METHOD FOR DETECTING MATERIAL INHOMOGENEITIES

Description

[0001] The present invention relates to a method for detecting inhomogeneities, in particular non-metallic inclusions, in a workpiece. State of the art

[0002] The strength of a metallic component depends largely on the purity of the material. Non-metallic inclusions or other defects are weak points where the material is more likely to fail under high stress. Therefore, there is a need to monitor the purity of the material during production.

[0003] Microscopic examination of metallographic sections allows defects to be detected with a spatial resolution that corresponds to the resolution of the microscope used. Because appropriate sample preparation is required, the examination is comparatively time-consuming.

[0004] Therefore, ultrasonic immersion testing is a common method for testing metallic materials. In this method, a test specimen made of the material in a coupling liquid, such as water, is irradiated with ultrasonic waves. The ultrasonic waves are partially reflected at the interfaces with defects. A quasi-standard for this testing method is described in the Stahl-Eisen test sheet SEP 1927, which is available from Stahl-Eisen-Verlag. Corresponding testing devices are known, for example, from DE 40 36 005 A1, EP 1 475 633 A1, US Pat. No. 6 318 178 B1, and WO 2006 120 875 A1.

[0005] DE 196 25 763 A1 discloses a device for testing the quality of tires. In this device, a tire is exposed to a field. An evaluation device detects a change in the field, which is a measure of the tire's condition.

[0006] DE 10 2009 043 001 A1 discloses a method for determining defects in a material transparent to electromagnetic waves, in particular for optical purposes.

[0007] US 3,766,387 discloses a non-destructive testing device that detects material defects using radiation.

[0008] GB 2521767 A discloses a scanning method that uses gamma radiation and radiation detectors to detect defects in underwater pipelines.

[0009] The applicant's tests have shown that even non-metallic inclusions that are too small to be reliably detected using current ultrasonic immersion technology can significantly weaken the strength of the material and lead to component failures. Disclosure of the invention

[0010] Within the scope of the invention, a method was developed for detecting a marker indicating an inhomogeneity in the material of a workpiece. This marker is evaluated from a signal with which the workpiece responds and / or has responded to a receiver in response to interrogation radiation from a transmitter. The interrogation radiation can, in particular, comprise sound waves, electromagnetic radiation, and / or ionizing radiation.

[0011] The perfect tense "answered" should be understood to mean that not only "fresh" measurement data recorded specifically for the purpose of the method can be used for evaluation, but also existing measurement data can be re-evaluated using the method. This is particularly advantageous if the workpiece is no longer available for further measurement data acquisition, for example, because it has since been subjected to destructive testing.

[0012] The term "workpiece" is not limited to test specimens manufactured specifically for material testing. A workpiece can also be a component or a pre-product intended for further processing into a component.

[0013] According to the invention, the marker is evaluated from a portion of the signal which changes periodically with the angle φ upon rotation and / or pivoting of the workpiece, and / or the transmitter, and / or the receiver by an angle φ.

[0014] This means that the part of the signal that changes periodically with the angle φ is first separated and that the marker is then evaluated from this part.

[0015] This separation can be performed offline, for example, after the complete acquisition of the measurement data. However, to further improve sensitivity, a rapid rotation or tilting at a specified angular frequency ω can also be performed during data acquisition, for example, and a signal that changes at this angular frequency ω can be preferentially detected, for example, using a lock-in amplifier.

[0016] It was discovered that by evaluating the signal component dependent on the angle φ, signals emanating from actual defects in the workpiece can be separated from background noise and false detections much more accurately than was possible with the previous state of the art. Especially when examining with ultrasonic immersion technology, sensitivity can be significantly improved, allowing significantly smaller defects to be reliably detected.

[0017] The invention is based on the basic idea that the workpiece is not more or less homogeneously interspersed with defects, but that such defects occur sporadically. Within the fixed workpiece, each defect is stationary. If the workpiece is then rotated or pivoted by the angle φ, the relative geometry between the transmitter, the defect, and the workpiece changes. This change affects the registered signal, which thus receives a component that depends on the angle φ with some kind of continuous function. The sources of background noise, on the other hand, are statistically distributed in terms of position and orientation within the workpiece and therefore, on average, show no functional dependence on the angle φ.

[0018] In a particularly advantageous embodiment of the invention, a cylindrical workpiece is selected, and the rotation occurs around the cylindrical axis of the workpiece, or it occurred in this way when the signal was recorded. If, for example, the interrogation radiation is coupled into the cylindrical surface at a predetermined location and the signal is coupled out of the cylindrical surface at the same location or at a different location in the direction of the receiver, the signal should not change as a result of the rotation for a completely homogeneous workpiece. A component of the signal that correlates with the angle φ of the rotation most likely originates from an inhomogeneity in the workpiece material.

[0019] In a further particularly advantageous embodiment of the invention, a query radiation is selected which is reflected at the inhomogeneity and which is attenuated by the material of the workpiece only to such an extent that the reflected signal can still be detected by the receiver. Ultrasound is particularly suitable for this purpose. At the interfaces to defects, the local speed of sound changes and thus also the refractive index for the query radiation, so that the query radiation is partially reflected. The speed of sound lies in a range in which it is technically comparatively easy to deduce the depth at which the signal originates in the workpiece, i.e. the defect, from the travel time T of the received signal.The principle remains unchanged if electromagnetic interrogation radiation is chosen, but measuring the time of flight T is more complex because the speed of light is significantly higher than the speed of sound. If the interrogation radiation is ionizing radiation emitted according to a statistical time program, a coincidence circuit, for example, can be used to measure the time of flight.

[0020] In a particularly advantageous embodiment of the invention, the component of the signal which varies with the angle φ is evaluated from a representation of the signal in which an amplitude A of the signal is plotted as a function of the angle φ on the one hand and a propagation time T of the signal and / or a quantity q derived from the propagation time T of the signal on the other hand.

[0021] The propagation time T is, in particular, a measure of the distance traveled by the interrogation radiation from the transmitter to a defect within the workpiece material, and the distance traveled by the signal generated at the defect on its way to the receiver within the workpiece material. This distance depends on the relative geometry between the transmitter, the receiver, and the defect. In the above-mentioned plot against the angle φ on the one hand and the propagation time T, or the derived quantity q, on the other hand, a genuine defect in the workpiece should therefore lead to particularly high signal amplitudes A at values ​​of T or q that are periodic with the angle φ. In this way, a genuine defect can be reliably distinguished not only from background noise, but also from other false detections.

[0022] Particularly advantageously, the distance X between the origin of the signal in the workpiece and a given location P where the signal exits or has exited the workpiece is chosen as the derived quantity q. This distance X is an approximation for the distance traveled by the interrogation radiation, or by the signal, in the material of the workpiece with a comparatively simple functional dependence on the angle φ.

[0023] In order to separate the part of the signal which varies with the angle φ, in a particularly advantageous embodiment of the invention the course of the amplitude A of the signal is determined along at least one trajectory curve which describes the T or q coordinate of a fixed origin of the signal as a function of the φ coordinate of this origin.

[0024] The relative geometry between the defect, the transmitter, and the receiver depends particularly on the position of the defect within the workpiece. For example, if the workpiece is a cylinder and an approximately disc-shaped subvolume of this cylinder is being examined, the decisive degrees of freedom for the position of the defect are the radial distance from the center of the disc and the azimuthal angular position on the disc. These two quantities influence, for example, the functional dependence of the distance X on the angle φ. Since the defect is precisely what the method is intended to detect in the first place, the position of the defect is unknown.In order to detect the existence of the defect and at the same time obtain information about its position, in a further particularly advantageous embodiment of the invention, an approach with a set of free parameters is established for the trajectory, such as the two degrees of freedom for the position of the defect. The determination of the course of the amplitude A of the signal is then repeated for a large number of value vectors of the parameters. If the course of the amplitude A of the signal for one of these value vectors particularly plausibly agrees with the hypothesis that a defect is present in the workpiece, then it is also plausible that the value vector contains the realistic values ​​for the free parameters. Thus, in the example mentioned, the polar coordinates of the defect are fixed within the disk-shaped partial volume under investigation.

[0025] In another particularly advantageous embodiment of the invention, the sum G sum of the signal amplitudes A along the corresponding trajectory curve in the signal representation is determined for each parameter value vector. This allows one to quantify the degree to which the signal amplitude A curve corresponds to the hypothesis that a defect is present and, at the same time, whether the value vector contains the realistic values ​​for the free parameters.

[0026] To reliably determine that a defect is present in the workpiece, the amplitude A curve, which periodically depends on the angle φ, should stand out from the background noise with a certain minimum degree of contrast. Therefore, in another particularly advantageous embodiment of the invention, the value vector for which the sum G sum reaches its maximum G max is considered a marker or a candidate for a marker if the ratio of the maximum G max to the mean or median G mean of all amplitudes A in the signal representation reaches or exceeds a predetermined threshold value S.

[0027] For example, it can be concluded from the examination of a disk-shaped partial volume of a cylindrical workpiece that the workpiece contains a defect. However, this finding can also be advantageously verified by examining another partial volume of the workpiece. Therefore, in a further particularly advantageous embodiment of the invention, at least two representations of signals plotted against the travel time T or the derived variable q on the one hand and against the angle φ on the other hand, which relate to adjacent and / or overlapping partial volumes of the workpiece, are evaluated. It is then advantageously possible, in particular, to conclude that there is a marker for a defect if the candidates determined from the representations belonging to both partial volumes lie within a predetermined distance D from one another.

[0028] In this way, the size, extent, and course of the defect in the workpiece can also be determined in all three spatial dimensions. For example, the defect can begin in the center of a first disc-shaped sub-volume and migrate further and further into the upper left quadrant of the disc in the subsequent sub-volumes along the cylinder axis. This more precise knowledge can be used, for example, to investigate the cause of the defect.

[0029] The detection of a signal component dependent on angle φ is not limited to the method described here. Any suitable pattern recognition method can be used. In particular, the signal can be transformed into another mathematical space in which the component of the signal dependent on angle φ and / or periodic at angle φ becomes more apparent. For example, the Hough transform is particularly suitable for detecting straight lines, circles, or any other parameterizable geometric figures in black and white images. The Fourier transform specifically emphasizes periodic signal components. However, a neural network can also be trained to detect a periodic signal component, for example.

[0030] This method allows a device for non-destructive material testing of workpieces to be upgraded to detect smaller inhomogeneities than was previously possible. The invention is as defined in the claims.

[0031] According to the invention, means are provided which are designed to carry out a method according to the invention.

[0032] The method can also be carried out on an unmodified ultrasonic testing system. Based on the above, existing data can also be re-evaluated, allowing new insights into inhomogeneities in the material of a workpiece to be gained without handling the workpiece again. Thus, the method can be implemented, in particular, partially or even completely, in software, and this software is a standalone, saleable product whose use does not require the presence of a physical testing system. The invention therefore also relates to a computer program product with machine-readable instructions which, when executed on a computer, cause the computer to carry out a method according to the invention.

[0033] The main application of the invention is the quality control of steels used to manufacture highly stressed and / or safety-relevant vehicle components for mesoscopic non-metallic inclusions. Examples of such components include injectors for diesel injection systems, components for high-pressure pumps in diesel injection systems, components in the drive train of such high-pressure pumps, as well as axles and other chassis components. In injectors, for example, the nozzle and the high-pressure bore are subject to particularly high stress.

[0034] In the applicant's tests, three non-metallic inclusions with a minimum KSR size of more than 0.3 mm were detected in sample material made of tempered steel using the ultrasonic immersion technique according to test sheet SEP 1927. However, using the method according to the invention, 45 non-metallic inclusions with a minimum KSR size of more than 0.2 mm were detected in the same workpiece. Had the steel been processed into highly stressed components, these inclusions could have led to premature failure of the components.

[0035] Further measures improving the invention are presented in more detail below together with the description of the preferred embodiments of the invention with reference to figures. Examples of implementation

[0036] It shows: Figure 1 Basic principle of ultrasonic immersion technology for examining a workpiece 1; Figure 2Extraction of the signal resulting from an inhomogeneity 2a in the material 2 of the workpiece 1 from the measured signal 6; Figure 3 Detection of the periodic component 6a of the signal 6 at angle φ with a parameterized approach 6e for trajectories 6d, 6d'; Figure 4 Summary of the method 100 including plausibility check 126 of several candidates 7a, 7a' for a marker 7 indicating an inhomogeneity 2a in the material 2 of the workpiece 1.

[0037] According to the not to scale Figure 1 In an immersion tank 50 filled with water 51 as a coupling medium, ultrasonic waves 5a are radiated as interrogation radiation 5 from the transmitter 3 into the workpiece 1. Where the material 2 of the workpiece 1 has non-metallic inclusions 2a, the interrogation radiation 5 produces an echo signal 6, which is picked up by the receiver 4.

[0038] The test head 52 with the transmitter 3 and the receiver 4 is movable in the x-direction. At the same time, the cylindrical workpiece 1 is rotatable 110 about its cylinder axis 1a. For each x-position of the test head 52 and for each angular position φ of the workpiece 1, the amplitude A of the signal 6 can be determined as a function of the y-coordinate from which the reflected signal 6 originates. The y-coordinate results from the propagation time T of the signal 6. The plot of A against y, which is shown in Figure 1 shown as an example for a value of the axial position x of the test head 52 and a value of the angle φ in the diagram is called an A-scan.

[0039] The very strong echoes at the left and right edges of the diagram result from reflections at the interface between workpiece 1 and water 51. The inhomogeneity 2a in the material 2 of workpiece 1 causes a peak 6z in signal 6. This peak 6z lies in the background noise range, so it cannot be reliably inferred from it that the presence of the inhomogeneity 2a is present.

[0040] According to the current state of the art, only the maximum amplitude A was further evaluated if it exceeded a detection threshold S1. This maximum was plotted as a function of y and φ in a diagram called a C-scan. From this, a cumulative length of defects along the y-axis was determined. The ratio of this cumulative length to the test volume was the defect index, which was used as a measure of the purity of material 2 of workpiece 1.

[0041] The value of the detection threshold S1 is in Figure 1The small inhomogeneity 2a, which is shown in Figure 1 The situation shown is not recognized as such with the current evaluation method. Lowering the detection threshold S1 to a value S1' is also not a solution, as false detections would then accumulate. The (in Figure 1 The C-scan (not shown) becomes a filled area that is no longer meaningful.

[0042] In the Figure 1In the example shown, the probe 52 has a diameter of 6 mm and transmits at a frequency of 10 MHz. To calibrate the amplitude scale, a reference body is examined that has a 1 mm deep hole in its front. The gain is adjusted so that this hole results in an amplitude of 80% of the maximum and that the absorption of the interrogation radiation 5 by the material 2 is compensated. With the evaluation according to the state of the art, non-metallic inclusions 2a can be detected down to a minimum size of 0.3 mm. The spatial resolution is 1 mm in the x-direction and 0.25 mm in the circumferential direction of the workpiece 1.

[0043] Figure 2illustrates, using an exemplary cylindrical workpiece 1 with radius r, the material 2 of which contains only a single inhomogeneity 2a at a distance a from the cylinder axis 1a of the workpiece 1 and at an angular position φ to the connecting line between the cylinder axis 1a and the test head 52, the dependence of the signal 6 originating from this inhomogeneity 2a on the angle of rotation φ of the workpiece 1 about the cylinder axis 1a.

[0044] Figure 2a is a sectional view at the axial position x where the test head 52 is located, and explains in particular the determination of the distance X between the origin 6c of the signal 6 in the workpiece 1 and the location P at which the signal 6 emerges from the workpiece 1. The distance X is given by X = a 2 + r 2 − 2 ⋅ a ⋅ r ⋅ sin φ + φ 0 with a phase shift φ 0 .

[0045] Figure 2bshows a representation 6b in which the amplitude of the signal 6 is plotted against the distance X and the angle φ. In addition to the component 6a, which changes depending on the angle φ, the signal 6 also has a component 6x, which results from the reflection at the interface between the water 51 and the workpiece 1, as well as a component 6y, which results from the reflection at the opposite interface between the workpiece 1 and the water 51. The components 6x and 6y are independent of the angle φ. A further signal component 6w, which results from an apparent detection, depends on the angle φ, but not in a periodic manner.

[0046] Thus, the component 6a can be separated from the signal 6, for example, by using a representation 6b as shown in Figure 2b As shown, a signal periodic at an angle φ is searched for. Any pattern recognition method can be used for this purpose.

[0047] The representation 6b can, for example, be generated and visualized as a gray value data matrix. An 8-bit gray value G can then be calculated from the corresponding amplitude A, for example, with the scaling G = 255 − A ⋅ 255 100 % The grayscale image can be contrast-enhanced in any way using software to achieve greater inspection sensitivity.

[0048] Figure 3 shows schematically such a grayscale image, which comprises m=360 pixels in the φ coordinate and n=512 pixels in the X coordinate, as representation 6b of the signal 6. A portion 6a of the signal 6 which is periodic at the angle φ and a portion 6w of the signal 6 which is non-periodic and dependent on the angle φ and which originates from an apparent detection, stand out so clearly from the background noise that they are recognizable with the naked eye. Figure 3 also explains the automatic detection of the periodic component 6a of the signal 6 in the diagram 6b.

[0049] For this purpose, equation (1) is transformed into a coordinate equation 6e, which links the coordinates (i,j) of the pixels belonging to the trajectory 6d for all trajectories 6d: i = round a 2 + n 2 2 − 2 ⋅ a ⋅ n 2 ⋅ sin j + φ 0

[0050] Here, n is the same for all trajectories 6d, while the radial distance a of the inhomogeneity 2a from the cylinder axis and the phase shift φ 0 form a set 6f of free parameters. All possible discrete values ​​a = [0,..., n / 2] for a and φ 0 = [0,..., 360°] for φ 0 are scanned, and for each possible value vector 6g = (a, (φ 0 ) ), the corresponding trajectory 6d is determined.

[0051] Along each trajectory 6d, the gray values ​​G(i,j) are now determined for all pixels with coordinates (i,j) belonging to the trajectory 6d and added to the sum G sum a φ 0 = ∑ i , j G i j i j ∈ 6 d a φ 0 summed up.

[0052] The agreement of the Figure 3The trajectory curve 6d shown as an example with the periodic component 6a of signal 6 is not yet particularly good. The trajectory curve 6d', however, which was created for a value vector 6g=(a 1 ,φ 0,1 ), exactly reproduces the course of the periodic component 6a of signal 6. Accordingly, the sum G sum (a,φ 0 ) for this value vector also assumes its maximum G max.

[0053] In the present example, the ratio of this maximum G max to the mean value G mean of all gray values ​​in the representation 6b is greater than the specified threshold value S, so that the value vector 6g=(a 1 ,φ 0,1 ) is evaluated as a candidate 7a for a marker 7.

[0054] Figure 4 summarizes the method 100 for the signal evaluation once again schematically and at the same time explains how the candidate 7a for the marker 7 determined in the manner described can be further verified.

[0055] When the probe 52 is in a first position x, a first disk-shaped partial volume 1b of the cylindrical workpiece 1 is examined. The workpiece 1 is rotated 110 once completely around its axis 1a, and the signal 6 is recorded.

[0056] When the probe head 52 is in a second position x', a second disk-shaped partial volume 1b' of the cylindrical workpiece 1 is examined. The probe head 52 is shown in this position x' with the dashed line and the reference symbol 52'. The workpiece 1 is rotated 110 once completely around its axis 1a, and the signal 6' is recorded.

[0057] The signals 6 and 6' are now further evaluated in block 120=125.

[0058] From the signal 6 or 6', the representation 6b or 6b' is obtained, which indicates the amplitude A of the signal 6 or 6' as a function of X and the angle of rotation φ. In step 123, a parameterized approach 6e with free parameters 6f for trajectories 6d of functions periodic at the angle of rotation φ is established, and by inserting all possible value vectors 6g for the free parameters 6f, all possible trajectories 6d are determined.

[0059] Along each of these trajectories 6d, the profile of the amplitude A of the signal 6 or 6' is determined in step 121. This profile is condensed in step 122 into the gray value sum G sum, which is assigned to a specific amplitude profile and thus also to a specific trajectory 6d. The maximum G max of the gray value sums G sum formed across all trajectories 6d is assigned to the trajectory 6d that best matches the periodic component 6a or 6a' of the signal 6 or 6'. This trajectory 6d thus specifies the periodic component 6a or 6a' of the signal 6 or 6'. The values ​​a 1 and φ 0,1 , or a 2 and φ 0,2 , by inserting which as free parameters 6f the trajectory curve 6d was obtained, are evaluated in step 124 as candidates 7a and 7a' respectively for the marker 7 indicating the inhomogeneity 2a in the material 2 of the workpiece.

[0060] Both candidates 7a and 7a' are checked against each other in step 126. For this purpose, it is checked whether the two candidates 7a and 7a' for the marker 7 are within a specified distance D from each other, i.e. whether in both adjacent partial volumes 1b and 1b' of the cylindrical workpiece 1 the inhomogeneity 2a is ultimately detected at the same location in the material 2 of the workpiece 1. In the Figure 4 In the example shown, it is checked whether both |a 1 -a 2 | < Δa and φ 0.1 -φ 0.2 < Δφ are met, with independent threshold values ​​Δa and Δφ. In the example shown in Figure 4 In the example shown, both conditions are met, and a marker 7 is concluded for the presence of the inhomogeneity 2a in the material 2 of the workpiece 1.

[0061] If no periodic component 6a' at an angle φ is detected in the representation 6b' determined from the signal 6', then no trajectory curve 6d can be aligned with such a periodic component 6a'. There is then neither an a 2 nor a φ 0.2 , so the plausibility check fails.

[0062] In the same way, the procedure can be continued for further partial volumes of the workpiece 1, so that at the end a result matrix is ​​created in which all inhomogeneities 2a of the material of the workpiece 1 are mapped in depth, angle and axial position.

Claims

1. Method (100) for detecting a marker (7), which indicates an inhomogeneity (2a) in the material (2) of a workpiece (1), in a signal (6) with which the workpiece (1) responds and / or has responded to a receiver (4) in response to interrogation radiation (5) from a transmitter (3), characterized in that a portion (6a) which is periodic at an angle ϕ and changes periodically, when the workpiece (1) and / or the transmitter (3) and / or the receiver (4) is / are rotated and / or pivoted (110) by the angle ϕ, with the angle ϕ, is first separated from the signal (6) and then the marker (7) from this portion (6a) is evaluated (120).

2. Method (100) according to Claim 1, characterized in that a cylindrical workpiece (1) is selected and the rotation (110) takes place or has taken place around the cylinder axis (1a) of the workpiece (1).

3. Method (100) according to one of Claims 1 to 2, characterized in that interrogation radiation (5) is selected, which is reflected at the inhomogeneity (2a) and which is attenuated by the material (2) of the workpiece (1) only to such an extent that the reflected signal (6) can still be detected by the receiver (4).

4. Method (100) according to one of Claims 1 to 3, characterized in that the portion (6a) of the signal (6), which varies with the angle ϕ, is evaluated (120) from a representation (6b) of the signal (6), in which an amplitude A of the signal (6) is plotted as a function of the angle ϕ, on the one hand, and a propagation time T of the signal (6), and / or a variable q derived from the propagation time T of the signal (6), on the other hand.

5. Method (100) according to Claim 4, characterized in that the distance X between the origin location (6c) of the signal (6) in the workpiece (1) and a predefined location P, at which the signal (6) exits or has exited the workpiece (1), is selected as the derived variable q.

6. Method (100) according to one of Claims 4 to 5, characterized in that the course of the amplitude A of the signal (6) along at least one trajectory (6d), which describes the T or q coordinate of a fixed origin location (6c) of the signal (6) in the representation (6b) on the basis of the ϕ coordinate of this origin location (6c), is determined (121).

7. Method (100) according to Claim 6, characterized in that an approach (6e) with a set of free parameters (6f) is set up for the trajectory (6d) and the determination (121) of the course of the amplitude A of the signal (6) is repeated (123) for a multiplicity of value vectors (6g) of the parameters (6f).

8. Method (100) according to Claim 7, characterized in that the sum Gsum of the amplitudes A of the signal (6) along the associated trajectory (6d) in the representation (6b) of the signal (6) is determined (122) for each value vector (6g) of the parameters (6f).

9. Method (100) according to Claim 8, characterized in that that value vector (6g) for which the sum Gsum assumes its maximum Gmax is classified (124) as a marker (7) or as a candidate (7a) for a marker (7) if the ratio of the maximum Gmax to the mean value or median Gmean of all amplitudes A in the representation (6b) of the signal (6) reaches or exceeds a predefined threshold value S.

10. Method (100) according to Claim 9, characterized in that at least two representations (6b, 6b') of signals (6, 6') that are plotted against the propagation time T or the derived variable q, on the one hand, and against the angle ϕ, on the other hand, and refer to adjacent and / or overlapping partial volumes (1b, 1b') of the workpiece (1) are evaluated (125).

11. Method (100) according to Claim 10, characterized in that a marker (7) is inferred (126) if the candidates (7a, 7a') determined from the representations (6b, 6b') are within a predefined distance D to each other.

12. Device for detecting a marker (7), which indicates an inhomogeneity (2a) in the material (2) of a workpiece (1), in a signal (6), comprising a transmitter (3) for interrogation radiation (5) and a receiver (4) for the signal (6), with which the workpiece (1) responds and / or has responded to the interrogation radiation (5), characterized in that means (120) are provided and first separate a portion (6a), which is periodic at an angle ϕ and changes periodically, when the workpiece (1) and / or the transmitter (3) and / or the receiver (4) is / are rotated and / or pivoted (110) by the angle ϕ, with the angle ϕ, from the signal (6) and then evaluate the marker (7) from this portion (6a).