WO2023117079A1

WO2023117079A1 - System and method for determining a size of a defect in a component

Info

Publication number: WO2023117079A1
Application number: PCT/EP2021/087288
Authority: WO
Inventors: Georg Bodammer; Matthias Goldammer; Hubert Mooshofer; Karsten SCHÖRNER
Original assignee: Siemens Aktiengesellschaft
Priority date: 2021-12-22
Filing date: 2021-12-22
Publication date: 2023-06-29

Abstract

The invention relates to a method and a device for determining a size of a defect in an object to be inspected on the basis of a SAFT method. In an inspection measurement INSPM, in a SAFT-based manner, a probe head records a plurality of A-scan images in different actual poses of the probe head. In an overall reference measurement, on a test object corresponding to the object to be inspected and having a plurality of defects DEFj with known parameters, in a SAFT-based manner, a plurality of A-scan images in different actual poses of the probe head are recorded for each of the artificial defects DEFj in a respective reference measurement. On the basis of a SAFT analysis of a selection of A-scan images of the reference measurements and on the basis of the known parameters of the defects DEFj, an assessment database BEW is created. On the basis of a SAFT analysis of a selection of A-scan images of the inspection measurement, amplitude sums for the spatial elements of the region to be inspected in the object to be inspected are determined. On the basis of the assessment database BEW, the amplitude sums of the spatial elements are converted into the size of the defect to be determined.

Description

System and method for determining a size of a defect in a component

The invention relates to determining the size of a defect in a component using a SAFT method.

Test objects such as e.g. Machine components can have defects near the surface or deep in the component, such as Exhibit cavities or cracks, which may have arisen during manufacture and/or during regular use of the component. The presence of such defects can lead to component failure. Consequently, in particular those components that are prone to the occurrence of defects, e.g. after their manufacture and/or as part of interim maintenance work, etc. checked to see if they have any defects.

Various non-destructive testing methods are known for this purpose, such as Eddy current testing, dye penetrant testing, magnetic particle testing, thermographic testing or ultrasonic testing.

Electrically insulating and non-ferromagnetic components can also be tested using ultrasonic testing, i .e . H . the respective test procedure is more independent of the material of the test object. Typically, in a non-destructive testing of a test object using ultrasound, an ultrasonic test head is moved in two spatial dimensions over the surface of the test object. As a result, local information about the local state of the test object is determined for each spatial position of the ultrasonic test head. The collected individually determined local information is then brought together so that an overall picture of the condition of the test object results. For this it is important to match the local information to the associated spatial positions of the ultrasonic probe to assign it, d . H . the spatial position of the ultrasonic probe must be known for each individual measurement. Only in this way is it possible to further process the results of the ultrasonic test, e.g. by means of diaphragms, B-images, C-images or also by means of other reconstruction methods. However, compared to the other approaches mentioned, conventional ultrasonic testing methods have indirect imaging that is more difficult to interpret. The spatial resolution capability is also lower in conventional ultrasonic testing of the component, with the orientation of the defect with respect to the direction of propagation of the ultrasound also being decisive for its detectability.

The analysis technique "SAFT" (Synthetic Aperture Focusing Technique) is known for better localization and separation of defects in non-destructive testing with ultrasound, which advantageously, at least in certain test scenarios, has improved lateral resolution and thus both improved sensitivity and also improved Defect separation brings with it as well as an improved signal-to-noise ratio SAFT is an approach that is applied as a post-processing step or, if necessary, directly to ultrasonic measurement data, whereby SAFT is based on the so-called "SAR" approach (Synthetic Aperature Radar) based on radar technology. There, signals corresponding to different radar positions are calculated with one another in such a way that a large virtual, synthetic aperture results. The thus artificially enlarged aperture provides the higher lateral resolution. At SAFT, this approach is adapted to ultrasound: The actual inspection measurement, i . H . the examination of the component to detect any defects is carried out in the same way as in a conventional ultrasonic test, even if the data recorded are recorded without rectification of the voltage signal from the measurement. For this measurement, the test volume is divided into small volume elements or voxels, and for each voxel those echo signal components from different positions of the probe are added which, according to their propagation time, could originate from this very voxel. In the SAFT analysis of the measurement data consisting of the various echo signals, amplitude sums are determined from the large number of echo signals for each voxel of the test object, ie. H . non-rectified ultrasonic data from different probe positions in the test object are superimposed in the correct phase. The SAFT signal or also the SAFT amplitude sum is thus obtained as a superimposition of the echo signals for different probe positions.

Ultrasonic testing can be carried out with the aid of SAFT analysis, e.g. in the case of a manual movement of a test head which emits the ultrasonic pulses and which receives the echo signals corresponding to the emitted ultrasonic pulses. Such SAFT exams are, for example. described in EP2932256B1 and in EP2898322B1.

In certain scenarios and applications, not only the possible presence of a defect is of interest in an inspection measurement of a test object, but also its size, i. H . its spatial extent. In the case of defects that are large compared to the ultrasonic wavelength used, the result of the SAFT evaluation can be measured directly. On the other hand, an assessment of defects that are small compared to the wavelength is only possible to a limited extent when using the SAFT method.

In DE102013211616A1, an approach to assessing the size of defects is presented, in which actually recorded SAFT measured values are compared with ultrasonic signals or ultrasonic signals simulated in a specified test scenario. so that corresponding echoes for defects of different sizes in a test object can be compared. In the event that real measurement data and simulation data match within certain limits, the size of a real defect in the test object can be concluded. However, this approach is not ideal insofar as the simulation is complex and may not take all practical parameters of the measurement into account, for example by neglecting certain physical effects, so that the accuracy of the approach suffers.

It is therefore an object of the present invention to specify an improved possibility of determining the sizes of defects in a test object when using a SAFT method, even if the respective defect is small compared to the ultrasonic wavelength.

This object is achieved by the method described in claim 1 and by the device described in claim 12. The dependent claims describe advantageous configurations.

The method for determining the size of a defect in a test area of a test object using a SAFT method provides that in an inspection measurement INSPM of the test object SAFT-based with a probe, a large number of A-images IMAA_INSP(p) in different actual poses PE_INSP (p) of the test head (210), ie the poses PE_INSP(p) are the real poses of the test head during a respective measurement of an A-scan. The probe (210) is positioned at different poses PE_INSP(p) for different A-scans. In this case, a pose PE_INSP(p) and an associated A-image IMAA_INSP(p), ie one recorded at this pose PE_INSP(p), are unambiguously assigned to one another. All poses PE_INSP(p) of the inspection measurement INSPM taken together form a set m_insp of poses. In a total reference measurement REFM on a test object corresponding to the test object, which has a number J>1 of defects DEFj with j = l,...,J with known parameters (TIEj, GRj ) in the test object with regard to their size and depth also SAFT-based for each of the artificial defects DEFj, i.e. for each j, in a number Vj>l of A-images IMAA_REF(j) in different actual poses PE_REF ( , n , m ) of the test head (210) are recorded for a respective reference measurement REFM(j). For different A-scans IMAA_REF ( , n , m ), the probe (210) is positioned at different poses PE_REF ( j , nj , mj ), with each pose PE_REF ( j , nj , mj ) and an associated one, ie at this Pose PE_REF ( j , nj , mj ) captured A-scan IMAA_REF ( j , nj , mj ) are uniquely associated with each other. All poses PE_REF(j, nj, mj) of a respective j or a respective reference measurement REFM(j) taken together form an entity m_ref(j) of poses.

In a first step, an evaluation database BEW is now based on a SAFT analysis of a selection IMAA_REF_SEL (j) of A-images of the reference measurements REFM(j) and based on the known parameters of the defects DEFj SAFT analysis resulting amplitude sum and the defect parameters together. In a second step, AMPL amplitude sums for the spatial elements or voxels of the test area of the SAFT method in the test object (100) are determined based on a SAFT analysis of a selection of A images of the inspection measurement INSPM. The amplitude sums AMPL of the spatial elements can then be converted into the size of the defect to be determined using the evaluation database BEW.

In the first step, the assessment database BEW is created using the amplitude sums resulting from the SAFT analysis and using the known parameters of the known defects DEFj. More precisely, to create the evaluation database BEW for each j, a SAFT analysis based on the A-images of the selection IMAA_REF_SEL (j) is carried out for the respective j in a reconstruction step SAFTREC(j), resulting in amplitude sums for the spatial elements or voxels of a test area of the SAFT method in the test object (100T), in particular for the space elements containing the known defects DEFj. The evaluation database BEW is then based on of the resulting amplitude sums and the related parameters known for the known defects DEFj.

For a respective reference measurement REFM(j), a test grid RAST(j) with intended measurement poses PE(j) is specified, in the best possible proximity of which the test head for recording a respective A-scan IMAA_REF ( j , nj , mj ) of the reference measurement REFM( j) to be positioned.

In the first step mentioned, A-scans for the respective selection IMAA_REF_SEL (j) for a respective j in a first selection step SEL1 (j) are selected from a total m_insp of poses PE_INSP(p) of the inspection measurement INSPM to form a selection m_insp_sel ( j) those poses PE_INSP(p') are selected and thus assigned to m_insp_sel(j) which, due to their spatial arrangement, come closest to the intended measurement poses PE(j) of a test grid RAST(j) defined for the respective reference measurement REEM(j). In a second selection step SEL2 (j), those actual poses PE_REF (j, nj , mj) are selected and thus assigned to m_ref_sel(j) that are spatially closest to the poses PE_INSP(p') of the entity m_insp_sel(j) determined in the first selection step SEL1(j). Finally, in an A-scan selection step IMAASEL(j), the A-scans IMAA_REF(j) of the respective reference measurement REFM(j) assigned to the actual poses of the entity m_ref_sel(j) selected in the second selection step SEL2(j) are used to form the selection IMAA_REF_SEL ( j ) selected.

In the first selection step, a pose PE_INSP(p) of the whole m_insp is then selected for m_insp_sel(j) if its spatial distance APOS_INSP(p) from the nearest intended measurement posePE(j) of the test grid RAST(j) is less than a predetermined threshold value , especially smaller ones is a fraction 1/n of the distance APOS between two adjacent measuring poses PE(j) of the test grid RAST(j). Ie a position POS_INSP(p) is only selected for the whole m_insp_sel ( j ) if APOS_INSP (p) < APOS ( j ) /n applies. For example, n=3 can apply to n.

In the second step mentioned, if the ensembles m_insp_sel(j) are largely identical for different j, the selection of A-scans for the SAFT analysis includes those A-scans of the inspection measurement INSPM that have the poses PE_INSP(p) a associated with any of the m_insp_sel( ) entities.

The property that "the ensembles m_insp_sel ( ) are largely identical for different j" means that an ensemble m_insp_sel ( ) comprises essentially the same poses as an ensemble m_insp_sel ( k) with kVj . The poses do not have to be completely identical , but can deviate from each other to a small extent.As is explained in detail in the description of the figures, this can result from the fact that test grids RAST(j) for the reference measurements REFM(j) are so similar to one another for different j that ultimately the wholes m_insp_sel ( k) and m_insp_sel ( ) are largely identical to each other, so that the entirety of the associated A-images in IMAA_INSP_SEL ( j ) and in IMAA_INSP_SEL ( k) are also identical, i.e. IMAA_INSP_SEL ( j ) =IMAA_INSP_SEL ( k) =IMAA_INSP_SEL .

In the second step mentioned, for the other case that the sets m_insp_sel ( ) for different j are at least partially different or, in other words, not largely identical, a SAFT analysis of those A images IMAA_INSP( p) the inspection measurement INSPM is carried out, which is assigned to the poses PE_INSP(p) of the respective selection m_insp_sel(j), resulting in the respective individual amplitude sums AMPL(j) for the spatial elements or voxels of the test area of the SAFT method in the test object (100) . Then either which forms a mean value from the individual amplitude sums AMPL(j) for all j, which represents the amplitude sums AMPL to be determined, which are then converted into the size of the defect to be determined using the evaluation database BEW. Alternatively, respective individual amplitude sums AMPL(j) are converted separately into individual defect sizes using the evaluation matrix BEW, and the size of the defect to be determined is determined as the mean value of the individual defect sizes.

In the second selection step SEL2 (j), for each j in a position correction step PCORR(j), a respective pose PE_REF of the selection m_ref_sel (j) can be assigned that pose PE_INSP from the totality m_insp_sel (j) which is spatially closest to it for each pair PP of poses formed in this way, a difference in the sound paths between the respective pose PE_REF, PE_INSP and the defect (110) is determined and the A-scan which is assigned to the pose PE_REF of this pair PP is corrected using the determined difference.

Alternatively or additionally, for each j in a coupling correction step ACORR(j), amplitudes of ultrasonic echoes UECHO(i) of the respective reference measurement REFM(j) can be compared with amplitudes of corresponding ultrasonic echoes UECHO(i), i.e. recorded in comparable measurement situations, of the inspection measurement INSPM . In the event that the compared amplitudes deviate from one another by more than a predetermined factor, the amplitudes of the A-scans of the corresponding individual reference measurement REFM(j) can be corrected based on the determined deviation.

Preferably, moving the test head to different poses is done manually.

A corresponding device for determining a size of a defect in a test area of a test object using a A SAFT process has a control unit that is set up to carry out such a process.

For this purpose, the device or the SAFT system has a transmitter for transmitting ultrasonic pulses UIMP(i) into the test object and a receiver for receiving ultrasonic echoes UECHO(i) corresponding to the transmitted ultrasonic pulses UIMP(i). The measurement signals received by the receiver form the A-scans recorded in the reference measurements REFM(j) or in the inspection measurement INSPM.

The transmitter and receiver are preferably housed in a common test head, the test head being movable manually to carry out the SAFT method.

A detection device is preferably provided for determining a respective pose of the transmitter and/or the receiver.

In contrast to the solutions proposed in EP2932256B1 and in EP2898322B1, the solution presented here offers the advantage that it not only offers a qualitative result in the form of the reconstructed SAFT images, but also gives the reconstructed voxels a defect size or Substitute assigns error size and thus represents a quantitative method. In contrast to the also quantitative but simulation-based defect size assessment in DE102013211616A1, the solution presented here also offers the advantage that the simulation of the ultrasonic signals can be omitted or. is replaced by a corresponding reference measurement of artificial defects. This approach, based on real reference measurements, promises higher accuracy than using a simulation, since the latter usually cannot take all physical effects into account. Further advantages and embodiments result from the drawings and the corresponding description.

The invention and exemplary embodiments are explained in more detail below with reference to drawings. There are possibly the same components in different figures are identified by the same reference symbols. It is therefore possible that, in the description of a second figure, there are no more detailed explanations for a specific reference sign that has already been explained in connection with another, first figure. In such a case, it can be assumed in the embodiment of the second figure that the component marked there with this reference symbol has the same properties and functionalities as explained in connection with the first figure, even without further explanation in connection with the second figure . Furthermore, for the sake of clarity, not all of the reference symbols are shown in all of the figures, but rather only those to which reference is made in the description of the respective figure.

Show it :

1 shows a system for inspecting a test object,

2 shows a perspective view of a test object with artificial defects of different sizes,

3 shows a perspective view of a test object with artificial defects at different depths,

4 shows the sequence of a method for creating the evaluation database BEW,

5 shows a plan view of a test grid,

6 shows a plan view of a test grid and poses of

inspection measurement,

7 shows a top view of a test grid and poses of the inspection measurement and a reference measurement,

8 shows a side view of a test object with poses of inspection measurement and reference measurement. FIG. 1 shows a simplified example of a test object 100 which has a defect 110 . The defect 110 can, for example. be a cavity unintentionally created during the production of the test object 100 .

1 also shows the basic architecture of a device 200 for ultrasonic testing of the test object 100 to find the defect 110 and to evaluate the size G of this defect 110 . In the example presented here, the device 200 is in the form of a SAFT system 200 whose mode of operation and functioning is based in principle on the SAFT method explained in the introduction, for example corresponding to the statements in EP2932256B1 and/or in EP2898322B1.

A probe 210 of the SAFT system 200 is designed, ultrasonic signals UIMP, for example. sending ultrasonic pulses out .

For this purpose, the probe 210, for example. be implemented as a round or rectangular single-oscillator probe or also as a round or rectangular phased array probe. The probe 210 is also designed to receive ultrasonic signals UECHO, these received ultrasonic signals UECHO being in particular the ultrasonic echoes UECHO corresponding to the transmitted ultrasonic pulses UIMP in the course of a SAFT measurement, which are reflected by walls and/or irregularities of the test object 100. Such an irregularity is e.g. the defect 110 represents .

The test head 210 can, in particular also manually, freely on or along a surface 101 of the Prüfobj ect 100 and are held in such a way that the probe 210 transmits the ultrasonic pulses UIMP into the Prüfobj ect 100 during the movement along the surface 101 on the one hand and the corresponding ultrasonic echoes UECHO are detected or detected on the other hand. measures .

The SAFT system 200 also includes a control unit 220 which is connected to the test head 210 via a wired or wireless connection 225 . The control unit Unit 220 is set up on the one hand to control the test head 210 via the connection 225 in such a way that it generates the ultrasonic pulses UIMP to be sent out for examining the test object 100 . On the other hand, the control unit 220 is set up to further process the ultrasonic echoes UECHO measured by the probe 210 and corresponding to the emitted ultrasonic pulses UIMP in the manner of a conventional SAFT analysis in order to finally generate images of a test area of the test object 100 and display them on a display device 240 . For this purpose, the corresponding measured time signals of the ultrasonic echoes UECHO are transmitted from the probe 210 to the control unit 220 via the connection 225 . There, the measured ultrasonic echoes UECHO, d. H . From the raw data of the measurement, so-called A-scans are generated first and thus before the actual conventional SAFT analysis. Such an A-scan ultimately represents a time series of the measured amplitudes of the ultrasonic echoes UECHO after an ultrasonic pulse UIMP, i. H . it shows the times and the corresponding amplitudes of the ultrasonic echoes UECHO received at the test head 210 corresponding to a previously radiated ultrasonic pulse UIMP. As part of the SAFT analysis, the A-scans are processed further, with the corresponding ultrasound amplitude contributions of the corresponding ultrasound echoes UECHO from the raw data being summed up individually for each voxel, e.g. in the form of superimposition and averaging of the amplitude values of the received ultrasonic echoes UECHO for each voxel.

In summary, the term "conventional SAFT method" includes conventional data acquisition and conventional SAFT analysis. Conventional data acquisition essentially involves moving the probe 210 along the surface 101 of the test object 100, emitting the ultrasonic pulses UIMP in the test object 100 by means of the test head 210 during the movement, the receiving of respective ultrasonic echoes UECHO corresponding to the transmitted ultrasonic pulses UIMP by means of the test head 210 and the generation of the corresponding A images SAFT analysis includes the creation of an image of a predetermined test area of the test object 100, consisting of a large number of voxels, based on a summation of amplitude values of the received ultrasonic echoes UECHO, in particular based on a superimposition and averaging of the corresponding amplitude values of the received ultrasonic echoes UECHO, using the Control unit 220. Accordingly, the conventional SAFT analysis supplies a SAFT amplitude sum for each observed voxel of the test area, which can finally be further processed for imaging and/or for other purposes.

In addition to these features of the conventional SAFT method, it can be provided that the SAFT system 200 uses a corresponding detection device 230 to determine the positions and possibly the orientations of the test head 210 during the measurement process. For example, the detection device 230 can be designed according to one of the embodiments explained in EP2932256B1 and can, for example, detect the position POSI (i) and orientation ORI (i) of the probe 210 at the time an ultrasonic pulse UIMP(i) is emitted by the test head 210, where i=1,2,... counts those successive ultrasonic pulses UIMP(i) for which corresponding ultrasonic echoes UECHO(i) are received and further processed. Typically, such an ultrasonic echo UECHO (i) comprises a plurality of individual echoes which have arisen at different structures in the test object 100 and which are reflected as a corresponding plurality of peaks in the associated A-scan. The respective position POSI (i) can, for example, be specified relative to a starting position POST (1) of the measurement, where, for example, POST ( 1 )=( 0 , 0 , 0 ) can apply in a Cartesian coordinate system K. The orientation ORI ( i) can relate, for example, to the normal on the surface 101, to one of the axes of the coordinate system K or again for i=1 to an orientation ORI(1) for the first ultrasonic pulse UIMP(1). As described in EP2932256B1, a current position POSI(i) and orientation ORI(i) of the probe 210 at the time T(i) of each ultrasonic pulse UIMP(i) can be determined from the measured positions and orientations and a respective time reference T(i) and at of the SAFT analysis to determine a respective distance between the reconstructed respective voxel and the measurement position. Using the detected position POSI (i) and orientation ORI (i) of the probe 210, a center position of the active aperture of the probe 210 can be determined when emitting the ultrasonic pulses UIMP(i) and taken into account when generating the image of the test area of the test object 100. The active aperture is to be understood as meaning that part of the test head 210 which serves as an effective transmission or reception surface. A spatial offset between the respective position measurement and the position of the test head 210 can be calculated using the recorded information about the test head orientation.

Ultimately and more generally expressed, the image of the test object 100 can be created by the control unit 220 based on the ultrasonic echoes UECHO(i) depending on the respectively detected positions POSI (i) and orientations ORI (i) of the test head 210 .

The inspection measurement INSPM or examination of a test object 100 with the SAFT system 200 presented here should in particular also include an evaluation of the size GR of a defect 110 that may be present in the test object 100 . Based on the conventional SAFT method described above, including the conventional data acquisition and the conventional SAFT analysis as well as the determination and use of poses PE(i) of the probe 210 in the course of a measurement, an evaluation of a size GR of a defect is explained as follows 110 in a test object 100 possible. The pose PE(i) should in particular include the position POSI(i) and possibly also the respective orientation ORI(i). The ORI (i) is particularly advantageous when the probe 210 does not generate a sound cone that is rotationally symmetrical. E.g. in the case of rotational symmetry, the orientation ORI(i) can be omitted. For the sake of simplicity, it is assumed below that only the positions, but not the orientations, are determined and used.

In general, the transmission of the UT pulse is not spatially uniform, i . H . the sound pressure depends on the angles phi , the-ta .

To determine the size GR of a defect, an overall reference measurement REFM is now typically, but not necessarily, carried out on a test object 100T before the actual inspection measurement INSPM of the test object 100 to be examined using the SAFT system 200 . As will be explained in more detail below, the overall reference measurement REFM on the test object 100T results in an evaluation database BEW, which relates SAFT amplitude sums from a conventional SAFT method to corresponding defect sizes. A regular inspection measurement INSPM on the test object 100 to be examined, which is based on a SAFT method, also produces SAFT amplitude sums, which can finally be converted into defect sizes GR based on the evaluation database BEW of the overall reference measurement REFM. The evaluation database BEW can therefore be used to replace the amplitude sums for different voxels of the test area, which result from the SAFT analysis, with a more practicable parameter, which allows a statement to be made about the size GR of a defect found there. For this purpose, the so-called Circular disk reflector (KSR) is used and the size GR of a defect is specified in "mmKSR", i.e. the measured amplitude sums are replaced by mmKSR from the evaluation database BEW. In the evaluation database BEW, which is thus e.g. Matrix is used, are therefore for this purpose the location or

Depth of a voxel containing a defect, the size GR of this defect and the corresponding sum of amplitudes linked to each other. Although the evaluation database BEW can be created based on simulations, as explained for example in the already mentioned DE102013211616A1, this brings with it the disadvantages mentioned. An advantageous alternative to creating the evaluation database BEW is therefore proposed below. In an inspection measurement INSPM of a real test object 100, the size GR of the defect 110 found in this way can be determined in mmKSR from the evaluation database BEW from an amplitude sum measured there and the depth in the test object 100 also determined during the measurement. In the event that the evaluation database BEW does not have an explicit entry for the exact combination of values consisting of depth, defect size and/or amplitude sum, a size GR in mmKSR can nevertheless be determined by interpolating the closest values in the evaluation database BEW.

The test object 100T largely corresponds to the test object 100 to be examined with regard to its internal and external structure, including the materials used. However, as visualized in FIG. 2 and FIG DEFj arranged with j = l,2,...,J, ie for each individual artificial defect DEFj its size GRj , which ultimately corresponds to the size in mmKSR read from the evaluation database BEW, and its depth TIEj in the test object 100T are known, which will later also be used to create the evaluation database BEW. J denotes the total number of artificial defects DEFj in the test object 100T. In particular, the different depths TIEj are equivalent to different signal propagation times of the ultrasonic signals UIMP, UECHO. FIG. 2 shows an embodiment in which the artificial defects DEFI, DEF2, DEF3 lie at the same depth TIEI below the surface 101 of the test object 100, but have different sizes GR1, GR2, GR3. Ie only the magnitudes GRj are varied, but not the depths TIEj. In contrast, FIG. 3 shows an embodiment in which where the artificial defects DEFI, DEF2, DEF3 lie at different depths TIEI, TIE2, TIE3 below the surface 101 of the test object 100, but have the same sizes GR1=GR2=GR3. Ie only the depths TIEj are varied, but not the sizes GRj of the defects DEFj. A preferred embodiment, which is not shown here, could be designed in such a way that both the sizes GRj and the depths TIEj of the artificial defects DEFj in the test object 100T are varied.

The overall reference measurement REFM on the test object 100T is carried out with the same settings of the SAFT system 200 as the actual inspection measurement INSPM of the test object 100 to find any defects in the test object 100, i.e. the same aperture and the same ultrasonic pulses UIMP(i) are defined, for example used by voltage, square pulse width, filter etc. As again symbolized by dots in FIG. 2 and FIG. 3, a multiplicity of A images IMAA_REF(j) are recorded at different locations in a known manner for each of the artificial defects DEFj. For this purpose, a typically rectangular test grid RAST(j) with measuring points provided for Vj=Nj*Mj is initially defined for each artificial defect DEFj. The correspondingly provided measuring positions POS (j) of these provided measuring points are identified in FIG. 2 and FIG. 3 by the dots and in FIG. 5 and also in FIG. 6 and FIG. 7 by the crossing points of the horizontal and vertical lines there. In the overall reference measurement REFM to be explained below, which provides a single reference measurement REFM(j) for each artificial defect DEFj, these intended measurement points of the test grid RAST(j) represent the positions POS(j) at which or in their best possible Proximity of the probe 210 is to be positioned to generate a data set for a respective A-scan. The test grid RAST(j), ie the arrangement of these provided measurement positions POS(j) of the reference measurement REFM(j), is selected such that a respective measurement covers and records the volume around the respective artificial defect DEFj as well as possible. Furthermore, the test grid RAST(j) is defined in such a way that the reference measurement REFM(j) with the greatest possible spatial sampling or with the finest possible sampling on the surface 101T of the test object 100T, since this increases the conversion accuracy that can ultimately be achieved.

The following explanations in connection with FIGS. 5-8 relate in part to j=1 as an example. It should be noted that the same considerations can also be applied to the other j=2, J without further ado.

FIG. 5 shows an example of j=l, a plan view of the test grid RAST(j=l) on the surface 101T of the test object 100T for the reference measurement REFM(l) at the defect DEF1. In the example chosen here with j=l, Nj=Nl=12 and Mj=Ml=12, i.e. the test grid RAST(l) includes Vl=12*12=144 intended measurement positions POS (j=l) , characterized by the crossing points of the horizontal and vertical lines of the grid RAST(l) .

FIG. 5 also shows, by way of example, the positions POS_REF (=1) identified by dots, at which the test head 210 for the reference measurement REFM(1) for the artificial defect DEF1 is actually positioned. All of these actual positions POS_REF(1) in the reference measurement REFM(1) for the defect DEF1 are referred to below as m_ref (j=1). On the one hand, it can be seen that the actual positions POS_REF (1) deviate from the intended measuring positions POS (1), ie the intended measuring positions POS (1) are not always met exactly in practice, e.g. because the positioning of the test head 210 is hand is not arbitrarily precisely possible. On the other hand, it may be possible that, as indicated in FIG. 5, a smaller number of individual measurements are carried out in the reference measurement REFM (1) than is provided according to the test grid RAST (1) made up of horizontal and vertical lines. FIG. 5 shows an example of how only 9*12=108 measurements are carried out or positions POS_REF (1) are approached for the artificial defect DEF1, ie on the grid RAST (1), while the test grid RAST (1) provides a number Vl=144 of intended measurement positions POS (1).

During the reference measurement REFM (1) for the artificial defect DEF1, the probe 210 is positioned as precisely as possible at one of the N1*M1 measuring points POS (1) of the test grid RAST (1) and generates the data for the corresponding A Image IMAA_REF(1,nl,ml) with nl = l,...,NI and ml = l,...,Ml. The test head 210 is then moved as precisely as possible to the next measuring point of the N1*M1 measuring points POS(1) of the test grid RAST(1) for the artificial defect DEF1 and again records the data for generating the corresponding A-image IMAA_REF there. An A-scan IMAA_REF (1, nl, ml) is recorded for each measurement point (nl, ml) provided for each grid RAST (1) or at least for a sufficient proportion of the measurement points provided there.

As soon as all Vj=Nj*Mj or at least a sufficient proportion of the intended measurement positions POS (j) of the respective test grid RAST(j) for the artificial defect DEFj have been processed, the corresponding process on the test grid RAST(k) with k^j and k= l, J of the next artificial defect DEFk executed.

If the test object now has 100T artificial defects DEFj with j=l,

J, where J specifies the number of artificial defects, the total reference measurement REFM accordingly generates J reference data sets REFDAT(j) with j = l, ..., J each with Vj=Nj*Mj A-images IMAA_REF (j, nj , mj ) . For example, J=S*D can apply to the number J, it being assumed that the test object has 100T defects DEFj at D different but known depths TIEd with d=1, . . . , D and with S different but known sizes GRs with s=l,...,S. Nj and Mj are preferably identical for different j, ie N1=N2=... and M1=M2=..., ie the test grids RAST(j), RAST(k) for different artificial defects DEFj, DEFk are preferably likewise identical, ie RAST ( j ) =RAST ( k) =RAST for j^k. Simultaneously with the individual measurements for generating a respective A-scan IMAA_REF (j, n, m) of the reference measurement REFM(j), the poses PE_REF (j, n, m) comprising the positions POS_REF (j, nj , mj ) and, if necessary, the orientations ORI_REF (j, nj , mj ) of the test head 210 are determined and stored for each individual measurement of the reference measurement REFM(j). All of the poses PE_REF(j, nj, mj) of a reference measurement REFM(j) on a respective defect DEFj are referred to below as m_ref(j). Consequently, in each case an A-scan IMAA_REF (j, nj , mj ) and the corresponding pose PE_REF (j, nj , mj ) are uniquely associated with one another. This means that after the reference measurement REFM(j) for each artificial defect DEFj, the associated poses m_ref(j) of the test head 210 are also available in addition to the respective A images IMAA_REF(j, nj, mj) itself. In this case, the orientations ORI_REF can be represented in particular by a rotation PHI_REF around the respective vertical axis or normal to the surface 101T.

Accordingly, from the totality REFM of the reference measurements REFM(j), in addition to poses m_ref(j) and A-images IMAA_REF(j, nj, mj), as already described, the sizes GRj and the depths TIEj of the respectively examined artificial defects DEFj are also known.

As part of the actual inspection measurement INSPM of the test object 100, during which any defects 110 in the test object 100 and, if applicable, their size GR are to be determined, the test head 210 is again preferably manually set in various poses PE_INSP(p) with p=1.2, ...,P positioned on the surface 101 in order to record the data for generating a respective A-scan IMAA_INSP(p) in a known manner in a respective pose PE_INSP(p), e.g. by insonifying an ultrasonic pulse UIMP(p) and receiving the echoes UECHO(p) corresponding thereto, resulting in corresponding A-frames IMAA_INSP(p) . In this case, P ultimately gives the number of individual measurements of the inspection measurement INSPM at. As with the reference measurements REFM(j), a respective pose PE_INSP(p) includes the position POS_INSP(p) of the test head 210 and possibly its orientation ORI_INSP(p). The poses PE_INSP(p) in or at which the test head 210 is actually positioned for a respective measurement process for recording a respective A-image IMAA_INSP(p) are detected with the detection device 230, for example. Consequently, an A-scan IMAA_INSP(p) and the corresponding pose PE_INSP(p) are in turn uniquely assigned to one another. Ideally, as already mentioned, the grids RAST(j)=RAST of the reference measurements REEM(j) are largely identical and the positions POS_INSP(p) are distributed on the surface 101 in such a way that they correspond to the intended measurement positions POS of the test grid RAST defined for the reference measurements are equivalent to. The totality of all actual poses PE_INSP(p) of the inspection head 210 during the inspection measurement INSPM, on which an individual measurement is carried out to generate a respective A image IMAA_INSP(p), is referred to below as m_insp.

Finally, the corresponding plurality P of A-images IMAA_INSP(p) as well as the total m_insp of the corresponding poses PE_INSP(p) result from the totality of the P individual measurements of the inspection measurement INSPM.

For the sake of simplicity, it is assumed below both with regard to the reference measurements REFM(j) and with regard to the inspection measurement INSPM that the poses PE_REF(j) and PE_INSP(p) and thus the corresponding totalities m_ref(j) and m_insp only include the positions POS_REF(j) or POS_INSP(p), but not the orientations ORI_REF(j) or ORI_INSP(p) .

For each artificial defect DEFj, ie for each j=l, 2 data generated in the process, the evaluation to create database BEW. The method ERBEW is again explained in the following as an example for j=1, but can be used in an analogous way for j=2, J.

In a first selection step SELl (j), those positions POS_INSP(p') of the inspection measurement INSPM are now selected for a respective j from the totality m_insp which, due to their spatial arrangement, correspond to the measurement positions POS (j) provided for the corresponding reference measurement REFM(j ) defined test grid RAST(j) come closest. The entirety of the measurement positions POS_INSP(p′) selected in this way for a respective j is denoted by m_insp_sel(j).

Assuming that the grids RAST(j) are identical for different j, i.e. RAST ( k) =RAST ( j ) for kV j , it can be assumed that the ensembles m_insp_sel ( k) and m_insp_sel ( j ) are also identical are to each other or include the same measurement positions POS_INSP(p').

In order to clarify this procedure, FIG. 6 shows a plan view of the test grid RAST(j=1) on the surface 101 of the test object 100, comparable to FIG. 5 for j=1. In addition to the test grid RAST(1), FIG 6 exemplarily shows the entirety m_insp of the positions POS_INSP1, ..., POS_INSP10 of the inspection head 210 actually approached during the inspection measurement INSPM, the individual positions POS_INSP(p) of the entirety m_insp, at which the inspection head 210 was actually positioned for the inspection measurement INSPM crosses are marked. It can be seen here that the positions POS_INSP(p) of the inspection measurement INSPM marked by crosses do not always correspond to the intended measurement positions POS (j=l) of the test grid RAST (1), which are still in the crossing points of the horizontal and vertical Lines of the test grid RAST (1) lie. Depending on how far a position POS_INSP(p) deviates from these crossing points POS (1), this position POS_INSP(p) becomes for m_insp_sel ( j =1 ) selected or not for the subsequent steps of the method ERBEW.

A criterion could be, for example, that a position POS_INSP(p) can only be selected for m_insp_sel ( j ) if its distance APOS_INSP(p) from the nearest crossing point POS (j) of the test grid RAST(j) is smaller than e.g .a fraction 1/n of the distance APOS between two adjacent intersections. I.e. a position POS_INSP(p) can only be selected for the whole m_insp_sel ( j ) if APOS_INSP (p) < APOS ( j ) /n applies. For example, n=3 can apply to n.

This or a similar requirement ultimately also addresses the fact that, as illustrated for j=l in FIG in the example shown only those marked with POS_INSP3, ..., POS_INSP8. However, POS_INSP(p) positions outside of REL (1) do not meet the condition APOS_INSP (p) <APOS (1) /n, and are therefore not considered for m_insp_sel (1).

In other words and expressed more generally, for a respective j for m_insp_sel ( j ), those positions POS_INSP(p') are selected from the total m_insp to form m_insp_sel ( j ) which, due to their spatial arrangement, correspond to the previously defined test grid RAST(j) come closest and which are at the same time in the coverage area REL(j). m_insp_sel ( j ) is a possibly spurious subset of m_insp, since m_insp_sel ( j ) and m_insp can be identical to one another in extreme cases.

For the comparison of the positions of the entirety m_insp with the intended measurement positions PE(j) of the reference measurement REFM(j) made in the respective first selection step SELl (j) to determine the closest positions, it can be assumed that the coordinate system of the Inspection measurement INSPM and the respective reference measurement REFM (j) in a known relationship to each other. This can be achieved, for example, in that the coverage area REL(j) of the test grid RAST(j) for the inspection measurement INSPM is projected onto the surface of the test object, e.g. by the detection device 230, and the probe 210 only within the area marked in this way is moved .

In a second selection step SEL2(j), those positions POS_REF(j) that correspond to the positions POS_INSP( p′) are spatially closest to all m_insp_sel (j) of the first selection step SEL1 (j), i.e. have the smallest spatial distance, e.g. according to the absolute value minimum of the corresponding difference vector. This is in turn visualized in FIG. 7 as an example for j=l, in which, as in FIG are close to the positions POS_INSP(p) are marked by dots. The positions POS_REF (1) and POS_INSP(p) that are closest to each other are assigned to one another and thus form corresponding pairs PP(j)_prj with prj = l, ..., PR(j) , where PR(j) is the number such pairs PP for the respective j. For example, according to FIG. 7, for j=l, the positions POS_REF(j=l)_1 and POS_INSP3 are associated with one another because they are close together and form a pair PP(j=l)_l. Likewise, positions POS_REF (1) _2 and POS_INSP4 are associated with each other and form a pair PP (1) _2. The same applies to POS_REF (1) _3 with POS_INSP5, POS_REF (1) _4 with POS_INSP6, POS_REF (1) _5 with POS_INSP7 and POS_REF (1) _6 with POS_INSP8. Hence PR(j)=6 for j = l and consequently prl = l, ..., 6.

The entirety of the selected in this course in the second selection step SEL2 (j) for a respective j from m_ref (j) th positions POS_REF(j) is referred to below as m_ref_sel(j).

For the comparison in the respective second selection step SEL2 (j) of the positions of the entirety m_ref (j) of the respective reference measurement REFM(j) with the positions of the entirety m_insp_sel (j) determined in the first selection step SEL1 (j) to determine the spatially am For the closest positions, it can also be assumed that the coordinate system of the inspection measurement INSPM and that of the respective reference measurement REFM(j) are in a known relationship to one another.

Utilizing the previously mentioned unique assignment of each A-scan from IMAA_REF(j) to a corresponding pose from PE_REF(j) or position from POS_REF(j) of the reference measurement REFM(j), in an A-scan selection step IMAASEL (j) the A-scans IMAA_REF(j) of the reference measurement REFM(j) associated with the measurement positions m_ref_sel (j) selected in the second selection step SEL2 (j) are selected, i.e. the associated A-scan is selected for each measurement position in m_ref_sel (j). selected from IMAA_REF(j). The A-scans selected based on the measurement positions in m_ref_sel (j) from IMAA_REF(j) form a set IMAA_REF_SEL ( j ) , which therefore includes those A-scans IMAA_REF (j, nj , mj ) from IMAA_REF(j) that for the artificial defect DEFj were measured at those positions POS_REF (j, n j , mj ) of the probe 210 from POS_REF(j) which best correspond to the positions contained in m_insp_sel ( j ).

In the ideal case, in the respective pairs PP(j)_l, ..., PP(j)_6 one of the assigned positions POS_INSP3, ..., POS_INSP8 from the totality m_insp_sel ( j ) and POS_REF ( j ) _1 , ... , POS_REF(j)_6 from the set m_ref_sel (j) match exactly. In practice, however, this will generally not be the case: FIG. 7 again shows an example of j=1, comparable to FIG. 5 and FIG Positions POS_REF(j=1), POS_INSP of the two sets m_insp_sel(j=1) and m_ref_sel(j=1). It can be seen that mutually associated positions POS_REF(1), POS_INSP of a respective pair PP(j)_prj are close to one another, but do not match.

These deviations of the mutually associated positions of the inspection measurement INSPM and the respective reference measurement REEM(j) are accompanied by different sound paths of the ultrasonic pulses UIMP and the corresponding ultrasonic echoes UECHO. This is visualized in FIG. There, as an example for j=l in a view along the surface 101, the positions of the two entities m_insp_sel (j=1) are marked with crosses and m_ref_sel (j=1) with dots. The sound paths to a defect 110 are shown as an example for the mutually assigned positions of the pair PP(1)_5 comprising the positions POS_REF(1)_5 and POS_INSP7. The sound paths from the other positions are not shown for clarity. Due to the offset of the position POS_REF (1) _5 compared to the position POS_INSP7, the corresponding distances or sound paths between the respective position and the defect 110 differ by AL (j=l, prj=5) = L ( POS_INSP7 ) - L ( POS_REF ( j =1 ) _5) , where L stands for the respective distance between the relevant position of the probe 210 and the defect 110 .

Taking this up, an associated sound path correction AL (j, pr j ) can be determined, which is used in each case to correct the A-scan IMAA_REF belonging to the corresponding position POS_REF. This correction consists of a displacement of the echo signals corresponding to the sound path correction AL(j, pr j ) in that A image IMAA_REF(j) which is uniquely assigned to the position POS_REF(j) of the pair PP(j)_prj. In addition or as an alternative to the position correction in PCORR(j), for each j in a rotation correction RCORR(j), which is also optional, it can be taken into account that, particularly with a flat test surface 101 during the inspection measurement INSPM, the test head 210 rotates PHI around the vertical axis compared to the alignment can occur in the reference measurements REFM(j). In the event that the sound field of the probe 210 is not rotationally symmetrical, which can be the case, for example, when using a non-circular aperture and/or when measuring with insonification angles V0°, the effect of this rotation PHI can be calculated based on a 3D sound f field simulation, as described, for example, in "Sound Field Calculation for Rectangular Sources", Kenneth B. Ocheltree, Leon A. Frizzell, IEEE Transactions on Ultrasonics, Vol. 36, No. 2, 1989, or in "Beam focusing behavior of linear phased arrays", Azar et al., NDET&E International, Elsevier, 2000, p. 189 -198. This is also ideally carried out before the actual inspection measurement INSPM in order to save computing time. The 3D sound field simulation can be used to calculate the sound pressure for both situations, where psound (j, n , m , PHI_REF(j) ) describes the sound pressure psound for the individual measurement (n , m ) of the respective reference measurement REFM(j) and psound (PHI_INSP, nj , mj ) stands for the sound pressure psound for the individual measurements (nj, mj ) of the inspection measurement INSPM. The resulting correction factor , with which the amplitudes of the A images IMAA_REF (j, nj , mj ) are corrected by multiplication, is calculated according to FKR (j, nj , mj ) =psound (PHI_INSP, nj , mj) ² /psound (j, nj , mj , PHI_REF ( j ) ) ² .

In addition or as an alternative to the position correction in step PCORR(j) and/or to the rotation correction in step RCORR(j), effects of coupling differences can be corrected for each j in a likewise optional coupling correction ACORR(j). For this purpose, in the reference measurements REFM(j) as well as in the inspection measurement INSPM in certain correct, comparable measurement situations, amplitudes of the corresponding ultrasonic echoes UECHO(i) are compared with one another in order to monitor the coupling quality/strength or to compensate for differences between the reference measurement REFM(j) and the inspection measurement INSPM. Such a specific measurement situation, in which the amplitudes A_INSPM of the echoes in the inspection measurement INSPM and the amplitudes A_REEM of the corresponding echoes in the reference measurements REFM(j) should be essentially the same, can be, for example, the scanning of the rear wall RW of the examined test object 100 or test object 100T. In the event that A_INSPM_RW and A_REFM_RW deviate from one another by more than a tolerance value, for example 5%, the amplitudes of the A-scans IMAA_REF of the reference measurements REFM(j) can be corrected by multiplying them by a factor FKA=A_INSPM_RW/A_REFM_RW.

The optional correction step CORR(j), possibly including the position correction PCORR(j), the rotation correction RCORR(j) and/or the coupling correction ACORR(j), processes or corrects the selected A-images IMAA_REF_SEL as input data for a respective j ( j ) and in turn supplies corrected A-scans IMAA_REF_SEL ( j ) .

Finally, in a reconstruction step SAFTREC(j), for each j, i.e. for each artificial defect DEFj , a conventional SAFT analysis based on the quantity IMAA_REF_SEL ( j ) generated in this way of possibly corrected A-images is carried out, which is carried out in particular for the artificial Defective DEFj mapping voxels corresponding amplitude sums. These amplitude sums and the quantities GRj and depths TIEj associated therewith, which are known for the artificial defects DEFj, are ultimately used to create the evaluation database BEW described above.

The evaluation database BEW, which is thus created individually for the test object 100T, is finally used to calculate the values during the inspection measurement INSPM of the corresponding test object. jekts 100 to convert recorded data into representative mmKSR values.

For this purpose, a conventional SAFT analysis is first carried out based on those A-images in the set IMAA_INSP_SEL( ) which are uniquely assigned to the selected measurement positions m_insp_sel ( ) of the inspection measurement INSPM, as repeatedly emphasized above.

In the event that, as explained above, the grids RAST(j) for different j are typically largely identical, i.e. RAST ( k) = RAST ( ) for kV j , it can be assumed that the wholes m_insp_sel ( k) and m_insp_sel ( ) are identical to one another or include the same measurement positions POS_INSP(p'), so that the entirety of the associated A-scans in IMAA_INSP_SEL ( ) and in IMAA_INSP_SEL ( k) are identical, i.e.

IMAA_INSP_SEL ( ) =IMAA_INSP_SEL (k) =IMAA_INSP_SEL . Consequently, the conventional SAFT analysis is performed based on precisely these A-frames of this IMAA_INSP_SEL ensemble.

However, if test grids RAST(j) , RAST(k) for kVj differ so much from each other, i.e. RAST ( k) VRAST ( j ) , that the ensembles m_insp_sel ( k) and m_insp_sel ( j ) and thus also the ensembles of the associated A -Images in IMAA_INSP_SEL ( j ) and in IMAA_INSP_SEL ( k ) are different, i.e. m_insp_sel ( k ) Vm_insp_sel ( j ) and IMAA_INSP_SEL ( k ) VIMAA_INSP_SEL ( j ) , for each j separately a conventional SAFT analysis is performed based on the respective ensemble IMAA_INSP_SEL ( j ) executed.

The SAFT analysis based on IMAA_INSP_SEL results in corresponding amplitude sums for the voxels of the examined test area of the test object 100. For example, for a voxel which contains the defect 110 of the test object 100, the amplitude sum corresponding to this voxel can be be converted into a defect size in mmKSR using the evaluation database BEW.

As already mentioned, the overall reference measurement REFM comprising the individual reference measurements REFM(j) is typically carried out before the actual inspection measurement INSPM of the test object 100 to be examined. This means that the information needed to determine the size is already available at the time of the inspection measurement INSPM. However, it is also conceivable to carry out the overall reference measurement REEM after the inspection measurement INSPM. In this case, the results of the sizing are only available afterwards, but this can be acceptable depending on the application of the SAFT examination.

The probe 210 can, for example. be designed as a so-called "phased array probe". A phased array probe allows test objects to be scanned not only mechanically but also electronically, i.e. by a type of electronic shifting of the active zone of the probe several Measurements can be carried out in a defined test grid With a stationary probe, data recorded with the same electronic scan can be evaluated with the SAFT analysis This works both with a stationary probe and with a probe moving during the electronic scan if the transmit and receiving positions as well as insonification angles and focusing are known at the time of reconstruction.

reference sign

100 test obj ect

100T test obj ect 110 defect

101 surface

200 device for ultrasonic testing

210 probe

220 control unit 225 connection

230 detector

240 display device

DEFI , DEF2 , DEF3 artificial defect

Claims

32 patent claims patent claims

1. A method for determining the size of a defect (110) in a test area of a test object (100) using a SAFT method, wherein

- in an inspection measurement INSPM of the test object (100) based on SAFT with a probe (210), a large number of A-images IMAA_INSP(p) are recorded in different actual poses PE_INSP(p) of the probe (210),

- In an overall reference measurement REFM on a test object (100T) corresponding to the test object, which has a number J>1 of known defects DEFj with j=l, J with known parameters (TIEj, GRj), SAFT-based for each of the artificial defects DEFj in a respective reference measurement REFM(j), a number Vj>l of A-images IMAA_REF(j) is recorded in different actual poses PE_REF (j, n j , mj ) of the test head (210), wherein

- in a first step, based on a SAFT analysis of a selection IMAA_REF_SEL (j) of A-scans of the reference measurements REFM(j) and based on the known parameters of the defects DEFj, an evaluation database BEW is created,

- in a second step, based on a SAFT analysis of a selection of A-images of the inspection measurement INSPM, AMPL amplitude sums for the spatial elements of the test area of the SAFT method in the test object (100) are determined and the AMPL amplitude sums of the spatial elements are calculated using the evaluation database BEW in the size of the defect to be determined can be converted.

2. Method according to claim 1, characterized in that in the first step the evaluation database BEW is created using amplitude sums resulting from the SAFT analysis and using the known parameters of the known defects DEFj. 33

3. The method according to any one of claims 1 to 2, characterized in that to create the evaluation database BEW

- for each j in a reconstruction step SAFTREC(j) for the respective j a SAFT analysis based on the A-images of the selection IMAA_REF_SEL ( j ) is carried out, resulting in amplitude sums for the spatial elements of a test area of the SAFT method in the test object ( 100T) , in particular for the space elements containing the known defects DEFj, and

- the evaluation database BEW is created using the resulting amplitude sums and the associated parameters known for the known defects DEFj.

4. The method according to any one of claims 1 to 3, characterized in that for a respective reference measurement REFM (j) a test grid RAST (j) with provided measurement poses PE (j) is specified, in the best possible proximity of the probe (210) for recording of a respective A-scan IMAA_REF ( j , nj , mj ) of the reference measurement REFM(j) is to be positioned.

5. The method according to any one of claims 1 to 4, characterized in that in the first step for the selection of A images for the respective selection IMAA_REF_SEL (j) for a respective j

- in a first selection step SELl (j), from a totality m_insp of the poses PE_INSP(p) of the inspection measurement INSPM to form a selection m_insp_sel (j), those poses PE_INSP(p') are selected which, based on their spatial arrangement, are intended measurement poses PE(j ) come closest to a test grid RAST(j) defined for the respective reference measurement REFM(j),

- in a second selection step SEL2 (j) from a total m_ref (j) of actual poses PE_REF ( j , nj , nj ) of the respective reference measurement REFM (j) to form a selection m_ref_sel (j) those actual poses PE_REF ( j , nj , mj ) are selected that are spatially closest to the poses PE_INSP(p') of the set m_insp_sel ( j ), - in an A-scan selection step IMAASEL(j), the A-scans IMAA_REF(j) associated with the selected actual poses of the entity m_ref_sel(j) of the respective reference measurement REFM(j) are selected to form the selection IMAA_REF_SEL(j).

6. The method according to claim 5, characterized in that in the first selection step a pose PE_INSP(p) of the whole m_insp is then selected for m_insp_sel(j) if its spatial distance APOS_INSP(p) from the nearest intended measurement posePE(j) of the test grid RAST(j) is smaller than a predetermined threshold value, in particular smaller than a fraction 1/n of the distance APOS between two adjacent measurement poses PE(j) provided for the test grid RAST(j).

7. The method according to any one of claims 5 to 6, characterized in that in the second step, in the event that the sets m_insp_sel ( j ) are largely identical for different j, the selection of A images for the SAFT analysis those A inspection measurement images INSPM associated with the poses PE_INSP(p) of any one of the sets m_insp_sel( ).

8. The method according to any one of claims 5 to 7, characterized in that in the second step, if the sets m_insp_sel ( ) are at least partially different for different j,

- For each j, a SAFT analysis of those A-images IMAA_INSP(p) of the inspection measurement INSPM is carried out separately, which are assigned to the poses PE_INSP(p) of the respective selection m_insp_sel ( j ), resulting in respective individual amplitude sums AMPL(j) for the Space elements of the test area of the SAFT process in the test object (100), where

- Either a mean value is formed from the individual amplitude sums AMPL(j), which represents the amplitude sums AMPL to be determined, which are then calculated using the Evaluation database BEW can be converted into the size of the defect to be determined, or

- respective individual amplitude sums AMPL(j) are converted separately into individual defect sizes using the evaluation matrix BEW and the size of the defect to be determined is determined as the mean value of the individual defect sizes.

9. The method according to any one of claims 5 to 8, characterized in that for each j in the second selection step SEL2 (j) in a position correction step PCORR(j) of a respective pose PE_REF of the selection m_ref_sel (j) that pose PE_INSP from the whole m_insp_sel ( j ) which is spatially closest to it, where for each pair PP of poses formed in this way

- a difference in the sound paths between the respective pose PE_REF, PE_INSP and the defect (110) is determined and

- That A-scan is corrected based on the determined difference, which is assigned to the pose PE_REF of this pair PP.

10. Method according to one of claims 5 to 9, characterized in that for each j in a coupling correction step ACORR(j)

- amplitudes of ultrasonic echoes UECHO(i) of the respective reference measurement REFM(j) are compared with amplitudes of corresponding ultrasonic echoes UECHO(i) of the inspection measurement INSPM and

- in the event that the compared amplitudes deviate from one another by more than a predetermined factor, the amplitudes of the A-scans of the corresponding individual reference measurement REFM(j) are corrected based on the determined deviation.

11. The method as claimed in one of claims 1 to 10, characterized in that the test head is moved manually to different poses. 36

12. Device (200) for determining a size of a defect (110) in a test area of a test object (100) using a SAFT method, comprising a control unit (220) which is set up to implement a method according to any one of claims 1 to 10 to execute.

13. Device (200) according to claim 12, comprising

- A transmitter (210) for sending ultrasonic pulses UIMP(i) into the test object (100),

- A receiver (210) for receiving with the transmitted ultrasonic pulses UIMP (i) corresponding ultrasonic echoes UECHO (i), wherein the receiver (210) received measurement signals in the reference measurements REFM (j) or in the inspection measurement INSPM recorded A- form images.

14. Device (200) according to any one of claims 12 to 13, characterized in that the transmitter (210) and the receiver (210) are housed in a common test head (210), the test head (210) for executing the SAFT

Proceedings can be moved manually.

15. Device (200) according to any one of claims 12 to 14, comprising a detection device (230) for determining a respective pose of the transmitter (210) and / or the receiver (210).