JPH075155A - Ultrasonic automatic flaw type identification method - Google Patents

Abstract

PURPOSE:To provide an ultrasonic method for automatically deciding the type of flaw accurately and quickly. CONSTITUTION:An echo from a material 1 under inspection is received by a pulser receiver 3 through a probe 2 and subjected to A/D conversion through an A/D conversion board 4 before being fed to a circuit 5 for deciding the presence/absence of waveform feature amount. The decision circuit 5 makes a decision whether a waveform S/N ratio is exceeded and when a decision is made that the waveform feature amount is present, the feature amount of that waveform is sampled and a waveform feature amount measuring circuit 7 is actuated to extract the waveform feature amount. The measuring circuit 7 conducts predetermined operation and delivers a data to a neural circuit 8. The neural circuit 8 is previously provided with the data of feature amount for each type of defect and collectively learns the data fed to the neural circuit 8 based on the previously provided data thus deciding the type of defect.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an ultrasonic flaw detection method for detecting cracks or inclusions in a material to be inspected using ultrasonic waves.

[0002]

2. Description of the Related Art A conventional method for determining the type of defects or inclusions in a material to be inspected, such as a steel plate, is to propagate ultrasonic waves simultaneously or continuously from a plurality of different angles, and to determine the amplitude of echoes by the ultrasonic waves. The evaluation is performed based on the confirmed echo characteristics. Signal intervals, signal durations, etc. are used for this echo characteristic.

[0003]

However, in the conventional method, it is necessary to perform flaw detection from at least two directions of, for example, 0 to 45 degrees and 45 to 90 degrees with respect to one reflection factor, and to inspect the entire surface of the material to be inspected. However, there is a problem that the inspection time is long. The present invention has been made in view of such circumstances, and collects characteristics of center frequency, peak frequency, rise time, fall time, duration, and wave height and a defect area, and based on these characteristics, a neuro circuit. It is an object of the present invention to provide an ultrasonic automatic flaw type discriminating method capable of discriminating with only one whole surface inspection by discriminating the type of the reflection factor.

[0004]

An ultrasonic automatic flaw type determining method according to the present invention uses ultrasonic waves to perform flaw detection on the surface of a material to be inspected, and echoes due to reflection factors such as defects and inclusions. In the ultrasonic automatic flaw type discriminating method for discriminating the type of the reflection factor, the characteristics of the echo center frequency, peak frequency, rise time, fall time, duration and wave height and the defect area are collected. The collected characteristic is given to a neuro circuit to which characteristic data for each reflection factor is given in advance, and the type of the reflection factor is discriminated by the neuro circuit.

[0005]

In the present invention, the rebounding ultrasonic wave (echo)
Based on the characteristics of center frequency, peak frequency, rise time, fall time, duration, and wave height, the neuro circuit determines the type of the reflection factor, so that only one full inspection can accurately measure the entire surface of the inspected material. The defect distribution and its defect type can be discriminated and the inspection time can be shortened. Further, discrimination from pseudo echo can be automatically performed.

[0006]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be specifically described below with reference to the drawings showing the embodiments. FIG. 1 is a block diagram showing a configuration of an apparatus used for carrying out an ultrasonic automatic flaw type determining method according to the present invention. In the figure, reference numeral 1 denotes a material to be inspected, such as a steel plate, and an incident angle of an ultrasonic wave is fixed to the material to be inspected 1 and a probe 2 for flaw detection of the material to be inspected 1 is installed. A scanner 10 and a controller 11 are connected to the probe 2,
Two-dimensional scanning of the entire surface of the material to be inspected and its probe position are detected. The echoes propagated from the surface to the bottom of the material 1 to be inspected, and the reflected echoes during that time are received by the pulsar receiver 3 via the probe 2. The signal received by the pulsar receiver 3 is A / D converted by the A / D conversion board 4 and sent to the waveform characteristic amount presence / absence determination circuit 5 so that the presence / absence of the waveform characteristic amount is determined. is there. The waveform feature amount presence / absence determination circuit 5 digitally determines whether or not the waveform exceeds the S / N ratio, and determines the waveform feature amount to be “present” or “not present”.

If the waveform characteristic amount presence / absence determining circuit 5 determines that the waveform characteristic amount is "absent", no data is stored. On the other hand, when the waveform feature amount presence / absence determination circuit 5 determines that the waveform feature amount is “present”, the feature amount of the waveform determined as “present” is sampled and the waveform feature amount extraction circuit 7 is activated. , 6 types of waveform feature quantities are extracted. Further, after the prescribed calculation is performed in the waveform feature amount extraction circuit 7, this data is given to the neuro circuit 8.

The position signal detected by the controller 11 is given to the signal processing device 9, and a start signal and a trigger signal are also given from a device (not shown). Then, only when the waveform feature quantity presence / absence determination circuit 5 determines “present”, the signal processing device 9 is activated to perform a process of determining a defect central portion based on the above-mentioned signal, and Area, defect center position) neuro circuit 8
It is designed to give to Specifically, when the waveform feature amount presence / absence determination circuit 5 determines that “present”, the position information and the other five waveform feature amount data are stored in the memory in the signal processing device 9. That is, when the two-dimensional scanning of the entire surface of the inspected material is completed, the defect area and the defect center position can be determined from the position distribution in which the information in the memory is written, and the waveform feature amount data at that position can be extracted. The neuro circuit 8 judges the flaw type from this data and the data from the waveform characteristic amount extraction circuit 7, and the result is stored in the disk 6.

The six types of waveform feature amounts extracted by the above-mentioned waveform feature amount extraction circuit 7 will be specifically described below. 1. Peak amplitude (%) 2. Rise time (μsec.) 3. Fall time (μsec.) 4. Envelope length (μsec.) 5. Peak frequency (MHz) 6. Center frequency (MHz)

These characteristic quantities will be described using the following graphs. FIG. 2 is a graph showing a rectified waveform, with the vertical axis representing the received signal strength and the horizontal axis representing time. The “peak amplitude” is the received signal strength at the peak in FIG. FIG. 3 is a graph showing a waveform envelope in which the vertical axis represents the received signal strength and the horizontal axis represents time. The “rise time” is the time (ba) from the point a to the point b in FIG. 3, and the “fall time” is
It is the time (dc) from the point c to the point d in FIG. The “envelope length” is the time (d−a) from point a to point d in FIG.

FIG. 4 is a graph showing the energy spectrum, with the vertical axis representing the received signal strength and the horizontal axis representing the frequency. The “peak frequency” is the frequency at point g in FIG. The “center frequency” is an average value ((f + h) / 2) of the frequency at the point f and the frequency at the point h.

Next, the features of the above-mentioned waveform feature amount will be described. The ultrasonic wave has a characteristic that the phase is inverted when the ultrasonic wave hits from a hard one to a soft one, so that the rising time (2) and the falling time (3) tend to be deviated in a cracking defect or a bubble. In the case of inclusions, it does not change much, but between alumina-based and manganese-based ones, the alumina-based one is characterized by a larger change. The center frequency (6) and the peak frequency (5) have high intensity at low frequencies in the case of inclusions, and have high intensity at higher frequencies in the case of bubbles. Among the inclusions, manganese-based inclusions tend to have high strength at particularly low frequencies.

The peak frequency and duration (4. envelope length) are affected by the defect length and tend to be longer in the case of alumina-based inclusions. This is because the alumina system allows ultrasonic waves to pass easily and has a small wavelength. The highest peak value of the echo and the area where the defect signal appears are influenced by the direction and shape of the defect. The planar cracking defect has a small area and a high echo frequency height. In addition, the inclusions are characterized by a large area and a low echo frequency height. The peak amplitude (1) is influenced by the defect area and shape, and has a feature that it becomes higher for planar defects.

Since it is difficult to determine the defect type with one of these pieces of information, the defect type is comprehensively learned and judged by the neuro circuit 8 based on the above information. judge.

As a result of distinguishing three different kinds of defects (Al ₂ O ₃ , MnO ₂ , cracks, etc.) by using the method of the present invention, the flaw-predictive ratio was 80% and the number of samples was 40. .
Moreover, as a result of making the above-described inclusions and bubbles, the result was that the flaw-predictive value was 90% and the number of samples was 50. Thus, according to the method of the present invention, it is possible to accurately determine the flaw type by measuring from one angle.

[0016]

As described above, according to the ultrasonic flaw detection method of the present invention, the reflection factor in the neuro circuit is based on the echo characteristics of the center frequency, peak frequency, rise time, fall time, duration and wave height. By discriminating the type, the present invention has an excellent effect such that the flaw type can be promptly and accurately discriminated by only one whole surface inspection.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of an apparatus used for carrying out an ultrasonic automatic flaw type determination method according to the present invention.

FIG. 2 is a graph showing a rectified waveform.

FIG. 3 is a graph showing a waveform envelope.

FIG. 4 is a graph showing an energy spectrum.

[Explanation of symbols]

 1 material to be inspected 2 probe 3 pulser receiver 4 A / D conversion board 5 waveform feature quantity presence / absence judgment circuit 6 disk 7 waveform feature quantity extraction circuit 8 neuro circuit 9 signal processing device 10 scanner 11 controller

Claims (1)

[Claims] 1. An automatic ultrasonic flaw for determining the type of the reflection factor by performing flaw detection on the surface of the material to be inspected using ultrasonic waves and receiving a reflection echo caused by a reflection factor such as a crack or inclusion. In the species discriminating method, characteristics of the center frequency, peak frequency, rising time, falling time, duration, and wave height of the echo and the defect area are sampled, and the sampled characteristics are given in advance by characteristic data for each reflection factor. An automatic ultrasonic flaw type discriminating method, characterized in that the type of the reflection factor is discriminated by applying it to a given neuro circuit.