Bending control device, bending device, and bending method

Abstract

To provide a technology for correcting by using a measurement result of a flexure-molded long member.SOLUTION: A flexure molding control device comprises: a movement control part for control the movement of a flexure tool, which flexure-molds a long member by being pressed against one face of the long member; a position determination part, which determines a flexure tool position by using a prediction model in which a candidate of a flexure tool position for indicating a position of the flexure tool at the time of flexure molding is mapped with a prediction curvature of a long member after flexure molding by the candidate; a measurement part, which measures an actual curvature that is the curvature of a long member actually flexure-molded by a flexure tool position determined by the position determination part; an error computing part, which computes a curvature error of an actual curvature and a prediction curvature matching the flexure tool position at which the actual curvature has been obtained; and a correction part, which corrects a matching relation of each candidate in a prediction model with a prediction curvature by each candidate by using the curvature error.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a bending control device, a bending device, and a bending method. [Background technology]

[0002] Various techniques have been proposed for improving the accuracy of machine tools that shape workpieces into desired shapes. For example, Patent Document 1 discloses a technique for suppressing dimensional changes in the cross section associated with bending in a manufacturing device that bends long members to manufacture curved members, by correcting the three-dimensional shape of the portion of the long member that will be bent later based on measurements of the three-dimensional shape of the bent portion. Patent Document 2 also discloses a technique for determining the occurrence of chatter vibrations that degrade machining accuracy in a numerical control device that numerically controls a machine tool by inputting an inference data set to an inference unit that has a trained inference model that has undergone machine learning. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2013-252567 [Patent Document 2] International Publication No. 2020 / 208893 Summary of the Invention [Problem to be solved by the invention]

[0004] The technology disclosed in Patent Document 1 assumes that the degree of bending when bending a long member is constant and that the cross section of the long member to be bent is constant. In other words, the technology disclosed in Patent Document 1 cannot be applied when the degree of bending or the cross section before bending of the bent portion of the long member (the portion to be measured) differs from the degree of bending or the cross section of the portion to be bent later. For this reason, there is room for improvement in the technology for performing correction using the measurement results of the bent long member. Furthermore, the technology disclosed in Patent Document 2 determines the occurrence of chatter vibration in a machine tool, but does not take into consideration the bending of the long member.

[0005] The present invention has been made to solve at least part of the above-mentioned problems, and aims to improve the technology for performing corrections using measurement results of a bent long member. [Means for solving the problem]

[0006] The present invention has been made to solve at least part of the above-mentioned problems, and can be realized in the following aspects.

[0007] (1) One aspect of the present invention provides a bending control device that includes: a movement control unit that controls movement of a bending tool that is pressed against one surface of a long member to bend the long member; a position determination unit that determines a bending tool position using a prediction model that associates candidate bending tool positions representing positions of the bending tool during bending with predicted curvatures of the long member after bending using the candidate positions; a measurement unit that measures an actual curvature, which is the curvature of the long member actually bent using the bending tool positions determined by the position determination unit; an error calculation unit that calculates a curvature error between the actual curvature and the predicted curvature corresponding to the bending tool position from which the actual curvature was obtained; and a correction unit that uses the curvature error to correct the correspondence between each of the candidates in the prediction model and the predicted curvature for each candidate.

[0008] This configuration uses the curvature error between the measured actual curvature and the predicted curvature corresponding to the bending tool position at which the actual curvature was obtained to correct the correspondence between each candidate bending tool position in the prediction model and the predicted curvature of the elongated workpiece after bending using each candidate. Therefore, the curvature error calculated using the actual curvature at the candidate bending tool position used in the actual bending can be used to correct not only the correspondence between the candidate bending tool position used in the actual bending and the predicted curvature at that candidate, but also the correspondence between other candidate bending tool positions different from the candidate bending tool position and the predicted curvature at that candidate. Therefore, the next time the workpiece is actually bent using the other candidate bending tool position, the curvature error is likely to be smaller. In other words, this configuration improves not only the accuracy of the next bending at the bending tool position at which the actual curvature was obtained, but also the accuracy of the next bending at other bending tool positions. This configuration therefore improves the technique for performing corrections using the measurement results of a bent elongated workpiece.

[0009] (2) In the bending forming control device of the above form, a storage unit is further provided that stores the actual curvature and the predicted curvature corresponding to the bending tool position at which the actual curvature was obtained in correspondence with each other, and the correction unit may use the actual curvature and the predicted curvature stored in the storage unit to correct the correspondence between each of the candidates in the prediction model and the predicted curvature for each of the candidates. With this configuration, the storage unit can store not only the actual curvature measured immediately before correcting the prediction model and the predicted curvature corresponding to that actual curvature, but also actual curvatures measured in the past and the predicted curvatures corresponding to those actual curvatures. Therefore, the actual curvatures and predicted curvatures stored in the storage unit can be used to correct the prediction model, improving the accuracy of the correction of the prediction model. As a result, the accuracy of bending a long workpiece into a desired shape (curvature) can be improved.

[0010] (3) In the bending control device of the above form, the measurement unit measures a partial actual curvature, which is the actual curvature of a portion of the long member that has been bent during bending, the error calculation unit calculates a curvature error between the partial actual curvature and the predicted curvature corresponding to the bending tool position where the partial actual curvature was obtained, the correction unit uses the curvature error calculated based on the partial actual curvature to correct the correspondence between each of the candidates in the prediction model and the predicted curvature for each of the candidates, and the movement control unit may move the bending tool to the bending tool position determined using the corrected prediction model when performing bending of a portion of the long member for which the partial actual curvature has been measured that has not yet been bent. According to this configuration, the prediction model is corrected using a curvature error calculated based on the partial actual curvature of the bent portion of the long member that is currently being bent. The corrected prediction model is then used to bend the remaining portion of the long member. Therefore, the corrected prediction model can be used during the bending of the long member to reduce the curvature error in the bending of the remaining portion of the long member. Therefore, even during the bending process, the accuracy of bending the remaining portion of the long member into the desired shape (curvature) can be improved.

[0011] (4) In the bending forming control device of the above form, the measurement unit measures the section actual curvature, which is the actual curvature in each section into which the long-shaped member is divided after bending forming is completed, the error calculation unit calculates the curvature error between the section actual curvature and the predicted curvature corresponding to the bending tool position from which the section actual curvature was obtained, the correction unit uses the curvature error calculated based on the section actual curvature to correct the correspondence between each of the candidates in the prediction model and the predicted curvature for each of the candidates, and the movement control unit may move the bending tool to the bending tool position determined using the corrected prediction model when performing bending forming of a new long-shaped member. According to this configuration, the prediction model is corrected using a curvature error calculated using the actual sectional curvatures for each section into which the long member has been divided after bending. Then, the corrected prediction model is used to perform bending of a new long member. The actual sectional curvatures obtained by measuring the long member after bending tend to be more reliable than the actual local curvatures obtained by measuring the bent portion of the long member during bending. Therefore, since the prediction model is corrected using a curvature error calculated based on the highly reliable actual sectional curvatures, a prediction model with high prediction accuracy can be constructed. As a result, the accuracy of bending a new long member into a desired shape (curvature) can be improved.

[0012] (5) In the bending forming control device of the above form, the error calculation unit calculates, in addition to the curvature error between the actual section curvature and the predicted curvature corresponding to the bending tool position at which the actual section curvature was obtained, the curvature error between the actual partial curvature measured when the long-shaped member was in the middle of bending after bending was completed and the predicted curvature corresponding to the bending tool position at which the actual partial curvature was obtained, the correction unit uses each curvature error calculated based on the actual partial curvature and the actual section curvature to correct the correspondence between each of the candidates in the prediction model and the predicted curvature for each of the candidates, and the movement control unit may move the bending tool to the bending tool position determined using the corrected prediction model when performing bending of a new long-shaped member. According to this configuration, the curvature error calculated based on the actual partial curvature and the actual sectional curvature is used to correct the correspondence between each candidate bending tool position in the prediction model and the predicted curvature of the long workpiece after bending using each candidate. Therefore, since the prediction model is corrected using each curvature error calculated based on various actual curvatures measured at different times, a prediction model with high prediction accuracy can be constructed.

[0013] (6) In the bending forming control device of the above form, the correction unit may assign a weight to the curvature error calculated based on the section actual curvature greater than a weight to the curvature error calculated based on the partial actual curvature, thereby correcting the correspondence between each of the candidate bending tool positions in the prediction model and the predicted curvature. According to this configuration, when correcting a prediction model, the curvature error calculated based on the actual section curvature, which tends to be highly reliable, can be reflected more strongly than the curvature error calculated based on the actual partial curvature, making it possible to construct a prediction model with high prediction accuracy.

[0014] (7) According to one aspect of the present invention, there is provided a bending apparatus, the bending apparatus including the bending control device according to the above aspect and the bending tool. This bending device makes it possible to precisely form a long member into a desired shape (curvature).

[0015] The present invention can be realized in various forms, for example, in the form of a bending forming control device, a bending forming method, a system including a bending forming control device, a computer program executed in these devices, methods, and systems, a server device for distributing the computer program, a non-transitory storage medium on which the computer program is stored, etc. [Brief explanation of the drawings]

[0016] [Figure 1] 1 is an explanatory diagram illustrating the configuration of a bending control device according to a first embodiment. [Figure 2] FIG. 10 is an explanatory diagram showing a state in which a long member is being bent. [Figure 3] FIG. 10 is an explanatory diagram showing an example of a prediction model. [Figure 4] 10 is a flowchart illustrating an example of a procedure for a prediction model construction process. [Figure 5] FIG. 10 is an explanatory diagram showing a portion on a bent elongated member. [Figure 6]FIG. 10 is an explanatory diagram illustrating an example of a target radius of curvature at a portion on an elongated member. [Figure 7] FIG. 10 is an explanatory diagram showing an example of a corrected prediction model. [Figure 8] FIG. 10 is an explanatory diagram showing an example of a corrected prediction model. [Figure 9] FIG. 10 is an explanatory diagram showing the position of a bending tool when bending a long member. [Figure 10] FIG. 10 is an explanatory diagram showing the measurement of the curvature of a long member after bending. [Figure 11] FIG. 10 is an explanatory diagram showing an example of a corrected prediction model. [Figure 12] FIG. 10 is an explanatory diagram showing the measurement results of a bent long member. [Figure 13] FIG. 10 is an explanatory diagram showing the shape of the long member after bending. [Figure 14] FIG. 10 is an explanatory diagram showing the shape of the long member after bending. [Figure 15] FIG. 10 is an explanatory diagram showing the maximum error in the shape of a long member after bending. DETAILED DESCRIPTION OF THE INVENTION

[0017] First Embodiment Fig. 1 is an explanatory diagram illustrating the configuration of a bending control device 1 according to one embodiment of the present invention. As shown in Fig. 2, which will be described later, the bending control device 1 controls a series of steps for bending a tubular long member LG, which is the object of processing, by pressing a bending tool 33 against the long member LG while feeding the long member LG in the longitudinal direction. The bending control device 1 includes a CPU 10, an input unit 21, a display unit 23, and a storage unit 25.

[0018] The input unit 21 is composed of a keyboard and mouse that accepts input from an operator and transmits the input information to the CPU 10. The display unit 23 is a monitor that can display various images, and displays various information in accordance with information transmitted from the CPU 10. The storage unit 25 stores data such as a prediction model, actual curvature, and predicted curvature, which will be described in detail later.

[0019] The CPU 10 executes a program stored in a read-only memory (ROM) (not shown) by loading the program into a random access memory (RAM). The CPU 10 functions as a position determination unit 11, a movement control unit 13, a measurement unit 15, an error calculation unit 17, and a correction unit 19.

[0020] FIG. 2 is an explanatory diagram showing a state in which a long-shaped member LG is being bent. FIG. 2 illustrates mutually orthogonal X, Y, and Z axes. In FIG. 2, the long-shaped member LG is fed in the feed direction indicated by the white arrow (the +X-axis direction in FIG. 2) by a feed tool (not shown) while being movably held between multiple holding tools 31 arranged opposite to each other. A bending tool 33, located on the +X-axis side of the holding tool 31, is configured to be movable along the Z-axis direction. The position of the bending tool 33p indicates the position of the bending tool 33 at its initial position before starting to move in the +Z-axis direction. The bending tool 33 moves toward the +Z-axis direction and is pressed against one surface of the long-shaped member LG (the surface facing the -Z-axis direction in FIG. 2), thereby bending the long-shaped member LG fed in the +X-axis direction. The bending tool 33 and the bending control device 1 of FIG. 1 can be collectively referred to as a bending device. A measuring mechanism 35 is configured to be movable along the X-axis direction and measures the curvature of the long-shaped member LG.

[0021] Returning to the explanation of Figure 1, the following describes each function of the CPU 10. The CPU 10 functions as the position determination unit 11, and determines the bending tool position using a prediction model in which candidates for the bending tool position, which indicate the position of the bending tool 33 during bending, are associated with the predicted curvature of the long member LG after bending using the candidates.

[0022] FIG. 3 is an explanatory diagram showing an example of a prediction model M1. The vertical axis in FIG. 3 indicates the bending tool position, and the horizontal axis in FIG. 3 indicates the curvature (predicted curvature). The same applies to the vertical and horizontal axes in FIGS. 7, 8, and 11, which will be described later. As shown in FIG. 3, when one of the curvature and the bending tool position is input and applied to the prediction model M1, the other of the curvature and the bending tool position corresponding to the input is determined. In this way, when the CPU 10, functioning as the position determination unit 11, receives an input of a target shape (curvature) from the operator via the input unit 21 (e.g., an input of the curvature indicated by point D1 in FIG. 3), it applies the input as a predicted curvature to the prediction model, thereby calculating and determining the bending tool position for bending the long-shaped workpiece LG to the predicted curvature. The predicted curvature and bending tool position are displayed on the display unit 23 for the operator to confirm. Then, when the operator inputs permission to bend the long-shaped workpiece LG using the displayed bending tool position, the CPU 10 functions as the movement control unit 13, which will be described later. On the other hand, when the CPU 10, functioning as the position determination unit 11, receives input of a bending tool position from an operator via the input unit 21 (for example, input of the bending tool position indicated by point D1 in FIG. 3), it determines the input as the bending tool position for bending the long workpiece LG and applies the input to a prediction model to calculate a predicted curvature of the long workpiece LG when bent at that bending tool position. The predicted curvature and bending tool position are displayed on the display unit 23 for the operator to confirm. Then, if the operator confirms the displayed predicted curvature and then inputs permission to bend the long workpiece LG at the determined bending tool position, the CPU 10 functions as the movement control unit 13, which will be described next.

[0023] The CPU 10 functions as the movement control unit 13 to control the movement of the bending tool 33. Specifically, the CPU 10 bends the long material LG by moving the bending tool 33 to the bending tool position when bending is permitted and pressing it against one surface of the long material LG. During bending, in addition to controlling the bending tool position, the CPU 10 also controls the feed amount of the long material LG that is fed in the feed direction (the +X-axis direction in FIG. 2 ) with the bending tool 33 positioned at that bending tool position.

[0024] Next, CPU 10 functions as measurement unit 15 to measure the actual curvature of the long member LG that has actually been bent using the bending tool position determined by position determination unit 11. Specifically, CPU 10 measures the actual curvature of the long member LG that has actually been bent by moving measurement mechanism 35 along the X-axis direction. Each time this actual curvature is measured, it is associated with the predicted curvature that corresponds to the bending tool position from which the actual curvature was obtained, and the results are stored in storage unit 25.

[0025] Next, the CPU 10 functions as the error calculation unit 17 to calculate the curvature error between the actual curvature measured by the measurement mechanism 35 and the predicted curvature corresponding to the bending tool position where the actual curvature was obtained. Here, the predicted curvature corresponding to the bending tool position where the actual curvature was obtained is the predicted curvature on the prediction model that corresponds to the bending tool position where the long member LG was bent to the actual curvature. The curvature error is the difference between the predicted curvature when bending is performed at a candidate bending tool position corresponding to the predicted curvature so that the bending condition of the long member LG matches the predicted curvature and the curvature (actual curvature) of the long member LG that is actually bent.

[0026] Next, the CPU 10 functions as the correction unit 19 to correct the prediction model using the curvature error calculated when functioning as the error calculation unit 17. The correction of the prediction model will be described in detail later.

[0027] 4 is a flowchart showing an example of the procedure for the prediction model construction process. The prediction model construction process is a process for constructing a prediction model with high prediction accuracy by correcting the prediction model using a curvature error calculated each time bending is repeated on a long-shaped member LG. In the bending control device 1 of this embodiment, in the prediction model construction process, a prediction model is constructed by actually bending a long-shaped member LG. The prediction model construction process is started by a start input from an operator via the input unit 21.

[0028] When the predictive model construction process starts, CPU 10 first receives input from the worker (step S10). At this time, the input received by CPU 10 from the worker via input unit 21 is the target shape (curvature) or the bending tool position. Next, CPU 10 functions as position determination unit 11, and determines the bending tool position from the information input by the worker, with reference to a predictive model (e.g., predictive model M1 illustrated in FIG. 3) stored in storage unit 25 (step S20). The determined bending tool position is displayed on display unit 23.

[0029] Next, CPU 10 determines whether bending is permitted (step S25). Specifically, CPU 10 determines whether bending is permitted depending on whether input permitting bending is received from the worker via input unit 21. If bending is not permitted (step S25: NO), CPU 10 executes the process of step S10 again.

[0030] If bending is permitted (step S25: YES), the CPU 10 functions as the movement control unit 13, and moves the bending tool 33 to the bending tool position determined during the processing of step S20 and presses it against one surface of the long member LG, thereby bending the long member LG (step S30). The feed amount of the long member LG at this time may be constant, or may be set at any timing before the processing of step S30 is executed.

[0031] Next, CPU 10 functions as measurement unit 15 to measure the actual curvature of the long member LG bent by the processing of step S30 (step S40). After measuring the actual curvature, CPU 10 functions as error calculation unit 17 to calculate the curvature error between the actual curvature and the predicted curvature corresponding to the bending tool position from which the actual curvature was obtained (step S50). After calculating the curvature error, CPU 10 functions as correction unit 19 to correct the prediction model (step S60). At this time, the prediction model updated by the correction is stored in storage unit 25.

[0032] Next, CPU 10 determines whether to continue bending (step S65). Specifically, CPU 10 determines whether to continue bending depending on whether input to continue bending has been received from the operator via input unit 21. If bending is to be continued (step S65: YES), CPU 10 executes the process of step S10 again, receives new input from the operator, and then executes processing using the corrected prediction model in step S20. If bending is not to be continued (step S65: NO), CPU 10 ends the prediction model construction process.

[0033] Fig. 5 is an explanatory diagram showing portions P1, P2, and P3 on a bent long member LG. The portions P1, P2, and P3 are portions that are bent by repeating the series of steps S10 to S60 in the prediction model construction process three times. Note that in Fig. 5, portion P1 is the first section to be bent.

[0034] Figure 6 is an explanatory diagram illustrating the target radii of curvature for portions P1, P2, and P3 on the long-shaped member LG. The vertical axis of Figure 6 represents the radius of curvature, and the horizontal axis of Figure 6 represents the feed amount of the long-shaped member LG. Of the feed amount of the long-shaped member LG, the range of 0 to 360 mm corresponds to portion P1, and the ranges of 360 to 720 mm and 720 to 1080 mm correspond to portions P2 and P3, respectively. As shown in Figure 6, the target radii of curvature for portions P1, P2, and P3 are 600 mm, 1200 mm, and 2400 mm, respectively.

[0035] FIG. 7 is an explanatory diagram showing an example of a corrected prediction model M2. The prediction model M2 is a prediction model constructed by correcting the prediction model M1 (shown by a dashed line in FIG. 7) during bending of the portion P1 (the processing of steps S10 to S60 in FIG. 4). The portion P1 was bent using the prediction model M1 at the bending tool position indicated by point D1. However, the curvature of the portion P1 actually bent was not the curvature indicated by point D1 (predicted curvature) but the curvature indicated by point R1 (actual curvature). In this case, in step S50 during bending of the portion P1, the CPU 10 functions as the error calculation unit 17 to calculate the curvature error between the curvature indicated by point R1 (actual curvature) and the curvature indicated by point D1 (predicted curvature), and then functions as the correction unit 19 to correct the prediction model M1 using this curvature error to construct the prediction model M2. That is, the CPU 10 functions as the correction unit 19 to correct the correspondence between each candidate bending tool position in the prediction model and the predicted curvature of the long member LG after bending at each candidate bending tool position. Specifically, as shown in FIG. 7, the CPU 10 corrects the prediction model M1 to construct a prediction model M2, thereby correcting not only the correspondence between the bending tool position indicated by point D1 used in the actual bending and the predicted curvature, but also the correspondence between the bending tool position and the predicted curvature at points on the prediction model M1 other than point D1. Note that a correction coefficient for the proportional term is used in correcting the prediction model M1. The prediction model M2 constructed in this manner is used to perform bending of the portion P2 (the processing of steps S10 to S60 in FIG. 4).

[0036] FIG. 8 is an explanatory diagram showing an example of a corrected prediction model M3. The prediction model M3 is a prediction model constructed by correcting the prediction model M2 (shown by a two-dot chain line in FIG. 8) during bending of the portion P2 (the processing of steps S10 to S60 in FIG. 4). The portion P2 was bent using the prediction model M2 at the bending tool position indicated by point D2. However, the curvature of the portion P2 actually bent was not the curvature indicated by point D2 (predicted curvature) but the curvature indicated by point R2 (actual curvature). Therefore, in step S50 during bending of the portion P2, the CPU 10 corrects the prediction model M2 using the curvature error between the curvature indicated by point R2 (actual curvature) and the curvature indicated by point D2 (predicted curvature) to construct the prediction model M3. Note that the correction of the prediction model M2 uses correction coefficients for the proportional term and the exponential term.

[0037] As explained with reference to Figures 5, 7, and 8, the CPU 10, as the measurement unit 15, measures the partial actual curvature, which is the actual curvature of the portion of the long material LG that has been bent while it is being bent. Referring to Figure 5, when bending of the portion P1 (or portion P2) is completed, the CPU 10 measures the partial curvature of the portion P1 (or portion P2) of the long material LG that has been bent. Furthermore, the CPU 10, as the error calculation unit 17, calculates the curvature error between the measured partial actual curvature (point R1 in Figure 7, point R2 in Figure 8) and the predicted curvature (point D1 in Figure 7, point D2 in Figure 8) corresponding to the bending tool position where the partial actual curvature was obtained, and as the correction unit 19, uses the curvature error to correct the previous prediction model and construct a new prediction model. After constructing the new prediction model, the CPU 10, as the movement control unit 13, moves the bending tool 33 to the bending tool position determined using the new prediction model when bending an unbent portion of the long member LG for which the actual partial curvature has been measured. That is, referring to FIG. 5 , after bending portion P1, bending portion P2 is performed using prediction model M2, and after bending portion P2, bending portion P3 is performed using prediction model M3. In this way, the CPU 10 corrects the prediction model using the curvature error calculated each time partial bending of the long member LG is repeated, thereby improving the prediction accuracy of the prediction model.

[0038] FIG. 9 is an explanatory diagram showing the bending tool position when bending portions P1, P2, and P3 on a long-shaped member LG. The ranges of portions P1, P2, and P3 relative to the feed amount of the long-shaped member LG are the same as those in FIG. 6. The solid line in FIG. 9 indicates the bending tool position when bending portions P1 to P3 into a target shape using the prediction model M1 shown in FIG. 3. The dashed line in FIG. 9 indicates the bending tool position when bending portions P2 and P3 into a target shape using the prediction model M2 shown in FIG. 7. The dashed line in FIG. 9 indicates the bending tool position when bending portion P3 into a target shape using the prediction model M3 shown in FIG. 8. As described above, the CPU 10 corrects the prediction model each time partial bending is repeated on the long-shaped member LG, so that in part P1, bending is performed at a bending tool position determined using the prediction model M1 (solid line), in part P2, bending is performed at a bending tool position determined using the updated prediction model M2 (dashed line), and in part P3, bending is performed at a bending tool position determined using the updated prediction model M3 (dotted line).

[0039] FIG. 10 is an explanatory diagram showing the measurement of the curvature of the long member LG after bending. In FIG. 10, region MR of the long member LG indicates the bent region. Region MR is the combined region of portions P1, P2, and P3 on the long member LG. The CPU 10 also measures the actual curvature of the long member LG after bending. Specifically, the CPU 10 measures the sectional actual curvature, which is the actual curvature of each section into which region MR is divided in the long member LG after bending. In FIG. 10, the sectional actual curvature of each section into which region MR is divided in the long member LG is measured. When this sectional actual curvature is measured, the region MR of the long member LG is fed in the feed direction and the actual curvature is measured while the load from the holding tool 31 and bending tool 33 is removed. Therefore, a more reliable actual curvature can be obtained compared to measuring the region MR while the load is applied. The measurement mechanism 35 measures the actual curvature of the elongated member LG in the region MR by moving along the X-axis direction. The position of the measurement mechanism 35p indicates the position of the measurement mechanism 35 at its initial position before starting to move in the +X-axis direction. The CPU 10 functions as the error calculation unit 17, thereby calculating the curvature error between the measured actual section curvature and the predicted curvature corresponding to the bending tool position at which the actual section curvature was obtained. The CPU 10 functions as the correction unit 19, thereby correcting the previous prediction model and constructing a new prediction model using the curvature error calculated based on the actual section curvature.

[0040] FIG. 11 is an explanatory diagram showing an example of a corrected prediction model M4. The prediction model M4 is a prediction model constructed by correcting the prediction model M3 shown in FIG. 8 (shown by a dashed line in FIG. 11) based on the curvature error calculated using the actual sectional curvature after bending. As described above, after bending of the long-shaped member LG is completed, the CPU 10 measures the actual sectional curvature and then corrects the prediction model using the curvature error calculated based on the actual sectional curvature. When a load is applied from the holding tool 31 and the bending tool 33 to the region MR of the long-shaped member LG (i.e., when the long-shaped member LG is in the middle of being bent), the bending tool position and actual curvature (partial actual curvature) when portions P1 and P2 (shown in FIG. 5) in the region MR are bent are indicated by points R1 and R2 (shown in FIG. 11). On the other hand, when the loads from the holding tool 31 and the bending tool 33 on the region MR are removed (i.e., when the long-shaped member LG has been bent), the bending tool position and actual curvature (actual section curvature) when bending the section corresponding to portions P1 and P2 (shown in FIG. 5) in the region MR are indicated by points R3 and R4 (shown in FIG. 11). Here, since the prediction model at the time the actual section curvature is measured is the prediction model M3, the curvature indicated by points R1 and R2 corresponds to the predicted curvature corresponding to the bending tool position at which the actual section curvature was obtained for the actual section curvature indicated by points R3 and R4. Therefore, after bending is completed, the CPU 10 corrects the prediction model M3 using the curvature error between the actual section curvature indicated by points R3 and R4 and the predicted curvature indicated by points R1 and R2 to construct a prediction model M4. Note that the correction of the prediction model M3 uses correction coefficients for the proportional and exponential terms. After constructing the prediction model M4, the CPU 10, as the movement control unit 13, moves the bending tool 33 to the bending tool position determined using the prediction model M4 when performing bending of a new long-shaped member after the long-shaped member LG.

[0041] FIG. 12 is an explanatory diagram showing the measurement results of a long workpiece LG bent by the bending control device 1. As described in FIG. 6, the target radii of curvature for portions P1, P2, and P3 are 600 mm, 1200 mm, and 2400 mm, respectively. In FIG. 12, "Comparison" indicates a bending control device that uses bending tool positions determined by the prediction model M1 (FIG. 3) for all bending of portions P1, P2, and P3 of the long workpiece LG. In FIG. 12, "Measurement values ​​during bending" indicates the values ​​of the radii of curvature (the reciprocal of the actual curvature of the portion) measured for portions P1, P2, and P3 of the long workpiece LG during bending. In FIG. 12, "Overall shape measurement values" indicates the values ​​of the radii of curvature (the reciprocal of the actual curvature of the portion) measured for sections corresponding to portions P1, P2, and P3 of the long workpiece LG after bending has been completed. In Fig. 12, "first" indicates the first long member LG bent by the bending control device 1. In Fig. 12, "second" indicates the second long member LG bent by the bending control device 1. The three-digit and four-digit numbers shown in Fig. 12 indicate the values ​​of the radius of curvature.

[0042] During the bending of the first long member LG by the bending control device 1, the actual curvature of each bent portion P1, P2, and P3 is measured each time the bending of each portion is completed, and the prediction model is repeatedly updated (corresponding to prediction models M1, M2, and M3). After the bending of the first long member LG by the bending control device 1, the actual curvature of each section corresponding to each portion P1, P2, and P3 is measured, and the prediction model is further updated (corresponding to prediction model M4). Then, when the bending control device 1 bends portion P1 of the second long member LG, the bending tool position determined by the prediction model (corresponding to prediction model M4) updated based on the actual curvature of each section is used. Thereafter, as with the first long member, during the bending of the second long member LG, the actual curvature of each bent portion is measured each time the bending of each portion is completed, and the prediction model is repeatedly updated.

[0043] As shown in Fig. 12, compared to the bending results of the comparative example, the bending results of the bending control device 1 have a reduced error from the target radius of curvature for both the first and second pieces. Furthermore, when comparing the bending results of the first and second pieces using the bending control device 1, the error from the target radius of curvature is generally even reduced for the second piece.

[0044] Figure 13 is an explanatory diagram showing the shape of the long member LG after bending for comparison. The vertical axis of Figure 13 indicates the position on the Z coordinate, and the horizontal axis of Figure 13 indicates the position on the X coordinate. These X and Z coordinates correspond to the coordinates represented by the XYZ axes shown in Figure 2. The dashed line in Figure 13 shows the shape of the long member LG when it is bent to the target radius of curvature. The solid line in Figure 13 shows the shape of the long member LG after bending for comparison.

[0045] FIG. 14 is an explanatory diagram showing the shape of a long member LG after bending using the bending control device 1. The vertical and horizontal axes in FIG. 14 are the same as those in FIG. 13. The dashed line in FIG. 14, like FIG. 13, indicates the shape of the long member LG when it is bent to the target radius of curvature. The solid line in FIG. 14 indicates the shape of the first long member LG after bending using the bending control device 1. The dashed line in FIG. 14 indicates the shape of the second long member LG after bending using the bending control device 1. The shapes of each long member LG shown in FIGS. 13 and 14 also show that bending using the bending control device 1 has a smaller error from the target radius of curvature than bending using the comparative example, and that bending using the bending control device 1 generally has an even smaller error for the second member than for the first member.

[0046] FIG. 15 is an explanatory diagram showing the maximum error of the shape of the long member LG after bending relative to the target radius of curvature. For the long member LG bent by the comparative example, the maximum error relative to the target radius of curvature was 193 mm. On the other hand, for the first long member LG bent by the bending control device 1, the maximum error was 63 mm, and for the second long member LG, the maximum error was 15 mm. The results shown in FIG. 15, as with FIGS. 13 and 14, also show that the error decreases in the order of bending by the comparative example, bending the first long member by the bending control device 1, and bending the second long member by the bending control device 1.

[0047] As described above, the bending control device 1 of the first embodiment uses the curvature error between the measured actual curvature and the predicted curvature corresponding to the bending tool position at which the actual curvature was obtained to correct the correspondence between each candidate bending tool position in the prediction model and the predicted curvature of the long workpiece LG after bending using each candidate. Therefore, the curvature error calculated using the actual curvature at the candidate bending tool position used in the actual bending can be used to correct not only the correspondence between the candidate bending tool position used in the actual bending and the predicted curvature at that candidate, but also the correspondence between other candidate bending tool positions different from the candidate and the predicted curvature at that candidate. Therefore, the next time actual bending is performed using the other candidate bending tool position, the curvature error is likely to be smaller. In other words, the bending control device 1 of the first embodiment can improve not only the accuracy of the next bending at the bending tool position at which the actual curvature was obtained, but also the accuracy of the next bending at other bending tool positions. Therefore, the bending control device 1 of the first embodiment can improve the technique for performing correction using the measurement results of the bent long member.

[0048] Furthermore, in the bending control device 1 of the first embodiment, the prediction model is corrected using a curvature error calculated based on the partial actual curvature of the portion of the long member LG that has been bent during bending. Then, the corrected prediction model is used to perform bending of the portion of the long member LG that has not yet been bent. Therefore, by using the prediction model corrected during bending of the long member LG, it is possible to reduce the curvature error in bending the portion of the long member LG that has not yet been bent. Therefore, even when bending is in progress, it is possible to improve the accuracy of bending the portion of the long member LG that has not yet been bent into a target shape (curvature).

[0049] Furthermore, in the bending control device 1 of the first embodiment, the prediction model is corrected using a curvature error calculated from the actual sectional curvature of each section into which the long member LG is divided after bending. Then, the corrected prediction model is used to perform bending of a new long member LG. The actual sectional curvature obtained by measuring the long member LG after bending tends to be more reliable than the actual local curvature obtained by measuring the portion of the long member LG that has already been bent during bending. Therefore, since the prediction model is corrected using a curvature error calculated based on the highly reliable actual sectional curvature, a prediction model with high prediction accuracy can be constructed. As a result, the accuracy of bending a new long member LG into a desired shape (curvature) can be improved.

[0050] Furthermore, the bending device including the bending control device 1 and the bending tool 33 of the first embodiment can accurately form the long member LG into a desired shape (curvature).

[0051] Second Embodiment The bending control device of the second embodiment differs from the bending control device 1 of the first embodiment in the timing of correcting the prediction model. In the bending control device 1 of the first embodiment, during bending of the long material LG, the actual local curvature of the bent portion is measured each time bending of each portion (e.g., portions P1, P2, P3) is completed, and the prediction model is corrected each time the actual local curvature is measured. Furthermore, even after bending of the long material LG is completed, the actual local curvature of each section corresponding to portions P1, P2, P3 is measured, and the prediction model is further corrected based on the actual local curvature. On the other hand, in the bending control device of the second embodiment, during bending of the long material LG, the actual local curvature of the bent portion is measured each time bending of each portion (e.g., portions P1, P2, P3) is completed, but the prediction model is not corrected in response to the measurement of the actual local curvature (the measured actual local curvature and the predicted curvature corresponding to the actual local curvature are stored in the storage unit 25). In the bending forming control device of the second embodiment, after bending of the long-shaped member LG is completed, the actual section curvatures of the sections corresponding to each of the parts P1, P2, and P3 are measured, and the prediction model is corrected based on the actual section curvatures and the actual section curvatures measured during bending.

[0052] Specifically, in the second embodiment, the CPU 10, as the error calculation unit 17, calculates the curvature error between the measured actual sectional curvature and the predicted curvature corresponding to the bending tool position at which the actual sectional curvature was obtained, as well as the curvature error between the partial actual curvature measured while the long workpiece LG was being bent after bending was completed and the predicted curvature corresponding to the bending tool position at which the partial actual curvature was obtained.The CPU 10, as the correction unit 19, then uses the curvature errors calculated based on the partial actual curvature and the actual sectional curvature to correct the previous prediction model and construct a new prediction model.That is, in the second embodiment, not only the actual sectional curvature measured after bending was completed and the predicted curvature corresponding to that actual curvature, but also the partial actual curvature and the predicted curvature corresponding to that partial actual curvature measured during an earlier bending process and stored in the storage unit 25, are used to correct the prediction model.

[0053] After constructing the new prediction model, the CPU 10, as the movement control unit 13, moves the bending tool 33 to the bending tool position determined using the corrected prediction model when bending a new long workpiece. That is, referring to FIGS. 5 and 7 , the bending control device of the second embodiment uses the prediction model M1 to perform bending of each of the portions P1, P2, and P3, and measures the actual partial curvatures of each portion after bending. After bending, the prediction model M1 is measured to measure the actual sectional curvatures of the sections corresponding to the portions P1, P2, and P3. Then, the prediction model M1 is corrected using the curvature error calculated based on these actual partial curvatures and the actual sectional curvatures. Here, since the prediction model at the time the actual partial curvatures and the actual sectional curvatures are measured is the prediction model M1, the predicted curvature used to calculate the curvature error is the predicted curvature calculated by applying the bending tool positions at which the actual sectional curvatures were obtained and the bending tool positions at which the actual partial curvatures were obtained to the prediction model M1.

[0054] Furthermore, in the second embodiment, the CPU 10 functioning as the correction unit 19 corrects the prediction model by weighting the curvature error calculated based on the actual sectional curvature more heavily than the curvature error calculated based on the actual partial curvature. That is, when building a new prediction model, the curvature error calculated based on the actual sectional curvature is reflected with priority over the curvature error calculated based on the actual partial curvature.

[0055] As in the first embodiment, the bending control device of the second embodiment described above can correct not only the correspondence between the bending tool position of a candidate used in actual bending and the predicted curvature for that candidate, but also the correspondence between the bending tool position of other candidates different from the candidate and the predicted curvature for that other candidate. Furthermore, the bending control device of the second embodiment can use the partial actual curvatures and their corresponding predicted curvatures stored in the storage unit 25, along with the sectional actual curvatures and their corresponding predicted curvatures, to correct the prediction model. In other words, since the prediction model is corrected using each curvature error calculated based on various actual curvatures measured at different points in time, a prediction model with high prediction accuracy can be constructed.

[0056] In addition, in the bending forming control device of the second embodiment, when correcting the prediction model, the curvature error calculated based on the actual section curvature, which tends to be more reliable, can be reflected more strongly than the curvature error calculated based on the actual partial curvature, and therefore, for this reason as well, a prediction model with high prediction accuracy can be constructed.

[0057] <Modification of this embodiment> The present invention is not limited to the above-described embodiment, and can be embodied in various forms without departing from the spirit of the invention. For example, the following modifications are also possible.

[0058] [Variation 1] In the above embodiment, the object to be processed by the bending control device is a tubular elongated member, but this is not limiting. For example, the object to be processed may be a plate-like elongated member.

[0059] [Variation 2] In the above embodiment, the correction coefficient of the proportional term is used only in correcting the prediction model M1, but this is not limiting. For example, in correcting the prediction model, whether to use the correction coefficient of the proportional term or the correction coefficient of the proportional and exponential terms may be determined arbitrarily by an operator.

[0060] [Variation 3] In the first embodiment, during bending of the long member LG, the actual local curvature of the bent portion is measured each time bending of the long member LG is completed, and the prediction model is repeatedly updated. After bending of the long member LG is completed, the actual local curvature of the section corresponding to each portion is measured, and the prediction model is further updated. However, this is not limited to this. For example, the prediction model may be updated only during bending of the long member LG. That is, the actual local curvature of the bent portion may be measured each time bending of the long member LG is completed, and the prediction model may be repeatedly updated using only the curvature error calculated based on the actual local curvature.

[0061] [Variation 4] In the second embodiment, the prediction model is corrected by weighting the curvature error calculated based on the actual section curvature more heavily than the curvature error calculated based on the actual partial curvature. However, this is not limiting. For example, the prediction model may be corrected by weighting the curvature error calculated based on the actual section curvature less heavily than the curvature error calculated based on the actual partial curvature. Alternatively, the prediction model may be corrected by weighting the curvature error calculated based on the actual section curvature equally to the curvature error calculated based on the actual partial curvature. The magnitude of each weight may be adjusted based on the experience of the operator, or may be adjusted based on the correlation between the reliability of the curvature error calculated based on the actual section curvature and the curvature error calculated based on the actual partial curvature.

[0062] [Variation 5] In the above embodiment, the prediction model is constructed by actually bending the long-sized member LG, but this is not limiting. For example, the prediction model may be constructed by bending the long-sized member LG in a forming simulation using a forming simulation model such as LS-DYNA.

[0063] This aspect has been described above based on embodiments and modifications. However, the above-described embodiments are intended to facilitate understanding of this aspect and are not intended to limit this aspect. This aspect may be modified or improved without departing from the spirit and scope of the claims, and equivalents thereof are included in this aspect. Furthermore, if a technical feature is not described as essential in this specification, it may be deleted as appropriate. [Explanation of symbols]

[0064] 1...Molding control device 10...CPU 11...Positioning section 13...Movement control unit 15...Measuring unit 17…Error calculation section 19...Correction section 21...Input section 23...Display section 25…Storage department 31...Holding tool 33...Bending tool 35...Measuring mechanism

Claims

1. A bending control device, a movement control unit that controls movement of a bending tool that is pressed against one surface of the elongated member to bend the elongated member; a position determination unit that determines the bending tool position using a prediction model in which candidates for bending tool positions that represent the positions of the bending tool during bending are associated with predicted curvatures of the elongated member after bending using the candidates; a measuring unit that measures an actual curvature, which is a curvature of the long member actually bent at the bending tool position determined by the position determining unit; and an error calculation unit that calculates a curvature error between the actual curvature and the predicted curvature corresponding to the bending tool position at which the actual curvature is obtained; a correction unit that corrects a correspondence between each of the candidates in the prediction model and the predicted curvature for each of the candidates using the curvature error; the measuring unit measures a partial actual curvature, which is the actual curvature of a portion of the elongated member that has been bent, in a state in which a load is being applied from a holding tool that movably holds the elongated member and the bending tool; the error calculation unit calculates a curvature error between the partial actual curvature and the predicted curvature corresponding to the bending tool position at which the partial actual curvature is obtained; the correction unit corrects a correspondence relationship between each of the candidates in the prediction model and the predicted curvature for each of the candidates using the curvature error calculated based on the partial actual curvature; The movement control unit is a bending forming control device that moves the bending tool to the bending tool position determined using the corrected prediction model when performing bending of a portion of the long-shaped member for which the partial actual curvature has been measured that has not yet been bent.

2. The bending control device according to claim 1, further comprising: a storage unit that stores the actual curvature and the predicted curvature corresponding to the bending tool position from which the actual curvature was obtained, in association with each other; A bending forming control device in which the correction unit corrects the correspondence between each of the candidates in the prediction model and the predicted curvature for each of the candidates using the actual curvature and the predicted curvature stored in the storage unit.

3. The bending control device according to claim 1 or 2, the measuring unit measures an actual section curvature, which is the actual curvature in each section into which the elongated member is divided after bending is completed; the error calculation unit calculates a curvature error between the actual section curvature and the predicted curvature corresponding to the bending tool position at which the actual section curvature is obtained; the correction unit corrects a correspondence relationship between each of the candidates in the prediction model and the predicted curvature for each of the candidates using the curvature error calculated based on the actual section curvature; The movement control unit is a bending forming control device that moves the bending tool to the bending tool position determined using the corrected prediction model when performing bending of a new long-shaped member.

4. The bending control device according to claim 3, The error calculation unit In addition to the curvature error between the actual section curvature and the predicted curvature corresponding to the bending tool position from which the actual section curvature was obtained, calculating a curvature error between the actual partial curvature measured when the elongated member was in the middle of being bent after bending has been completed and the predicted curvature corresponding to the bending tool position at which the actual partial curvature was obtained; the correction unit corrects a correspondence relationship between each of the candidates in the prediction model and the predicted curvature for each of the candidates, using each curvature error calculated based on the partial actual curvature and the interval actual curvature; The movement control unit is a bending forming control device that moves the bending tool to the bending tool position determined using the corrected prediction model when performing bending of a new long-shaped member.

5. The bending control device according to claim 4, The correction unit corrects the correspondence between the candidate bending tool positions and the predicted curvatures in the prediction model by assigning a weight to the curvature error calculated based on the actual section curvature greater than the weight to the curvature error calculated based on the actual partial curvature.

6. A bending device, The bending control device according to any one of claims 1 to 5, a bending tool; and a bending forming apparatus comprising:

7. A bending method, comprising: a movement control step of controlling the movement of a bending tool that is pressed against one surface of the elongated member to bend the elongated member; a position determination process for determining a bending tool position using a prediction model in which candidates for bending tool positions representing positions of the bending tool during bending are associated with predicted curvatures of the elongated member after bending using the candidates; a measuring step of measuring an actual curvature of the elongated member actually bent at the bending tool position determined in the position determining step; an error calculation step of calculating a curvature error between the actual curvature and the predicted curvature corresponding to the bending tool position at which the actual curvature was obtained; a correction step of correcting a correspondence between each of the candidates in the prediction model and the predicted curvature for each of the candidates using the curvature error; In the measuring step, a partial actual curvature is measured, which is the actual curvature of a portion of the elongated member that has been bent, while a load is being applied from a holding tool that movably holds the elongated member and the bending tool; In the error calculation step, a curvature error between the actual partial curvature and the predicted curvature corresponding to the bending tool position at which the actual partial curvature is obtained is calculated, In the correction step, a correspondence relationship between each of the candidates in the prediction model and the predicted curvature for each of the candidates is corrected using the curvature error calculated based on the partial actual curvature; In the movement control process, when performing bending of a portion of the long-shaped member for which the partial actual curvature has been measured that has not yet been bent, the bending tool is moved to the bending tool position determined using the corrected prediction model.

Images