JP5217221B2 - Method for detecting surface defect shape of welded portion and computer program - Google Patents

Description

  The present invention relates to a method and apparatus for detecting a surface defect shape in a welded portion of a metal such as a steel material, and in particular, is a technique suitable for quantitatively detecting the depth and size of the defect.

  For example, after a rectangular steel plate is folded into a U-shape, it is folded into an O-shape, and the opposite sides of the rectangle are butt welded to produce a cross-sectional profile (inner side (upper surface) or outer side of the cross section). (Refers to the outer shape of the lower surface), the rise of the welded portion is called a weld bead (hereinafter referred to as a bead) or a surplus. The boundary between the bead and the base material is called a toe portion. Hereinafter, the bead and the periphery of the toe portion are collectively referred to as a welded portion.

  Incidentally, it is known that a plurality of types of defects occur in the bead. FIG. 3 is a perspective view of the vicinity of the welded portion showing a defective portion called an undercut 111, which is a deep cut generated in the toe portion 11 on the inner surface of the UO steel pipe, together with a bead portion 10 and a base material (steel plate) 12. . FIG. 4 is a sectional view in which the vicinity of the welded portion of the UO steel pipe of FIG. 3 is cut along a cutting line A-A ′. FIG. 5 is a view schematically showing a defective portion called an under bead 101, which is a depression generated in the bead portion 10 on the inner surface of the UO steel pipe, and FIG. 6 is a cutting line near the welded portion of the UO steel pipe in FIG. It is the figure which showed the cross section cut by BB '. Welding defects such as undercuts and underbeads have a significant effect on the strength of the weld and cause cracks in the weld during pipe expansion forming, so the size and shape of the defect must be quantitatively grasped prior to pipe forming. This is important for quality control of UO steel pipes. In addition, for steel materials other than steel pipes, it is often necessary to quantitatively detect the size (width and depth) and shape of defects in the weld.

  By the way, various methods and apparatuses for automatically detecting the shape of the weld have been proposed. Among them, the optical method is often useful because it can measure the shape in a non-contact manner with a relatively simple apparatus. The optical method disclosed in Patent Document 1 is to detect the quality of the weld bead shape. After the weld bead is imaged from the top and side surfaces, the analog signal is converted into a digital signal representing the density of the image, A digital signal processing such as differentiation processing is performed on a digital signal for one line of an image to detect a point at which the shading of the image changes abruptly to be a defect. Since this method does not require a special light source and only uses camera imaging, the apparatus configuration is simple. However, since a difference in brightness at the defect portion is assumed, the defect may not be detected due to a difference in the surface properties of the bead portion and the base material portion which is not originally related to the surface defect. Furthermore, as a matter of course, the depth and shape of the defect could not be detected.

  In addition, a light source image irradiated on the inspected part is captured by an imaging unit such as a camera to obtain an image of the inspected part, and based on the pixel position of the captured light source image in the image, the steel tube As a method for measuring the cross-sectional profile, a method using a light cutting method as disclosed in Patent Document 2 or a triangulation method using a laser distance meter or the like is known. In the following, for convenience of explanation, not only the outer shape of a steel pipe or the like but also the data of the measurement result of the cross-sectional profile of the steel pipe or the like is also referred to as a cross-sectional profile.

  The technique disclosed in Patent Document 2 deals with a method for obtaining the height and rising angle of a bead of an electric resistance welded steel pipe classified as one of welded steel pipes. In general, the surface of the welded portion is rougher than the base metal portion, and the cross-sectional profile obtained by an optical method often has jagged noise. For this reason, in the technique disclosed in Patent Document 2, on the cross-sectional profile of the ERW steel pipe, the region corresponding to the bead portion is divided into two regions on the left and right, and the region of the base material portion outside the bead portion is divided. The cross-sectional profile is approximated in a piecewise manner using different functions, and the height and rise angle of the bead are obtained from the intersections and differential coefficients of these functions. In the case of an electric resistance steel pipe, the bead portion has a gentle slope and a steep portion due to pressure welding, and the bead portion has an irregular shape. Therefore, the approximation is higher when the bead portion is approximated by a plurality of functions.

  On the other hand, unlike the case of an electric resistance welded steel pipe, in the case of a UO steel pipe, the bead portion exhibits a gentle cross-sectional profile due to the difference in the groove shape at the time of welding for welding. When the gentle bead portion is approximated by dividing it into left and right regions as in the method disclosed in Patent Document 2, the approximation degree is good in each of the left and right regions, but the slope of the approximate curve near the apex of the bead where both approximate curves intersect Becomes discontinuous, and on the contrary, the difference from the original cross-sectional profile of the approximate curve increases. Moreover, the UO steel pipe before pipe expansion forming may exhibit a linear cross-sectional profile at the base material portion, and the left and right base material portions may exhibit a step at the butt portion. Therefore, when the surface shape of each of the left and right base metal parts is linear and has a step in the vicinity of the welded part like a UO steel pipe, the base material part is approximated by one function by the method of Patent Document 2. The degree of approximation gets worse. As described above, when the bead portion is gentle and the surface shapes of the left and right base metal portions are linear and have a step, the approximation method according to Patent Document 2 cannot be applied as it is.

  By the way, when the depth of the defect is derived from the cross-sectional profile measured by the optical means as exemplified above, it is necessary to determine the reference line that is the origin of the depth. Since it is not a fixed shape, it is not possible to set a reference line for the depth of the defect in advance. For this reason, when the cross-sectional profile in the width direction is gentle like a thin plate, an approximate curve obtained by polynomial approximation of the cross-sectional profile by the least square method is often used as a reference line for measuring the depth of the defect. Is called. However, when the object to be inspected is a welded steel pipe, the degree of change in the cross-sectional profile differs greatly between the base metal part and the welded part, and the cross-sectional profile cannot be approximated by one function. As described above, the cross-sectional profile may be approximated by a plurality of functions.

  However, when the size of the weld is similar to that of a weld defect such as an underbead, or approximately a fraction of that of the weld bead, the cross-sectional profile is approximated. If the weld defect part is included in the vicinity of the weld, the approximate curve representing the required cross-sectional profile is greatly influenced by the value of the cross-sectional profile of the defect part, and the size and shape of the defect are underestimated. There was a problem that.

JP-A-60-135705 JP 2004-181471 A JP 7-40049 A

  In view of the problems of the prior art described above, the present invention is capable of measuring at least one of the size and shape such as the depth and width of the defect generated in the welded portion with higher accuracy than conventional from the measurement data of the cross-sectional profile. And it makes it a subject to detect simply.

According to a first aspect of the present invention, there is provided a surface defect shape detecting method for a welded portion, wherein a member having a welded portion is used as an inspection object, and a surface defect shape detecting method for a welded portion for detecting a defect on the surface of the welded portion. Including a step of deriving a cross-sectional profile of a predetermined section L in the width direction of the bead of the weld including, a degree-of-change calculation step of deriving a degree of change that is a change in inclination with respect to the width direction of the bead for the cross-sectional profile; Among a plurality of sudden change sections in which the degree of change of the cross-sectional profile is larger than a predetermined threshold, the leftmost and rightmost toes are designated as the leftmost and rightmost sudden change sections of the predetermined section L, respectively. The cross-sectional profile in the inner section M sandwiched between the toe part detecting step to detect as a section section and the detected left toe section section and right toe section section is defined as the inner section. A bead portion approximation step of approximating a first function predetermined by using the data of the cross-sectional profile that does not belong to the sudden change interval of the M, the left outer than the left toe portion section, and the right toe the cross-sectional profile of the outer right from part section, a second function that respective predetermined, and a base metal approximation step of approximating the third function, the intersection between said second function and said first function A left toe point, a toe point calculating step for obtaining an intersection point of the first function and the third function as a right toe point, and the left toe point and the right toe point as the dividing points of the section L, Of the cross-sectional profile, the first function is only extrapolated to the entire section sandwiched between the left toe point and the right toe point, or extrapolated and interpolated, and the second function is Extrapolate to the left outer section from the left toe point, The number is expanded by extrapolating the section to the right outside the right toe point, and the first, second, and third functions are connected to approximate the cross-sectional profile over the entire section L. An approximate cross-sectional profile calculating step for deriving a piecewise approximate curve, a deviation data calculating step for deriving deviation data that is a difference between the cross-sectional profile and the piecewise approximate curve, and a defect size based on the deviation data Or a defect shape calculation step for evaluating at least one of the shapes.

In the method for detecting a surface defect shape of a welded portion according to the second aspect of the present invention, the degree-of-change calculation step performs second-order differential processing on the y-coordinate point sequence S (n) (n is a natural number) that is the cross-sectional profile. The degree of change is derived.

A computer program for detecting a surface defect shape of a welded part according to a third aspect of the present invention is a computer program for detecting a surface defect shape of a welded part that detects a defect on the surface of the welded part using a member having the welded part as an inspection object The process of deriving a cross-sectional profile of a predetermined section L in the width direction of the bead of the weld including the base material portion, and the degree of change that is a change in the inclination of the cross-sectional profile with respect to the width direction of the bead. Derivation degree calculation processing to be derived, and among a plurality of sudden change sections in which the change degree of the cross-sectional profile is larger than a predetermined threshold, a sudden change section on the left outermost side and a sudden change section on the right outermost side of the predetermined section L A toe detection process for detecting the left and right toe sections, and an inner section sandwiched between the detected left toe section and the right toe section, respectively. The cross-sectional profile in M, and the bead portion approximation process for approximating a first function predetermined by using the data of the cross-sectional profile that does not belong to the sudden change interval among the inner section M, the left toe A base material part approximating process for approximating the cross-sectional profile on the left outer side from the section and the right outer side from the right toe section by a predetermined second function and a third function, respectively; A toe point calculation process for obtaining an intersection point of a function and the second function as a left toe point and an intersection point of the first function and the third function as a right toe point, the left toe point and the right toe point With the end point as the division point of the section L, the first function of the cross-sectional profile is extrapolated only over the entire section sandwiched between the left stop point and the right stop point, or extrapolation and interpolation. And the second function is changed to Extrapolate to the left outer section from the toe point, and extend the third function by extrapolating to the right outer section from the right toe point. These first function, second function, and second Approximate section profile calculation processing for deriving a piecewise approximate curve that approximates the section profile over the entire section L by connecting the three functions, and deriving deviation data that is the difference between the section profile and the piecewise approximate curve And a defect shape calculation process for evaluating at least one of the size or shape of the defect based on the deviation data.

  The present invention approximates the cross-sectional profile of the normal part using the measurement data of the cross-sectional profile excluding the defective part on the weld, and uses this approximate curve as a reference line for measuring the depth of the defect. It is possible to detect a defect by evaluating at least one of the size or shape of the surface defect of the welded part with a higher accuracy and more easily than the conventional one, which is approximately the same as the welded part or about a fraction of that of the welded part. Become.

  DESCRIPTION OF EMBODIMENTS Hereinafter, embodiments for carrying out a defect detection method and apparatus for welds according to the present invention will be described in detail with reference to the drawings. In each figure, the same reference numerals are used to indicate the same parts, and the overlapping of the signs is avoided to clarify the description of the figures.

  In this embodiment, a method for deriving a cross-sectional profile of a welded part using an optical cutting method to measure the surface shape of the welded part will be described as an example. The cross-sectional profile of the weld is preferably derived in a non-contact manner by an optical method such as a light cutting method or a laser distance meter, but is measured by another method such as a contact-type shape meter. It is clear that it is also good.

<Capture of cut image>
FIG. 1 is a schematic diagram showing the configuration of an optical cutting method optical system (perspective view) for measuring a cross-sectional profile of a welded portion by a light cutting method and a signal processing system, taking a welded steel pipe as an example to be inspected. is there. 1 shows a welded steel pipe of the object to be inspected (only a part is shown), 2 creates a linear light projection pattern 3 in a direction crossing the inspection part (that is, the region including the bead) which is a welded part of the object to be inspected Projecting means (projecting apparatus) 5, imaging means (imaging apparatus) for copying the projected pattern and outputting image data, and 6, a surface on which the projected pattern is irradiated by image processing the image data A cross-sectional profile deriving means (image processing apparatus) for deriving a cross-sectional profile of the part, and 7 is a signal processing control apparatus for executing arithmetic processing for deriving a defect shape such as a defect size and depth based on the cross-sectional profile. is there. FIG. 1 shows an arrangement in which the light projecting means 2 and the imaging means 5 are placed inside the welded steel pipe 1 to measure the cross-sectional profile of the inner surface. In the method for deriving the cross-sectional profile in this embodiment, for convenience of calculation, a coordinate system is used in which the pipe axis direction of the welded steel pipe 1 of the object to be inspected is the z-axis and the cross section perpendicular to the pipe axis direction is the xy plane. Although the coordinate system is used, the coordinate system may be determined as appropriate in accordance with the shape of the inspection unit when it is implemented. The xy plane is overlapped with the plane 4 generated by the trajectory of light emitted from the light projecting means 2 in a fan shape, and the center direction of the fan-shaped emission spread in the plane 4 is taken as the y axis. In addition, it is preferable in terms of profile measurement to arrange the beads so that they are close to the center of the light projection pattern 3. The position of the origin O of the xyz coordinate is the intersection of the optical axis direction of the photographing means 5 and the xy plane.

  FIG. 2 is a diagram showing a positional relationship between the light projecting means 2, the imaging means 5, and the origin O of the xyz coordinates when the optical arrangement in the optical system for light cutting method shown in FIG. 1 is viewed in the x-axis direction. 51 is the imaging surface of the imaging means 5, Op is the lens center (front principal point position) of the imaging lens (hereinafter simply referred to as a lens) of the imaging means 5, and O is the coordinate axis on the object (object) side of the coordinate axis (Referred to as the object-side origin). The X axis of the imaging surface 51 is taken in parallel with the x axis, and the axis perpendicular to the X axis in the imaging surface 51 is taken as the Y axis. Oc is the center of the imaging surface 51 and is referred to as the image side coordinate origin as the coordinate origin of the imaging surface 51XY plane. As described above, an extension of a straight line connecting the image side coordinate origin Oc and the lens center Op passes through the object side coordinate origin O, and the object side x axis and the image side X axis are parallel to each other. Further, the optical axis direction OOc of the imaging means (imaging system) 5 is a predetermined angle θ with respect to the y axis that is the optical axis direction of the light projecting means 1, and it is easy to detect the shape change of the light projection pattern 3 due to the welded part shape. Arrange as follows. For example, when measuring on a horizontal plane with the welded portion of the welded steel pipe facing down, the light projecting means 2 is arranged vertically above the approximate center in the width direction of the welded portion with the horizontal plane as a reference plane. It is convenient for calculation if the coordinate system is set so that the center direction of the emitted light, that is, the y-axis direction is the vertical direction.

  As the light projecting means 2, for example, a laser or lamp is used as a light emitting light source, and light emitted from the light emitting light source is spread and emitted in a fan shape using a cylindrical lens, a prism, or the like. A light source that converges to the linear projection pattern 3 at the irradiation position is used (hereinafter referred to as a linear light source). In addition, for example, a light beam (referred to as a point beam) that converges in a dot shape at an irradiation position using the above-described light-emitting light source using a lens or the like is rotated to a bead 10 of the welded steel pipe 1 by a polygon mirror. A scanning point light source that scans in the x-axis direction orthogonal to the tube axis direction (z-axis) and obtains a light projection pattern 1 with a linear trajectory may be used as the light projecting means 1 (hereinafter referred to as scanning). This is referred to as an expression point light source). That is, when a linear light source is used as the light projecting means 1, 4 in FIG. 1 shows a linear beam spread in a fan shape, and follows the surface shape when irradiated to the curved portion of the welded steel pipe 1 which is an object to be inspected. Thus, a light section image as the light projection pattern 3 is formed. When a scanning point light source is used as the light projecting means 1, 4 in FIG. 1 shows the locus of a point beam scanned in the width direction, and the light projection pattern 3 is a point image at an irradiation position on the welded steel pipe 1. Shows the trajectory. In this application, all the light projection patterns 1 are described as a light cut image.

(First Embodiment)
<Section profile derivation>
For example, an area sensor such as a CMOS camera or a CCD camera is used as the image pickup means 5 to pick up a light projection pattern (light-cut image) 3 on the inspection object and output it as image data. Note that the imaging unit 5 may capture the light section image in synchronization with the light emission timing of the light projecting unit 2. In the cross-sectional profile deriving unit 6 in FIG. 1, image data including position information (position in the image) and luminance information (brightness) of the light section image is taken in from the imaging unit 5. The cross-sectional profile deriving means 6 derives a cross-sectional profile of the object to be inspected based on the image data. The pixel position (relative position information of each pixel in the image) of the light section image 3 is expressed by the barycentric position in the Y-axis direction with the luminance distribution in the Y-axis direction as a weight at each X coordinate on the imaging surface, for example. An example of the specific method will be described below.

  FIG. 11 shows a light section image captured by the image pickup means 5 and taken into the image input unit, and a luminance distribution in the C-C ′ transverse direction at a certain horizontal position n on the image. The lower left of the captured image in which the light-cut image is captured is taken as the origin, and the coordinate axes on the image corresponding to the X axis and the Y axis on the imaging surface are defined as the IX axis and the IY axis, respectively. If the number of pixels in the IX direction is IH and the number of pixels in the IY direction is IV, the captured image is composed of IH × IV pixels.

  If the luminance at the pixel (n, k) (n = 1, 2,..., K = 1, 2,...) In the captured image is Br (n, k), the IY axis at the horizontal position n The center-of-gravity position G (n) in the IY-axis direction using the luminance distribution in the direction as a weight is obtained as follows.

However, Σ is the sum for the pixels in the IY axis direction, but is within a predetermined range centered on the position where the luminance in the IY axis direction is maximum, and the luminance Br (n, k) is equal to or greater than a predetermined threshold value Δ (n,
k) is 1, and 0 if this condition is not satisfied.

  The above processing is performed at each position n (n = 0, 1,..., IH−1) on the IX axis, and the coordinates (n, G (n)) can be used as a cross-sectional profile on the image. G (n) is not an integer, but is left as a real number without being converted into an integer for conversion to XY coordinates on the imaging surface.

The cross-sectional profile (X (n), Y (n)) on the XY plane on the imaging surface corresponding to the cross-sectional profile (n, G (n)) on the captured image is on the image with the number of pixels IH × IV. Cross-sectional profile (n,
G (n)) is multiplied by pixel sizes dX and dY in the X direction and Y direction of the imaging surface to obtain the following equation.

  The cross-sectional profile (X (n), Y (n)) on the XY plane on the imaging surface was obtained by projecting the cross-sectional profile (x (n), y (n)) on the real plane xy plane. Therefore, the latter can be obtained by coordinate conversion of the former. The coordinate conversion method determined by the optical arrangement of the light projecting means 2, the imaging means 5, and the object to be inspected 1 as shown in FIG. 2 will be described in detail.

From the lens imaging formula in geometrical optics, if the distance from the lens center Op to the object-side origin O in FIG. 2 is L and the focal length of the lens is f, the object-side origin O is for the image captured on the imaging surface 51. The magnification (lateral magnification) M of the object is represented by M = (L−f) / f. Strictly speaking, at the point Q at the height of y in the y-axis direction from the object side origin O, the exact magnification M ′ is M ′ = M · (L−f -Y ・ cosθ) /
(L−f). However, in the case of the present embodiment, it is assumed that the field of view width, that is, the width of the region to be imaged is several times the bead width, and the range of the bead and the surrounding undulation y within the field width is small, so the magnification M May be M = (L−f) / f.

  The projection (x, ysinθ) of the point Q (x, y) on the xy plane onto the plane perpendicular to the imaging plane passing through the origin O is M times the corresponding image (X, Y) on the imaging plane XY. Therefore, a coordinate conversion equation from the image side (imaging surface side) to the object side (inspected object side) is obtained as the following equation.

  From (2), (3), (4), and (5), the cross-sectional profile coordinates (x (n), y (n The conversion to)) is as follows.

  Further, in the present embodiment, the cross-sectional profile deriving unit 6 recognizes the brightness from the luminance value distribution of the captured image obtained by the imaging unit 5, and based on the luminance value distribution information, the light section image Is also controlled such as the irradiation time of the light projecting means 1, the exposure time of the imaging means 5, and the exposure timing.

<Detection of irregularities on surface defects>
In the following, the surface profile unevenness, size, and shape performed by each data processing unit constituting the signal processing control device 7 after the cross-sectional profile deriving means 6 derives the measured value of the cross-sectional profile of the welded steel pipe 1. A derivation procedure and a derivation method to be detected will be described.

FIG. 7 shows the outline of each data processing configuration for detecting irregularities, sizes, and shapes of surface defects, and a flowchart of the flow of each data processed in the signal processing control device 7. Here, with respect to the set of discrete points constituting the cross-sectional profile of digital = data obtained as described above, the coordinates of each point are expressed as a coordinate point sequence (x (n), y (n)) (n = 0 ,
1, 2,..., N−1: N is represented by the number of data constituting the cross-sectional profile N = IH), and the y coordinate point sequence y (n) is also represented by S (n). Hereinafter, description will be made with reference to FIGS. In the figure, the coordinate point sequence, which is a set of discrete points, is represented by a continuous line because the interval is narrow.

<Outline of processing flow>
First, as described above, the cross-sectional profile deriving means 6 uses the coordinate point sequence (x (n), S) of the cross-sectional profile of a predetermined section in the width direction of the bead of the welded part (bead and its peripheral part) including the base material part. (n)) is derived as a cross-sectional profile (S601, cross-sectional profile deriving step). It is preferable that the predetermined section includes a bead and a range equal to or larger than the bead width on both sides of the bead in order to perform the function approximation described below with high accuracy. Thereafter, the following processing is executed.

  (S701) In the degree-of-change calculation step, the degree-of-change point sequence F (n) representing the degree of change in the inclination of the y-coordinate point sequence S (n) of the derived cross-sectional profile with respect to the bead width direction (that is, n) calculate. As the degree of change of the y coordinate point sequence S (n), for example, F (n) obtained by applying a secondary differential filter used in the field of digital signal processing to the y coordinate point sequence S (n) can be taken. The absolute value of the degree-of-change point sequence F (n) takes a large value at the sudden change portion where the change of the y-coordinate point sequence S (n) is large.

  As an example of this second-order differential filter, 2a + 1 discrete points including S (n) centered on 2a + 1 y-coordinate point sequence S (n) are expressed by the following equation: There is a differential filter. Here, a is set to an appropriate value according to the degree of change to be detected.

  However, if the y coordinate point sequence S (n), which is a raw measurement value, is used, the influence of noise is large. Therefore, as a pre-processing of the second-order difference processing, S (n) is once smoothed and further the second-order difference processing is performed. Various methods can be considered, such as setting the degree of change F (n) to be the value obtained after execution, or F (n) for the point sequence obtained by smoothing as post-processing after the second-order difference processing. It is done. There is also an example in which processing equivalent to the second-order difference is performed by combining the first-order difference and smoothing as in Patent Document 3 above. There is also a method such as Savitzky-Golay method that can simultaneously realize smoothing and differentiation with one filter. Here, what applied the process equivalent to the secondary differentiation in a continuous function with respect to the cross-sectional profile which consists of a discrete point will be interpreted in a broad sense, and will be called a 2nd-order difference process.

  (S702) In the toe portion detection step, a peak search of the degree-of-change point sequence F (n) is performed from the left and right outer sides of the bead, and, for example, as shown in FIG. The section is detected as two left and right sections (denoted as toe section sections) IL and IR corresponding to the left and right toe sections.

(S703) Using the cross-sectional profile data (x (n), y (n)) in the section corresponding to the inner bead portion sandwiched between the left and right toe portions IL and IR in the bead portion approximation step. The bead portion is approximated by least square approximation. However, data belonging to a sudden change interval in which the absolute value of F (n) is larger than a predetermined value is not used for the least square approximation. Assuming a function form of the first function f ₁ (x) suitable for approximating the cross-sectional profile of the normal bead part without defects, the first function is determined by the least square method using only the cross-sectional profile data of the normal part. To do. The first function form for approximating the normal bead part and the method of the least square method will be described in detail later.

(S704) Cross section profile data (x (n), y (n) in the range WL (left side) and WR (right side) which are the base metal parts outside the two sections IL and IR of the toe part by the base metal part approximating step. )) To approximate the left and right base metal parts by least square approximation. Assuming a function form of the second and third functions f ₂ (x) and f ₃ (x) suitable for approximating the cross-sectional profiles of the left and right base metal parts, the second and third functions are obtained by the least square method. decide. The second and third function forms that approximate the left and right base metal parts and the method of least squares will be described in detail later.

  (S705) In the toe point calculation step, the intersection of the first approximate function and the second approximate function is the left toe point A (reference numeral 13), and the intersection of the first approximate function and the third approximate function is right-stopped. The respective coordinates are calculated as the end point B (reference numeral 14). In addition, a toe point means the typical point which shows the position of a toe end part (welding toe end part).

  (S706) In the approximate cross-section profile calculation step, the first and second defined by the section I1 sandwiched between the left and right toe points, the section I2 on the left outer side from the left toe point, and the right outer section I3 from the right toe point, respectively. Using the third function, an approximate function value T (n) in the x coordinate point sequence x (n) corresponding to the y coordinate point sequence S (n) is obtained as follows.

  That is, in this step, the first, second, and third functions are respectively set to the left outside section I2 from the left toe point, the right outside section I3 from the right toe point, and the left and right toe parts, if any. Extrapolate or interpolate between sudden intervals between intervals to determine the value of each function. Then, T (n) is derived as a piecewise approximate curve.

  (S707) In the deviation data calculation step, deviation data D (n) = S (n) −T (n) between S (n) and T (n) is calculated.

  (S708) In the defect shape calculation step, an area where the magnitude of the deviation data D (n) is equal to or larger than a predetermined threshold is detected as a defect, and the shape is evaluated.

  The procedure and processing results of an example in which the above processing flow is applied to a normal part having no defect, an undercut, and an underbead will be specifically described with reference to the drawings.

<Normal part>
FIG. 8 schematically shows the processing result of a normal part having no defect in an easy-to-understand manner. FIG. 8A shows a cross-sectional profile having the extracted S (n) as a y-coordinate point sequence. FIG. 8 (b) schematically shows a y-coordinate point sequence F (n) obtained by second-order differential processing of S (n), and F (n) is large at the toe where the shape changes. Indicates the value. In FIG. 8 (b), a section having a peak equal to or greater than a predetermined threshold th of F (n) is searched and detected from the left and the right to be the left and right toe sections IL and IR. In the inner region WC sandwiched between IL and IR, approximation is performed using the first function by the least square method using S (n). In the left and right regions WL and WR outside the WC, approximation is performed by the second function and the third function, respectively, by the least square method.

  Next, intersection points of curves represented by a first function that approximates the bead portion and second and third functions that approximate the left and right base metal portions are obtained as left and right toe points A and B, respectively. The section to the left of the left toe point is the second function, the section to the right of the right toe point is the third function, and the section between the left and right toe points is the first function. The y coordinate point sequence on the piecewise approximation function shown in 8 (c) is defined as T (n).

  The difference D (n) between the y-coordinate point sequence S (n) shown in FIG. 8 (a) representing the raw cross-sectional profile and the y-coordinate point sequence T (n) on the approximate function shown in FIG. 8 (c). ) = S (n) −T (n), and finally, deviation data D (n) shown in FIG. 8D is obtained. In the case of the normal part, the absolute value of D (n) is small and is equal to or less than the threshold value U corresponding to the defect to be detected, so no defect is detected.

<Undercut>
FIG. 9 schematically shows the processing result when there is an undercut in an easy-to-understand manner. FIG. 9A shows a cross-sectional profile having the extracted S (n) as a y-coordinate point sequence. FIG. 9B schematically shows a y-coordinate point sequence F (n) obtained by second-order difference processing using S (n). Since an undercut occurs at the toe, in the case of FIG. 9B, F (n), which is the result of the second-order difference process, shows a larger peak value at the left toe with an undercut.

  In FIG. 9 (b), the left and right toe sections IL and IR having a peak equal to or greater than the threshold value th are searched in the same manner as in the normal part, and IL and IR at the toe part having a large change in cross-sectional profile. By using S (n) in the section excluding, a y coordinate point sequence T (n) on the piecewise approximation function shown in FIG. 9C is obtained.

  When the deviation data D (n) = S (n) −T (n) is obtained in the same manner as in the normal portion in FIG. 9D, D (n) having a large absolute value is obtained in the portion with the undercut. . The undercut can be extracted by using the width w of the section where the absolute value of the deviation data D (n) is equal to or greater than the preset threshold UC and the maximum value h of the absolute value of D (n) as the undercut width and depth, respectively.

<Underbead>
FIG. 10 schematically shows the processing result when there is an underbead in an easy-to-understand manner. FIG. 10A shows a cross-sectional profile having the extracted S (n) as a sequence of y coordinate points. FIG. 10 (b) schematically shows a y-coordinate point sequence F (n) obtained by subtracting S (n) from the second floor, and F (n) corresponding to the underbead in the bead portion other than the toe portion. The part where the absolute value of) shows a large value can be seen.

  After obtaining the toe sections IL and IR in the same way as in the normal section, the outer sections WL and WR are the same until the base material section is approximated by the second and third functions, respectively. The bead portion where the bead exists is obtained by using the cross-sectional profile data S (n) in the sections WC1 and WC2 in which the absolute value of F (n) indicates a small value less than a predetermined threshold value (for example, the absolute value of th '). Approximate by the first function.

  When deviation data D (n) = S (n) −T (n) is obtained in the same manner as in the normal part in FIG. 10D, D (n) having a large absolute value is obtained in a portion where there is an underbead. Underbeads can be extracted by using the width w and the maximum value h of the absolute value of D (n) as the width w and the depth of the section where the absolute value of the deviation data D (n) is equal to or greater than a preset threshold UB.

<Details of function approximation method by least square method>
Next, a function approximation method by the least square method performed by the bead portion approximation circuit that performs the bead portion approximation step (S703) and the base material portion approximation circuit that performs the base material portion approximation step (S704) will be described. The x-axis and y-axis are the steel pipe width direction and height direction, respectively, and the coordinates of the discrete data representing the cross-sectional profile are (xi, yi) and (yi = S (i)). According to the least square method, for example, assuming an n-order polynomial as the form of the approximate function f (x) as follows,

An approximate function can be obtained by solving the following simultaneous equations concerning the coefficients {a _m }, (m = 0, 1,..., N).

  Assuming an arc represented by the following equation as an approximate function:

  An approximate function can be obtained by solving the following simultaneous equations for the coefficients {a, b, c}.

  In the case of UO steel pipe before pipe expansion forming, the cross-sectional profile shows a gradual change in the bead part and a linear change in the base material part, so here the bead part is approximated by a circular arc and the base material part is approximated by a straight line. .

  According to the method of the present invention, since the sudden change portion of the cross-sectional profile at the toe portion and the defect portion is detected and the cross-sectional profile is approximated by a function excluding the sudden change portion, the cross-sectional profile is not affected by the data of the defect portion. Can be approximated. Using the approximate curve data T (n) as a reference line for measuring the depth of the defect, subtracting T (n) from the cross-sectional profile data S (n), the depth of the defect D (n) = S (n) −T ( n) can be obtained accurately.

  1 including the deviation data string obtained in the longitudinal direction by performing binarization processing in which the sign of D (n) is negative and the absolute value is equal to or greater than the defect detection threshold value is 1 and other parts are 0. This part is connected. Defect portions that are two-dimensionally connected in the width direction and the longitudinal direction can be extracted, and at least one of the size or shape of the defect such as width, depth, and length can be correctly recognized.

(Second Embodiment)
The cross-sectional profile deriving unit 6 includes an image capturing board that captures a captured image from the image capturing unit 5 and converts it into digital data, an internal memory, an external recording device such as an HDD or a DVD-RAM, an input device such as a keyboard or a mouse, and the like. A network = interface for transmitting / receiving data to / from an information terminal = an interface, other I / O boards, and a computer = a computer having a display (not shown) can be used.

  Similarly, the signal processing control device 7 can be configured by a computer. Furthermore, the cross-sectional profile deriving means 6 and the signal processing device 7 may be integrated to constitute a single computer = system.

  Software for causing the computer to execute each process for deriving the above-described cross-sectional profile (S601), each process of S701 to S708 performed by the signal processing apparatus 7, and a process for controlling the imaging apparatus. By loading and executing the above-mentioned HDD or built-in memory, a surface defect shape detection device for a desired welded portion can be realized, and a surface defect shape detecting method for a welded portion can be implemented. . The measurement result of the surface defect shape obtained by the above processing may be displayed on an attached computer = display, or may be transmitted to an external welding operation management computer through a network = interface.

  12, 13, and 14 show the processing results when there is a normal part without a defect, an undercut, and an underbead.

  With a camera having a width direction pixel number IH = 656, a height direction pixel number IV = 484, and a pixel size dX = dY = 10 μm, the camera is arranged so that the shooting angle θ = 45 ° and the working distance L = 207 mm. The cross-sectional profile of the inner surface of a steel pipe having an inner diameter of 895 mm was measured. The bead has a width of 20 mm and a height of about 2 mm, the undercut has a width of 1.5 mm and a depth of about 0.5 mm, and the underbead has a width of about 3 mm and a depth of about 0.7 mm.

  (A) of each figure shows the cross-sectional profile derived by the cross-sectional profile deriving means 6 and the second-order difference result which is the degree of change output from the degree-of-change calculating circuit performing the degree-of-change calculating step (S701). ing.

  (b) shows the approximate cross-section profile output from the approximate cross-section profile calculation circuit performing the approximate cross-section profile calculation step (S706) together with the cross-sectional profile.

  (c) shows the deviation data output from the deviation data calculation circuit performing the deviation data calculation step (S707) together with the cross-sectional profile.

  The normal bead without the heel has no unevenness in the deviation data, but the undercut and the underbead can calculate the deviation data corresponding to the dent in the heel. Even under comparatively large wrinkles such as under beads, it is possible to approximate the cross-sectional profile excluding the buttock without being affected by the cross-sectional profile of the buttock, and by calculating the deviation from the original cross-sectional profile, it corresponds to the buttock The concave portion could be accurately detected.

  As described above, the uneven portion corresponding to the defect can be calculated as the deviation data, and if the deviation data is arranged in the longitudinal direction, the shape of the defect can be captured three-dimensionally.

  As mentioned above, although preferred embodiment of this invention was described referring an accompanying drawing, it cannot be overemphasized that this invention is not limited to the example which concerns. It will be apparent to those skilled in the art that various changes and modifications can be made within the scope of the claims, and these are naturally within the technical scope of the present invention. Understood.

It is a figure which shows the outline of the whole structure by the optical cutting method using a linear light source or a point light source which is embodiment of the surface defect shape detection apparatus of the welding part of this invention. It is a figure which shows the optical arrangement | positioning by the optical cutting method using a linear light source or a point light source, and the coordinate system of an object side and an image side in the measuring method of a cross-sectional profile in embodiment of this invention. It is a figure which shows an example of the undercut which arises in a toe part. It is sectional drawing of an example of the undercut produced in a toe part. It is a figure which shows an example of the under bead which arises in a bead part. It is sectional drawing of an example of the under bead produced in a bead part. It is a flowchart of each process in embodiment of the surface defect shape detection method of the welding part of this invention. In embodiment of the surface defect shape detection method of this weld part, it is a figure explaining the process result of the normal part without a defect. In embodiment of the surface defect shape detection method of this welding part, it is a figure explaining the process result in case there exists an undercut. In embodiment of the surface defect shape detection method of this welding part, it is a figure explaining the process result in case there exists an under bead. It is a figure explaining the relationship between the light cut image and luminance distribution in embodiment of the surface defect shape detection method of this welding part. It is a figure explaining the signal processing result of the normal part without a defect in the Example of the surface defect shape detection method of this welding part. In the Example of the surface defect shape detection method of this welding part, it is a figure explaining the signal processing result in case there exists an undercut. In the Example of the surface defect shape detection method of this welding part, it is a figure explaining the signal processing result in case there exists an under bead.

Explanation of symbols

1 Welded steel pipe (inspected object)
2 Projector (Linear laser light source or scanning point laser light source)
3 Light cut image (light projection pattern)
4 Trace of linear laser beam or scanning point beam (in the air)
DESCRIPTION OF SYMBOLS 5 Imaging device 6 Image processing device 7 Signal processing control device 10 Bead part 11 Stop end part 12 Base material (periphery part of bead outer side)
13 Left toe point 14 Right toe point 51 Imaging surface 101 Underbead 111 Undercut S601 Cross-sectional profile derivation step S701 Change degree calculation step S702 Toe part detection step S703 Bead part approximation step S704 Base material part approximation step S705 Toe point calculation step S706 Approximate cross section profile calculation step S707 Deviation data calculation step S708 Defect shape calculation step

Claims (3)

In the method for detecting the surface defect shape of a welded part, which detects a defect on the surface of the welded part, using a member having a welded part as an inspection object,
Deriving a cross-sectional profile of a predetermined section L in the width direction of the bead of the weld including the base material portion;
About the cross-sectional profile, a degree-of-change calculation step for deriving a degree of change that is a change in inclination with respect to the width direction of the bead;
Among a plurality of sudden change sections in which the degree of change of the cross-sectional profile is larger than a predetermined threshold, the leftmost and rightmost toes are designated as the leftmost and rightmost sudden change sections of the predetermined section L, respectively. A toe part detecting step to detect as a section,
The cross-sectional profile in the inner section M sandwiched between the detected left toe section and the right toe section section is the cross-sectional profile data not belonging to the sudden change section in the inner section M. A bead part approximating step for approximating with a first function predetermined using
A base material part approximating step for approximating the cross-sectional profiles on the left outer side from the left toe section and the right outer side from the right toe section by a predetermined second function and a third function, respectively. When,
A toe point calculation step for obtaining an intersection point between the first function and the second function as a left toe point, and obtaining an intersection point between the first function and the third function as a right toe point;
Using the left toe point and the right toe point as a division point of the section L, the first function of the cross-sectional profile is only extrapolated to the entire section sandwiched between the left toe point and the right toe point. Alternatively, extrapolation and interpolation are performed, and the second function is extrapolated to the left outer section from the left toe point, and the third function is extrapolated to the right outer section from the right toe point. Approximate cross-section profile calculating step for deriving a piecewise approximate curve that approximates the cross-sectional profile over the entire section L by connecting the first function, the second function, and the third function. When,
A deviation data calculating step for deriving deviation data that is a difference between the cross-sectional profile and the piecewise approximate curve;
And a defect shape calculation step for evaluating at least one of the size and shape of the defect based on the deviation data.   2. The welding according to claim 1, wherein the degree of change calculation step derives the degree of change by performing second-order difference processing on a y-coordinate point sequence S (n) (n is a natural number) that is the cross-sectional profile. Method for detecting the surface defect shape of a part. A computer program for detecting a surface defect shape of a welded part, which detects a defect on the surface of the welded part, using a member having a welded part as an inspection object
A process of deriving a cross-sectional profile of a predetermined section L in the width direction of the bead of the weld including the base material part;
About the cross-sectional profile, a degree-of-change calculation process for deriving a degree of change that is a change in inclination with respect to the width direction of the bead;
Among a plurality of sudden change sections in which the degree of change of the cross-sectional profile is larger than a predetermined threshold, the leftmost and rightmost toes are designated as the leftmost and rightmost sudden change sections of the predetermined section L, respectively. Toe detection processing to detect as a section,
The cross-sectional profile in the inner section M sandwiched between the detected left toe section and the right toe section section is the cross-sectional profile data not belonging to the sudden change section in the inner section M. A bead part approximation process for approximating with a first function determined in advance using
Base material part approximating process for approximating the cross-sectional profile of the left outer side from the left toe section and the right outer side from the right toe section by a predetermined second function and third function, respectively. When,
A toe point calculation process for obtaining an intersection point of the first function and the second function as a left toe point, and obtaining an intersection point of the first function and the third function as a right toe point;
Using the left toe point and the right toe point as a division point of the section L, the first function of the cross-sectional profile is only extrapolated to the entire section sandwiched between the left toe point and the right toe point. Alternatively, extrapolation and interpolation are performed, and the second function is extrapolated to the left outer section from the left toe point, and the third function is extrapolated to the right outer section from the right toe point. Approximate cross-section profile calculation processing for deriving a piecewise approximation curve that approximates the cross-section profile over the entire section L by connecting and extending the first function, the second function, and the third function. When,
Deviation data calculation processing for deriving deviation data that is a difference between the cross-sectional profile and the piecewise approximate curve;
Defect shape calculation processing for evaluating at least one of the size or shape of the defect based on the deviation data;
A computer program for causing a computer to execute.

Images