the tube expansion process, which is the output of the roundness prediction model, is obtained by dividing the outer diameter shape of the steel pipe after the tube expansion process obtained by the roundness offline calculation unit 112 into 1080 points in the circumferential direction, and the difference between the maximum diameter Dmax and the minimum diameter Dmin is used.

[0222] As described above, the roundness prediction model for the steel pipe after the expansion process in this embodiment includes the final spacing of the U-shaped stamping support and the U-shaped stamping reduction as operating parameters for the U-shaped stamping process, and the O-shaped stamping reduction as operating parameters for the O-shaped stamping process. Furthermore, the yield stress is included in the operating condition dataset as property information of the steel plate, and the expansion die radius and expansion ratio are included in the operating condition dataset as operating parameters for the expansion process. The data required for the finite element analysis of the forming process, including such an operating condition dataset, is provided to the roundness offline calculation unit 112, and the roundness of the steel pipe after the expansion process is calculated. The learning dataset obtained through calculation is accumulated in the database 120, and a roundness prediction model M is generated using the roundness prediction model generation unit 130.

[0223] In the roundness offline calculation unit 112, a two-dimensional plane strain element is set after the steel plate, which will become the raw material before the C-shaped stamping process, is divided into 720 sections in the width direction and 18 sections in the thickness direction. Finite element analysis is performed in the order of C-shaped stamping process, U-shaped stamping process, O-shaped stamping process, welding process, and tube expansion process. The finite element analysis solver used in this embodiment is Abaqus2019, and the calculation time for each case is approximately 3 hours. During the stage of accumulating 300 data points in the database 120, a roundness prediction model M is generated. The roundness prediction model M is a machine learning model that includes the property information of the steel plate, the operating parameters of the U-shaped stamping process, the operating parameters of the O-shaped stamping process, and the operating parameters of the tube expansion process as input. For the machine learning model, a random forest is used, with a maximum depth of 3 for the decision tree level and a maximum number of leaves of 220 in the generated decision tree.

[0224] The roundness prediction model M, generated as described above, was used for online processing in the steel pipe manufacturing process. After the O-ring stamping process, the roundness of the steel pipe after the expansion process was predicted. In predicting the roundness of the steel pipe after the expansion process, steel pipes meeting the aforementioned manufacturing conditions were used as the target. Operational performance data for the steel pipe manufacturing process were obtained, including the yield stress of the steel plate, the final spacing of the U-ring stamping support, the U-ring stamping reduction, and the O-ring stamping reduction. Furthermore, the expansion die radius and expansion rate, which are operational parameters for the expansion process, were obtained from the control computer, generating an operational condition dataset that serves as the input to the roundness prediction model M. Finally, as the output of the roundness prediction model M, the predicted value of the roundness of the steel pipe after the expansion process was calculated and compared with the actual value of the roundness of the steel pipe after the expansion process (the actual roundness value). As a result, the difference between the roundness prediction value output by the roundness prediction model M and the actual roundness value is an average error of 0.4% and a standard deviation of 4.0%, confirming that the roundness prediction model M can accurately predict the roundness after the tube expansion process.

[0225] [Example 7]

[0226] In this embodiment, Figure 10The basic data acquisition unit 110 shown includes an operation condition dataset containing the following operation parameters. The steel pipe that becomes the product after the pipe expansion process is defined as having an API grade of X42, a thickness of 44.5 mm, and an outer diameter of 609.6 mm. First, the steel plate's property information is set as a thickness of 44.5 mm and a width of 1761 mm. The yield stress is included in the operation condition dataset as a property information of the steel plate, corresponding to a tensile strength of 400–600 MPa. It should be noted that the width is set after the pre-processing step. Regarding the operation parameters for the C-shaped stamping process, the radius of curvature of the forming surface of the upper die is set to R190 mm. The calculation conditions for end bending processing are set as the end bending processing width, extending from the end in the width direction for a range of 120–140 mm, and are included in the operation condition dataset.

[0227] The operating parameters of the U-shaped stamping process are based on the U-shaped stamping process of the Caesar-type U-shaped stamping device, and a finite element model of the corresponding U-shaped stamping process is generated. Regarding the shape information of the U-shaped stamping tool, the base radius (R) is 246mm, and the base radius angle... The angle is 120 degrees, and the horizontal radius (r) is calculated with a condition of 160 mm. Furthermore, as operating parameters in the U-shaped stamping process, the final interval of the U-shaped stamping support is varied within a range of 474 ± ​​30 mm, and the U-shaped stamping reduction is varied within a range of 757.2 ± 12.7 mm; these are included in the operating condition dataset for the U-shaped stamping process.

[0228] On the other hand, regarding the operating parameters for the O-ring stamping process, the upper and lower dies are set with an O-ring stamping die radius R301mm, an arc depth of 301mm for the upper die, and an arc depth of 289mm for the lower die. Furthermore, the calculation conditions are modified to make the distance between the uppermost point of the upper die and the lowermost point of the lower die (O-ring stamping reduction) 602±4mm, and this is included in the operating condition dataset for the O-ring stamping process. Moreover, in this embodiment, as operating parameters for the tube expansion process, the tube expansion die is based on a die with a shape having a circumferentially divided 10 sections and a tube expansion die radius of 254mm, and the tube expansion rate is varied within the range of 0.6% to 1.4%, and this is included in the operating condition dataset for the tube expansion process.

[0229] Regarding the welding process, the behavior of binding the open tube from the surroundings to achieve a zero joint gap is numerically analyzed using the finite element method, and the results are used as the stress / strain state after the welding process. Specifically, for the open tube after the O-ring stamping process, finite element analysis was performed on the behavior of pressing the rollers into a circle with a diameter of 905.3 mm by positioning seven rollers at the joint gap and at distances of 36, 84, and 132 degrees from the joint gap. However, the heat effect during welding was not considered. The stress / strain state of the open tube with the joint gap at zero after pressing in the rollers was used as the stress / strain state of the steel pipe after the welding process and sent to the finite element model generation unit for the subsequent tube expansion process. It should be noted that the roundness of the steel pipe after the tube expansion process, which is the output of the roundness prediction model, is obtained by dividing the outer diameter shape of the steel pipe after the tube expansion process obtained by the roundness offline calculation unit 112 into 1080 points in the circumferential direction, and using the difference between the maximum diameter Dmax and the minimum diameter Dmin.

[0230] As described above, the roundness prediction model for the steel pipe after the expansion process in this embodiment includes the final spacing of the U-shaped stamping support and the U-shaped stamping reduction as operating parameters for the U-shaped stamping process, and the O-shaped stamping reduction as operating parameters for the O-shaped stamping process. Furthermore, the yield stress is included in the operating condition dataset as property information of the steel plate, and the end bending processing width is included as an operating parameter for the C-shaped stamping process. Moreover, the expansion rate is included in the operating condition dataset as an operating parameter for the expansion process. The data required for the finite element analysis of the forming process, including such an operating condition dataset, is provided to the roundness offline calculation unit 112, and the roundness of the steel pipe after the expansion process is calculated. The learning dataset obtained through calculation is stored in the database 120, and a roundness prediction model M is generated using the roundness prediction model generation unit 130.

[0231] In the roundness offline calculation unit 112, a two-dimensional plane strain element obtained by dividing the steel plate, which will become the raw material before the C-shaped stamping process, into 720 divisions in the width direction and 18 divisions in the thickness direction is set. Finite element analysis is performed in the order of C-shaped stamping process, U-shaped stamping process, O-shaped stamping process, welding process, and tube expansion process. The finite element analysis solver used in this embodiment is Abaqus2019, and the calculation time for each case is approximately 3 hours. During the stage of accumulating 500 data points in the database 120, a roundness prediction model M is generated. The roundness prediction model M is a machine learning model that includes the property information of the steel plate, the operating parameters of the C-shaped stamping process, the operating parameters of the U-shaped stamping process, the operating parameters of the O-shaped stamping process, and the operating parameters of the tube expansion process as inputs. As a machine learning model, Gaussian process regression with radial basis functions as basis functions is used.

[0232] The roundness prediction model M, generated as described above, was used for online processing in the steel pipe manufacturing process. After the O-forming stamping process, the roundness of the steel pipe after the expansion process was predicted. In predicting the roundness of the steel pipe after the expansion process, steel pipes meeting the aforementioned manufacturing conditions were used as the target. Operational performance data for the steel pipe manufacturing process were obtained, including the yield stress of the steel plate, the end bending width in the C-forming stamping process, the final spacing of the U-forming stamping support, the U-forming stamping reduction, and the O-forming stamping reduction. Furthermore, the expansion rate, which is an operational parameter of the expansion process, was obtained from the control computer, generating an operational condition dataset that became the input to the roundness prediction model M. Finally, as the output of the roundness prediction model M, the predicted value of the roundness of the steel pipe after the expansion process was calculated and compared with the actual value of the roundness of the steel pipe after the expansion process (the actual roundness value). As a result, the difference between the roundness prediction value output by the roundness prediction model M and the actual roundness value is an average error of 0.2% and a standard deviation of 5.0%, confirming that the roundness prediction model M can accurately predict the roundness after the tube expansion process.

[0233] The foregoing has described embodiments using the invention made by the inventors, but the present invention is not limited to the description and drawings that constitute a part of the disclosure of the present invention. That is, all other embodiments, examples, and applications made by those skilled in the art based on the present embodiments are included in the scope of the present invention.

[0234] Industrial availability

[0235] According to the present invention, a method for generating a roundness prediction model for a steel pipe is provided, capable of generating a high-precision roundness prediction model for the steel pipe after the pipe expansion process in the manufacturing process of a UOE steel pipe including multiple processes. Furthermore, according to the present invention, a method and apparatus for predicting the roundness of a steel pipe after the pipe expansion process in the manufacturing process of a UOE steel pipe including multiple processes are provided, capable of generating a high-precision roundness prediction model for the steel pipe. Additionally, according to the present invention, a method for controlling the roundness of a steel pipe and a manufacturing method for producing UOE steel pipes with good roundness are provided.

[0236] Explanation of reference numerals in the attached figures

[0237] 10 C-shaped stamping device

[0238] 11. Transportation agencies

[0239] 12A and 12B stamping mechanisms

[0240] 13. Upper mold

[0241] 13a Forming surface

[0242] 14 Lower mold

[0243] 14a Pressing surface

[0244] 15 Tool Holders

[0245] 16 Hydraulic cylinders

[0246] 17 Clamping Mechanism

[0247] 20 Frames

[0248] 21 Lifting Cylinder

[0249] 22, 37 U-shaped stamping tools (U-shaped punches)

[0250] 23 Suspension components

[0251] 24 Lower bed surface

[0252] 25 strokes

[0253] 26 Sliding Cylinder

[0254] 27 Sliding Block

[0255] 28. Seat frame (pedestal)

[0256] 29 Linkages

[0257] 30 Rotation center

[0258] 31 arms

[0259] 32 Brake Rollers

[0260] 33 Frame Section

[0261] 34 U-shaped bending support

[0262] 35 Lower mold (rocker arm mold)

[0263] 36. Rotation pivot

[0264] 38. Buffer

[0265] 40 Lower punch die

[0266] 41 Upper punch die

[0267] 51. Pipe Expanding Die

[0268] 52 Cone outer circumference

[0269] 53 Pull rod

[0270] 60 arms

[0271] 61a and 61b displacement gauges

[0272] 62 Rotation Angle Detector

[0273] 63 Rotating Arm

[0274] 64a, 64b Press Rollers

[0275] A device for generating a roundness prediction model for 100 steel pipes.

[0276] 110 Basic Data Acquisition Department

[0277] 111 Operational Conditions Dataset

[0278] 112 Offline Roundness Calculation Department

[0279] 112a~112c Finite Element Model Generation Section

[0280] 112d Finite Element Analysis Solver

[0281] 120 Database

[0282] 130 Roundness Prediction Model Generation Unit

[0283] 140 Operating Condition Resetting Section

[0284] 150 host computers

[0285] 160 steel pipe roundness prediction device

[0286] 161 Operation Parameter Acquisition Unit

[0287] 162 Storage Department

[0288] 163 Roundness Prediction Department

[0289] 164 Output Section

[0290] 165 Input Section

[0291] 166 Display Department

[0292] P steel pipe

[0293] S-shaped steel plate.

Claims

1. A method for generating a roundness prediction model for steel pipes, comprising a U-shaped stamping process that forms a U-shaped cross-section from a steel plate using a U-shaped stamping tool, an O-shaped stamping process that forms the U-shaped cross-section into an open pipe, and an expansion process that performs expansion-based forming on the steel pipe formed by joining the width-direction ends of the open pipe together, wherein, include: The basic data acquisition step involves calculating the roundness information of the steel pipe after the tube expansion process by numerical analysis corresponding to a series of manufacturing processes up to the tube expansion process, based on the set values ​​of the operating condition dataset. This calculation is performed multiple times while changing the set values ​​of the operating condition dataset, thereby generating multiple sets of data containing the roundness information of the steel pipe after the tube expansion process corresponding to the operating condition dataset as learning data. The operating condition dataset includes one or more operating parameters selected from the operating parameters of the U-shaped stamping process and one or more operating parameters selected from the operating parameters of the O-shaped stamping process. The series of manufacturing processes includes the U-shaped stamping process, the O-shaped stamping process, and the tube expansion process. The roundness prediction model generation step uses multiple learning datasets generated in the basic data acquisition step to generate a roundness prediction model through machine learning, which takes the operating condition dataset as input data and the roundness information of the steel pipe after the pipe expansion process as output data.

2. The method for generating the roundness prediction model of steel pipes according to claim 1, The basic data acquisition steps include using the finite element method to calculate the roundness information of the steel pipe after the pipe expansion process based on the set values ​​of the operating condition dataset.

3. The method for generating the roundness prediction model of steel pipes according to claim 1 or 2, The roundness prediction model, as part of the operating condition dataset, includes one or more parameters selected from the attribute information of the steel plate.

4. The method for generating the roundness prediction model of steel pipes according to claim 1 or 2, The roundness prediction model, as part of the operating condition dataset, includes one or more operating parameters selected from the operating parameters of the tube expansion process.

5. The method for generating the roundness prediction model of steel pipes according to claim 1 or 2, The manufacturing process of the steel pipe includes a C-shaped stamping process, which is formed by bending the end of the steel plate in the width direction before the U-shaped stamping process. The roundness prediction model, as the operating condition dataset, includes one or more operating parameters selected from the operating parameters of the C-shaped stamping process.

6. The method for generating the roundness prediction model for steel pipes according to claim 1 or 2, The operating parameters of the U-shaped stamping process include one or more of the following: the shape information of the U-shaped stamping tool, the U-shaped stamping reduction amount, the initial interval of the U-shaped stamping support, and the final interval of the U-shaped stamping support.

7. The method for generating the roundness prediction model of steel pipes according to claim 1 or 2, As the machine learning method described, machine learning methods selected from neural networks, decision trees, random forests, Gaussian process regression, and support vector regression are used.

8. A method for predicting the roundness of a steel pipe, comprising: The step of obtaining operating parameters involves obtaining, online, an operating condition dataset set as the operating conditions for the manufacturing process of the steel pipe, as input to the steel pipe roundness prediction model generated by the method of generating the steel pipe roundness prediction model according to any one of claims 1 to 7. and The roundness prediction step involves inputting the operational condition dataset obtained in the operational parameter acquisition step into the roundness prediction model to predict the roundness information of the steel pipe after the pipe expansion process.

9. A method for controlling the roundness of a steel pipe, comprising the following steps: Using the roundness prediction method for steel pipes according to claim 8, before the start of a resetting target process selected from a plurality of forming processing processes constituting the manufacturing process of the steel pipe, the roundness information of the steel pipe after the pipe expansion process is predicted, and based on the predicted roundness information of the steel pipe, at least one or more operating parameters selected from the operating parameters of the resetting target process or one or more operating parameters selected from the operating parameters of the forming processing process downstream of the resetting target process are reset.

10. A method for manufacturing a steel pipe, comprising the step of manufacturing the steel pipe using the roundness control method of claim 9.

11. A device for predicting the roundness of a steel pipe, comprising a U-shaped stamping process for forming a U-shaped cross-section from a steel plate using a U-shaped stamping tool, an O-shaped stamping process for forming the U-shaped cross-section into an open pipe, and an expansion process for forming a steel pipe by joining the width-direction ends of the open pipe together, wherein the roundness of the steel pipe after the expansion process is predicted. have: The basic data acquisition unit performs numerical calculations based on the set values ​​of the operating condition dataset, which are used to calculate the roundness information of the steel pipe after the pipe expansion process through numerical analysis corresponding to a series of manufacturing processes up to the pipe expansion process. This process is repeated multiple times while changing the set values ​​of the operating condition dataset, thereby generating multiple sets of data containing the roundness information of the steel pipe after the pipe expansion process corresponding to the operating condition dataset as learning data. The operating condition dataset includes one or more operating parameters selected from the operating parameters of the U-shaped stamping process and one or more operating parameters selected from the operating parameters of the O-shaped stamping process. The series of manufacturing processes includes the U-shaped stamping process, the O-shaped stamping process, and the pipe expansion process. The roundness prediction model generation unit uses multiple learning data generated in the basic data acquisition unit to generate a roundness prediction model through machine learning, taking the operating condition dataset as input data and the roundness information of the steel pipe after the pipe expansion process as output data. The operation parameter acquisition unit acquires, online, an operation condition dataset set as the operation conditions for the manufacturing process of the steel pipe; and The roundness prediction unit uses the roundness prediction model generated in the roundness prediction model generation unit to predict the roundness information of the steel pipe after the expansion process, which corresponds to the operating condition dataset obtained by the operating parameter acquisition unit, in an online manner.

12. The roundness prediction device for steel pipes according to claim 11, The device includes a terminal unit that receives input information based on user operations and a display unit that displays the roundness information. The operation parameter acquisition unit updates part or all of the operation condition dataset in the steel pipe manufacturing process based on the input information obtained by the input unit. The display unit shows the roundness information of the steel pipe predicted by the roundness prediction unit using the updated operating condition dataset.

Images