JPH0245823B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a method for detecting defects using an ultrasonic method. The purpose is to determine the degree of harmfulness of the defect.

In cast materials, horizontal hole-like cavities may occur in the material due to gas trapped inside during the casting process. In many cases, the shape of the defect and its harmful effect on the strength of the material are clarified depending on the purpose of use. regulated by size. However, due to the directionality of the stress applied to the material, the level of harm is usually different depending on whether a horizontal hole-like defect of the same length is located in the X direction or in the Y direction. Conversely, the size of defects that can be tolerated in terms of material strength usually differs between the X direction and the Y direction. Under such circumstances, it is necessary to accurately recognize the shape (direction, inclination, size) of the horizontal hole-like defect.

However, as a method for detecting such internal defects,
2. Description of the Related Art Ultrasonic methods that transmit an ultrasonic beam into the interior of a material to be inspected and receive echoes from defects have been conventionally provided and put into practical use. However, in the conventional ultrasonic method, the probe is simply moved in a predetermined direction within the flaw detection area of the test material and the echoes are captured, so the echoes cannot be determined from the flaw detection pattern displayed on the display medium. The number of elements was extremely limited, and it was difficult to completely capture the shape of the hole-like defect.

Therefore, the present inventors used an angle probe that transmits and receives an ultrasonic beam in a flaw detection area inside the test material so as to intersect the center of rotation, and rotated the angle probe around the center of rotation. The ultrasonic beam is transmitted and received from each direction of 360 degrees, and the peak value of the echo generated within the flaw detection gate is displayed as a flaw detection pattern on the display medium in correspondence with the incident direction of the ultrasonic beam.
We have invented a method to determine the direction, inclination, size, and depth of defects from this flaw detection pattern. Unlike the conventional ultrasonic method, this method has the advantage of being able to reliably recognize the shape, ie, direction, inclination, size, and depth of defects inside the specimen. However, since there is only one probe, in order to identify the depth, it is necessary to repeat the same operation several times while changing the flaw detection depth, which has the disadvantage that the flaw detection time is somewhat longer. In addition, since the degree of toxicity of defects is determined by the worker, there is no particular problem when detecting small materials, but it cannot be used when detecting particularly large materials, and the degree of toxicity is determined by on-site flaw detection online. I cannot say that I have judged it.

In view of these problems, the present invention uses a plurality of angle probes with different flaw detection depths at the same time and obtains flaw detection patterns of multiple channels, thereby simultaneously recognizing the depth. The defect information included in the flaw detection pattern is decoded using a correlation method, and the degree of toxicity is determined online in a very short time. When detecting defects using the ultrasonic method, an angle probe with multiple channels of different depths is used to transmit and receive an ultrasonic beam so as to intersect the rotation center in the detection area inside the material being tested. While rotating the bevel probe of each channel around the center of rotation, ultrasonic beams are transmitted and received from each direction of 360°, and the peak value of the echo generated within the flaw detection gate of each channel is measured in each direction of incidence for each channel. By creating multiple flaw detection patterns corresponding to the The point is to judge the degree of harm.

Next, the principle of the present invention will be explained. FIGS. 1A and 1B show an ultrasonic beam 4 when detecting a horizontal hole-like defect 3 with an inclination η in a test material 2 by the oblique ultrasonic flaw detection method using an oblique probe 1 with a refraction angle θ. It shows the route of However, it is assumed to be θη. As shown in Fig. 1A and B, the probe 1 is moved 1 in the direction of the arrow from the starting line position (the position in the longitudinal direction when the defect is projected onto the material surface) around the center of rotation.
When the ultrasonic beam 4 is transmitted while rotating,
There are two directions in which the ultrasonic beam 4 is perpendicularly incident on the defect 3: positions rotated by α ₁ and α ₂ from the starting line.
This is shown using the symbols in the figure: P→Q→R→
It is shown as Q→P and P'→Q'→R'→Q'→R'. Therefore, it can be seen from a simple geometrical calculation that the following relationship exists between α, η, and θ.

cos (α ₂ − α ₁ /2) = tan η / tan θ Actually, while rotating the probe 1,
The ultrasonic beam 4 is transmitted along the trajectory C of the incident point shown in
When the peak value of the echo obtained at that time is plotted against the incident direction α, a flaw detection pattern as shown in FIG. 2 is obtained. The angle-related characteristics of such a flaw detection pattern are determined by a formula, and the maximum peak value is determined by the size of the defect 3. For the relationship between the size of the defect 3 and the echo height, a conventional flaw detection equation can be used, or an actual value from an artificial defect can be used. For this reason, the flaw detection pattern shown in Fig. 2 is based on the direction and inclination of the defect 3.
It turns out that it contains size information.

Usually, the direction of the starting line is selected to be a special direction of the shape of the material 2 to be inspected, such as the longitudinal direction, so the direction of the defect 3 is measured from such the starting line. At that time, the direction of defect 3 is given by the following equation.

β ₁ +β ₂ /2 (β ₂ −β ₁ 180°) β ₁ +β ₂ /2+180° (β ₂ −β ₁ 180°) The slope η of the defect 3 can be obtained from the following equation.

cos (β ₂ - β ₁ /2) = tan η / tan θ (β ₂ - _{β 1} 180°) cos (180 ° - β _{2 - β 1} /2) = tan η / tan θ (β ₂ - β ₁
180°) Such operations can be performed by people who have the ability to recognize shapes, but it takes a lot of time. On the other hand, from a practical point of view, it is sufficient if the inclination of the defect 3 can be detected and determined from 0° to 45° with an error of ±2.5°. Therefore, we used a correlation method as a means of having a machine perform the above-mentioned judgments.

The cross-correlation process is given by the following equation.

C(S)=1/360∫ ³⁶⁰ ₀ r(α) f(α+S) dα Here, f(α) and r(α) are arbitrary functions defined by 0α360. In recent years, cross-correlation processing has been applied to pattern recognition technology in image processing, error correction technology in communication technology, etc., but the present invention considers the shape of the flaw detection pattern and compares an unknown flaw detection pattern with a known reference pattern. The pattern is then considered similar to the reference pattern with the highest similarity.

A brief explanation of this process is as follows. FIG. 3 shows an example of one flaw detection pattern and five types of reference patterns r ₁ (α) to r ₅ (α). Humans can compare the number of mountains, the distance between them, and the slope of the mountains using their natural pattern recognition capabilities. Then, check the overlap between the detection pattern and the reference pattern, and find the one with the greatest degree of overlap r ₄ (α)
is determined to be the same as Such an operation is a correlation process, and in order to quantitatively determine the degree of overlap, a formula has been introduced as a value proportional to the area of the overlapped portion. S in the equation is the amount of shift on the horizontal axis when two patterns are superimposed as described above.

In this case, the reference pattern is originally created from the horizontal hole-like artificial defect by actual measurement or by calculation according to a formula, and since the characteristics (direction, inclination) of these artificial defects are known, the amount of deviation can be calculated. S
gives the deviation between the direction of the artificial defect and the direction of the defect just detected.

Applying the above principle, we created 10 reference patterns with the same direction, length, and depth of the reflective surface, and with inclinations of 0, 5, 10, 15, 20, 25, 30, 35, 40, and 45 degrees.
The direction and inclination of the defect can be detected by preparing data obtained from various types of side-hole artificial defects and correlating them with the flaw detection pattern just obtained. Defect classification can be best achieved by keeping the maximum value of the reference pattern constant at this time. Furthermore, the correlation method is also effective when the flaw detection pattern as shown in FIG. 4 contains a lot of noise, as it is often used for signal processing to improve S/N. For flaw detection patterns containing such noise (which is the usual case in actual flaw detection),
Operations such as measuring the number of mountains and the distance between them are extremely difficult for machines.

Next, an embodiment of the defect detection device according to the present invention will be described. 5 and 6 show a detection device using a four-probe type local water immersion method, 5 is a support plate provided to be able to scan in the X-axis and Y-axis directions, and the lower surface of this support plate 5. , a water tank 6 that can be moved freely over the test material 2
A probe rotation mechanism 7 is disposed within the water tank 6, and water 8 is supplied therein. The probe rotation mechanism 7 includes a cylindrical rotating shaft 12 rotatably held in a bearing case 11 on a support plate 5 via bearings 9 and 10, and a probe mounted on the lower end of this rotating shaft 12. holder 13, and the probe holder 13 has four probes arranged at a pitch of 90° in the circumferential direction.
Angle angle probes 14, 15, 16, and 17 are incorporated. Although the probes 14, 15, 16, and 17 are arranged so that they can detect flaws within a range of 20 mm from the surface of the test material 2 at a pitch of 8 mm, the present invention is not limited thereto. For example, each probe 14, 15, 16, 17 has an ultrasonic beam 14a, 15a, with a refraction angle θ of 45°.
The beam diameter of the ultrasonic beams 16a and 17a is 9.2 mm, and as shown in FIG.
5a, 16a, 17a are depth 2.5, 7.5, 12.5,
They are arranged at slightly different distances from the rotation center so that they intersect with the rotation center at a position of 17.5 mm. Therefore, the probes 14, 15, 16, 17 are
When rotated, the respective flaw detection gates 14b and 15
If echoes within b, 16b, and 17b are received, the detection area becomes an abacus bead shape (the shaded area) as shown in FIG. 7, and echoes of defects within this range can be captured. 18 is the probe rotation mechanism 7
19 is a rotary encoder that detects the rotation angle of the probe rotation mechanism 7. These are mounted on the support plate 5, and are interlocked and connected to the rotation shaft 12 via belt transmission mechanisms 20 and 21. has been done. 22 each probe 14 of the probe rotation mechanism 7;
This is a slip ring for transmitting and receiving ultrasonic signals to and from terminals 15, 16, and 17.

FIG. 8 shows a block diagram of the signal processing device, and the operation during flaw detection will be explained with reference to FIG. 23 is a scanning mechanism that scans the probe rotation mechanism 7 in the X-axis and Y-axis directions, and 24 is a controller that controls this scanning mechanism 23. 25, 26, 27,
28 is a pulser receiver provided corresponding to the probes 14, 15, 16, and 17 of 4 channels CH ₁ , CH ₂ , CH ₃ , and CH ₄ , and 29 , 30 , 31 , and 32 are analog peak hold circuits. , each probe 14,1 while rotating the probe rotation mechanism 7 during flaw detection.
When the ultrasonic beams 14a, 15a, 16a, 17a are transmitted to the specimen 2 by the probes 5, 16, 17, the reflected echo from the defect 3 in the specimen 2 is transmitted to the probes 14, 15, 16, 17. are received by the pulser receivers 25, 26, 27, 28 through the analog peak hold circuit 2, and the peak values of the signals generated at the flaw detection gates 14b, 15b, 16b, 17b are transferred to the analog peak hold circuit 2.
9, 30, 31, 32 hold. 33,3
4, 35, 36 are A/D converters for peak values, 3
7, 38, 39, 40 are shift registers,
Peak value from 128 incident directions every (360/128)°,
That is, the flaw detection pattern is temporarily stored. Note that the direction of incidence is detected by the rotary encoder 19, and it goes without saying that a flaw detection pattern is created for each channel corresponding to the direction of incidence. 4
1 is a memory such as RAM or ROM, which stores 10 types of reference patterns as described later. 42 is a data selector, and 43 is a digital correlator, which sequentially performs correlation processing between the flaw detection pattern sent for each channel through the data selector 42 and the reference pattern stored in the memory 41. A digital peak detector 44 detects the maximum correlation value between the flaw detection pattern of each channel and the reference pattern. Also, the amount of deviation S at this time is detected by the data latch 45. The maximum value Cmi (i=1, 2, 3, 4) of the correlation value for each channel detected in this way and the deviation amount Si
(i=1, 2, 3, 4) and the current position (x, y) of the scanning mechanism 23 are ordered by the data selector 46 and stored in the shift register 47.
Note that the signal processing after the digital correlator 43 is as follows.
Although it can be executed using software using a computer as described later, it usually takes a long time in that case.

FIG. 9 shows the D/A converted output of the digital correlator 43, and FIG. 10 shows the flaw detection pattern at that time. The defect 3 at this time is as shown in FIGS. 11A and 11B. The slope η of defect 3 is 24°, and it can be seen from FIG. 9 that it has the maximum correlation with the reference pattern of 25°. Furthermore, since the peak value of the correlation is shifted from the center (=0°) by +60°, it can be seen that the direction of this defect 3 is 60° with respect to the specific direction of the specimen 2.

Such a determination can actually be easily and quickly executed by transferring the data in the shift register 47 to a computer (not shown) through the interface 48. The processing within the computer will be described below.

As shown in FIG. 7, the ultrasonic beams 14a, 15a, 16a, and 17a of each channel overlap in the test material 2, so one defect 3 has a significant correlation value Cmi for multiple channels. appears. Therefore, in that case, the depth of the reflective surface of the defect 3 is calculated using the following formula.

Next, the deviation of the reflecting surface of the defect 3 from the center line of the ultrasonic beam is corrected.

C′mi={1+k|d−di|/di}Cmi Here, k is a proportionality constant depending on the probe diameter, frequency, etc. The length of the defect 3 can be determined from C′mi obtained in this way. 12th
The figure shows the actual measured value (marked with an X) of the maximum correlation value between a flaw detection pattern obtained from a defect with a diameter of φ2, a length of l, and an inclination of 35° and a reference pattern with an inclination of η = 35°.
From such data, it is possible to calculate the effect of the received sound pressure shift on the correlation value and determine k.

During flaw detection, the XY position of the probe rotation mechanism 7 (corresponding to the XY position of the defect) is stored in the shift register 47 from the controller 24 via the data selector 46, and is stored in the shift register 47 via the interface 48.
The data is transferred to the computer along with data such as the maximum correlation value. Then, this data is later displayed on the CRT as position information of the defect 3 when displaying the detection data of the harmful defect. 49 in Figure 8
is a timing controller. The probe rotation mechanism 7 can also be replaced with an electronic scanning probe.

FIG. 13 is a block diagram showing the flow of signals when the shape of the defect 3 is recognized on the software of the computer 50. Components with the same names as those in FIG. 8 are denoted by the same reference numerals, and explanations thereof will be omitted.
In this case, the pulsar receivers 25, 26, 2
7, 28, analog peak hold circuit 29, 3
0, 31, 32 and A/D converters 33, 34, 3
5, 36, probes 14, 15, 16, 17
The peak values of the echoes obtained corresponding to the rotation angles are collected for one rotation and stored in the memory 51 to create a flaw detection pattern. The memory 51 serves as an external memory for the computer 50, and the probes 14,1
5, 16, and 17 transmit and receive the first ultrasonic beam in the next rotation, the flaw detection pattern in the memory 51 is transferred to the internal memory 53 via the interface 52 of the computer 50 by DMA. The computer 50 includes a CPU 54, a CRT display 55 for displaying flaw detection data, a floppy disk 56 for storing flaw detection data, and the like.

Computer 50 processes the signals according to the flowchart shown in FIG. In Fig. 14, the flaw detection pattern is indicated by Ai (α), and its subscript i
is the channel and α is the rotation angle. The reference pattern is denoted by Rj (α), where j indicates the slope of the defect.
The correlation equation between Ai (α) and Rj (α) is given by Rji (S) = 1/360∫ ³⁶⁰ ₀ Ai (α) Rj (α + S) dα. This formula is generally provided as software.

A case will be described in which the present invention is applied to flaw detection of a cast prober 57. The flaw detection area 58 of this propeller 57 is 1200 x 800 as shown in the shaded area in Fig. 15.
mm/wing, and the flaw detection depth is 20 mm from the surface. The horizontal hole-shaped defects 3 included in this flaw detection area 58 are detected, and only harmful defects are output from them. However, the harmful defects of the propeller 57 are determined by the projected length in the circumferential direction, taking into account the direction in which stress is applied. Evaluate by slc.
Further, the degree of harmfulness also differs depending on the depth q of the tip of the defect 3. Therefore, the degree of harmfulness of defect 3 is as shown in Fig. 16 A and B.
It is defined by the defect projection length lc shown in FIG. 17 and the defect tip depth q shown in FIG. From this, the computer 50 performs the following as final processing.

(i) Calculation of defect projected length lc lc = lcosη＊sinα where l is the actual length of the defect, η is the slope of the defect, α
is the direction of the defect starting from the radial direction.

(ii) Calculation of defect tip depth q q=Max{0, (d-1/2lsinη)} Here, d is the center depth of the defect (iii) Determination of whether to output the defect 0≦q≦10 When lc≧4mm When 0≦q≦20, lc≧8mm These are defined as allowable projected defect dimensions in two stages.

Therefore, the computer 50 performs the calculation as shown in the flowchart of FIG.
on the printer when any of the conditions are met.
Output as a harmful defect along with XY position information.

As described in detail in the embodiments above, according to the method of the present invention, bevel probes with multiple channels with different flaw detection depths are used, and ultrasonic beams are transmitted and received while rotating these probes at the same time. Since the flaw detection pattern is obtained by capturing the peak value of the echo generated within the flaw detection gate of the channel, it is not necessary to detect the same location several times while changing the flaw detection depth, unlike when using a single probe. You can work efficiently without any problems. In addition, multiple flaw detection patterns are created in which the peak value of the echo generated within the flaw detection gate of each channel corresponds to the incident direction, and the correlation between each of these flaw detection patterns and a preset reference pattern is calculated. By determining the direction, slope, size, and depth of the defect,
The directionality of defects can be easily and reliably deciphered by applying the formula shown on page 7, and by skillfully utilizing known reference patterns,
By simple digital calculation of the formula ~ shown on page 16,
It becomes possible to determine the direction, inclination, and size of defects at high speed from patterned flaw detection signals. Moreover, by calculating the degree of harmfulness of a defect using the formula shown on page 19, it is possible to calculate the degree of harmfulness of a defect extremely quickly with simple signal processing. Therefore, compared to the case of making a judgment based on the flaw detection pattern itself, it is easier to make a judgment by a machine using a computer or hardware logic circuit, and in this respect as well, speeding up can be significantly promoted.

[Brief explanation of the drawing]

The drawings illustrate an embodiment of the present invention, and FIG. 1A and B are explanatory diagrams showing the path of an ultrasonic beam, FIG. 2 is a waveform diagram of a flaw detection pattern, and FIG. 3 is an explanation of correlation processing. Figure 4 is a waveform diagram of an actual flaw detection pattern, Figure 5 is a sectional view of the probe rotation mechanism,
Fig. 6 is a bottom view of the probe holder, Fig. 7 is an explanatory diagram of the flaw detection area, Fig. 8 is a block diagram, Fig. 9 is a waveform diagram when the output of the digital correlator is D/A converted, Figure 10 is a waveform diagram of a flaw detection pattern obtained from a defect with an inclination η≒24°, Figures 11A and B are explanatory diagrams of the defect, and Figure 12 is the relationship between the length of the defect and the ratio of the maximum correlation value. Fig. 13 is a block diagram, Fig. 14 is a flowchart, Fig. 15 is a partial plan view of the propeller, Fig. 16 A and B are explanatory drawings of harmful defects, Fig. 17 is the depth of the tip of the defect. FIG. 3 is an explanatory diagram showing the relationship between the projection defect length and the allowable projected defect length. 1, 14, 15, 16, 17...bevel angle flaw detector,
2... Test material, 3... Defect, 4, 14a, 15
a, 16a, 17a... Ultrasonic beam, 7... Probe rotation mechanism, 14b, 15b, 16b, 17b
...Flaw detection gate, 25, 26, 27, 28... Pulsar receiver, 29, 30, 31, 32... Analog peak hold circuit, 37, 38, 39, 4
0...A/D converter, 41, 51...memory,
43...digital correlator, 50...computer.

Claims (1)

[Claims] 1. When detecting defects inside a material to be inspected using the ultrasonic method, an angle probe is used in which an ultrasonic beam is transmitted and received so as to intersect the center of rotation in the flaw detection area inside the material to be inspected, and multiple channels with different detection depths are used. The bevel probe of each channel is rotated around the center of rotation while transmitting and receiving ultrasonic beams from each direction of 360°, and the peak value of the echo generated within the flaw detection gate of each channel is calculated. A plurality of flaw detection patterns are created for each channel, each corresponding to the direction of incidence, and by determining the correlation between each of these flaw detection patterns and a preset reference pattern, the direction and inclination of the defect can be determined.
A defect detection method using an ultrasonic method, which is characterized by deciphering the size and depth and determining the degree of harmfulness of the defect from the results.