JP2007298468A - Program, processing device, and processing method for processing ultrasonic flaw detection data - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a processing method of ultrasonic flaw detection data by improved digital operation processing, and to suppress dispersion of evaluation of a flaw caused by an ability difference of an engineer. <P>SOLUTION: An ultrasonic flaw detection data processing program makes an operation device to execute a shape discrimination processing step (S05) for performing processing for erasing a shape image corresponding to the shape of a specimen from an ultrasonic image acquired from flaw detection data, to the flaw detection data; an identity determination step (S06) for recognizing defect images appearing on the ultrasonic image from which the shape image is erased, determining whether one defect image among the defect images is caused by the same defect as another defect image or not, by a prescribed reference, and performing matching between the defect image and a defect existing in the specimen; and a dimension identification step (S07) for identifying the dimension of the same defect from the defected image matched with the same defect. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a program, a processing apparatus, and a processing method for processing ultrasonic flaw detection data, and more particularly, to digital calculation processing of ultrasonic flaw detection data.

  As a technique for nondestructively inspecting a bridge and other structures, an ultrasonic flaw detection technique is known. The ultrasonic flaw detection technique is a technique for inspecting a structure using the fact that an ultrasonic wave is reflected by the defect when a defect exists inside the structure. In ultrasonic flaw detection, an ultrasonic beam is incident on the structure while being scanned, and the intensity (echo height) of the reflected wave obtained from each position of the structure is displayed as an ultrasonic image. When a defect exists in the structure, an echo image (defect image) corresponding to the defect appears in the ultrasonic image. The presence of the defect can be known by the fact that the defect image is reflected in the ultrasonic image. In addition, the size of the defect can be roughly determined from the size of the defect image. As described above, the ultrasonic flaw detection technique is one of effective methods for evaluating defects.

  One problem with the current ultrasonic flaw detection technology is that considerable skill is required for engineers who evaluate defects in order to correctly evaluate defects from ultrasonic images. This is due to the fact that the ultrasonic image does not directly represent the defect structure. First, not only a defect image but also an echo image (shape image) corresponding to the shape of the structure appears in the ultrasonic image. For example, since the ultrasonic wave is reflected by the surface of the structure, an echo image corresponding to the surface of the structure appears in the ultrasonic image. In order to correctly evaluate the defect, it is necessary to distinguish each of the echo images reflected in the ultrasonic image into a defect image and a shape image. However, when the defect image and the shape image are close to each other or overlap, it is difficult to correctly distinguish them.

  In addition, the shape of the defect image does not indicate the defect structure itself. This is because the direction in which the ultrasonic waves are reflected depends on the direction of the defect. Depending on the orientation of the defect, the ultrasonic wave reflected by the defect may not be detected by the probe. For example, when the extension direction of a defect changes three-dimensionally, one defect may appear as being separated into a plurality of defect images in an ultrasonic image. Even if there is actually one large defect, a plurality of small defect images may appear in the ultrasonic image.

  An engineer engaged in ultrasonic flaw detection technology considers the shape and arrangement of the echo image reflected in the ultrasonic image, the structure of the structure to be inspected, and various other factors, and compares the shape image and the defect image. It is required to correctly identify and determine the size of the defect by correctly estimating the shape of the actually existing defect. However, it is difficult for engineers with little experience to make such a judgment correctly. For this reason, at present, it is difficult to avoid variations due to differences in the ability of engineers in evaluating defects by ultrasonic flaw detection.

  In order to suppress the variation in defect evaluation due to differences in the ability of engineers, it is effective to evaluate defects by processing the data (flaw detection data) obtained by ultrasonic flaw detection using specific digital arithmetic processing. is there. Japanese Patent Laying-Open No. 2005-274444 discloses an ultrasonic flaw detection data processing method for evaluating defects by digitally processing flaw detection data. The known ultrasonic flaw detection data processing method detects a pair of echo images appearing in an ultrasonic image, reads luminance peak coordinates indicating a luminance peak in the pair of echo images, and outputs y of the luminance peak coordinates. The indicated height is calculated from the coordinate difference and the pixel resolution.

However, according to the inventors' investigation, there is room for further improvement in the ultrasonic flaw detection data processing method disclosed in this publication.
JP 2005-274444 A

  Accordingly, an object of the present invention is to provide a method for processing ultrasonic flaw detection data by improved digital arithmetic processing, and thereby to suppress variation in defect evaluation due to a difference in engineer's ability.

  In order to achieve the above object, the present invention employs the means described below. In the description of the means, in order to clarify the correspondence between the description of [Claims] and the description of [Best Mode for Carrying Out the Invention], [Best Mode for Carrying Out the Invention] ] Is added with the numbers and symbols used in []. However, the added number / symbol should not be used to limit the technical scope of the invention described in [Claims].

The ultrasonic flaw detection data processing program according to the present invention is a program for processing flaw detection data (20) obtained by performing ultrasonic flaw detection on a subject by digital arithmetic processing. The program is
A shape identification processing step (S05) in which processing for erasing the shape image corresponding to the shape of the subject (2) from the ultrasonic image obtained from the flaw detection data (20) is performed on the flaw detection data (20);
The defect image (51) appearing in the ultrasonic image from which the shape image has been erased is recognized, and whether one defect image in the defect image (51) is caused by the same defect as the other defect image Identity determination step (S06) for determining whether the defect image (51) and the defect existing in the subject (2) are associated with each other based on a predetermined criterion;
Dimension identification step for identifying the dimension of the same defect from the defect image (51) associated with the same defect (S07)
Are executed by the arithmetic unit (17).

The shape identification processing step (S05)
A defect candidate area defining step (S22) for defining a defect candidate area (42) which is a rectangular area surrounding the echo image (41) with respect to the echo image (41) appearing in the ultrasonic image;
An extraction step (S23) for extracting a region overlapping the shape identification gate region (34) set based on the shape data of the subject (2) from the defect candidate region (42) as the shape candidate region (43). ,
When the echo image (41) included in the shape candidate region (43) has a plurality of echo height peaks (45a, 45b), the shape candidate region (43) is converted into the valley of the echo height distribution. A division step (S25 to S28) for dividing the portion so as to be a boundary;
A representative point setting step (S29) for setting a representative point (46) representing the position of the shape candidate region (43) after the dividing step (S25 to S28);
When the representative point (46) is included in the shape identification gate region (34) set based on the shape data of the subject (2), the echo image (41) is determined to be a shape image to be erased. Shape determination step (S32)
It is preferable to comprise.

The representative point setting step (S29)
Recognizing an echo peak region (47) corresponding to a peak having an echo height equal to or greater than a predetermined value for the echo image (41) (S41, S42);
Step of setting the center of the echo peak region (47) as the representative point (46) (S45)
It is preferable to comprise.

Instead, the representative point setting step (S29)
The distribution of the path direction average value which is an average value of the effective value data which is not “0” among the echo height data inside the shape candidate region (43), with respect to the path direction of the beam used for the ultrasonic flaw detection is calculated. And steps to
Calculating a distribution of vertical average values that are average values in a vertical direction perpendicular to the path direction of the effective value data;
The representative point (46) is an intersection of a straight line extending in the path direction at the peak position of the path direction average value distribution and a line extending in the vertical direction at the peak position of the vertical direction average value. It is also preferable to comprise the step of defining.

  The dividing step (S25 to S28) preferably includes a path direction separating step (S25) for dividing the shape candidate region (43) in the path direction of the beam used for the ultrasonic flaw detection. In this case, the dividing step (S25 to S28) further includes a vertical direction for further dividing the shape candidate region (43) in the direction perpendicular to the path direction in the B scope image after the path direction separating step (S25). It is preferable to include a separation step (S27).

In this case, the vertical direction separation step (S27)
A maximum value curve extracting step of extracting a maximum value curve indicating a vertical distribution of the maximum value of the echo height in the path direction of the shape candidate region (43) perpendicular to the path direction;
It is preferable that the shape candidate region (43) includes a step of dividing a portion where the maximum value curve shows a minimum as a boundary.

The dividing step (S25 to S28)
A vertical separation step (S25 ′) for dividing the shape candidate region in a direction perpendicular to the path direction of the beam used for the ultrasonic flaw detection in the B-scope image;
It is also preferable to include a vertical direction separation step (S27 ') for further dividing the shape candidate region in the path length direction after the vertical direction separation step.

In this case, the path direction separation step (S27 ′)
A maximum value curve extracting step of extracting a maximum value curve indicating the distribution in the path direction of the maximum value of the echo height in the vertical direction of the shape candidate region; and
It is preferable that the step includes dividing the shape candidate region with a valley portion of the maximum value curve as a boundary.

The identity determination step (S06)
A defect image representative point setting step (S52) for determining a representative point (53) representing the position of the defect image (51);
Judgment step of determining whether one defect image of the defect images (51) is caused by the same defect as the other defect images from the distance between the representative points (53) (S53, S54)
It is preferable to comprise.

  In this determination step (S53, S54), it is determined whether or not the two defect images (51) are caused by the same defect from the distance between the representative points of the two defect images (51) of the same cross section. The two defect images (51) are identical from the distance between the representative points of the two defect images (51) of different cross-sections in the in-plane direction parallel to the cross-section. Determining whether or not the defect is caused by a defect.

The defect image representative point setting step (S52) includes a step of defining a rectangular area (52) so as to surround an echo peak area having an echo height of a predetermined value or more in the defect image (51), and a rectangular area (52 And a step of defining an intersection of diagonal lines as the representative point (53). Instead, in the defect image representative point setting step (S52), the path height direction average value which is the average value of the echo height inside the defect image (51) with respect to the path direction of the beam used for the ultrasonic flaw detection. Calculating a distribution;
Calculating the distribution of the average value in the vertical direction, which is the average value in the vertical direction perpendicular to the path direction, of the echo height inside the defect image;
The representative point (53) is an intersection of a straight line extending in the path direction at the peak position of the path direction average value distribution and a line extending in the vertical direction at the peak position of the vertical direction average value. It is also preferable to comprise the step of defining.

  In the ultrasonic flaw detection, when flaw detection data (20) is acquired for a plurality of cross sections while scanning the probe (6) along a predetermined scanning direction, the ultrasonic flaw detection data processing program further includes: Between the cross sections for correcting the flaw detection data (20) of the target cross section among the plurality of cross sections by using flaw detection data of a predetermined number of adjacent cross sections adjacent to the front of the target cross section in the scanning direction and / or the rear of the scanning direction. It is preferable to cause the arithmetic unit (17) to execute the enhancement processing step (S10). In this case, in the shape identification processing step (S05), processing for deleting the shape image corresponding to the shape of the subject (2) is performed on the corrected flaw detection data (20). Instead, in the inter-section enhancement processing step (S10), the flaw detection data (20) after the shape identification processing step may be corrected using flaw detection data of adjacent cross sections.

  In the inter-section enhancement processing step (S10), the flaw detection data of the target cross section is corrected so that the echo height data at each position of the target cross section is increased according to the echo height at the same position of the adjacent cross section. It is preferred that

  When the cross-section enhancement processing step (S10) is performed, in the dimension identification step (S07), the number of continuous cross-sections in which the defect image (51) corresponding to the same defect appears, and the flaw detection data of the target cross-section. It is preferable that the length of the same defect is identified based on the number of the adjacent cross-sections used for the correction.

  The ultrasonic flaw detection data processing program further executes a coupling check step for determining whether or not the coupling between the probe (6) and the subject (2) is correctly maintained in the arithmetic processing. It is preferable to make it. In this case, in the coupling check step, the probe (6) is automatically scanned by comparing the echo height and the threshold value at each position within the predetermined monitoring range (32) of the subject (2). And determining whether or not the coupling between the probe (6) and the subject (2) has been properly maintained during the measurement, the threshold value being used for the control of the probe (6) In a state in which the coupling with the structure is obtained, the ultrasonic wave is incident on the control structure by the probe (6), and the predetermined sound portion (31) of the control structure is measured. It is preferably determined from the echo height at each position. In this case, the control structure and the subject (2) are preferably the same.

The ultrasonic flaw detection data processing apparatus according to the present invention is an ultrasonic flaw detection data processing apparatus for processing flaw detection data (20) obtained by performing ultrasonic flaw detection on a subject (2) by digital arithmetic processing. is there. The ultrasonic flaw detection data processing apparatus performs processing for deleting the shape image corresponding to the shape of the subject (2) on the flaw detection data (20) from the ultrasonic image obtained from the flaw detection data (20). The shape identification processing means (17, 18) and the defect image (51) appearing in the ultrasonic image from which the shape image has been erased are recognized, and one defect image in the defect image (51) is another defect. An identity determination unit (17) that determines whether or not the defect is caused by the same defect as the image based on a predetermined criterion and associates the defect image (51) with the defect present in the subject (2). , 18) and dimension identification means (17, 18) for identifying the dimension of the same defect from the defect image (51) associated with the same defect.
The ultrasonic flaw detection data processing method according to the present invention is a program for processing flaw detection data (20) obtained by performing ultrasonic flaw detection on a subject by digital arithmetic processing. The ultrasonic flaw detection data processing method is as follows:
A shape identification processing step (S05) in which processing for erasing the shape image corresponding to the shape of the subject (2) from the ultrasonic image obtained from the flaw detection data (20) is performed on the flaw detection data (20);
The defect image (51) appearing in the ultrasonic image from which the shape image has been erased is recognized, and whether one defect image in the defect image (51) is caused by the same defect as the other defect image Identity determination step (S06) for determining whether the defect image (51) and the defect existing in the subject (2) are associated with each other based on a predetermined criterion;
Dimension identification step for identifying the dimension of the same defect from the defect image (51) associated with the same defect (S07)
And.

  ADVANTAGE OF THE INVENTION According to this invention, the processing method of the ultrasonic flaw detection data by the improved digital arithmetic processing is provided, and, thereby, the variation in the defect evaluation by the difference in an engineer's ability can be suppressed.

(First embodiment)
In the first embodiment, the ultrasonic flaw detection data processing method according to the present invention is applied to ultrasonic flaw detection as described below.

  FIG. 1 shows a configuration of an automatic ultrasonic flaw detection system 1 and a configuration of a subject 2 used in the present embodiment. In the present embodiment, the subject 2 to be flawed by the automatic ultrasonic flaw detection system 1 is two steel plates 3 and 4 welded by T-shaped fillet welding. In the following, a description will be given of a case where flaw detection is performed on the welded portion 5 of the steel plates 3 and 4 and the periphery thereof, but it will be obvious to those skilled in the art that the subject 2 is not limited to such a structure.

  In this embodiment, the automatic ultrasonic flaw detection system 1 includes a probe 6, a scanning device 7 that scans the probe 6 in a desired scanning direction, a control device 8, a display device 9, and ultrasonic flaw detection data. A processing apparatus 10 is provided. In the present embodiment, the scanning direction of the probe 6 is determined in the direction along the weld line of the steel plates 3 and 4. In the following description, it should be noted that an xyz orthogonal coordinate system is used in which the scanning direction is the z-axis direction, the direction perpendicular to the scanning direction and the direction parallel to the surface of the steel plate 3 is the x-axis direction.

  The probe 6 makes an ultrasonic beam incident on the subject 2 in response to the electrical signal supplied from the control device 8, and further receives the reflected wave from the subject 2 to generate an electrical signal corresponding to the reflected wave. To do. The probe 6 is composed of a plurality of ultrasonic transducers arranged in a line in a plane parallel to the xy plane, and the probe 6 is in a plane parallel to the xy plane. It is possible to function as a phased array that electronically scans the ultrasonic beam.

  The scanning device 7 includes a rail 11 and a probe holding mechanism 12. The rail 11 extends in the scanning direction, that is, the direction of the weld line of the subject 2. The probe holding mechanism 12 holds the probe 6. The probe holding mechanism 12 is placed on the rail 11 so as to be movable in the flaw detection direction, and the probe 6 is scanned by moving the probe holding mechanism 12 on the rail 11. Is called. The position in the scanning direction of the probe 6 can be detected by an encoder (not shown) provided on the rail 11.

  The control device 8 controls the probe 6 and performs various calculations for detecting the subject 2. Specifically, the control device 8 supplies an electric signal to the probe 6 to cause the probe 6 to generate an ultrasonic beam. Further, the control device 8 stores flaw detection data 20 which is data corresponding to the waveform of the reflected wave received by the probe 6 in a storage device (not shown). The control device 8 further has a function of generating various ultrasonic images, that is, an A scope image, a B scope image, a C scope image, and a D scope image from the flaw detection data 20 and displaying them on the display device 9. .

  The ultrasonic flaw detection data processing device 10 receives flaw detection data 20 from the control device 8 via a communication line, performs various digital arithmetic processing on the received flaw detection data, thereby analyzing defects present in the subject 2, In particular, the defect size is calculated. FIG. 2 is a block diagram showing the configuration of the ultrasonic flaw detection data processing apparatus 10. As shown in FIG. 2, the ultrasonic flaw detection data processing apparatus 10 includes an input device 13 such as a keyboard and a mouse, a display device 14, a communication device 15 for communicating with the control device 8, and a storage device 16. And an arithmetic unit 17. The input device 13 and the display device 14 are used as a human interface for an engineer to operate the ultrasonic flaw detection data processing device 10. The communication device 15 is used for receiving flaw detection data 20 from the control device 8. The storage device 16 stores an ultrasonic flaw detection data processing program 18 for processing flaw detection data 20 and various data used for processing flaw detection data. As the ultrasonic flaw detection data processing program 18 is executed by the arithmetic unit 17, desired processing is performed on the flaw detection data. The data stored in the storage device 16 includes healthy part flaw detection data 19a, which is flaw detection data acquired for a portion where no defect exists, and shape information data 19b indicating the shape of the subject 2. As will be described later, the healthy part flaw detection data 19a is used to confirm whether or not the coupling between the subject 2 and the probe 6 has been correctly established. The shape information data 19b is used when processing depending on the shape of the subject 2 is performed. Note that the supply of the flaw detection data 20 from the control device 8 to the ultrasonic flaw detection data processing device 10 can be performed not by a communication line but by a storage medium (for example, an optical disk).

  As shown in FIG. 3, the automatic ultrasonic inspection system 1 of the present embodiment is configured to electronically scan an ultrasonic beam in the xy plane and mechanically scan in the z-axis direction. Has been. The ultrasonic beam can be incident in any direction from any position on the contact surface between the probe 6 and the subject 2. Specifically, among a plurality of ultrasonic transducers constituting the probe 6, a predetermined number of ultrasonic transducers arranged in succession are selected, and the selected ultrasonic transducers are excited to generate an ultrasonic beam. Is incident on the subject 2. The combination of excited ultrasonic transducers is often called “focal law”. For example, when six ultrasonic transducers are excited simultaneously, the sixth ultrasonic transducer from the left end of the probe 6 constitutes one “focal low”, and the second from the left. Up to the seventh ultrasonic transducer constitutes another “focal low”. The position where the ultrasonic beam is incident is determined by the combination of ultrasonic transducers to be excited (ie, focal law) among the ultrasonic transducers constituting the probe 6, and the direction in which the ultrasonic beam is incident. Is determined by the phase difference of the electrical signal supplied to the excited ultrasonic transducer.

  The path characteristic of the ultrasonic beam is determined by the combination of excited ultrasonic transducers and the incident direction of the ultrasonic beam (that is, the phase difference of the electrical signal). For example, as shown in FIG. 3, the sixth ultrasonic transducer from the left end is selected as the ultrasonic transducer to be excited, and an electrical signal having a predetermined phase difference is supplied to these ultrasonic transducers. As a result, an ultrasonic beam of focal low “1” is generated. The waveform of the reflected wave of the ultrasonic beam is acquired as A scope data of focal law “1”. Similarly, the second to seventh ultrasonic transducers from the left are selected as the ultrasonic transducers to be excited, and an electrical signal having a predetermined phase difference is supplied to these ultrasonic transducers, thereby causing the focal A low "2" ultrasonic beam is generated. In this way, by changing the combination of excited ultrasonic transducers (ie, generating an ultrasonic beam with a different focal law), the ultrasonic beam is scanned in the xy plane.

  Acquisition of the flaw detection data 20 by the automatic ultrasonic flaw detection system 1 is performed according to the following procedure. With reference to FIG. 4, first, the probe 6 is aligned with the desired cross section i of the subject 2 by the scanning device 7. The focal row having the cross section i is selected, and A scope data is acquired for the focal row. The flaw detection data 20 acquired by the automatic ultrasonic flaw detection system 1 indicates an identifier for identifying a focal law and A scope data of each focal law (that is, a waveform of a reflected wave of an ultrasonic beam generated for each focal law). Data).

  As shown in FIG. 5A, in the present embodiment, the A scope data is described as the echo height (that is, the signal level of the reflected wave) in each path of the focal law. FIG. 5 shows a plurality of A scope data 21-1 to 21 -N of focal rows “1” to “N”. In the A scope data, the echo height is a numerical value in a predetermined numerical range, and is expressed by a numerical value of 0 or more and 100 or less in the present embodiment. However, when the signal level of the reflected wave is a predetermined value or more, as shown in FIG. 5B, the echo height described in the A scope data is determined as the upper limit value of the numerical range. By the same procedure, A scope data is acquired for all the focal lows defined by the cross section i.

  Subsequently, the probe 6 is moved in the scanning direction (z-axis direction) and aligned with the adjacent section i + 1. Similarly, for the section i + 1, A scope data is acquired for all of the defined focal rows. Similar procedures are repeated for all desired cross sections of the subject 2, whereby flaw detection data 20 is acquired for the entire subject 2. In this embodiment, the focal law from which A scope data is acquired is the same for all cross sections. That is, the echo height data of the same path of the same focal law represents the echo height data of the same position in all the cross sections.

  Various ultrasonic images, that is, a B scope image, a C scope image, and a D scope image can be obtained by performing arithmetic processing on the acquired flaw detection data 20. For example, from the A scope data obtained for a certain cross section, a B scope image of the cross section can be generated. FIG. 6 is a diagram conceptually showing the contents of the B scope image, and FIG. 7 is a conceptual diagram showing an example of the B scope image. Referring to FIG. 6, the B scope image is an image that represents the echo height at each position of a cross section parallel to the xy plane of the subject 2 (that is, perpendicular to the scanning direction) by color tone or gradation. The B scope image specifies the position of the cross section where the echo is generated from the path of the ultrasonic beam generated from each focal law and the incident angle of the ultrasonic beam, and further, the gradation and color of the position in the B scope image. Is determined according to the echo height of the echo. For example, as shown in FIG. 6, a shape echo 22 generated due to the shape of the subject 2 and a defect echo 23 generated due to a defect present in the subject 2 are caused by the probe 6. If detected, as shown in FIG. 7, the B scope image includes a shape image 24 that is an echo image corresponding to the shape echo 22 and a defect image that is an echo image corresponding to the defect echo 23. 25 appears. In FIG. 7, the horizontal axis indicates the position where the ultrasonic beam is incident (that is, the combination of excited ultrasonic transducers), and θ indicates the incident direction of the ultrasonic beam. A B-scope image of a cross-section allows a visual indication of the location where a defect is considered to be present in the cross-section.

  As shown in FIG. 7, two types of echo images: a shape image 24 and a defect image 25 appear mainly in the B scope image obtained by performing ultrasonic flaw detection on the subject 2. The shape image 24 is an echo image representing the shape of the subject 2, and the defect image 25 is an echo image corresponding to a defect present in the subject 2. Providing a technique for correctly identifying the shape image 24 and the defect image 25 and further correctly determining the size of the defect from the defect image 25 is the subject of the ultrasonic flaw detection data processing method of the present embodiment. One. Hereinafter, the ultrasonic flaw detection data processing method of the present embodiment will be described in detail.

  FIG. 8 is a flowchart showing a method for processing ultrasonic flaw detection data in the present embodiment. In the ultrasonic flaw detection data processing method in this embodiment, flaw detection data reading processing (step S01), coupling check processing (step S02), threshold cut processing (step S03), evaluation gate setting processing (step S04), shape Identification processing (step S05), defect identity determination processing (step S06), automatic defect position / dimension identification, defect listing processing (step S07), engineer correction processing (step S08), final A defect list creation process (step S09) is performed. These processes are performed by the ultrasonic flaw detection data processing program 18. Each process will be described in detail below.

Step S01: Reading of flaw detection data In the processing of ultrasonic flaw detection data in the present embodiment, first, flaw detection data 20 is read into the ultrasonic flaw detection data processing apparatus 10. The flaw detection data 20 is sent from the control device 8 to the ultrasonic flaw detection data processing device 10 via the communication line and stored in the storage device 16. In the following, various processes are performed on the flaw detection data 20. However, the original flaw detection data 20 read into the ultrasonic flaw detection data processing apparatus 10 is stored in the storage device 16 and is referred to as necessary.

Step S02: Coupling Check Subsequently, the ultrasonic flaw detection data processing program 18 performs a coupling check process. The coupling check process is a process for confirming the coupling between the subject 2 and the probe 6. When ultrasonic flaw detection is performed in a state where the coupling between the subject 2 and the probe 6 is not correctly taken, the ultrasonic wave is incident on the subject 2 and the reflected wave from the subject 2 is correctly detected. Therefore, the correct flaw detection data 20 cannot be obtained. In the coupling check process, it is confirmed that the state in which the subject 2 and the probe 6 are correctly coupled is maintained while the probe 6 is being scanned.

  FIG. 9 is a diagram for explaining the procedure of the coupling check process. Before performing the automatic flaw detection of the subject 2, the sound part flaw detection data 19a is acquired in advance, and the acquired sound flaw detection data 19a is stored in the storage device 16 (step S11). The sound part flaw detection data 19a is flaw detection data used for determining a threshold value for the coupling check, and indicates that the coupling between the subject 2 and the probe 6 is correctly maintained. Acquired in a state confirmed by visual inspection by an engineer. When the sound detection data 19a is collected, an ultrasonic image is displayed on the display device 9, and the engineer correctly determines the coupling between the probe 6 and the subject 2 from the displayed ultrasonic image. Make sure it is preserved.

  As shown in FIG. 10, acquisition of the healthy part flaw detection data 19a is performed for a section of the subject 2 and at least for a range including the healthy part 31 defined in advance in the section. The incident angles of the focal law and the ultrasonic beam from which the sound detection data 19a is collected are determined so as to cover at least the whole sound 31. As the healthy part 31, a part that is considered to have no defect is selected. It is empirically possible to consider a portion where no defect exists from the shape of the subject 2. The healthy part 31 is selected so as to be separated from the surface of the subject 2 and the welded part 5. Since the healthy part 31 is a part that does not include a defect, all echoes obtained from each position of the healthy part 31 are forest echoes.

  As shown in FIG. 9, after collection of the sound detection data 19a, a threshold value is determined from the sound detection data 19a (step S12). More specifically, the ultrasonic flaw detection data processing apparatus 10 extracts the echo height of the echo obtained from each position of the sound portion 31 from the sound portion flaw detection data, and calculates the average value of the echo heights. The average value is determined as a threshold value.

  The acquisition (step S11) of the healthy part flaw detection data 19a and the determination of the threshold value (step S12) do not need to be performed every time the flaw detection data 20 is processed. Once the threshold is determined, it is possible to perform the coupling check process again for another flaw detection data 20 using the threshold.

  Subsequently, the ultrasonic flaw detection data processing program 18 performs a coupling check using the threshold value determined in step S12 (step S13). More specifically, as shown in FIG. 11, a monitoring range 32 is set in advance for the subject 2, and the ultrasonic flaw detection data processing program 18 performs echoes at each position in the monitoring range 32. Data indicating the height is extracted from the flaw detection data 20. Further, the ultrasonic flaw detection data processing apparatus 10 compares the data indicating the echo height at each position in the monitoring range 32 with the threshold value obtained in step S02, and performs a coupling check based on the comparison result.

  Specifically, the ultrasonic flaw detection data processing program 18 obtains the flaw detection data 20 for a focal low having a certain cross section based on the number of positions where the echo height in the focal low monitoring range 32 exceeds a threshold. It is determined whether or not the coupling between the probe 6 and the subject 2 has been correctly maintained. When there is a predetermined number or more of positions where the echo height exceeds the threshold value in the focal range monitoring area 32 having a certain cross section, the ultrasonic flaw detection data processing program 18 acquires the probe data when acquiring the flaw detection data of the focal law. 6, it is determined that the coupling between the incident position of the ultrasonic wave by the focal law 6 and the subject 2 is correctly maintained. Otherwise, the ultrasonic flaw detection data processing program 18 determines that a coupling failure has occurred.

  For example, as shown in FIG. 12A, when a predetermined number or more of positions where the echo height exceeds the threshold appears in the focal scope A scope data having a certain cross section, the ultrasonic flaw detection data processing program No. 18 determines that the coupling is correctly maintained. On the other hand, as shown in FIG. 12B, when the number of positions where the echo height exceeds the threshold is less than the predetermined number (in FIG. 5B, the number of positions where the echo height exceeds the threshold is 0), The ultrasonic flaw detection data processing program 18 determines that a coupling failure has occurred when acquiring the flaw detection data 20 of the focal law.

  When the coupling defect is detected, the ultrasonic flaw detection data processing program 18 generates coupling defect information indicating the cross section and the focal law where the coupling defect has occurred, and outputs the coupling defect information from the display device 14.

  The monitoring range 32 of each cross section is allowed to be set at a position different from the sound portion 31 of the cross section from which the sound detection data 19a is collected. However, it is preferable that the monitoring range 32 and the healthy portion 31 of each cross section are determined at the same position. That is, the projection of the monitoring range 32 of each cross section onto the xy plane is preferably the same range as the projection of the sound portion 31 onto the xy plane. By determining the monitoring range 32 in this way, the height of the forest echoes that appear in the flaw detection data obtained during automatic flaw detection with the coupling properly maintained, and the forest echoes that appear at the sound flaw detection data positions. Can be made the same height. This is suitable for checking the coupling reliably.

  The healthy part flaw detection data 19a can be acquired not from the subject 2 but from a structure formed of the same material as the subject 2 or a structure manufactured in the same manufacturing process as the subject 2. is there. Since the forest echo is generally determined by the material, the sound detection data 19a is acquired from another structure formed of the same material as the subject 2 or another structure manufactured in the same manufacturing process. Even so, it is possible to determine a generally appropriate threshold. However, it is preferable to acquire the healthy part flaw detection data by using the subject 2 itself to be inspected in that the threshold value can be determined even if there are variations in materials and manufacturing processes.

Step S03: Threshold Cut Process Following the coupling check process, the ultrasonic flaw detection data processing program 18 performs a threshold cut process. In the threshold cut processing, processing for replacing all data having an echo height smaller than a predetermined threshold with 0 is performed on the flaw detection data 20. This eliminates all background level echoes. This threshold cut processing also has a role of separating echo images appearing in the ultrasonic image from each other. When the echo height at a certain position of the ultrasonic image is not 0 (that is, not less than a predetermined threshold), the position is in an area where an echo image (for example, a shape image or a defect image) exists. On the other hand, if the echo height at a certain position is 0, the position belongs to a region where no echo image exists.

Step S04: Evaluation Gate Setting Following the threshold cut processing, the ultrasonic flaw detection data processing program 18 performs evaluation gate setting processing. The evaluation gate setting process is a process for setting the evaluation gate 33 on the subject 2. The evaluation gate 33 is a region where defects are evaluated. In the evaluation gate setting process, a process of replacing all the echo height data at positions outside the evaluation gate 33 in the flaw detection data 20 with 0 is performed. As a result, the defect is not evaluated for the region outside the evaluation gate 33 in the subject 2.

  FIG. 13 is a diagram illustrating an example of the evaluation gate 33. In the present embodiment, the upper end (evaluation gate upper end 33a) of the evaluation gate 33 is set at a position away from the front surface 3a of the steel plate 3 by a predetermined offset toward the inside of the subject 2, and the back surface of the steel plate 3 is further provided. 3b and the lower end of the evaluation gate 33 (evaluation gate lower end 33b) are set at a position away from the surface 5a of the welded portion 5 by a predetermined offset, and the region between the evaluation gate upper end 33a and the evaluation gate lower end 33b is evaluated. The gate 33 is defined. The evaluation gate upper end 33 a and the evaluation gate lower end 33 b of the evaluation gate 33 are determined based on the shape information data 19 b of the subject 2. When the shape information data 19b and the offset are given, the ultrasonic flaw detection data processing apparatus 10 automatically determines a region to be the evaluation gate 33 from the shape information data 19b and the offset. The echo height data at the position outside the evaluation gate 33 determined in the flaw detection data 20 is replaced with 0.

  Note that the order in which the threshold cut processing in step S03 and the evaluation gate setting processing in step S04 are performed may be reversed.

Step S05: Shape Identification Processing Following the evaluation gate setting processing, the ultrasonic flaw detection data processing program 18 performs shape identification processing. In the shape identification process, each of the echo images appearing in the ultrasonic image is identified as a shape image corresponding to the shape of the subject 2 or a defect image corresponding to the defect, and the shape image is determined from the ultrasonic image. It is a process to erase. When it is determined that a certain echo image appearing in the ultrasonic image is a shape image, the echo height data corresponding to the shape image in the flaw detection data 20 is replaced with zero. As a result, the shape image is erased from the ultrasonic image.

  FIG. 14 is a flowchart showing the procedure of the shape identification process. The ultrasonic flaw detection data processing program 18 first performs a process of setting a shape identification gate on the subject 2 (step S21). As shown in FIG. 15, the shape identification gate 34 is a region to be subjected to shape identification processing. The shape identification gate 34 is a region where a shape image is predicted to exist, and is determined based on the shape information data 19b indicating the shape of the subject 2.

  In the present embodiment, the shape identification gate 34 is set as follows. The shape identification gate upper end 34a is set at a position shifted from the back surface 3b of the steel plate 3 and the surface 5a of the welded portion 5 by a predetermined distance α in the direction of the surface 3a of the steel plate 3. Furthermore, the shape identification gate lower end 34b is set at a position shifted from the back surface 3b of the steel plate 3 and the surface 5a of the welded portion 5 by a predetermined distance β in the direction opposite to the surface 3a of the steel plate 3. A region between the shape identification gate upper end 34 a and the shape identification gate lower end 34 b is the shape identification gate 34.

  FIG. 16 is a diagram showing the relationship between the evaluation gate 33 and the shape identification gate 34 in each A scope data included in the flaw detection data 20. Before the shape identification process is performed, the A-scope data of each focal law includes an echo height in each path corresponding to a position inside the evaluation gate 33. The shape identification gate 34 is defined as a part of the evaluation gate 33.

  As shown in FIG. 14, following the setting of the shape identification gate 34, an intra-section area recognition process is performed (step S <b> 22). The intra-section area recognition process is a process of extracting an area where an echo image exists from the B scope image. FIG. 17A is a conceptual diagram illustrating the cross-sectional area recognition processing. The ultrasonic flaw detection data processing program 18 first extracts an echo image 41 inside the evaluation gate 33. Note that in FIG. 17A, only one echo image 41 is shown. In the present embodiment, the extraction of the echo image 41 is performed by recognizing one continuous region where the echo height is not 0 as one echo image 41. As described above, since the background level echo is removed by the threshold cut processing, it should be noted that some echo image exists at a position where the echo height is not zero.

  Furthermore, the ultrasonic flaw detection data processing program 18 defines a defect candidate region 42 that surrounds each extracted echo image 41. In the present embodiment, the defect candidate region 42 is defined to be a rectangle having two sides perpendicular to the path direction of the ultrasonic beam and two parallel sides in the B scope image. Instead, the contour of the echo image 41 can be extracted, and the region inside the contour can be defined as the defect candidate region 42.

  The echo image 41 extracted in step S22 may be a defect image or a shape image. In addition, as illustrated in FIG. 17A, a plurality of echo images that can be distinguished if a human sees the B scope image may overlap each other. Among these multiple echo images, one echo image may be a defect image and the other echo image may be a shape image. Therefore, in order to correctly evaluate the defect, it is necessary to separate the shape image and the defect image. As described below, the ultrasonic flaw detection data processing program 18 separates the overlapped echo images and performs a process for identifying each of the shape image and the defect image.

  In order to separate the shape image and the defect image, the ultrasonic flaw detection data processing program 18 first determines whether or not at least a part of the defect candidate region 42 is inside the shape identification gate 34 (step S23). . This process is for selectively separating the echo image 41 that can be reliably determined to be a defect image. Specifically, when the entire defect candidate region 42 is located outside the shape identification gate 34, the ultrasonic flaw detection data processing program 18 uses the echo image 41 included in the defect candidate region 42 as a defect image. (Step S24). On the other hand, when at least a part of the defect candidate region 42 overlaps the shape identification gate 34, the ultrasonic flaw detection data processing program 18 uses the defect candidate region 42 as a shape candidate that is a region that may include a shape image. The area 43 is extracted (step S23: YES). In the present embodiment, the shape candidate region 43 is defined to be a rectangle having two sides perpendicular to the path direction of the ultrasonic beam and two parallel sides, like the defect candidate region 42. It should be noted that when the defect candidate region 42 is defined as a region inside the contour of the echo image 41, the shape candidate region 43 is also defined as a region inside the contour of the echo image 41. In the example of the present embodiment illustrated in FIG. 17B, the defect candidate area 42 illustrated in FIG. 17A is extracted as the shape candidate area 43. For the extracted shape candidate region 43, the processing from step S25 is executed.

  Subsequently, the ultrasonic flaw detection data processing program 18 executes an A scope separation process that is a process of separating the echo image 41 included in the shape candidate region 43 in the path direction of the ultrasonic beam (step S25). FIG. 17B is a diagram for explaining the A scope separation process. In the example of FIG. 17B, of the A scope data 21-1 to 21-N corresponding to each of the focal rows “1” to “N”, there are two A scope data corresponding to the focal row located near the center. Has a peak. This suggests that the echo image 41 is composed of a plurality of echo images that are overlapped in line in the path direction of the ultrasonic beam. The ultrasonic flaw detection data processing program 18 divides the echo image 41 that satisfies these conditions in the path length direction.

  18A and 18B are diagrams for explaining the A scope separation process in more detail. FIG. 18A shows A scope data 44i of a certain focal law “i”. Two peaks 45a and 45b are recognized in the A scope data 44i. The peak 45a closer to the end point of the shape candidate region 43 (the peak 45a closer to the side farther from the probe 6 out of the two sides perpendicular to the path direction of the shape candidate region 43) is a shape image. The other peaks 45b are considered to be peaks corresponding to the defect image, but the two peaks 45a and 45b are connected at the base. That is, the shape image and the defect image are not separated. At this stage, it is also unclear whether the peaks 45a and 45b correspond to the shape image or the defect image.

  The ultrasonic flaw detection data processing program 18 extracts a peak in which the echo height has the maximum value inside the shape identification gate 34 as a shape candidate peak. In the example of FIG. 18A, the peak shape peak is extracted as the shape candidate peak. Hereinafter, the peak 45a may be referred to as a shape candidate peak 45a. Subsequently, the ultrasonic flaw detection data processing program 18 sets a predetermined ratio of the height of the shape candidate peak 45a as the cut threshold. The cut threshold can be set, for example, as a value that is -A (dB) relative to the peak value of the shape candidate peak 45a; where A is a positive constant. Further, the ultrasonic flaw detection data processing program 18 performs a truncation process of replacing the echo height data of each path whose echo height is smaller than the cut threshold in the A scope data 44i with 0.

  FIG. 18B shows the A scope data 44 ′ i after the truncation process is performed. The shape candidate peak 45a is separated from the peak 45b by the truncation process. Thereby, the echo image 41 composed of a plurality of overlapping echo images is divided in the path direction at the valley portions of the echo height distribution. However, it should be noted that at this stage, it is not determined whether each of the shape candidate peak 45a and the peak 45b is a peak corresponding to a shape image or a peak corresponding to a defect image.

  After the A scope separation process is performed, the ultrasonic flaw detection data processing program 18 executes the cross-sectional area recognition process again (step S26). Specifically, as shown in FIG. 17C, the ultrasonic flaw detection data processing program 18 extracts the echo images 41-1 and 41-2 so that the peaks adjacent in the path direction are separated, and The shape candidate regions 43-1 and 43-2 are defined in the echo images 41-1 and 41-2, respectively. As a result of the A scope separation processing, the echo images 41-1 and 41-2 have only a single peak in the path length direction. That is, in step S26, a plurality of echo images that are overlapped in the path direction are separated and extracted.

  Subsequent to the intra-section area recognition processing in step S26, the ultrasonic flaw detection data processing program 18 performs focal row direction separation processing (step S27). The focal law direction separation process is a process for separately recognizing a plurality of echo images that overlap while shifting in the focal law direction (that is, a direction perpendicular to the path length direction). Although the plurality of echo images overlapped in the path direction are separated by the A scope separation process in step S25, the echo images 41-1 and 41-2 extracted in step S26 are directions perpendicular to the path direction. There is a possibility that it is composed of a plurality of echo images that are overlapped with each other. In the focal low direction separation process, when the echo images 41-1 and 41-2 extracted in step S26 are composed of a plurality of echo images that are shifted and overlapped in the focal low direction, these echo images are separated. The process of recognizing as an object is performed.

  In the example of the present embodiment configured by echo images 41-3 and 41-4 in which the echo image 41-2 overlaps, as shown in FIG. 17D, the echo image 41 is obtained by the focal low direction separation process. -2 is separated into a shape candidate region 43-3 corresponding to the echo image 41-3 and a shape candidate region 43-4 corresponding to the echo image 41-4. .

  The focal row direction separation process will be described in more detail with reference to FIGS. 19A, 19B, and 19C. FIG. 19A shows an echo image 41-2 to be processed and a shape candidate region 43-2 defined for the echo image 41-2. As shown in the upper diagram of FIG. 19B, when a line Line1 is defined in the B scope image, which crosses the echo image 41-3 in the focal low direction (that is, extends in a direction perpendicular to the path direction). , One peak appears in the echo height distribution along the line Line1. Here, it should be noted that the line Line1 is a line segment that passes through the positions of the focal lows “1” to “N” that have a certain path value. Similarly, when another line Line2 that crosses the echo image 41-4 in the focal low direction is defined in the B scope image, as shown in the lower diagram of FIG. 19B, the echo height along the line Line2 is shown. One peak is observed in the distribution. However, after the A scope separation process, the peak position of the echo height distribution along the line Line1 is always different from the peak position of the echo height distribution along the line Line2. The echo image 41-2 is separated into echo images 41-3 and 41-4 using such a difference in peak position of the echo height distribution.

More specifically, the ultrasonic flaw detection data processing program 18 performs the following processing.
(A) The ultrasonic flaw detection data processing program 18 first performs maximum value extraction processing. Specifically, the ultrasonic flaw detection data processing program 18 extracts the maximum value of the echo height inside the shape candidate region 43-2 for each focal law related to the shape candidate region 43-2, and extracts the focal law. Then, a maximum value curve representing the correspondence between the echo heights inside the shape candidate region 43-2 is extracted. In other words, the maximum value curve is a line connecting the maximum values when the echo height at each position of the shape candidate region 43-2 is projected in the path length direction. FIG. 19C is a graph showing an example of such a maximum value curve. When the echo image 41-2 is composed of a plurality of echo images that are shifted and overlapped in the focal low direction, a peak corresponding to the number of the echo images appears in the maximum value curve.

(B) Subsequently, the ultrasonic flaw detection data processing program 18 divides the shape candidate region 43-2 with the valley portion of the maximum value curve (the portion near the position where the maximum value curve in FIG. 9C is minimized) as a boundary. By doing so, a shape candidate region is newly defined. FIG. 17D shows newly defined shape candidate regions 43-3 and 34-4.

  With reference to FIG. 20, the minimum part recognition process performed by the ultrasonic flaw detection data processing program 18 in the process (b) will be described. The ultrasonic flaw detection data processing program 18 performs recognition processing of the valley portion of the maximum value curve using one of the following three recognition methods.

In the first recognition method, condition (1):
H (i-1)> H (i) <H (i + 1)
Is satisfied, the shape candidate region 43 is separated in the focal law direction with the position corresponding to the focal law “i” as a boundary. H (i−1), (i), and H (i + 1) in the above equation are the echo heights of the maximum value curves corresponding to the focal lows “i−1”, “i”, and “i + 1”, respectively (ie, The focal law “i−1”, “i”, “i + 1” echo height maximum value inside the shape candidate region 43) is shown. This condition indicates that the maximum value curve becomes a valley in the focal law “i” as in the condition (1) indicated by the solid line in FIG. Thereby, the shape candidate area | region 43 is isolate | separated by the shallow trough of the maximum value curve. This condition is suitable for high-speed processing because the amount of calculation required for determination is small.

In the second recognition method, condition (2):
H (i-1)> H (i) <H (i + 1) and H (i-2)> H (i) <H (i + 2)
Is satisfied, the shape candidate region 43 is separated in the focal law direction with the position corresponding to the focal law “i” as a boundary. As shown in the condition (2) of FIG. 20, this condition is that the maximum value of the echo height of the focal low “i” is the focal low “i−1”, “i + 1” and It indicates that the focal heights “i−2” and “i + 2” adjacent to the outside of them are smaller than the maximum echo height. Thereby, the shape candidate area | region 43 is isolate | separated by the slightly shallow trough of the maximum value curve. This condition (2) is more suitable for reducing the influence of noise than the condition (1).

In the third recognition method, condition (3):
H (i-1)> H (i) <i + 1, and H (i-2)> H (i-1) and i + 1 <i + 2.
Is satisfied, the shape candidate region 43 is separated in the focal law direction with the position corresponding to the focal law “i” as a boundary. As shown in condition (3) of FIG. 20, this condition (3) is a total of five focal lows, that is, two focal lows on the left and right adjacent to the focal low “i” and the focal low “i”. Of these, the focal law “i” means the smallest value in the maximum value curve. Thereby, the shape candidate area | region 43 is isolate | separated by the deep valley of a maximum value curve. This condition is suitable when the boundary between the shape image and the defect image seems to be relatively clear.

  Subsequently, the ultrasonic flaw detection data processing program 18 sets a peak in which the echo height is maximum in each of the shape candidate regions 43-2 and 43-3 newly defined by the focal row direction separation processing. Extraction is performed, and a predetermined percentage of the height of the extracted peak is set as the cut threshold. Further, the ultrasonic flaw detection data processing program 18 replaces the data of the echo height of each path whose echo height is smaller than the cut threshold in the A scope data related to the shape candidate areas 43-2 and 43-3 with 0. Perform truncation processing. Thereby, the echo image 41-2 is separated into echo images 41-3 and 41-4.

  After the focal row direction separation is performed, the ultrasonic flaw detection data processing program 18 executes the cross-sectional area recognition process again (step S28). As a result, as shown in FIG. 17E, echo images 41-3 and 41-4 that overlap while shifting in the focal low direction are separated and recognized. Further, echo images 41-3 and 41-4 For each, the shape candidate region 43-3 and the shape candidate region 43-4 are defined separately.

  Subsequently, the ultrasonic flaw detection data processing program 18 sets region representative points for each of the shape candidate regions 43 defined at this point. The area representative point is a point representing the position of the shape candidate area 43. One of the two setting methods described below is used as a method for setting the region representative points.

  In the first setting method, the ultrasonic flaw detection data processing program 18 determines the region representative point 46 according to the position where the echo height of the shape candidate region 43 takes the maximum value. FIG. 21 is a flowchart showing a first setting method for setting region representative points. First, the ultrasonic flaw detection data processing program 18 detects the peak point of the shape candidate region 43 (that is, the point where the echo height takes the maximum value inside the shape candidate region 43) (step S41). As described with reference to FIG. 6, since the upper limit value (100 in the present embodiment) is set in the numerical range of the echo height of the A scope data, the echo height is high overall. Note that multiple peak points can be detected by saturating the echo height to the upper limit.

  When the detected peak point is one (step S42: NO), the ultrasonic flaw detection data processing program 18 sets the peak point as the region representative point 46 (step S43).

  On the other hand, when there are two or more detected peak points (step S42: YES), the ultrasonic flaw detection data processing program 18 executes the processing after step S44.

  As shown in FIG. 22A, the ultrasonic flaw detection data processing program 18 extracts an echo peak region 47 from the shape candidate region 43 (step S44). The echo peak region 47 is a continuous region in which the echo height has an upper limit value at all positions inside it. When there are two or more peak points, there is a continuous region where the echo height is saturated to the upper limit value. The ultrasonic flaw detection data processing program 18 extracts this area as an echo peak area 47.

  Further, the ultrasonic flaw detection data processing program 18 defines the center of the echo peak area 47 as the area representative point 46 (step S45). In one embodiment, the center of gravity of the echo peak region 47 may be defined as the region representative point 46. In another embodiment, as shown in FIG. 22B, a peak rectangular region 48 circumscribing the echo peak region 47 is defined, and the intersection of the diagonal points can be determined as the region representative point 46.

  In the second setting method, as shown in FIG. 22C, the ultrasonic flaw detection data processing program 18 determines the distribution of the average value of the path direction in the effective value data and the vertical value of the effective value data for each of the shape candidate regions 43. A direction average value distribution is calculated, and an area representative point 46 is determined from the calculated path direction direction average value distribution and vertical direction average value distribution. Here, the effective value data is data that is not “0” among the echo height data at each position in the shape candidate region 43.

ここで、Ｎ_１は、フォーカルロー”ｉ”の各路程のうち、形状候補領域４３の内部にあり、且つ、エコー高さが”０”でない路程の数であり、Ｅｃｈｏ（ｐ）は、路程ｐのエコー高さであり、Σ_ｐは、形状候補領域４３の内部にあり、且つ、エコー高さが”０”でない路程についての和を表している。有効値データの路程方向平均値が形状候補領域４３に関与する全てのフォーカルローについて算出され、これにより、有効値データの路程方向平均値の分布が得られる。 The path direction average value of valid value data is a value defined for each focal law, and the path direction average value of valid value data of a certain focal row for a certain shape candidate region 43 is the value of the focal law. This is the average value of the effective value data of the path corresponding to the position inside the shape candidate region 43. It should be noted that the data whose echo height is “0” is not used in the calculation of the average value in the path direction of the effective value data. When the shape candidate region 43 is defined as the contour of the echo image 41, the path height direction average value of the echo height is the path length average value of the effective value data as it is. For example, the path direction average value Ave ₁ (i) of valid value data of a certain focal law “i” is defined by the following formula:

Here, N ₁ is the number of paths within the shape candidate region 43 and the echo height is not “0” among the paths of the focal law “i”, and Echo (p) is the path length. p is an echo height, the sigma _p, is inside the shape candidate area 43, and the echo height represents the sum of the path length is not "0". The average value in the path direction of the effective value data is calculated for all focal rows related to the shape candidate region 43, and thereby the distribution of the average value in the path direction of the effective value data is obtained.

ここで、Ｎ_２は、ラインＬｉｎｅｊ上の点のうち、形状候補領域４３の内部にあり、且つ、エコー高さが”０”でない点の数であり、Ｅｃｈｏ（ｑ）は、ラインＬｉｎｅｊ上の点ｑのエコー高さであり、Σ_ｑは、形状候補領域４３の内部にあり、且つ、エコー高さが”０”でない点についての和を表している。有効値データの垂直方向平均値が形状候補領域４３に関与する全てのラインについて算出され、これにより、有効値データの垂直方向平均値の分布が得られる。 On the other hand, the vertical average value of the effective value data is a value defined for each line defined perpendicular to the path direction, and the vertical average value of the effective value data of a certain line for a certain shape candidate region 43. Is the average value of the effective value data of each point corresponding to the position inside the shape candidate region 43 on the line. It should be noted that even in the calculation of the average value in the vertical direction of the effective value data, data having an echo height of “0” is not used. When the shape candidate region 43 is defined as the contour of the echo image 41, the vertical average value of the echo height is the same as the average value in the path direction of the effective value data. For example, the path direction average value Ave ₂ (j) of the effective value data of a certain line Linej is defined by the following formula:

Here, N ₂ is the number of points on the line Linej that are inside the shape candidate region 43 and the echo height is not “0”, and Echo (q) is on the line Linej. an echo height of the point q, the sigma _q, is inside the shape candidate area 43, and the echo height represents the sum of the points is not "0". The vertical average value of the effective value data is calculated for all the lines related to the shape candidate region 43, and thereby the distribution of the vertical average value of the effective value data is obtained.

  A region representative point 46 is determined from the calculated distribution of the average value in the path direction and the distribution of the average value in the vertical direction. The area representative point 46 is defined as the intersection of the straight line extending in the path direction at the peak position of the path direction average value distribution and the line at the peak position of the vertical direction average value distribution in the shape candidate area 43. Is done.

  When the setting of the region representative points 46 for each of the shape candidate regions 43 is completed, as shown in FIG. 14, the ultrasonic flaw detection data processing program 18 sets each region representative point 46 inside the shape identification gate 34. It is determined whether it is present or outside (step S30). The ultrasonic flaw detection data processing program 18 extracts the shape candidate region 43 in which the region representative point 46 is located outside the shape identification gate 34 (step S30: NO), and the echo image 41 included in the shape candidate region 43. Is determined to be a defect image derived from a defect (step S31). In the example of FIG. 24, since the region representative point 46-6 corresponding to the shape candidate region 43-6 is located outside the shape identification gate 34, the echo image included in the shape candidate region 43-6 corresponds to the defect. It is determined as a defect image.

  Further, the ultrasonic flaw detection data processing program 18 extracts the shape candidate region 43 in which the region representative point 46 is located inside the shape identification gate 34 (step S30: YES), and the echo image included in the shape candidate region 43. Is determined to be a shape image corresponding to the shape of the subject 2. In the example of FIG. 24, the echo images included in the shape candidate regions 43-5 and 43-7 are the shape images because the region representative points 46-5 and 46-7 are located inside the shape identification gate 34. Determined. The ultrasonic flaw detection data processing program 18 performs processing for deleting the shape image from the ultrasonic image. Specifically, the ultrasonic flaw detection data processing program 18 sets the echo height data at a position corresponding to the shape image of each A scope data of the flaw detection data 20 to zero. Thereby, the shape image is erased from the B scope image, and the shape identification process (step S05) is completed. FIG. 17F shows an example of the B scope image obtained by the shape identification process. The echo image 41-3 determined to be a shape image is erased from the B scope image, and only the echo images 41-1 and 41-4 corresponding to the defect are selectively left in the B scope image.

  The shape identification process described above is performed for all cross sections. Only the defect image corresponding to the defect appears in the B scope image generated from the flaw detection data 20 after the shape identification processing is performed.

  In order to reduce the amount of calculation, it is possible not to execute the intra-section area recognition process (step S28) after the focal row direction separation process. In this case, a region representative point 46 is determined for the shape candidate region 43 defined by the focal row direction separation process. It should be noted that the shape candidate region 43 defined by the focal row direction separation process includes a shape candidate region obtained by dividing one shape candidate region 43.

  In the shape identification process described above, as shown in FIG. 14, after the process of separating the echo image in the path direction (A scope separation process) is performed in step S25, the echo image is converted into the focal low direction in step S27. A process of separating in the (direction perpendicular to the path length direction) (focal row direction separation process) is performed. Instead, as shown in FIG. 23A, after the process of separating the echo image in the focal low direction (step S25 ′) is performed, the process of separating the echo image in the path length direction (step S27 ′) is performed. Can be done.

  In this case, in step S25 ', the echo image is separated in the focal low direction by the same process as the A scope separation process except that the direction in which the echo image is separated is different. That is, in the process in step S25 ′, as shown in FIG. 23B, instead of the A scope data 44i, the echo height distribution data along each line defined in the direction perpendicular to the path length direction is used. The same processing as the above-described A scope separation processing is performed. Accordingly, as shown in FIG. 23C, the echo image is separated in the focal low direction at the valley portion of the echo height distribution. In the example of FIG. 23C, the echo image 41-1 is separated into two echo images 41-4 and 41-5 adjacent in the focal law direction. The echo image 41-5 is composed of two echo images 41-1 and 41-3 that are arranged side by side in the path direction, but is not separated by the process of separating the echo image in the focal low direction in step S25 '. After the process of separating the echo image in step S25 'in the focal law direction, the cross-sectional area recognition process is executed again (step S26).

  After the cross-sectional area recognition process in step S26, a process (step S27 ') for separating the echo image in the path direction is performed. The echo image is separated in the path direction by the same process as the focal law separation process except that the direction in which the echo image is separated is different. In other words, in the processing in step S27 ', processing similar to the above-described focal row separation processing is performed using A scope data instead of the echo height distribution data along each line. More specifically, for each line defined in the shape candidate area 43, the maximum value of the echo height inside the shape candidate area 43 is extracted, and the correspondence between the line and the echo height inside the shape candidate area 43 is determined. The maximum value curve representing is extracted. In other words, the maximum value curve is a line connecting the maximum values when the echo height at each position of the shape candidate region 43 is projected in the focal low direction. Further, as shown in FIG. 23D, the shape candidate region 43 is divided with the valley portion of the maximum value curve (the portion near the position where the maximum value curve of FIG. Is newly defined. In the example of FIG. 23D, the shape candidate region 43-5 is divided in the path direction, and the shape candidate regions 43-1 and 43-3 are newly defined. Further, for each of the shape candidate regions 43 newly defined by the process of separating the echo image in step S27 ′ in the path direction, a peak having the maximum echo height is extracted inside, and the extracted peak height is extracted. A predetermined percentage of the height is set as the cut threshold. Further, the ultrasonic flaw detection data processing program 18 performs a truncation process of replacing the echo height data of each path whose echo height is smaller than the cut threshold of the A scope data related to the shape candidate region 43 with zero. As a result, the echo images are separated in the path direction. For example, in the example of FIG. 23E, the echo image 41-5 is separated into echo images 41-1 and 41-4.

  After step S27 'is performed, the cross-sectional area recognition process is executed again (step S28). As a result, as shown in FIG. 23E, the echo images 41-1 and 41-3 which are overlapped while being displaced in the direction of the path are separated and recognized, and each of the echo images 41-1 and 41-3 is further recognized. The shape candidate region 43-1 and the shape candidate region 43-3 are defined separately. The subsequent processing is the same as the shape identification processing described above.

Step S06: Defect Identity Determination Process As shown in FIG. 8, after the shape identification process is completed, the ultrasonic flaw detection data processing program 18 performs a defect identity determination process (step S06). . This process is performed to associate the defect image that appears in the B-scope image of each cross section with the defect that actually exists. The association between the defect image and the actual defect is necessary for the following two reasons. First, in order to identify the dimension of the defect, it is necessary to associate the defect image appearing in the B-scope image of the adjacent cross section. Specifically, in order to identify the designated length (that is, the length of the defect in the scanning direction), the defect image appears in the B-scope image of one cross-section and the B-scope image of another cross-section. It is necessary to determine whether or not the defective image is caused by the same defect. Secondly, the defect image that appears in the B scope image does not necessarily correspond one-to-one with the defect that actually exists in the subject 2. That is, even if two defect images appear in the same cross section, they may actually be attributed to one defect. In such a case, it is necessary to determine whether or not two defect images are associated with the same defect.

  In the present embodiment, the ultrasonic flaw detection data processing program 18 performs defect identity determination processing according to the following procedure. FIG. 25 is a flowchart showing a defect identity determination process in the present embodiment. First, the ultrasonic flaw detection data processing program 18 identifies a defect image appearing in the B scope image from which the shape image has been erased by the shape identification process (step S51). The defect image is identified by the following procedure. First, a position where the echo height is not 0 is searched on the B scope image. When a position where the echo height is not 0 is found, the echo height is scanned for all the internal positions including the position by scanning the B scope image in both the vertical direction and the horizontal direction starting from the position. A continuous region with a non-zero value is recognized. The area is recognized as a defect image area.

  Subsequently, as shown in FIG. 26A, a region representative point 53 is determined for each of the defect images 51 appearing in the B scope image of each cross section (step S52). The area representative point 53 is a point representing the position of the defect image 51.

  In one embodiment, as shown in FIG. 26B, the rectangular area 52 is defined so as to circumscribe the echo peak area 54 of the defect image 51, and the intersection of the diagonal lines of the rectangular area 52 is the area. The representative point 53 may be determined. The echo peak area 54 is a continuous area in which the echo height has an upper limit value at all the positions inside. As described above, it should be noted that an upper limit value is set for the numerical range of the echo height of the A scope data. When the echo height at a certain position of the subject 2 is higher than a predetermined value, the A scope data describes that the echo height at that position is the upper limit value.

  In another embodiment, as shown in FIG. 26C, for each defect image 51, the distribution of the average direction value of the echo height and the distribution of the average value of the vertical direction of the echo height are calculated and calculated. The region representative point 53 is determined from the distribution of the average values in the path direction and the distribution of the average values in the vertical direction.

  The path height average value of the echo height is a value defined for each focal law, and the path height average value of the echo height of a certain focal law for a certain defect image 51 is the defect of the focal law. This is the average value of the echo height of the path corresponding to the position inside the image 51. A path direction average value of the echo height is calculated for all focal lows related to the defect image 51, and thereby a distribution of the average value of the echo height in the path direction is obtained.

  On the other hand, the vertical average value of the echo height is a value defined for each line defined perpendicular to the path direction, and the vertical average value of the echo height of a certain line for a certain defect image 51. Is an average value of echo heights of the respective points corresponding to positions inside the defect image 51 on the line. The vertical average value of the echo height is calculated for all the lines involved in the defect image 51, and thereby the distribution of the vertical average value of the echo height is obtained.

  A region representative point 46 is determined from the calculated distribution of the average value in the path direction and the distribution of the average value in the vertical direction. The region representative point 53 is defined as an intersection of a straight line extending in the path direction at the peak position of the path direction average value distribution and a line at the position of the vertical direction average value in the defect image 51.

  Subsequently, as shown in FIG. 25, a process of determining whether or not the plurality of defect images 51 appearing on the same cross section is caused by the same defect is performed (step S53). For this determination, the position of the area representative point 53 of the rectangular area 52 defined for each defect image 51 is used. As shown in FIG. 27, when the distance d between the area representative points 53 of the rectangular areas 52 of the two defect images 51 is larger than the predetermined value A, the two defect images 51 are different from each other. It is judged to be caused. On the other hand, when the distance between the region representative points 53 is equal to or less than the predetermined value A, it is determined that the two defect images 51 are caused by the same defect.

For example, if the coordinates of the area representative point 53 _j of each rectangular area 52 _j are (x _j , y _j ), the distance d _j between the area representative points 53 _{j and} 53 _{j + 1} of the rectangular areas 52 _j and 52 _{j + 1} respectively _{. j + 1} is represented by the following formula:
d ₁ (j, j + 1) = √ {(x _j −x _{j + 1} ) ² + (y _j −y _{j + 1} ) ² }.
In the example of FIG. 27, since d (j, j + 1) is larger than the predetermined value A, it is determined that the defect images 51 corresponding to the rectangular areas 52 _j and 52 _{j + 1} are caused by different defects. On the other hand, since the distance d ₁ (j + 1, j + 2) between the region representative points 53 _{j + 1 and} 53 _{j + 2} of the rectangular regions 52 _{j + 1} and 52 _{j + 2} is equal to or less than the predetermined value A, it corresponds to the rectangular regions 52 _j and 52 _{j + 1} , respectively. The defect image 51 to be determined is caused by the same defect.

  The ultrasonic flaw detection data processing program 18 associates the defect image 51 determined to be caused by the same defect with respect to each of the cross sections subjected to the ultrasonic flaw detection. This is performed by assigning the same defect ID to the defect image 51 caused by the same defect, and assigning a different defect ID to the defect image 51 caused by the different defect.

Subsequently, as shown in FIG. 25, a process of determining whether or not the defect image 51 appearing in the adjacent cross section is caused by the same defect is performed (step S54). Also in this determination, the position of the area representative point 53 of the rectangular area 52 defined for each of the defect images 51 is used. A defect image 51 of a cross-section, for all combinations of the defect image 51 of the cross-section adjacent thereto, the distance d ₂ in the plane direction between the area representative point 53 of the corresponding rectangular area 52 is calculated. The distance d ₂ in the plane direction, definitive when projected area representative point 53 of the rectangular region 52 of different cross-section in the xy plane, is that the distance between the projection points of the area representative point 53. The distance d ₍ in-plane direction) between the area representative point 53 _{i, j} of the rectangular area 52 _{i, j} of the cross section i and the area representative point 53 _{i + 1, k} of the rectangular area 52 _{i + 1, k} of the cross section i + 1 adjacent to the cross section i. _{i, j) (i + 1, k)} is represented by the following formula:
d _{i, i + 1} (j, k) = √ {(x _{i, j} −x _{i + 1, k} ) ² + (y _{i, j} −y _{i + 1, k} ) ² },
Here, (x _{i, j} , y _{i, j} ) is an xy coordinate of the region representative point 53 _{i, j} of the rectangular region 52 _{i, j} of the cross section i, and (x _{i + 1, k} , y _{i + 1, k} ). _Are the xy coordinates of the region representative point 53 _{i + 1, j} of the rectangular region 52 _{i + 1, k} of the cross section i + 1.

As shown in FIG. 28, the area representative point 53 _{i, j} of the rectangular area 52 _{i, j} of the cross section i, and the area representative point 53 _{i + 1, k} of the rectangular area 52 _{i + 1, k} of the cross section i + 1 adjacent to the cross section i _. for _k, the distance _{d 2} in the plane direction is larger than the predetermined value B, the rectangular region _{52 i} of the cross section _i, a defect image 51 corresponding to _j, the defect image corresponding to the cross-section i + 1 of the rectangular region _{52 i + 1, k} 51 is determined to be caused by a different defect. The predetermined value B is a constant value larger than the predetermined value A used in step S53 described above. On the other hand, when the distance between the region representative points 53 _{i, j} , 53 _{i + 1, k} is equal to or less than the predetermined value B, it is determined that the two defect images 51 are caused by the same defect.

In the example of FIG. 28, the defect image 51 corresponding to each of the rectangular area 52 _{i, j of} the cross section i, the rectangular area 52 _{i + 1, k of} the cross section i + _{1, and} the rectangular area 52 _{i + 1, m} of the cross section i + 2 is caused by the same defect. On the other hand, the defect image 51 corresponding to the rectangular area 52 _{i + 1, n} of the cross section i + 3 is determined to be caused by a different defect.

  The ultrasonic flaw detection data processing program 18 correlates the defect images 51 determined to be caused by the same defect. This is performed by assigning the same defect ID to the defect image 51 caused by the same defect, and assigning a different defect ID to the defect image 51 caused by the different defect.

  It should be noted that even if it is determined in step S53 that the two defect images 51 of a certain cross section are caused by different defects, the determination is corrected in step S54, and as a result, the same defect ID can be given. For example, even if it is determined in step S53 that the two defect images 51 of the cross section i are caused by different defects, in step S54, the defect image 51 having the cross section i adjacent to the defect image 51 is identical to the two defect images 51. When it is determined that the defect is caused by the defect, it is finally determined that the three defect images 51 are caused by the same defect, and the same defect ID is given.

Step S07: Automatic identification of defect position and dimension and defect listing When the defect identity determination process is completed, the ultrasonic flaw detection data processing program 18 automatically identifies the defect position and dimension, Pass / fail is determined for the defect, and a defect list in which the defect ID, position, size, and pass / fail determination result of each defect is listed is created (step S07). In one embodiment, the defect list includes a defect ID, a position X in the welding direction (z-axis direction) of the defect, a depth D of the defect, a position K in the lateral direction (x-axis direction), an indicated length L, and The pass / fail judgment result is described. The designated length L of the defect is determined based on the number of cross sections in which the defect images 51 associated with the same defect appear continuously. When the defect image 51 determined to be caused by the same defect (that is, given the same defect ID) appears on N consecutive cross sections, the indicated length is (N−1) · D It is judged that. Here, D is the distance between two adjacent cross sections. The determination of pass / fail of the defect is performed based on the maximum echo height of the defect image and the indicated length L. If the indicated length L of a certain defect is larger than a predetermined reference value and the maximum echo height of a series of defect images corresponding to the defect is larger than a predetermined reference value, the defect is regarded as a serious defect. To be judged. In this case, the ultrasonic flaw detection data processing program 18 describes information indicating that the subject 2 is defective due to the defect in the defect list as a pass / fail determination result.

Step S08: Correction Process by Engineer Once the defect list is created, the ultrasonic flaw detection data processing program 18 performs a process of providing a human interface for the engineer to check the defect list and correct the defect list if necessary. Perform (step S08). Specifically, the ultrasonic flaw detection data processing program 18 displays the defect list on the display device 14, and further displays a screen requesting the technician to correct the defect list. At this time, the ultrasonic flaw detection data processing program 18 is generated from the ultrasonic image generated from the original flaw detection data 20 or the flaw detection data 20 subjected to the shape identification process in response to the operation of the input device 13 by the engineer. The ultrasonic image thus displayed is displayed on the display device 14. The engineer inputs defect list correction data indicating the correction contents of the defect list to the input device 13 as necessary.

Step S09: Creation of Final Defect List When defect list correction data is input, the ultrasonic flaw detection data processing program 18 corrects the defect list and generates a final defect list (step S09). The ultrasonic flaw detection data processing program 18 displays a final defect list on the display device 14 and outputs the final defect list by a printer (not shown) when requested by a technician. Thus, the processing of the flaw detection data 20 is completed.

  According to the ultrasonic flaw detection data processing method described above, the position and size of a defect are automatically identified by digital calculation processing, thereby suppressing variation in defect evaluation due to a difference in engineer's ability. be able to.

(Second Embodiment)
When the extending direction of the defect changes three-dimensionally, the defect image may not be correctly associated between adjacent cross sections. FIG. 29 is a diagram for explaining the reason. As shown in FIG. 29, the defect 55 does not necessarily extend parallel to the surface of the subject 2. For example, in the cross sections i ₁ and i ₃ , the extending direction of the defect 55 is parallel to the surface of the subject 2, but in the cross section i ₂ between them, the extending direction of the defect 55 is oblique to the surface of the subject 2. There may be cases. In such a case, although the reflected wave returns to the probe 6 in the cross sections i ₁ and i ₃ , the reflected wave may not return to the probe 6 in the cross section i ₂ . That is, the height of the echo due to the defect 55 is high in the cross sections i ₁ and i ₃ , but the height of the echo due to the defect 55 may be low in the cross section i ₂ . In such cases, it may not appear defective image in the cross section i _2. For example, the echo height by defects 55 in the cross-section i ₂ is lower than the threshold used in the threshold cut processing (step S03), it does not appear defect image corresponding to the defect 55 in its cross-section i _2. In such a case, the defect image appearing in the cross-sections i ₁ and i ₃ is recognized as two defects despite the single defect 55, and as a result, the size of the defect is not correctly evaluated. .

  In order to prevent such a problem, the cross-section enhancement process is performed in the present embodiment. The cross-section enhancement processing is processing for correcting flaw detection data of a certain target cross section (that is, echo height data at each position of a certain target cross section) using flaw detection data of a cross section in the vicinity of the target cross section. . Hereinafter, the processing procedure of the digital arithmetic processing performed in this embodiment will be described in detail.

  FIG. 30 is a flowchart showing a processing procedure of digital arithmetic processing performed in the present embodiment. In the present embodiment, after the threshold cut process (step S03) is performed, the above-described inter-section enhancement process is performed (step S10). The processes after the evaluation gate setting process (steps S04 to S09) are performed on the flaw detection data 20 on which the cross-section enhancement process has been performed.

In the cross-section enhancement processing (step S10) of the present embodiment, as shown in FIG. 31, the flaw detection data of the cross-section i includes two cross-sections i-2, i-1 adjacent to the rear in the scanning direction, and Correction is performed using flaw detection data of cross sections i + 1 and i + 2 adjacent to the front in the scanning direction. Specifically, in the cross-section enhancement processing according to the present embodiment, the echo height data at a position where the cross section i exists is corrected to the maximum value of the echo height data at the same position of the five cross sections i−2 to i + 2. Is done. In the present embodiment the focal rows A-scope data is acquired is the same for all cross sections, the echo height data H _'i path length k of focal row j of section i after _{treatment, j, k} is the following Is represented as:
H ′ _{i, j, k} =
max [ _{Hi-2, j, k} , _{Hi-1, j, k} , _{Hi, j, k} , _{Hi + 1, j, k} , _{Hi + 2, j, k} ],
_{Here, max [x 1, x 2} , x 3, x 4, x 5] _is the maximum value of x 1 ~x _{5, H i, j, k} is the path length k of focal row j of section i The echo height.

  FIG. 32 is a graph showing a specific example of the cross-section enhancement processing. The upper part of FIG. 13 shows A scope data of the same focal row having the cross sections i−2 to i + 2, and the lower part shows A scope data of the focal row of the cross section i after the cross-section enhancement processing. . For example, the echo height of the path “4” in the section i-2 is “70”, and the echo height of the path “4” in the other sections i−1, i, i + 1, i + 2 is “0”. In this case, the data of the echo height of the path length “4” of the A-scope data of the focal low of the cross section i after the cross section enhancement processing is “70”. Similarly, the echo height of the path “5” in the cross section i−1 is “80”, and the echo height of the path “5” in the other cross sections i−2, i, i + 1, i + 2 is “0”. In some cases, the echo height data of the path length “5” of the A-scope data of the focal row of the cross section i after the cross section enhancement processing is “80”.

According to such cross-section enhancement processing, even if the echo height at a certain position of a target cross section is low, if the echo height at the same position of an adjacent cross section is high, the echo height at the position of the target cross section is high. The data is corrected to be higher. Therefore, even if the echo height of the target cross section is lowered due to a three-dimensional change in the shape of the defect, the defect image is not lost from the B scope image of the target cross section. This effectively prevents the defect image appearing in the cross-sections i ₁ and i _{3 from} being recognized as two defects despite being caused by the single defect 55.

  In the cross-section enhancement processing, when the echo height at the same position in the cross sections i-2, i-1, i + 1, i + 2 is higher than the echo height at the position where the cross section i exists, the echo height at the corresponding position in the cross section i is set. Different processes may be performed as long as the data is corrected so that the data is increased. For example, the echo height data at a position of the cross section i is corrected to the average value only when the level is lower than the average value of the echo height data at the same position of the five cross sections i−2 to i + 2. It is also possible. However, the process of correcting the echo height data at a position of the cross section i to the maximum value of the echo height data at the same position of the five cross sections i−2 to i + 2 reduces the size of the defect. It is preferable in that it is not evaluated.

  In the cross-section enhancement processing of this embodiment, flaw detection data of two cross sections adjacent to the front in the scanning direction and two cross sections adjacent to the rear in the scanning direction are used for correcting the flaw detection data of the target cross section. It will be apparent to those skilled in the art that the number of cross sections used for correction can be changed as appropriate. The flaw detection data of n cross sections adjacent to the front in the scanning direction (n is an integer of 1 or more) and n cross sections adjacent to the rear in the scanning direction can be used for correcting the flaw detection data of the target cross section. is there.

  Further, only flaw detection data of only n cross sections adjacent to the front in the scanning direction are used for correcting the flaw detection data of the target cross section, and n cross sections adjacent to the rear in the scanning direction are not used for correcting the flaw detection data of the target cross section. It is also possible. Similarly, only flaw detection data of only n cross sections adjacent to the rear in the scanning direction is used for correcting the flaw detection data of the target cross section, and n cross sections adjacent to the front of the scanning direction are used for correcting the flaw detection data of the target cross section. It is also possible not to be. However, from the viewpoint of preventing the defect size from being evaluated smaller than the actual size, the cross section of n sheets (n is an integer of 1 or more) adjacent to the front in the scanning direction and n sheets adjacent to the rear in the scanning direction. It is preferred that both cross-section flaw detection data are used to correct the target cross-section flaw detection data.

  By performing the above-described cross section emphasis processing, it is possible to effectively avoid the defect image defect caused by the three-dimensional change in the defect shape, while the defect indication length in step S07 (the length in the scanning direction). ), The indicated length is measured longer than the actual length. For example, when the flaw detection data of two cross sections adjacent to the front in the scanning direction and two cross sections adjacent to the rear in the scanning direction are used for correcting flaw detection data of the target cross section, the instruction length is The measurement is longer than the actual length by four times the interval (that is, the length corresponding to the two cross-sections in the front and rear of the scanning direction).

In order to avoid such overestimation of the instruction length, it is preferable that the instruction length is corrected in step S07 according to the number of cross sections used for correcting the flaw detection data of the target cross section. For example, when flaw detection data of n cross sections adjacent to the front in the scanning direction (n is an integer of 1 or more) and n cross sections adjacent rearward in the scanning direction are used for correcting the flaw detection data of the target cross section. the instruction length L _d is preferably calculated based on the following formula:
L _d = (N−2n−1) · D,
Here, N is the number of cross sections in which defect images corresponding to the same defect appear continuously, and D is the distance between two adjacent cross sections.

Further, when only n cross sections adjacent to the front in the scanning direction are used for correcting the flaw detection data of the target cross section, and only n cross sections adjacent to the rear in the scanning direction are used for correcting the flaw detection data of the target cross section. If so, the indicated length L _d is preferably calculated based on the following formula:
L _d = (N−n−1) · D.

  In the present embodiment, the cross-section enhancement processing in step S10 is performed after the threshold cut processing (step S03) and before the evaluation gate setting processing (step S04). Can be performed at various processing timings.

  For example, the cross-section enhancement processing can be performed before the threshold cut processing in step S03. However, from the viewpoint of reducing the amount of calculation, it is more preferable that the cross-section enhancement processing is performed after the threshold cut processing in step S01. When the cross-section enhancement processing is performed after the threshold cut processing, the position of a certain processing target (that is, the path length) is set for all cross-sections used for correcting the target cross-section and the flaw detection data 20 of the target cross-section. When the echo height is 0, it is not necessary to calculate the maximum value for the position.

From the viewpoint of further reducing the amount of calculation, the cross-section enhancement processing can be performed after the evaluation gate setting processing (step S04). In the evaluation gate setting process, all echo heights outside the evaluation gate 33 are set to zero. Therefore, there are more cases where it is not necessary to calculate the maximum value as compared with the case where the cross-section enhancement processing is performed after the threshold cut processing.

  Further, from the viewpoint of further reducing the calculation amount, the cross-section enhancement processing can be performed after the shape identification processing (step S05). In the shape identification process, the echo heights at the positions corresponding to the shape image are all zero, so there is no need to calculate the maximum value compared to the evaluation gate setting process cross-section enhancement process. Will become even more.

  However, in order to suppress a defect from being overlooked due to a three-dimensional change in the shape of the defect, it is preferable that the cross-section enhancement process is performed after the threshold cut process.

  In the above-described embodiment, scanning of the ultrasonic beam in the x-axis direction is performed by causing the probe 6 to function as a phased array. However, the probe 6 is mechanically moved in the x-axis direction. The flaw detection data may be acquired while scanning. It will be apparent to those skilled in the art that the present invention can be applied without depending on the scanning method of the ultrasonic beam.

FIG. 1 is a diagram showing the configuration of an automatic ultrasonic flaw detection system used in the first embodiment of the ultrasonic flaw detection data processing method according to the present invention. FIG. 2 is a block diagram showing a configuration of an ultrasonic flaw detection data processing apparatus included in the automatic ultrasonic flaw detection system. FIG. 3 is a conceptual diagram showing a procedure for acquiring flaw detection data by the automatic ultrasonic flaw detection system of FIG. FIG. 4 is a conceptual diagram showing a procedure for acquiring flaw detection data by the automatic ultrasonic flaw detection system of FIG. FIG. 5A is a conceptual diagram illustrating an example of A scope data included in flaw detection data. FIG. 5B is a conceptual diagram showing another example of A scope data included in flaw detection data. FIG. 6 is a diagram conceptually showing the contents of the B scope image. FIG. 7 is a conceptual diagram showing an example of a B scope image generated from flaw detection data. FIG. 8 is a flowchart showing a processing procedure of ultrasonic flaw detection data in the first embodiment. FIG. 9 is a flowchart showing the procedure of the coupling process. FIG. 10 is a cross-sectional view showing a procedure for collecting sound detection data for a healthy part. FIG. 11 is a cross-sectional view illustrating a monitoring range that is a target of the coupling check process. FIG. 12A is a graph showing an example of A scope data when the coupling is good. FIG. 12B is a graph illustrating an example of A scope data when the coupling is defective. FIG. 13 is a diagram illustrating evaluation gates set in the evaluation gate setting process. FIG. 14 is a flowchart showing the procedure of the shape identification process. FIG. 15 is a diagram showing a shape identification gate set in the shape identification process. FIG. 16 is a diagram illustrating the relationship between the evaluation gate and the shape identification gate. FIG. 17A is a diagram showing a defect candidate area defined for an echo image. FIG. 17B is a diagram illustrating shape candidate regions extracted from the defect candidate regions. FIG. 17C is a diagram showing echo images separated by the A scope separation process and shape candidate regions defined for the echo images. FIG. 17D is a diagram illustrating shape candidate regions divided by the focal row direction separation processing. FIG. 17E is a diagram showing shape candidate regions recognized by the in-cross-region recognition processing after the focal row direction separation processing. FIG. 17F is a conceptual diagram of the B scope image from which the shape image has been deleted. FIG. 18A is a diagram illustrating A scope data before A scope separation processing is performed. FIG. 18B is a diagram illustrating the A scope data after the A scope separation process is performed. FIG. 19A is a conceptual diagram illustrating the focal row direction separation process. FIG. 19B is a conceptual diagram illustrating the focal row direction separation process. FIG. 19C is a conceptual diagram illustrating the focal row direction separation process. FIG. 20 is a graph showing conditions for separation in the focal row direction. FIG. 21 is a flowchart illustrating a first method for setting region representative points in shape identification processing. FIG. 22A is a diagram showing an extracted echo peak region. FIG. 22B is a diagram showing a peak rectangular area and area representative points defined for the echo peak area. FIG. 22C is a diagram illustrating a second method for setting region representative points. FIG. 23A is a flowchart illustrating a procedure of shape identification processing according to another embodiment. FIG. 23B is a diagram showing a recognized echo image. FIG. 23C is a diagram showing echo images separated by the focal law separation process and shape candidate regions defined for the echo images. FIG. 23D is a diagram illustrating shape candidate regions divided by the path direction separation process. FIG. 23E is a diagram showing shape candidate regions recognized by the intra-section region recognition processing after the focal row direction separation processing. FIG. 24 is a conceptual diagram illustrating processing for identifying a shape image and a defect image. FIG. 25 is a flowchart showing the procedure of defect identity determination processing. FIG. 26A is a diagram illustrating region representative points defined in the defect image. FIG. 26B is a diagram illustrating one method for determining region representative points in a defect image. FIG. 26C is a diagram showing another method for determining the region representative points in the defect image. FIG. 27 is a diagram illustrating processing for determining whether or not a plurality of defect images appearing on the same cross section are caused by the same defect. FIG. 28 is a diagram illustrating processing for determining whether or not a defect image appearing in an adjacent cross section is caused by the same defect. FIG. 29 is a diagram for explaining the reason why the defect images may not be correctly associated between adjacent cross sections when the defect extension direction changes three-dimensionally. FIG. 30 is a flowchart illustrating a method for processing ultrasonic flaw detection data according to the second embodiment. FIG. 31 is a diagram for explaining cross-section enhancement processing in the second embodiment. FIG. 32 is a graph for explaining cross-section enhancement processing in the second embodiment.

Explanation of symbols

1: Automatic ultrasonic flaw detection system 2: Subject 3, 4: Steel plate 5: Welded part 6: Probe 7: Scanning device 8: Control device 9: Display device 10: Ultrasonic flaw detection data processing device 11: Rail 12: Probe holding mechanism 13: Input device 14: Display device 15: Communication device 16: Storage device 17: Computing device 18: Ultrasonic flaw detection data processing program 19a: Sound part flaw detection data 19b: Shape information data 20: Flaw detection data 21: A scope data 22: Shape echo 23: Defect echo 24: Shape image 25: Defect image 31: Sound part 32: Monitoring range 33: Evaluation gate 33a: Evaluation gate upper end 33b: Evaluation gate lower end 34: Shape identification gate 34a: Shape identification Gate upper end 34b: Shape identification gate lower end 41: Echo image 42: Defect candidate area 43: Shape candidate area 44i: A scope data 45a: peak (shape candidate peak)
45b: peak 46: region representative point 47: echo peak region 48: peak rectangular region 51: defect image 52, 52i, 52j: rectangular region 53, 53i, 53j: region representative point 54: echo peak region 55: defect

Claims (22)

An ultrasonic flaw detection data processing program for processing flaw detection data obtained by performing ultrasonic flaw detection on a subject by digital arithmetic processing,
A shape identification processing step for performing processing for deleting the shape image corresponding to the shape of the subject from the ultrasonic image obtained from the flaw detection data, on the flaw detection data;
A defect image appearing in the ultrasonic image from which the shape image has been erased is recognized, and whether or not one of the defect images is caused by the same defect as the other defect image is determined according to a predetermined criterion. An identity determination step for determining and associating the defect image with the defect present in the subject;
An ultrasonic flaw detection data processing program for causing an arithmetic device to execute a dimension identification step for identifying a dimension of the same defect from the defect image associated with the same defect. The ultrasonic flaw detection data processing program according to claim 1,
The shape identification processing step includes
A defect candidate area defining step for defining a defect candidate area surrounding the echo image with respect to the echo image appearing in the ultrasonic image;
An extraction step of extracting, as the shape candidate area, an area that overlaps the shape identification gate area set based on the shape data of the subject from among the defect candidate areas;
When the echo image included in the shape candidate region has a plurality of echo height peaks, a division step of dividing the shape candidate region so that the valley portion of the echo height distribution becomes a boundary;
After the dividing step, a representative point setting step for setting a representative point representing the position of the shape candidate region;
Ultrasonic flaw detection data processing comprising: a shape determination step for determining that the echo image is a shape image to be erased when the representative point is included in a shape identification gate region set based on the shape data of the subject. program. The ultrasonic flaw detection data processing program according to claim 2,
The representative point setting step includes:
Recognizing an echo peak region corresponding to a peak having an echo height equal to or greater than a predetermined value with respect to the echo image;
An ultrasonic flaw detection data processing program comprising: determining a center of the echo peak region as the representative point. The ultrasonic flaw detection data processing program according to claim 2,
The representative point setting step includes:
Calculating a path direction average value distribution, which is an average value of effective value data other than “0” among the echo height data inside the shape candidate region, with respect to the path direction of the beam used for the ultrasonic flaw detection. When,
Calculating a distribution of vertical average values that are average values in a vertical direction perpendicular to the path direction of the effective value data;
Defining the representative point as an intersection of a straight line extending in the path direction at the peak position of the path direction average value distribution and a line extending in the vertical direction at the peak position of the vertical direction average value An ultrasonic flaw detection data processing program comprising: The ultrasonic flaw detection data processing program according to claim 2,
The division step includes an ultrasonic flaw detection data processing program including a path direction separation step that divides the shape candidate region in a path direction of a beam used for the ultrasonic flaw detection. The ultrasonic flaw detection data processing program according to claim 5,
The division step further includes a vertical direction separation step of further dividing the shape candidate region in a direction perpendicular to the path direction in the B scope image after the path direction separation step. The ultrasonic flaw detection data processing program according to claim 6,
The vertical separation step includes:
A maximum value curve extracting step of extracting a maximum value curve indicating a vertical distribution of the maximum value of the echo height in the path direction of the shape candidate region, perpendicular to the path direction;
An ultrasonic flaw detection data processing program comprising: dividing the shape candidate region with a valley portion of the maximum value curve as a boundary. The ultrasonic flaw detection data processing program according to claim 2,
The dividing step includes
A vertical separation step of dividing the shape candidate region in a direction perpendicular to a path direction of a beam used for the ultrasonic flaw detection in the B scope image;
An ultrasonic flaw detection data processing program comprising, after the vertical direction separation step, a vertical direction separation step of further dividing the path direction direction shape candidate region. The ultrasonic flaw detection data processing program according to claim 8,
The path direction separation step includes
A maximum value curve extracting step of extracting a maximum value curve indicating the distribution in the path direction of the maximum value of the echo height in the vertical direction of the shape candidate region; and
An ultrasonic flaw detection data processing program comprising: dividing the shape candidate region with a valley portion of the maximum value curve as a boundary. The ultrasonic flaw detection data processing program according to claim 1,
The identity determination step includes
A defect image representative point setting step for determining a representative point representing the position of the defect image;
An ultrasonic flaw detection data processing program comprising: a determination step of determining whether one defect image of the defect images is caused by the same defect as another defect image based on a distance between the representative points. The ultrasonic flaw detection data processing program according to claim 10,
The determination step includes a step of determining whether or not the two defect images are caused by the same defect from a distance between the representative points of the two defect images of the same cross section. . The ultrasonic flaw detection data processing method according to claim 10,
The determining step is a step of determining whether or not the two defect images are caused by the same defect from a distance in an in-plane direction parallel to the cross section between the representative points of the two defect images of different cross sections. Including ultrasonic flaw detection data processing program. The ultrasonic flaw detection data processing method according to claim 10,
The defect image representative point setting step includes:
An ultrasonic flaw detection data processing program comprising: determining, as the representative point, the center of an echo peak area having an echo height of a predetermined value or more in the defect image. The ultrasonic flaw detection data processing program according to claim 10,
The previous defect image representative point setting step is:
Calculating a path direction average value distribution, which is an average value of the echo height inside the defect image with respect to the path direction of the beam used for the ultrasonic flaw detection;
Calculating the distribution of the average value in the vertical direction, which is the average value in the vertical direction perpendicular to the path direction, of the echo height inside the defect image;
Defining the representative point as an intersection of a straight line extending in the path direction at the peak position of the path direction average value distribution and a line extending in the vertical direction at the peak position of the vertical direction average value An ultrasonic flaw detection data processing program comprising: The ultrasonic flaw detection data processing program according to claim 1,
In the ultrasonic flaw detection, the flaw detection data is acquired for a plurality of cross sections while scanning the probe along a predetermined scanning direction.
The ultrasonic flaw detection data processing program further includes:
Cross-section enhancement processing for correcting the flaw detection data of the target cross section of the plurality of cross sections using flaw detection data of a predetermined number of adjacent cross sections adjacent to the front of the target cross section in the scanning direction and / or rearward in the scanning direction. Causing the arithmetic unit to execute steps;
An ultrasonic flaw detection data processing program in which in the shape identification processing step, processing for deleting a shape image corresponding to the shape of the subject is performed on the corrected flaw detection data. The ultrasonic flaw detection data processing program according to claim 1,
In the ultrasonic flaw detection, the flaw detection data is acquired for a plurality of cross sections while scanning the probe along a predetermined scanning direction.
The ultrasonic flaw detection data processing program further includes:
With respect to the flaw detection data subjected to the shape identification processing step, the flaw detection data of the target cross section of the plurality of cross sections is set to a predetermined number adjacent to the front of the target cross section in the scanning direction and / or the rear of the scanning direction. An ultrasonic flaw detection data processing program for causing the arithmetic unit to execute an inter-section emphasis processing step for performing correction processing using flaw detection data of adjacent cross sections. The ultrasonic flaw detection data processing program according to claim 15 or claim 16,
In the inter-section enhancement processing step, the flaw detection data of the target cross section is corrected so that the echo height data at each position of the target cross section is increased according to the echo height at the same position of the adjacent cross section. Ultrasonic flaw detection data processing program. The ultrasonic flaw detection data processing program according to claim 17,
In the dimension identification step, based on the number of consecutive cross sections in which defect images corresponding to the same defect appear and the number of adjacent cross sections used for correcting the flaw detection data of the target cross section, the same An ultrasonic flaw detection data processing program that identifies the length of defects. The ultrasonic flaw detection data processing program according to claim 1,
Further, the calculation processing is executed to determine whether or not the coupling between the probe used for the ultrasonic flaw detection and the subject has been correctly maintained,
The coupling check step includes:
By comparing an echo height and a threshold value at each position within a predetermined monitoring range of the subject, the probe is automatically scanned while the probe is being scanned. Comprising determining whether the coupling has been maintained correctly,
The threshold structure is measured by injecting ultrasonic waves into the control structure by the probe in a state where coupling between the probe and the control structure is obtained. An ultrasonic flaw detection data processing program determined from the echo height at each position of a predetermined healthy part of the body. The coupling check method according to claim 19,
The ultrasonic flaw detection data processing program, wherein the control structure and the subject are the same. An ultrasonic flaw detection data processing apparatus for processing flaw detection data obtained by performing ultrasonic flaw detection on a subject by digital arithmetic processing,
A shape identification processing means for performing processing for erasing a shape image corresponding to the shape of the subject from an ultrasonic image obtained from the flaw detection data, on the flaw detection data;
A defect image appearing in the ultrasonic image from which the shape image has been erased is recognized, and whether or not one of the defect images is caused by the same defect as the other defect image is determined according to a predetermined criterion. An identity determination unit that determines and associates the defect image with the defect present in the subject;
An ultrasonic flaw detection data processing apparatus comprising: a dimension identification unit that identifies a dimension of the same defect from the defect image associated with the same defect. An ultrasonic flaw detection data processing method for processing flaw detection data obtained by performing ultrasonic flaw detection on a subject by digital arithmetic processing,
A shape identification processing step for performing processing for deleting the shape image corresponding to the shape of the subject from the ultrasonic image obtained from the flaw detection data, on the flaw detection data;
A defect image appearing in the ultrasonic image from which the shape image has been erased is recognized, and whether or not one of the defect images is caused by the same defect as the other defect image is determined according to a predetermined criterion. An identity determination step for determining and associating the defect image with the defect present in the subject;
An ultrasonic flaw detection data processing method comprising: a dimension identification step for identifying a dimension of the same defect from the defect image associated with the same defect.

Images