JP2017049041A - Shape calculation method of analyte, press correction method of steel sheet, production method of steel sheet and steel sheet - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a shape calculation method of an analyte capable of calculating a shape of an analyte with practical accuracy and at high speed, a press correction method of a steel sheet using the same, a production method of a steel sheet and a steel sheet.SOLUTION: A shape calculation method of an analyte calculates a shape of an analyte using a smoothing spline method, based on sampling points on an analyte surface and data values containing errors corresponding to the respective sampling points, the method comprises a partial sampling point selection step S2 and a shape calculation step S4. In the partial sampling point selection step S2, as a distance from a reference point is longer, a partition area is set larger, as the distance from the reference point is larger, a point group density of the partial sampling points is set lower. In the shape calculation step S4, by a pretreatment matrix formed of a sampling measurement weighting smoothing cardinal spline curve surface group corresponding to all sampling point groups, an unknown parameter matrix is subjected to pretreatment.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a method for calculating the shape of a subject, a method for correcting the press of a steel plate using the method, a method for manufacturing a steel plate, and a steel plate.

  Conventionally, as an apparatus for automatically measuring the shape of a steel sheet, as proposed in Patent Document 1, for example, an apparatus for measuring distance data by multi-axis rotational scanning with a laser beam from a single laser light source (so-called tertiary) There is one that measures the shape of the surface of a steel plate that is stationary on a transport line using a former scanner.

  Patent Document 2 proposes a method for smoothing data in order to remove measurement errors contained in sampling point group data measured by the three-dimensional scanner.

  Here, the method proposed in Patent Document 2 is based on a sampling point weighted smoothing spline input from a sampling point on the surface of an object input from a three-dimensional scanner and a data value including an error corresponding to each sampling point. The shape of the subject is calculated by estimating the curved surface and solving the unknown parameters numerically. However, the method proposed in Patent Document 2 has a problem that the calculation amount becomes enormous with the increase in the number of sampling points, and the calculation becomes difficult.

  Therefore, in Non-Patent Document 1, in order to reduce the amount of calculation at the time of smoothing, a cardinal spline curved surface is calculated approximately using a partial sampling point group having a smaller number of points than the total sampling point group, and this cardinal spline. A method of performing smoothing using a curved surface has been proposed.

JP 2010-155272 A JP 2012-37313 A

R. K. Beatson, J. B. Cherrie and C. T. Mouat, `` Fast Fitting of Radial Basis Functions: Methods Based on Preconditioned GMRES Iteration '', Advances in Computational Mathematics, 11, 1999, p253-270

  Here, in the method proposed in Non-Patent Document 1, when a partial sampling point group is selected, first, a reference point in the sampling point group is determined, and only a sampling point in the vicinity of this reference point is in the vicinity. Only the sampling point and the farthest sampling point are selected as partial sampling points. Therefore, since the cardinal spline curved surface is calculated without considering sampling points other than the nearest and farthest points, there is a problem in accuracy. In the method proposed in Non-Patent Document 1, there is still a problem that the calculation amount becomes enormous as the number of sampling points increases.

  The present invention has been made in view of the above, and has an object shape calculation method capable of calculating the object shape at high speed with practical accuracy, and press correction of a steel plate using the object shape calculation method. It aims at providing a method, a manufacturing method of a steel plate, and a steel plate.

  In order to solve the above-described problems and achieve the object, the object shape calculation method according to the present invention includes a sampling point input from a three-dimensional scanner and data including an error corresponding to each sampling point. In the shape calculation method for calculating the shape of the subject using a smoothing spline method that estimates a regression surface defined as a curved surface that minimizes a predetermined functional from a value, each sampling point is a reference point A partial sampling point selection step of dividing each sampling point by a predetermined division area around each reference point, and selecting one of the sampling points included in the division area as a partial sampling point for each division area The reference point, the partial sampling point group selected from each segmented area, and the size of the segmented area The sampling object weighted smoothing cardinal spline surface corresponding to each sampling point is solved to solve the unknown parameter matrix of the sampling points of the partial sampling points indicated by A shape calculating step for calculating the shape of the partial sampling point selecting step, wherein in the partial sampling point selecting step, the portion is increased as the distance from the reference point is increased by increasing the sectioning area as the distance from the reference point is increased. The point parameter density of sampling points is reduced, and the unknown parameter matrix is preprocessed by a preprocessing matrix composed of smoothing cardinal spline curved surfaces with sampling measure weights corresponding to all sampling point groups in the shape calculation step. It is characterized by doing.

  In the above-described invention, the object shape calculation method according to the present invention is characterized in that the object is a steel plate, a slab, a steel piece, or a steel material before or after forging or during forging.

  In order to solve the above-mentioned problems and achieve the object, the steel sheet press correction method according to the present invention calculates the amount of strain from the shape of the steel plate calculated by the above-described specimen shape calculation method, and the strain The steel sheet is subjected to press correction according to the amount.

  In order to solve the above-described problems and achieve the object, the steel sheet manufacturing method according to the present invention is characterized in that the steel sheet after rolling or heat treatment is press-corrected by the above-described method for correcting the steel sheet press.

  In order to solve the above-described problems and achieve the object, the steel sheet according to the present invention is characterized by being press-corrected by the above-described method for correcting a steel sheet press.

  In order to solve the above-described problems and achieve the object, the steel sheet according to the present invention is manufactured by the method for manufacturing a steel sheet described above.

  According to the present invention, in the smoothing spline method, preprocessing is performed with a preprocessing matrix composed of a smoothing cardinal spline curved surface group with sampling measure weights, so that the shape of the subject can be obtained with practical accuracy and at high speed. Can be calculated.

FIG. 1 is a flowchart showing an example of processing by a subject shape calculation method according to an embodiment of the present invention. FIG. 2 is a diagram showing an example of a sampling point dividing method when a smoothing cardinal spline curved surface with a sampling measure weight is calculated in the object shape calculating method according to the embodiment of the present invention. FIG. 3 is a diagram schematically showing a curve conceptually showing a smoothing cardinal spline curved surface with a sampling measure weight in the object shape calculation method according to the embodiment of the present invention. FIG. 4 is a diagram schematically showing a curve conceptually showing a smoothing cardinal spline curved surface with a sampling measure weight in the method proposed in Non-Patent Document 1. FIG. 5 is a diagram showing plane shape point cloud data of a thick plate measured using a three-dimensional scanner. FIG. 6 is a diagram showing data obtained by smoothing the planar shape point cloud data of the thick plate shown in FIG. 5 with the sampling measure weighted smoothed cardinal spline curved surface according to the present invention. FIG. 7 is a table showing the number of iterations when the planar shape point cloud data of the thick plate shown in FIG. 5 is smoothed in the present invention and the prior art.

  Embodiments of a subject shape calculation method according to the present invention will be described below. In addition, this invention is not limited to the following embodiment. In addition, constituent elements in the following embodiments include those that can be easily replaced by those skilled in the art or those that are substantially the same.

  The object shape calculation method according to the present invention is based on large-scale point cloud data measured by a measuring device such as a three-dimensional scanner, for example, when measuring flatness and shape in the field of thick plate, forging, and shape steel. This is a method for executing flatness and shape smoothing processing practically and at high speed. Specifically, the object shape calculation method according to the present invention is based on a sampling point on the surface of the object (for example, a steel plate) input from a three-dimensional scanner and a data value including an error corresponding to each sampling point. The shape of the subject is calculated using a smoothing spline method that estimates a regression surface defined as a surface that minimizes the functional.

  The object shape calculation method according to the present invention includes, for example, various storage devices such as a CPU, a ROM, a RAM, a hard disk, a CD-ROM, a communication device, an output device, an input device, an interface device, etc. It can be implemented using hardware resources.

  In the following, first, the techniques proposed in Patent Document 2 and Non-Patent Document 1 serving as the basis of the present invention will be described, and then the present invention will be described.

In Patent Document 2, m sampling points arbitrarily distributed on the coordinate system (x, y)
And the data value including the error corresponding to each sampling point
A sampling measure weighted smoothing spline method for estimating a regression surface f (x, y) from the above has been proposed. In this method, a functional П that defines a smoothed spline curved surface f (x, y) that is a regression surface is given by the following equation (1).

Here, in the above formula (1),
It is.

Further, the sampling measure satisfies the following formula (2).

  Here, A in the above formula (2) is the area of a region (sampling region) on the coordinate system (x, y) including all sampling points. For example, when the sampling points are uniformly present in the sampling region, the sampling measure is given by the following equation (3).

  The smoothed spline curved surface is defined as a curved surface that minimizes the functional П in the above equation (1). Further, in the integrand of the right-hand side integral term of the above equation (1), the following equation (4) is obtained when l is an odd number, and the following equation (5) is obtained when l is an even number.

Then, in Patent Document 2, the unknown parameters {d _M} by solving the simultaneous linear equations represented by the following formula (6) _numerically, by obtaining the {c}, determine the sampling measure weighted smoothing spline surface, large It has been proposed to smooth scale point cloud data.

Here, the number of multiplications when the above equation (6) is solved directly is m ³ . Therefore, when a large number of sampling points related to the object surface are input from the three-dimensional scanner, the number of unknown parameters {d _M }, {c} becomes enormous and calculation is difficult.

  In order to numerically solve simultaneous linear equations with tens of thousands of unknown parameters in this way, iterative solving is more effective than direct solving. Therefore, it is preferable to apply the iterative solution to the above equation (6). However, preconditioning is important when applying the iterative solution method. If this preprocessing is not appropriate, the number of iterations (Iteration) increases, or convergence becomes impossible.

Here, for example, simultaneous linear equations shown in the following formula (7)
The preprocessing is performed by multiplying both sides of the above equation (7) by the preprocessing matrix [Q] from the left, and the preprocessing is expressed by the following equation (8). In Equation (7), [A] is a constant matrix, {x} is an unknown parameter, and {b} is a constant vector.

The preprocessing matrix [Q] in Equation (8) is preferably closer to the inverse matrix [A] ⁻¹ of the matrix [A]. That is, if the preprocessing matrix [Q] = [A] ⁻¹ , the unit matrix is included on the left side of the above equation (8), and the calculation becomes easy. Further, the number of multiplications when the simultaneous linear equations are solved by the iterative method is n _I m ² , and it can be seen that if the number of iterations n _I is small, the number of multiplications can be considerably reduced as compared with the direct solution method.

Here, as described above, in order to approximate the preprocessing matrix [Q] to the inverse matrix [A] ⁻¹ of the matrix [A], for example, a method using a cardinal spline curved surface can be considered. A cardinal spline curved surface is a spline in which, when a certain reference point is defined in each sampling point, the value at the reference point is 1 and the values at other sampling points (sampling points other than the reference point) are 0. It is a curved surface. A specific example of this cardinal spline curved surface will be described later (see FIGS. 3 and 4).

  When a cardinal spline curved surface is used, the following formula (9) is derived from the above formula (6). The solution of the following equation (9) becomes an unknown parameter of the smoothing cardinal spline curved surface with the sampling measure weight with respect to the i-th reference point.

Here, in the above formula (9),

From the above equation (9), the unknown parameter matrix of the smoothing cardinal spline curved surface with sampling measure weight with all sampling points as reference points is represented by the following equation (11). In the following formula (11), [I _{m × m} ] on the right side is a unit matrix of m rows and m columns.

  Here, the unknown parameter matrix shown in the above equation (11) satisfies the requirements of the preprocessing matrix, and the preprocessing matrix is expressed by the following equation (12).

However, as shown in the above equation (12), obtaining the unknown parameter matrix of the smoothing cardinal spline curved surface with the sampling measure weight with all sampling points as reference points is substantially the first on the left side of the above equation (6). Is equivalent to calculating the inverse of the matrix (number of multiplications m ³ ), does not contribute to reducing the amount of calculation, and increases the amount of calculation instead (number of multiplications m ³ + n _I m ² ). .

  Therefore, in Non-Patent Document 1, partial sampling with fewer points than the total sampling point group as a contrivance to reduce the calculation amount of the cardinal spline curved surface for the case where the smoothing parameter γ is 0 (interpolation problem). A method of calculating a cardinal spline curved surface approximately using a point cloud has been proposed.

From the above equation (9), the unknown parameter of the sampling measure weighted smoothed cardinal spline surface for the partial sampling point group S _i (S _i ⊂M) including the reference point i is expressed by the following equation (13).

Unknown parameter obtained by equation (13) above
Is the same as the number of points in the partial sampling point group, and is smaller than the number m of all the sampling point groups.

An element that does not correspond to the partial sampling point group is set to 0, and an unknown parameter corresponding to the partial sampling point group is set.
Unknown parameter using
Is defined from the above equation (12), the preprocessing matrix becomes the following equation (14).

Here, the multiplication number of the above equation (13) is s ³ (s is the average of the points of the partial sampling point group), and the multiplication number of the above equation (14) is ms ³ . As a result, the number of multiplications when the equation (14) is applied as a preprocessing matrix and the equation (13) is solved by an iterative solution method is ms ³ + n _I m ² . Accordingly, it can be seen that if the number of points of the partial sampling point group can be reduced, the number of multiplications can be considerably reduced as compared with the case where the above equation (6) is solved directly.

Furthermore, in Non-Patent Document 1, FMM (Fast Multipole Method) is used as means for reducing the number of multiplication operations on the left side of Equation (6), and theoretically the number of multiplications is ms ³ + n _I mlogm. It is reduced to. Thus, by eliminating the term m ^{2 in} the number of multiplications, the number of multiplications becomes the same level as algorithms such as fast Fourier transform and quick sort, and it can be applied practically.

  However, in the technique proposed in Non-Patent Document 1, as described above, the cardinal spline curved surface is calculated without considering sampling points other than the nearest and farthest points. There was still a problem that the amount of calculation became enormous as the number of points in the group increased.

  In order to solve the above problems, the present inventors have devised a shape calculation method capable of calculating the shape of a subject (particularly, a planar shape) from large-scale point cloud data at a practical level at high speed. The contents of the present invention will be described below with reference to FIGS.

  In the object shape calculation method according to the present invention, when the object shape is calculated using the smoothing spline method, first, as shown in FIG. Data values including sampling points on the sample surface and errors corresponding to the respective sampling points are input (step S1 in FIG. 1).

  Next, as a partial sampling point selection step, the sampling point group is divided and a partial sampling point is selected (step S2 in FIG. 1). Specifically, in this step, as shown in FIG. 2, each sampling point is set as a reference point, each sampling point is divided by a predetermined division area around each reference point, and each division area is divided into the division area. Is selected as a partial sampling point. FIG. 2 visualizes the partial sampling point selection process in an easy-to-understand manner, and the position of the sampling point and the shape and size of the segmented area are merely examples. In FIG. 2, the points indicated by double circles are reference points, the points indicated by black circles are neighboring sampling points, the points indicated by medium black double circles are remote sampling points, and points indicated by normal circles. Are other sampling point groups.

  Here, the reference point is a sampling point arbitrarily selected from each sampling point. In this step, a partial sampling point group is selected for each reference point using each sampling point as a reference point. For example, when the number of sampling point groups is 10,000, the partial sampling point group is selected 10,000 times using each sampling point as a reference point.

  The neighborhood sampling point group is a partial sampling point group constituted by sampling points in the vicinity of the reference point. The far sampling point group is a partial sampling point group constituted by sampling points far from the reference point. The other sampling point group is a sampling point group that is not selected as a partial sampling point group, and means a sampling point that is not used in the calculation of the cardinal spline curved surface.

  In this step, as shown in FIG. 2, each sampling point is divided by a predetermined dividing area with the reference point as the center. At that time, the classification is made finer in the vicinity of the reference point, and the classification is made coarser as the distance from the reference point increases. That is, for sampling points in the vicinity of the reference point, each sampling point is surrounded by a segmented region, and each is selected as a partial sampling point. On the other hand, for sampling points far from the reference point, a plurality of sampling points are collectively surrounded by a segmented area, and only one of the sampling points included in the segmented area is selected as a partial sampling point.

  Thus, in this step, as shown in FIG. 2, the farther the distance from the reference point, the larger the segmented region, and the farther the distance from the reference point, the lower the point group density of the partial sampling points. Thereby, the number of partial sampling points can be reduced according to the distance from the reference point. In selecting the partial sampling points, as shown in the figure, sampling points near the center of each segmented area are selected. Note that the area of the segmented area shown in FIG. 2 means the sampling measure of the sampling points described above.

  Here, also in the above-mentioned Non-Patent Document 1, it is introduced that the sampling point group is divided and the partial sampling point is selected. However, in the technique proposed in the same document, for example, the neighboring sampling point group in FIG. (15) and only the farthest far sampling point group (7) are selected as the partial sampling point group. That is, in the present invention, 39 of the 256 sampling points shown in FIG. 2 are selected as the partial sampling points except for the reference point, but in Non-Patent Document 1, only 22 at the maximum are partially selected. Selected as sampling point.

  Next, as a shape calculation step, preprocessing is performed with a sampling card weighted smoothed cardinal spline curved surface, and the unknown parameter is obtained to calculate the shape of the subject (steps S3 and S4 in FIG. 1).

  Specifically, in this step, each sampling point is composed of a reference point, a partial sampling point group selected from each segmented area, and a sampling measure of the partial sampling point indicated by the size of the segmented area. By solving the unknown parameter matrix of the corresponding sampling measure weighted smoothed cardinal spline curved surface by an iterative method, each sampling point is smoothed to calculate the shape of the subject. Further, in this step, when solving the unknown parameters, the unknown parameter matrix is preprocessed by a preprocessing matrix composed of the smoothing cardinal spline curved surface groups with sampling measure weights corresponding to all the sampling point groups.

  More specifically, in this step, the unknown parameter matrix of the smoothing cardinal spline curved surface with the sampling measure weight shown in the following equation (15) is preprocessed by the preprocessing matrix shown in the following equation (16).

  Here, the sampling measure-weighted smoothed cardinal spline curved surface used in this step is conceptually shown, as shown in FIG. 3, passing through the reference point or its vicinity, and a partial sampling point group (neighboring sampling point group and far sampling). It can be represented by a curve passing through the vicinity of the point cloud. In the figure, the value at the reference point is 1, and the other partial sampling values are 0. In addition, each point in the figure is illustrated in correspondence with the reference point, the neighborhood sampling point group, and the far sampling point group shown in FIG. Further, the horizontal axis in the graph of FIG. 3 indicates the distance from the reference point.

  As shown in FIG. 3, this curve does not necessarily pass through the reference point and other partial sampling points. However, as shown in the figure, the distance between the intersection of the curve and the horizontal axis and the partial sampling point in the vicinity thereof is configured to decrease as the distance from the reference point increases. Further, this curve is configured to pass as much as possible, for example, at the farthest partial sampling point.

  The distance between the intersection of such curves and the partial sampling point in the vicinity thereof is related to the area of the segmented region shown in FIG. As shown in the figure, as the distance from the reference point increases, the area of the segmented area, that is, the sampling measure increases, and the point group density of the partial sampling points decreases. Therefore, in FIG. 2, it is expressed that weighting is performed according to the distance of the partial sampling point with respect to the reference point (sampling measure weighting) on the smoothing cardinal spline curved surface with the sampling measure weight.

  If the cardinal spline curved surface proposed in Non-Patent Document 1 is conceptually shown in the same manner as in FIG. 3, it can be represented by a curve as shown in FIG. 4, for example. As shown in the figure, in Non-Patent Document 1, only the sampling points in the vicinity of the reference point (neighboring sampling point group) and the farthest sampling point (farthest far sampling point group) are selected as the partial sampling points. Thus, it becomes a conceptually different curve from the smoothing cardinal spline curved surface with the sampling measure weight according to the present invention.

  Hereinafter, the description will proceed to the preprocessing matrix. The preprocessing matrix represented by the above equation (16) corresponds to a sampling card weighted smoothed cardinal spline curved surface shown in FIG. 3 calculated for each sampling point and superposed thereof. For example, when the number of sampling points is 10,000, a preprocessing matrix represented by the above equation (16) is obtained by superimposing 10,000 million sampling measure weighted smoothed cardinal spline curved surfaces.

In the object shape calculation method for performing the above-described processing, the partial sampling point group can be uniquely determined through the partial sampling selection step described above. The average of the points of the partial sampling points selected in this partial sampling point selection step is logm. As a result, the number of multiplications when the equation (16) is applied as a preprocessing matrix and the equation (15) is solved by the iterative solution method is log ³ m + n _I mlogm, which is smaller than that of the prior art. In the present invention, the number of partial sampling points increases as the number of samplings increases. However, the increase is more gradual than in Non-Patent Document 1 and is not fatal.

  Thus, according to the shape calculation method of the subject according to the present invention, in the smoothing spline method, the preprocessing is performed by using the preprocessing matrix composed of the smoothing cardinal spline curved surface group with the sampling measure weight. The shape of the subject can be calculated with high accuracy and high speed.

  Hereinafter, the present invention will be described more specifically with reference to examples. In this embodiment, actual large-scale point cloud data input from a three-dimensional scanner is smoothed using each of the method according to the present invention and the method according to the prior art, and the number of iterations is compared. . The conventional technique taken up in the present embodiment is a technique in which Non-Patent Document 1 is applied to Patent Document 2 described above (however, sampling measure weight is not considered).

  FIG. 5 shows the point cloud data of the planar shape of the thick plate measured using a three-dimensional scanner. Here, in this example, a thick plate having a plate length of 6 m and a plate width of 2 m was used, and the measurement range on the z axis was ± 5 mm. Note that the scale of the thick plate and the measurement range are examples. The total number of point cloud data shown in FIG. 2 is 25691. However, if a three-dimensional scanner is used, the point cloud data can be acquired in about 30 seconds.

  Here, since the point cloud data shown in FIG. 5 includes a measurement error of about ± 2 mm, the measurement error is removed using each of the method according to the present invention and the method according to the prior art. In the method according to the present invention, smoothing is performed using a sampling measure-weighted smoothed cardinal spline curved surface based on the above equations (15) and (16). Further, in the method according to the prior art, smoothing is performed using a smoothed cardinal spline curved surface that does not consider sampling measure weights.

  FIG. 6 shows the result of smoothing by the sampling measure weighted smoothed cardinal spline curved surface according to the present invention. Since the total number of sampling points in FIG. 5 is 25691, it is practically impossible to calculate by the direct solution based on the above equation (6) due to restrictions on memory capacity and calculation time. The number of sampling points that can be calculated by the direct solution method is, for example, about 3000.

  FIG. 7 shows a case where a smoothed cardinal spline curved surface not considering sampling measure weights is used from the point cloud data shown in FIG. 5 (conventional technology) and a case where a smoothed cardinal spline curved surface with sampling measure weights is used (present invention). The number of iterations of calculation in the iterative method is shown. Here, as an iterative solution, a GMRES method (Generalized Minimal Residual Method) was used with reference to Non-Patent Document 1. In addition, in FIG. 1 and no. The point cloud data shown in FIG. It was created by thinning out 3 sampling points.

As shown in FIG. 7, in the present invention, the number of iterations is almost constant even if the number of sampling points increases, but in the prior art, it can be seen that the number of iterations increases as the number of sampling points increases. The result shown in the figure means that the present invention can achieve the theoretical multiplication number log ³ m + n _I mlogm. Further, in the conventional technique, for example, when the number of sampling points reaches several hundred thousand, the calculation load due to the increase in the number of iterations becomes fatal, and practical calculation becomes impossible.

  The method for calculating the shape of the subject according to the present invention has been specifically described above with reference to modes and examples for carrying out the invention. However, the gist of the present invention is not limited to these descriptions, and claims Should be interpreted broadly based on the description of the scope. Needless to say, various changes and modifications based on these descriptions are also included in the spirit of the present invention.

  For example, in FIG. 2, when the remote sampling point group is divided, four or sixteen sampling points are grouped into one, but the size of the divided area (number of sampling points to be divided) is However, the present invention is not limited to that shown in FIG. Further, although the segmented area is shown as a square in the figure, the segmented area may have a rectangular or other polygonal shape.

  In addition to the steel plate, the specimen shape calculation method according to the present invention may use steel materials, slabs, steel slabs, and the like before and after forging or during forging as specimens. That is, the present invention measures various shapes of steel materials before and after forging or during forging, slabs produced by continuous casting, steel slabs, etc., in addition to the use of calculating the planar shape of a steel sheet after rolling, cooling and heat treatment. -It can be used when calculating.

  The object shape calculation method according to the present invention measures the flatness when a steel plate after rolling, cooling and heat treatment is placed on a cooling bed (cooling bed), for example, and is online when there is no problem with the flatness. If there is a problem with the flatness, it is possible to use it for the purpose of adjusting the heat treatment conditions by feedback. Or it is also possible to utilize for the use which test | inspects the flatness of a steel plate at the time of product shipment.

  The subject shape calculation method according to the present invention can also be used, for example, when the steel sheet is press-corrected offline. In particular, it can be used when shape correction of thick steel plates is performed by pressing. When shape correction is required for a thick steel plate, it is effective and often practiced to correct the shape online using a so-called cold leveler, that is, using a roller leveler in the cold. However, when the shape of the thick steel plate is so bad that it cannot be loaded into the leveler, it is necessary to perform press forming offline. In this case, the amount of strain is calculated from the shape of the steel sheet calculated by the above-described method for calculating the shape of the subject, and the steel sheet is press-corrected according to the amount of strain. Conventionally, the operator directly measured the strain amount of the steel sheet and corrected the press according to the strain amount. By using the present invention, the strain amount is automatically calculated without using the operator. can do.

  Moreover, the press correction method of an above described steel plate can also be utilized when manufacturing a steel plate. In this case, the steel sheet after rolling or heat treatment is press-corrected by the above-described method for correcting the steel sheet press to produce a steel sheet.

Claims (6)

A smoothing spline method that estimates a regression surface defined as a curved surface that minimizes a predetermined functional from sampling points on the surface of an object input from a 3D scanner and data values that include errors corresponding to each sampling point In the shape calculation method for calculating the shape of the subject by using,
Each sampling point is set as a reference point, each sampling point is divided by a predetermined dividing area around each reference point, and for each divided area, one of the sampling points included in the divided area is set as a partial sampling point. A partial sampling point selection step to select;
Weighted sampling measure corresponding to each sampling point, comprising the reference point, a partial sampling point group selected from each segmented region, and a sampling measure of the partial sampling point indicated by the size of the segmented region Calculating a shape of the subject by smoothing each sampling point by solving an unknown parameter matrix of a smoothed cardinal spline curved surface, and
In the partial sampling point selection step, the farther the distance from the reference point is, the larger the division area is, and the farther the distance from the reference point is, the lower the point group density of the partial sampling point is,
The shape calculation step includes pre-processing the unknown parameter matrix with a pre-processing matrix composed of a smoothing cardinal spline curved surface group with sampling measure weights corresponding to all sampling point groups. Calculation method.   The method for calculating the shape of an object according to claim 1, wherein the object is a steel plate, a slab, a steel piece, or a steel material before or after forging or during forging.   A method for correcting the press of a steel sheet, comprising calculating a strain amount from the shape of the steel plate calculated by the method for calculating the shape of an object according to claim 1, and performing press correction of the steel plate in accordance with the strain amount. .   A method for producing a steel sheet, comprising: pressing a steel sheet after rolling or heat treatment by the method for straightening a steel sheet according to claim 3.   A steel plate that has been press-corrected by the method for correcting a press of a steel plate according to claim 3.   A steel plate manufactured by the method for manufacturing a steel plate according to claim 4.