JP2003066017A - Ultrasonic echo signal processing device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To accurately detect a malfunctioning spot in an object to be inspected by reducing a pseudo echo from an ultrasonic echo signal. SOLUTION: When inspecting a defective spot in an object to be inspected in accordance with the ultrasonic echo obtained from the object to be inspected, a waveform converting part 12 creates a conversion factor by applying rotary wavelet conversion to an echo signal. A significant component selecting part 13 selects the malfunctioning spot the conversion factor in accordance with the direction of an equiphase face, for example, it selects a conversion factor relating to an echo corresponding to an injury as a significant component. A significant component synthesizing part 14 synthesizes the significant component based on a weighting factor specified in advance, and outputs it as a filtering signal. As a result, a pseudo echo is reduced from the filtering signal.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a processing device for analyzing and processing an ultrasonic echo signal, and particularly to a nondestructive inspection of an object to be inspected by an ultrasonic flaw detection test.
The present invention relates to an ultrasonic echo signal processing device for analyzing and processing an ultrasonic echo signal obtained from an object to be inspected.

[0002]

2. Description of the Related Art Generally, an ultrasonic flaw detection test (UT) may be used when a non-destructive inspection of an object to be inspected. In the UT, at the time of nondestructive inspection of an object to be inspected, a so-called ultrasonic sensor is moved on the object to be inspected to obtain an echo (ultrasonic echo signal) reflected from a scratch or the like in the object to be inspected. Then, this ultrasonic echo signal is analyzed and processed to find a defective portion such as a scratch on the inspection object.
In the UT, flaw detection may be performed using sensors with different ultrasonic wave incident angles.

At the time of UT, there is an echo (pseudo echo) from, for example, a welded portion in addition to an echo from a flaw as an inspection target (hereinafter simply referred to as a flaw echo). In order to find a defective portion such as a scratch, it is desirable to distinguish the scratch echo from the pseudo echo and reduce the pseudo echo.

Conventionally, in UTs, in order to reduce noise, a bandpass filter (bandpass filter) is used to improve the signal-to-noise ratio (S / N ratio). If a bandpass filter is used, It is possible to effectively remove signals other than the ultrasonic signal frequency band oscillated from the acoustic wave sensor. That is, since the scratch echo is a signal having the ultrasonic signal frequency band, the band-pass filter allows the scratch echo to pass and blocks signals outside the ultrasonic signal frequency band.

[0005]

By the way, the above-mentioned pseudo echo also has an ultrasonic signal frequency band, and as a result,
By simply using a bandpass filter, it is possible to reduce signals other than the ultrasonic signal frequency band, for example, electrical noise as described above, but since the pseudo echo is in the same frequency band as the scratch echo, It is difficult to reduce pseudo echo with a pass filter.

Therefore, if the pseudo echo exists in addition to the scratch echo, it becomes extremely difficult to distinguish the scratch echo from the pseudo echo, and it is impossible to perform flaw detection with high accuracy, and a defective portion such as a scratch on the object to be inspected. Considering the pseudo echo when determining whether or not there is a flaw, a flaw detection test must be performed at a plurality of incident angles using various ultrasonic sensors.

As a result, there is a problem in that a non-destructive inspection of the object to be inspected by the UT is not only time-consuming, but also a defective portion such as a scratch cannot be accurately detected.

An object of the present invention is to provide an ultrasonic echo signal processing device capable of accurately detecting a defective portion of an object to be inspected.

Another object of the present invention is to provide an ultrasonic echo signal processing device capable of outputting an echo signal with reduced pseudo echo as a filtering signal.

[0010]

According to the present invention, an ultrasonic echo signal processing apparatus used when inspecting a defective portion of an object to be inspected according to an ultrasonic echo signal obtained from the object to be inspected. That is, a waveform conversion unit that generates a conversion coefficient by performing rotary wavelet conversion of the ultrasonic echo signal, and a conversion coefficient related to the echo corresponding to the defective portion from the conversion coefficient according to the direction of the equiphase surface. An ultrasonic echo signal processing device, comprising: a significant component selecting unit that selects as a significant component; and a coefficient synthesizing unit that synthesizes the significant component based on a weighting coefficient defined in advance and outputs as a filtering signal. can get.

As described above, since the conversion coefficient relating to the echo corresponding to the defective portion is selected as the significant component according to the direction of the equiphase surface, the echo other than the echo corresponding to the defective portion is detected from the ultrasonic echo signal ( (Pseudo echo) can be reduced, and as a result, the defective portion of the inspection object can be accurately detected.

Further, according to the present invention, there is provided an ultrasonic echo signal processing apparatus used when inspecting a defective portion of the inspection object according to the ultrasonic echo signal obtained from the inspection object, The ultrasonic echo signal is received as a sample signal by the rotary wavelet transform, which has a waveform conversion unit that generates a conversion coefficient, and obtains the ultrasonic echo signal obtained from the object to be inspected in which the position of the defective portion is specified in advance. A learning device that generates a weighting coefficient that minimizes a pseudo echo as a synthesis parameter according to a sample conversion coefficient obtained by performing the rotational wavelet conversion of the sample signal in the waveform conversion unit, and a super device obtained from an object to be inspected. The evaluation target conversion coefficient obtained by receiving the sound wave echo signal as the evaluation target signal and performing the rotational wavelet transform of the evaluation target signal in the waveform conversion unit is added to the sum. Ultrasonic echo signal processing device is obtained, characterized in that it comprises a signal generator outputting a combined and filtered signal based on the parameter.

As described above, since the object to be inspected whose position of the defective portion is known is inspected and the weighting coefficient (combining parameter) is learned from the inspection result, the object to be evaluated is based on the combining parameter. If the conversion coefficients are combined, the pseudo echo can be reduced with high accuracy. In other words, since the optimum synthesis parameter is learned using the sample signal, it is possible to reduce the false echo by emphasizing the scratch echo even for the echo signal for which it is difficult to physically assign the cause of occurrence and the conversion coefficient. it can.

[0014]

By thus combining the transform coefficients along the equiphase surface, the spread of the echo can be reduced and the resolution can be improved.

[0016]

BEST MODE FOR CARRYING OUT THE INVENTION The present invention will be described in detail below with reference to the embodiments shown in the drawings. However, the dimensions, materials, shapes, relative positions, etc. of the components described in this embodiment are not intended to limit the scope of the present invention thereto, unless there is a specific description, and are merely illustrative examples. Nothing more.

A first example of the ultrasonic echo signal processing apparatus according to the present invention will be described with reference to FIG.

The illustrated ultrasonic echo signal processing device (hereinafter simply referred to as a signal processing device) is used when processing an ultrasonic echo signal. When an ultrasonic signal is sent from the ultrasonic sensor (not shown) to the object to be inspected, this ultrasonic signal is an ultrasonic echo signal (hereinafter simply referred to as an echo signal) at a defective portion such as a scratch on the object to be inspected. Is called).
As described above, the ultrasonic signal is also reflected as a pseudo echo by the welded portion or the like. In the following description, the echo reflected at the defective portion is called a scratch echo, and the echo signal is a signal including a scratch echo and a pseudo echo. Then, the signal processing device reduces the pseudo echo in the echo signal and outputs it as a filtered signal as described later.

The illustrated signal processing device 11 includes a waveform conversion section 1
2, the significant component selecting unit 13 and the significant component synthesizing unit 14 are included, and the filtering signal is output from the significant component synthesizing unit 14 as described later. The echo signal described above is given to the waveform conversion unit 12 as an original signal, and the waveform conversion unit 12 performs a so-called rotational wavelet conversion on the original signal to generate a conversion coefficient.

The waveform conversion unit 12 is provided with a base generation unit 12a and a wavelet conversion unit 12b, and the base generation unit 12a generates a base (directional multidimensional base) as follows.

The mother wavelet of the one-dimensional wavelet transform is φ (t), the principal component direction scale is a1, the orthogonal direction scale is a2, the shift vector is b, the weighting function is w (t), and the rotation transformation of the angle θ is R ( θ, X) and the directional multidimensional basis is Φ (a1, a2, b, θ, x, y), Φ (a1, a2, b, θ, X) = R (θ, φ)
(X / a1) * w (y / a2)) + b. However,
X = (x, y).

That is, the directional multidimensional basis is a one-dimensional wavelet transform basis offset in a multidimensional space, weighted according to the offset distance, and further rotated by an angle θ. , Scale-shifted as a multi-dimensional basis mother wavelet.

The above-mentioned directional multidimensional basis is given to the wavelet transform unit 12b. Wavelet transform unit 12
In b, the convolution operation of the original signal and the directional multidimensional basis is performed to generate the transform coefficient. For example, if the original signal is U
(X) and the conversion coefficient is W (a1, a2, b, θ), W (a1, a2, b, θ) = ∫∫Φ (a1, a2
b, θ, X) · U (X−b).

As a result, the conversion coefficient W is the angle θ of the original signal.
The component of the scale (a1, a2) is extracted from the components of the direction.

By the way, in the ultrasonic flaw detection test (UT), at the observation point P, the signal observed at the time t (here, the pulse transmission time at each observation point is "0") t is changed to U (X ) (Where X = (t, P)). The equiphase surfaces of U (X) are arranged in parallel to each other at right angles to the t-axis in a portion where there is no defective portion (abnormality: scratch or noise) in the inspection object. Echoes due to scratches or noise (scratch echoes) have discontinuity in the equiphase surface due to the route of the ultrasonic waves, and their orientation also fluctuates. In many cases, as a result of the difference in the direction of the reflection source and the route of ultrasonic waves, a difference occurs in the direction of the equiphase surface. Therefore, the conversion coefficient that is the result of converting the echo signal U (X) by the above-described directional multidimensional basis is
The angle θ indicates a (different) distribution that can be discriminated between the scratch echo and the pseudo echo.

The above-mentioned conversion coefficient is given to the significant component selecting section 13. As described above, since the transformation coefficient, which is the result of transformation by the directional multidimensional basis, shows different distributions between the scratch echo and the pseudo echo depending on the angle θ, the significant component selection unit 13 uses the scratch echo and the pseudo echo distribution. From, the conversion coefficient related to the wound echo is selected as a significant component. Then, these significant components are provided to the significant component synthesis unit 14.

The significant line segment synthesizing unit 14 synthesizes the significant components and sends them as a filtering signal. The significant component is
It is represented by a complex number, and when synthesizing a significant component, the absolute value thereof is taken, and then the reciprocal of a predetermined weighting coefficient is multiplied by each absolute value and added. As a result, the component related to the pseudo echo is removed or reduced from the filtered signal.

As described above, in the example shown in FIG. 1, attention is focused on the characteristic points related to the equiphase surface in addition to the frequency characteristics of the time-series signal. Can be emphasized.

Next, a second example of the signal processing device according to the present invention will be described with reference to FIG.

In the signal processing apparatus shown in FIG. 2, the same components as those of the signal processing apparatus shown in FIG. 1 are designated by the same reference numerals and the description thereof will be omitted. The signal processing device shown in FIG. 2 includes a learning device 21 and a filtering signal generation unit 22, and the learning device 21 generates a coefficient vector as a synthesis parameter as described later. Then, the filtering signal generation unit 22 generates a filtering signal based on the synthesis parameter.

In the illustrated example, first, an echo signal (flaw detection signal: hereinafter referred to as a sample signal) when an object to be inspected whose scratch position is known in advance is inspected using the UT is given. This sample signal is applied to the waveform conversion unit 12 as described in connection with FIG.
Then, a conversion coefficient is generated according to the sample signal (hereinafter, this conversion coefficient is referred to as a sample conversion coefficient).

The sample conversion coefficient is given to the learning device 21,
The learning device 21 obtains a weighting coefficient (coefficient vector) that minimizes the pseudo echo according to the sample conversion coefficient. For example,
The learning device 21 obtains the weighting coefficient that maximizes the S / N ratio based on the sample conversion coefficient (as described above, in the sample signal, the position of the scratch on the inspection object is known in advance, so the sample conversion is performed). The weighting coefficient that maximizes the S / N ratio can be obtained according to the coefficient).

The sample signals described above are obtained for a plurality of objects to be inspected (objects to be inspected whose scratch positions are known in advance), and these sample signals are given to the signal processing apparatus shown in FIG. Then, as described above, the learning device 21 obtains the weighting coefficient that maximizes the S / N ratio in accordance with each sample conversion coefficient. That is, the learning device 21 learns the weighting coefficient that maximizes the S / N ratio based on the sample signal (that is, the sample conversion coefficient). When obtaining the weighting factor, for example, the least squares method is used (so that the pseudo echo becomes zero, that is, the sum of the linear combination (combined model) becomes zero). Instead of the least squares method, a so-called neural network may be used to learn the weighting coefficient (coefficient vector).

On the other hand, the object to be inspected whose scratch position is unknown is U
It is inspected by using T, and an echo signal (hereinafter referred to as an evaluation target signal) is given to the signal processing device. This evaluation target signal is the waveform conversion unit 1 as described in connection with FIG.
2, the waveform conversion unit 12 generates a conversion coefficient according to the evaluation target signal (hereinafter, this conversion coefficient is referred to as an evaluation target conversion coefficient).

The above-mentioned evaluation target transform coefficient is given to the filtering signal generator 22. On the other hand, the weighting coefficient vector (synthesis parameter) obtained by learning as described above is given to the filtering signal generation unit 22, and the filtering signal generation unit 22 synthesizes the evaluation target transformation coefficient based on the synthesis parameter. And outputs a filtered signal.

As described above, the object to be inspected whose scratch position is known is inspected in advance by using the UT, and the weighting coefficient (synthesis parameter) is learned from the inspection result. By synthesizing the transform coefficients to be evaluated on the basis of, it is possible to accurately reduce the pseudo echo. In other words, since the optimal synthesis parameter is learned using the sample signal, even for an echo signal whose physical meaning between the cause of occurrence and the conversion coefficient is difficult, the scratch echo is emphasized to reduce the pseudo echo. You can

Further, a third example of the signal processing device according to the present invention will be described with reference to FIG. In the example shown in FIG. 3, the same components as those of the signal processing device shown in FIG. 1 are designated by the same reference numerals, and the description thereof will be omitted.

As described with reference to FIG. 1, the echo signal (original signal) is given to the waveform converting unit 12, and the waveform converting unit 1
In step 2, conversion coefficients are generated according to the original signal. The signal processing device shown in FIG. 3 includes a coefficient processing unit 31 and a coefficient synthesizing unit 32, and the above-mentioned conversion coefficient is given to the coefficient processing unit 31. The coefficient processing unit 31 narrows the image representing the conversion coefficient. That is, the conversion coefficient is made to express the directionality remarkably. In other words, the coefficient processing unit 31 processes the conversion coefficient along the equiphase surface direction to obtain the direction vector.

The above-mentioned direction vector is given to the coefficient synthesizing unit 32. As described with reference to FIG. 1, since the conversion coefficient, which is the result of conversion by the directional multidimensional basis, shows different distributions in the scratch echo and the pseudo echo depending on the angle θ, the coefficient combining unit 31 first A directional vector related to the scratch echo is selected as a significant component from the echo and the pseudo echo distribution. Then, these significant components are synthesized. For example, the coefficient synthesizing unit 32 takes the correlation of the directional vectors, superimposes them, and outputs the filtered signal.

As described above, in the example shown in FIG. 3, since the conversion coefficients are combined along the equiphase surface, the spread of the echo can be reduced and the resolution can be improved.

[0041]

As described above, in the present invention, the echo signal is subjected to the rotary wavelet transform to generate the transform coefficient, and the transform coefficient corresponds to the defective portion (scratch) according to the direction of the equiphase surface. The conversion coefficient related to the echo is selected as the significant component, and the significant component is combined based on the weighting coefficient defined in advance and output as the filtering signal. Therefore, the false echo is reduced in the filtering signal, and as a result, the filtering is performed. There is an effect that a defective portion of the inspection object can be accurately detected using the signal.

Further, according to the present invention, an ultrasonic echo signal obtained from the object to be inspected in which the position of the defective portion is specified in advance is received as a sample signal, and the sample signal is converted into a sample conversion coefficient obtained by rotating wavelet transform. Accordingly, the weighting coefficient that minimizes the pseudo echo is generated as a synthesis parameter, the ultrasonic echo signal obtained from the object to be inspected is received as the evaluation target signal, and the evaluation target signal is obtained by rotating wavelet transform. Since the evaluation target conversion coefficient is synthesized based on the synthesis parameter and output as a filtering signal, it is possible to accurately reduce the pseudo echo, and as a result, the filtering signal is used to detect the defective portion of the inspection object. This has the effect of enabling accurate detection.

In addition, according to the present invention, the ultrasonic echo signal is subjected to the rotary wavelet transform to generate the transform coefficient, and the transform coefficient is processed along the equiphase plane direction to obtain the direction vector. The transform coefficient related to the echo corresponding to the defective portion is selected as the significant component from the direction vector according to the direction of the equiphase surface, and the significant components are combined according to the correlation between the significant components and output as a filtering signal. Therefore, there is an effect that not only the pseudo echo can be accurately reduced, but also the spread of the echo can be reduced and the resolution can be improved.

[Brief description of drawings]

FIG. 1 is a block diagram showing a first example of an ultrasonic echo signal processing device according to the present invention.

FIG. 2 is a block diagram showing a second example of the ultrasonic echo signal processing device according to the present invention.

FIG. 3 is a block diagram showing a third example of the ultrasonic echo signal processing device according to the present invention.

[Explanation of symbols]

11 Signal processing device 12 Waveform converter 12a Basis generator 12b Wavelet transform unit 13 Significant component selection section 14 Significant component synthesizer 21 Learning device 22 Filtering signal generator 31 Coefficient processing part 32 coefficient synthesis section

   ─────────────────────────────────────────────────── ─── Continued front page    (72) Inventor Yoshihisa Tada             3-1-1 Wadasaki-cho, Hyogo-ku, Kobe-shi             Hishi Heavy Industries, Ltd.Kobe Shipyard F-term (reference) 2G047 AA05 BC10 EA10 GG08

Claims (3)

[Claims] 1. An ultrasonic echo signal processing device used when inspecting a defective portion of an object to be inspected according to an ultrasonic echo signal obtained from the object to be inspected. A waveform transforming unit that generates a transform coefficient by performing a rotary wavelet transform, and a significant component selecting unit that selects, as a significant component, a transform coefficient associated with an echo corresponding to the defective portion from the transform coefficient according to the direction of the equiphase surface. And a coefficient synthesizing unit that synthesizes the significant component based on a predetermined weighting factor and outputs the synthesized signal as a filtering signal. 2. An ultrasonic echo signal processing device used when inspecting a defective portion of the inspection object according to the ultrasonic echo signal obtained from the inspection object, wherein the ultrasonic echo signal is The ultrasonic wave echo signal obtained from the object to be inspected in which the position of the defective portion is specified in advance is received as a sample signal by the waveform conversion unit. A learning device that generates a weighting coefficient that minimizes a pseudo echo as a synthesis parameter according to a sample conversion coefficient obtained by rotating the sample signal by the rotational wavelet transform, and an ultrasonic echo signal obtained from an object to be inspected as an evaluation target. The evaluation target transform coefficient obtained by subjecting the evaluation target signal to the rotational wavelet transform in the waveform transforming unit as a signal is combined based on the synthesis parameter. An ultrasonic echo signal processing device, comprising: a signal generation unit configured to generate and output a filtered signal. 3. An ultrasonic echo signal processing apparatus used when inspecting a defective portion of the inspection object according to the ultrasonic echo signal obtained from the inspection object, wherein the ultrasonic echo signal is A waveform transforming unit that generates a transform coefficient by performing a rotary wavelet transform, a coefficient processing unit that processes the transform coefficient along the direction of the equiphase surface to obtain a direction vector, and the direction according to the direction of the equiphase surface. A coefficient synthesizing unit that selects a transform coefficient associated with an echo corresponding to the defective portion from a vector as a significant component, synthesizes the significant component according to the correlation between the significant components, and outputs the synthesized signal as a filtering signal. Ultrasonic echo signal processing device.