Method, device and program for detecting defects in materials by ultrasound

Abstract

The present invention relates to a method for detecting defects in materials (MS) by means of ultrasound, transmitted by M transmitters of index i and sampling times nT e In the figure, N columns Y are formed by signals x(n,i,j) by M receivers with index j. n A sampling matrix (A Δ ), where the distance between the receiver with index j and the transmitter with index i is equal to the deviation Δ and the row X i,j is formed by the signal x(n,i,j), and the pair i,j is the row X i,j The steps and the K largest eigenvalues ​​(λ k ) corresponding to the K eigenvectors (V k ) for the matrix (A Δ ) row X i,j Projection of X i,j proj,k The covariance matrix of (C Δ ) and for each row X i,j K projections from X i,j proj,k to obtain a residual measurement signal x*(n,i,j) for detecting defects.

Description

[Technical Field]

[0001] The present invention relates to a method and device for detecting defects in materials by means of ultrasound.

[0002] The field of the invention relates to the non-destructive detection of defects by ultrasound, which can be used in the moldings of the primary circuits of nuclear power plants, and can also be applied to other components of the nuclear fleet, and to other industrial sectors such as aviation or naval. [Background technology]

[0003] The invention is advantageously applied to composite materials, by which is meant any type of material that generates spurious signals that can mask the detection of defects.

[0004] To detect potential defects in a material, a common method is to apply to the surface of the material a multi-element ultrasonic probe containing transducers alternately placed in transmission and reception modes of ultrasonic measurement signals that propagate within the material.

[0005] Spurious signals may correspond, for example, to lateral or near-surface waves, which may substantially interfere with the detection of defects in the first millimeters of the material. This region of the first millimeters below the surface of the material is sometimes considered a dead region, since defects that may be present there may be masked by ambient noise. Spurious signals may also correspond, for example, to structural noise, which arises from the inhomogeneous microstructure of the material. When the wavelength of the ultrasonic waves emitted by the probe is close to the average diameter of the grains of the material, this noise particularly interferes with the analysis of the obtained values, especially when this noise is higher than the level of the signal reflected by potential defects.

[0006] The present invention seeks to reduce or eliminate the effect of noise in acquired measurement signals.

[0007] A known method for processing this type of ultrasonic measurement signal is the so-called point focusing or total focusing method (TFM). This method generates, for each position of the probe, an image corresponding to a cross-section of the material under the probe. The presence of spurious noise (near-surface waves, structural noise) significantly degrades the quality of these images and therefore impairs the accurate detection of potential defects in the material.

[0008] Without possible post-processing for TFM, noise generated by surface waves and / or by microstructural inhomogeneities can become very significant, making defect detection difficult or, in some cases, even impossible, especially at shallow depths.

[0009] The spurious noise observed in images resulting from TFM exhibits spatial inhomogeneity, which varies substantially according to the point position, and tends to be particularly high in the vicinity of the oscillator / receiver array (the effect of surface waves).

[0010] Document FR-A-3085481 discloses a method for ultrasonically detecting and characterizing defects in dissimilar materials, which provides for post-processing of images obtained, for example by the TFM method, using the following statistics: a measure of the central tendency of the focused amplitude at the level of the probed points on different probe positions, a measure of a function that represents the variation of the focused amplitude at the level of the probed points on different probe positions. Another focusing method taught by document FR-A-3085481 is the plane wave imaging method, called PWI, in which different configurations are distinguished from one another by different delays applied to the transmission of ultrasound waves by the transmitting transducers, by sequentially exciting all transducers.

[0011] However, these statistics used in the method known from document FR-A-3085481 assume the presence of healthy areas (free from defects) in the image, which can therefore be trained in advance to constitute representative or reference areas of the material being inspected. These measurements then normalize the noise and allow for the identification of any amplitude gaps synonymous with defects. The statistics performed on healthy areas correspond to the training phase of the material being investigated. This training phase can only be carried out if a healthy representative area is available and serves as a reference area. If such a healthy representative area does not exist, a step must be carried out before the operation of the method of document FR-A-3085481, thus requiring the organization of further tests that increase the overall duration of the measurement. Summary of the Invention

[0012] One object of the present invention is to provide a method and apparatus for detecting defects in materials by means of ultrasound that overcomes the above-mentioned drawbacks and eliminates the need for such reference areas if they are not available.

[0013] A first subject of the invention is therefore a method for detecting defects in a material by means of ultrasound, said method comprising the following steps: a) continuously transmitting ultrasonic waves to a surface of a material by M transmitting ultrasonic transducers with index i of a multi-element probe, where i is a first natural number from 1 to M, and M is a second predetermined natural number equal to or greater than 2; Sampling time nT e receiving a measurement signal x(n,i,j) representing the amplitude of an ultrasonic wave propagating in a material by M receiving ultrasonic transducers with index j of a multi-element probe, where n is a third natural number from 1 to N, and N is a fourth predetermined natural number equal to or greater than 2; e is a predetermined sampling period, and j is the fifth natural number from 1 to M; b) By computer, for at least one predetermined gap Δ that is positive or zero, N columns Y n forming a sampling matrix having N columns Y for n from 1 to N n is formed by all measurement signals x(n,i,j) and N sampling instants nT e Corresponding to, Each column Y n is the sampling time nT e For the distance between the receiving ultrasonic transducer with index j and the transmitting ultrasonic transducer with index i, we have N columns Y n having all measurement signals x(n,i,j) equal to a given gap Δ that is the same for The sampling matrix is ​​the row X formed by all the measurement signals x(n,i,j). i,j and index i is for each row X i,j and the index j is the same for each row X i,j are identical in the pair i,j in row X i,j Each step is different, c) calculating by a computer a covariance matrix from the sampling matrix, the covariance matrix being a square symmetric matrix of dimensions N×N; d) calculating, by a computer, p eigenvectors and p eigenvalues ​​associated with the eigenvectors for the covariance matrix, where p is a sixth predetermined natural number greater than or equal to 2 and is a predetermined maximum number of calculated eigenvectors and calculated eigenvalues ​​less than or equal to N; e) Using a computer, find row X of the sampling matrix for the K eigenvectors corresponding to the K largest eigenvalues. i,j Projection of X i,j proj,k where K is a selected number less than the maximum number p of computed eigenvectors and computed eigenvalues; f) Using a computer, calculate row X of the sampling matrix. i,j From each of these rows X for the K eigenvectors ij Projection of Xi,j proj,k to obtain the residual defect detection row X* formed by the set of residual defect detection measurement signals x*(n,i,j). i,j The index i is the step of obtaining each residual defect detection row X* i,j and the index j is the same for each residual defect detection row X* i,j and the pair i,j is the residual defect detection row X* i,j Each step is different The present invention is characterized by comprising:

[0014] The present invention attenuates or eliminates spurious signals embedded with possible defects in the material being inspected, which allows these defects to be better detected, and therefore does not require the learning phase taught by document FR-A-3085481.

[0015] Embodiments of the present invention are described below, which are applicable to a method for detecting defects by ultrasound in accordance with the present invention, a device for detecting defects by ultrasound in accordance with the present invention, and a computer program for detecting defects by ultrasound in accordance with the present invention.

[0016] According to one embodiment of the present invention, M is a second predetermined natural number greater than or equal to three.

[0017] According to one embodiment of the present invention, e) Using a computer, calculate row X of the sampling matrix for the K eigenvectors corresponding to the K largest eigenvalues, for k from 1 to K. i,j Projection of X i,j proj,k where K is a selected number less than the maximum number p of computed eigenvectors and computed eigenvalues, and k is a seventh natural integer between 1 and K; f) Using a computer, calculate row X of the sampling matrix. i,j From each of these, for k from 1 to K, i,j K projections of X i,jproj,k to obtain the residual defect detection row X* formed by the set of residual defect detection measurement signals x*(n,i,j). i,j The index i is the step of obtaining each residual defect detection row X* i,j and the index j is the same for each residual defect detection row X* i,j and the pair i,j is the residual defect detection row X* i,j Different steps for each Further includes:

[0018] According to one embodiment of the present invention, the method comprises the following steps: g) performing post-processing by a computer to detect defects in the material from the residual defect detection measurement signal x*(n,i,j); Further includes:

[0019] According to one embodiment of the present invention, post-processing for defect detection includes an algorithm for focusing and generating an image of the residual defect detection measurement signal x*(n,i,j).

[0020] According to one embodiment of the present invention, the focusing algorithm is a focusing algorithm for focusing on all points, the focusing algorithm for focusing on all points comprising: By computer,

[0021]

number

[0022] According to one embodiment of the present invention, the focusing algorithm is a focusing algorithm for focusing on all points, the focusing algorithm for focusing on all points comprising: By computer (CAL),

[0023]

number

[0024] According to one embodiment of the present invention, the method comprises the following steps: h) performing bilateral filtering of the image by a computer. Further includes:

[0025] According to one embodiment of the present invention, By computer, λ K >m+2s and λ K+1 Calculate an integer K such that m+2s (λ k ) 1≦k≦Nrepresents the eigenvalue for the seventh natural number k between 1 and N, and m is the number of N eigenvalues ​​(λ k ) 1≦k≦N is the average of the N eigenvalues ​​(λ k ) 1≦k≦N The standard deviation of the step Further includes:

[0026] According to one embodiment of the present invention, the M transmitting ultrasonic transducers with index i and the M receiving ultrasonic transducers with index j are uniformly distributed over the surface of the material.

[0027] According to one embodiment of the present invention, the M transmitting ultrasonic transducers with index i form part of each of M ultrasonic transmitting-receiving units respectively positioned at M different predetermined positions within the multi-element probe, and the M receiving ultrasonic transducers with index j form part of each of M ultrasonic transmitting-receiving units respectively.

[0028] According to one embodiment of the present invention, M transmitting ultrasonic transducers with index i and M receiving ultrasonic transducers with index j are distributed in a plane.

[0029] According to one embodiment of the present invention, the M emitting ultrasonic transducers with index i and the M receiving ultrasonic transducers with index j are distributed along at least one linear axis.

[0030] According to one embodiment of the present invention, the method further includes the step of calculating, by a computer, the residual defect detection measurement signal x*(n,i,j) for several predetermined gaps Δ that are different from one another.

[0031] According to one embodiment of the present invention, the method further includes a step of calculating, by a computer, a residual defect detection measurement signal x*(n,i,j) for a predetermined gap Δ corresponding to all combinations of M transmitting ultrasonic transducers with index i and M receiving ultrasonic transducers with index j.

[0032] According to one embodiment of the present invention, M transmitting ultrasonic transducers with index i of the multi-element probe are matched with M receiving ultrasonic transducers with index j of the multi-element probe.

[0033] According to another embodiment of the present invention, the M transmitting ultrasonic transducers with index i of the multi-element probe are distinct from the M receiving ultrasonic transducers with index j of the multi-element probe.

[0034] A second subject of the invention is a device for detecting defects in materials by means of ultrasound, said device comprising: a multi-element probe including M transmitting ultrasonic transducers with index i, capable of continuously transmitting ultrasonic waves to a surface of a material, wherein i is a first natural number from 1 to M, and M is a second predetermined natural number equal to or greater than 2; The multi-element probe is designed to measure the sampling time nT e The method includes: receiving M ultrasonic transducers with index j that can receive a measurement signal x(n,i,j) representing the amplitude of an ultrasonic wave propagating in a material; n is a third natural number from 1 to N; N is a fourth predetermined natural number greater than or equal to 2; and T e is a predetermined sampling period, and j is the fifth natural number from 1 to M. Including, The device is For at least one given gap Δ that is positive or zero, n configured to form a sampling matrix having where Y is a set of N columns for n from 1 to N n is formed by all measurement signals x(n,i,j) and N sampling instants nT e Corresponding to, Each column Y n is the sampling time nT e For the distance between the receiving ultrasonic transducer with index j and the transmitting ultrasonic transducer with index i, we have N columns Yn having all measurement signals x(n,i,j) equal to a given gap Δ that is the same for The sampling matrix is ​​the row X formed by all the measurement signals x(n,i,j). i,j and index i is for each row X i,j and the index j is the same for each row X i,j are identical in the pair i,j in row X i,j Each time is different, configured to calculate a covariance matrix from the sampling matrix, where the covariance matrix is ​​a square symmetric matrix of dimensions N×N; configured to compute p eigenvectors and p eigenvalues ​​associated with the eigenvectors for the covariance matrix, where p is a sixth predetermined natural number greater than or equal to 2 and a predetermined maximum number of computed eigenvectors and computed eigenvalues ​​less than or equal to N; Row X of the sampling matrix for the K eigenvectors corresponding to the K largest eigenvalues i,j Projection of X i,j proj,k where K is a selected number less than the maximum number p of computed eigenvectors and computed eigenvalues; Row X of the sampling matrix i,j From each of these rows X for the K eigenvectors ij Projection of X i,j proj,k to obtain the residual defect detection row X* formed by the set of residual defect detection measurement signals x*(n,i,j). i,j where index i is the residual defect detection row X* i,j and the index j is the same for each residual defect detection row X* i,j and the pair i,j is the residual defect detection row X* i,j Each time is different, It is characterized by including a calculator.

[0035] According to one embodiment of the present invention, a computer Row X of the sampling matrix for the K eigenvectors corresponding to the K largest eigenvalues, for k from 1 to K i,j Projection of X i,j proj,k where K is a selected number less than the maximum number p of computed eigenvectors and computed eigenvalues, and k is a seventh natural integer between 1 and K; Row X of the sampling matrix i,j From each of these, for k from 1 to K, i,j K projections of X i,j proj,k to obtain the residual defect detection row X* formed by the set of residual defect detection measurement signals x*(n,i,j). i,j where index i is the residual defect detection row X* i,j and the index j is the same for each residual defect detection row X* i,j and the pair i,j is the residual defect detection row X* i,j Each time is different, Includes a calculator.

[0036] A third subject of the invention is a computer program for detecting defects by ultrasound, comprising code instructions for implementing, when executed by a computer, the method for detecting defects by ultrasound described above. [Brief explanation of the drawings]

[0037] The invention will be better understood on reading the following description, given by way of non-limiting example only, with reference to the following figures of the accompanying drawings, in which: [Figure 1] 1 shows a schematic perspective view of a device for detecting defects by ultrasound according to an embodiment of the present invention; [Figure 2] 1 illustrates a schematic cross-sectional view of a device for detecting defects by ultrasound according to an embodiment of the present invention; [Figure 3] 1 illustrates a schematic cross-sectional view of a device for detecting defects by ultrasound according to an embodiment of the present invention; [Figure 4]1 illustrates a schematic cross-sectional view of a device for detecting defects by ultrasound according to an embodiment of the present invention; [Figure 5] 3 illustrates a flowchart of a method for detecting defects by ultrasound in accordance with one embodiment of the present invention. [Figure 6] 3 illustrates a measurement signal obtained during a method for detecting defects by ultrasound according to an embodiment of the present invention. [Figure 7] 4A-4C represent images acquired without implementing the method and device for detecting defects according to one embodiment of the present invention. [Figure 8] 8 shows an image obtained by implementing a method and device for detecting defects according to an embodiment of the present invention from a measurement signal in FIG. 7. [Figure 9] 3A-3C represent images obtained by implementing a method and device for detecting defects according to an embodiment of the present invention, further including the application of bilateral filtering. [Figure 10] 1 is a graph depicting signal-to-noise ratio as a function of selected parameters of a method for detecting defects and a number of components of a device in accordance with an embodiment of the present invention. [Figure 11] 8 represents an image obtained by a method and device for detecting defects according to an embodiment of the present invention, from the same configuration as in FIG. 7, but with the additional criteria of the selection of the number of components and the bilateral filter. [Figure 12] 11 depicts an image obtained from the configuration used for the graph in FIG. 10 without implementing the method and device for detecting defects according to one embodiment of the present invention. [Figure 13] 10, representing images acquired by a method and device for detecting defects according to an embodiment of the present invention, with the selection of the number of components and the addition of a bilateral filter as criteria, from the configuration used for the graph in FIG. DETAILED DESCRIPTION OF THE INVENTION

[0038] A method for detecting defects by ultrasound, a device for detecting defects by ultrasound, and a computer program for implementing the method are described below with reference to FIGS. 1 to 5. The device 100 for detecting defects includes a multi-element probe 10 including ultrasonic transducers 14, 15, which may be ultrasonic emitters and / or receivers. The steps of the method for detecting defects by ultrasound are described with reference to FIG. 5. In a first step E1, the multi-element probe 10 is positioned at a specific position z on the surface S of the material MS to be inspected. For example, the multi-element probe 10 may include a coupling medium disposed between the surface S of the material MS to be inspected and the ultrasonic transducers 14, 15, allowing propagation of ultrasonic waves between the surface S of the material MS to be inspected and the ultrasonic transducers 14, 15. In one embodiment of the present invention, the coupling medium forms an integral part of the multi-element probe 10 and may be fixed to the ultrasonic transducers 14, 15. For example, the coupling medium may include a gel contained in a container fixed to the ultrasonic transducers 14, 15, in which case the multi-element probe 10 abuts or contacts the surface S of the material MS to be inspected. In another embodiment of the present invention, this coupling medium is not an integral part of the multi-element probe 10, is not fixed to the ultrasonic transducers 14, 15, but can be added between the surface S of the material MS to be inspected and the ultrasonic transducers 14, 15, and this coupling medium can be, for example, the height of water present between the surface S of the material MS to be inspected and the ultrasonic transducers 14, 15 when the material MS and the multi-element probe 10 are submerged in water, in which case the multi-element probe 10 is at a short distance from the surface S of the material MS to be inspected.

[0039] The material MS to be inspected can be any type of material, in particular a coarse-grained material, such as for example coarse-grained steel.

[0040] The multi-element probe 10 includes M transmitting ultrasonic transducers 14, each having an index i (a first natural number) ranging from 1 to M, where M is a second predetermined natural number greater than or equal to 2. In one embodiment, M is a second predetermined natural number greater than or equal to 3. The multi-element probe 10 includes M receiving ultrasonic transducers 15, each having an index j (a fifth natural number) ranging from 1 to M.

[0041] In a second step E2 following the first step E1, the M transmitting ultrasonic transducers 14 of the multi-element probe 10 with index i transmit signals at successive transmission times IE i At each time point IE, an ultrasonic signal S1 is continuously emitted to the surface S of the material MS. i (and at the transmission time IE of each of the next transmitting ultrasonic transducers 14 with index i+1) i+1 In response to each ultrasonic signal S1 emitted by each transmitting ultrasonic transducer 14 of index i (before the second step E2), M receiving ultrasonic transducers 15 of index j receive ultrasonic measurement signals x(n,i,j) representing the amplitude of the ultrasonic waves S2 propagating in the material MS, for j from 1 to M. Each receiving ultrasonic transducer of index j receives an ultrasonic measurement signal x(n,i,j) at a sampling instant nT e (each transmission time IE of the transmitting ultrasonic transducer 14 of index i) i and before each transmission time IEi+1 of the next transmitting ultrasonic transducer 14 with index i+1, these ultrasonic measurement signals x(n,i,j) are received, where n is a third natural number from 1 to N, N is a fourth predetermined natural number equal to or greater than 2, and T e is a predetermined sampling period (the reciprocal of the predetermined sampling frequency). The M transmitting ultrasonic transducers 14 with index i and the M receiving ultrasonic transducers 15 with index j of the multi-element probe 10 can have a first transmitting-receiving configuration described below. e is defined within a certain range relative to the initial time point.

[0042] According to one embodiment of the present invention, each ultrasonic transducer 14, 15 can alternately act as a transmitter or a receiver. Each ultrasonic transducer 14, 15 can be placed in an ultrasonic transmission mode or an ultrasonic reception mode. In this case, M transmitting ultrasonic transducers 14 with index i of the multi-element probe 10 correspond to M receiving ultrasonic transducers 15 with index j of the multi-element probe 10.

[0043] According to another embodiment of the present invention, the M transmitting ultrasonic transducers 14 with index i of the multi-element probe 10 are distinct from the M receiving ultrasonic transducers 15 with index j of the multi-element probe 10 .

[0044] According to one embodiment of the present invention, as shown by way of non-limiting example in Figures 2 and 3, M transmitting ultrasonic transducers 14 with index i and M receiving ultrasonic transducers 15 with index j are uniformly distributed within the multi-element probe 10. For example, as shown by way of non-limiting example in Figures 2 and 3, the M transmitting ultrasonic transducers 14 with index i each form part of M ultrasonic transmitting / receiving units 13 respectively arranged at M different predetermined positions in the multi-element probe 10, and the M receiving ultrasonic transducers 15 with index j each form part of these M ultrasonic transmitting / receiving units 13. Of course, any other distribution of the ultrasonic transducers 14, 15 can be provided.

[0045] According to one embodiment of the present invention, the M transmitting ultrasonic transducers 14 with index i and the M receiving ultrasonic transducers 15 with index j are distributed, for example, in a plane P parallel to the surface S of the material MS and may be located on the surface S of the material MS or at a non-zero distance from the surface S of the material MS, as shown as non-limiting examples in Figures 2 to 4.

[0046] According to one embodiment of the present invention, the M transmitting ultrasonic transducers 14 with index i and the M receiving ultrasonic transducers 15 with index j are distributed along one (or several) linear axes 16, as shown, for example, in Figures 2 and 3. In this case, the probe 10 may be a multi-element strip in which the M transmitting ultrasonic transducers 14 with index i and the M receiving ultrasonic transducers 15 with index j are aligned along this linear axis 16.

[0047] According to one embodiment of the present invention, the plane P in which the M transmitting ultrasonic transducers 14 with index i and the M receiving ultrasonic transducers 15 with index j are located is parallel to the plane of the surface S of the material MS to be inspected.

[0048] According to one embodiment of the present invention, one or some or all of the M transmitting ultrasonic transducers 14 with index i and the M receiving ultrasonic transducers 15 with index j may be at a distance from the surface S of the material MS.

[0049] According to another embodiment of the present invention, the plane P on which the M transmitting ultrasonic transducers 14 with index i and the M receiving ultrasonic transducers 15 with index j are located is inclined with respect to the surface S, and one or some or all of the M transmitting ultrasonic transducers 14 with index i and the M receiving ultrasonic transducers 15 with index j may be at a certain distance from the surface S of the material MS.

[0050] According to another embodiment of the present invention, the plane P on which the M transmitting ultrasonic transducers 14 with index i and the M receiving ultrasonic transducers 15 with index j are located is at a non-zero distance from the surface S and can be parallel to the surface S of the material MS or can be inclined.

[0051] According to one embodiment of the present invention, the M transmitting ultrasonic transducers 14 with index i are distributed according to M predetermined coordinates iL separated from each other by the same non-zero predetermined pitch L relative to each other, as shown by non-limiting examples in Figures 2 and 3 (for example, this coordinate can be an abscissa iL along one (or several) linear axes 16 having the linear pitch L therebetween, or in the case of the M receiving ultrasonic transducers 15 with index j distributed around this other axis having the angular pitch L therebetween, it can be an angle iL around another axis, etc.), and / or the M receiving ultrasonic transducers 15 with index j are distributed according to M predetermined coordinates jL separated from each other by a non-zero predetermined pitch L relative to each other (for example, this coordinate can be an abscissa jL along one (or several) linear axes 16 having the linear pitch L therebetween, or in the case of the M receiving ultrasonic transducers 15 with index j distributed around this other axis having the angular pitch L therebetween, it can be an angle iL around another axis). Of course, the M transmitting ultrasonic transducers 14 with index i may not be distributed according to the same non-zero predetermined pitch L relative to each other, and / or the M receiving ultrasonic transducers 15 with index j may not be distributed according to the same non-zero predetermined pitch L relative to each other.

[0052] For a given position of the probe 10 relative to the surface S of the material MS, a three-dimensional matrix of measurement signals x(n,i,j) is available, where n corresponds to the number of discrete time steps, i is the index of the emitting ultrasonic transducer 14 and j is the index of the receiving ultrasonic transducer 15.

[0053] In a third step E3 following the second step E2, a computer CAL forming part of the defect detection device 100 calculates a number of N columns Y n A sampling matrix A with Δ The sampling matrix A is formed as follows: Δgroups together all of the measurement signals x(n,i,j) acquired for n from 1 to N that have the same distance gap Δ between the receiving ultrasonic transducer 15 with index j and the transmitting ultrasonic transducer 14 with index i. n is formed by the measurement signal x(n,i,j) and is obtained at N sampling instants nT of M receiving ultrasonic transducers 15. e corresponds to. The computer CAL can form part of the probe 10 or be linked or connected to it. As computer CAL it can be provided, for example, as one or several computers, and / or one or several processors, and / or one or several servers, and / or one or several machines, which can be pre-programmed with a pre-recorded computer program for carrying out the method and which can include one or several permanent memories in which this program is pre-recorded. The computer CAL automatically carries out the steps of the defect detection method. A further computer linked to or integrated in the probe 10 can record the obtained values, and this further computer or computer CAL can then be processed by the method according to the invention.

[0054] Each column Y n is formed from the measurement signal x(n,i,j) where the distance d(i,j) between the receiving ultrasonic transducer 15 with index j and the transmitting ultrasonic transducer 14 with index i is equal to a predetermined gap Δ, and the predetermined gap Δ is the distance between the receiving ultrasonic transducer 15 with index j and the transmitting ultrasonic transducer 14 with index i. n (Thus, for a pair (i, j) different from the pair (i', j'), it is equal to the distance d(i', j') between the receiving ultrasonic transducer 15 with index j' and the transmitting ultrasonic transducer 14 with index i'). Each matrix A Δ The gap Δ for is defined in the computer CAL. This gap Δ can be positive or 0. Therefore, we obtain: matrix A ΔFor a pair (i,j) different from the pair (i,j) in matrix A Δ Row X i,j is formed from the measurement signal x(n,i,j) for n from 1 to N, where the index i is the i,j and the index j is the same for each row X i,j are identical in the pair i,j in row X i,j It varies depending on the time.

[0055] Therefore, the computer CAL is the sampling matrix A Δ which is defined by the following equation:

[0056]

number

[0057] For example, the sampling matrix A associated with Δ=5L Δ Each column of Yn In this case, for n from 1 to N, we get: A signal x(n,1,6) transmitted by a transmitting ultrasonic transducer 14 with index i=1 and received by a receiving ultrasonic transducer 15 with index j=6 A signal x(n,2,7) transmitted by a transmitting ultrasonic transducer 14 with index i=2 and received by a receiving ultrasonic transducer 15 with index j=7 A signal x(n,3,8) transmitted by a transmitting ultrasonic transducer 14 with index i=3 and received by a receiving ultrasonic transducer 15 with index j=8 ·and so on In steps E4 to E6, which follow the third step E3, the computer CAL calculates the sampling matrix A Δ A principal component analysis (PCA) algorithm is performed on the

[0058] In a fourth step E4 following the third step E3, the computer CAL calculates the sampling matrix A Δ The covariance matrix C corresponding to Δ Calculate the covariance matrix C Δ The coefficient c of the rth row and qth column of r,q is a natural number between 1 and N, and q is a natural number between 1 and N. r,q =cov(Y r ,Y q ) and Y r ,Y q is the sampling matrix A Δ The columns of the covariance matrix C Δ is the sampling matrix A for n from 1 to N according to the following formula: Δ It is calculated as a function of

[0059]

number

[0060] In a fifth step E5 following the fourth step E4, the computer CAL, for k from 1 to p, for the covariance matrix C Δ corresponding to the sampling matrix A Δ computes p eigenvectors V k and p eigenvalues λ k associated with the eigenvector V k . The number p is a sixth predetermined natural number greater than or equal to 2, and the predetermined maximum number of the computed eigenvectors V k (and the predetermined maximum number of the computed eigenvalues λ k ). For example, p = N. In another example, p ≤ N can be set. In another example, p < N can also be set (eigenvectors fewer than the number N of columns of the covariance matrix C Δ can be computed). When p = N, the eigenvectors V k form a basis, and any row of the matrix A Δ can be decomposed as a linear combination of N eigenvectors V k for k from 1 to N.

[0061] In a sixth step E6 following the fifth step E5, the computer CAL, for K < p, for the covariance matrix C Δ corresponding to the sampling matrix A Δ computes the projection X k of the rows X k of the sampling matrix A Δ onto the K eigenvectors V i,j corresponding to the K largest eigenvalues λ i,j projk for k from 1 to K. The eigenvalues of the decomposition of the covariance matrix into eigenvectors are positive real values because this covariance matrix is square and symmetric. The number K is defined by or determined by the computer CAL, and this number is smaller than the maximum number p of the eigenvectors V k and eigenvalues λ k . To do so, the computer, for the covariance matrix C Δ corresponding to the sampling matrix A ΔThe eigenvalue λ of k and the K largest eigenvalues ​​λ k You can select.

[0062] The measurement signal x(n,i,j) can contain several pieces of information: information related to surface waves, information related to structural noise, and possibly also information related to defects in the material MS, i.e., defects that one wishes to detect. Composite materials MS generate noise due to different phenomena of wave scattering. Furthermore, noise is ubiquitous in the measurement signal x(n,i,j), which does not necessarily mean that defects are present. Even if defects are present, this does not represent a large amount of information for all signals of interest. Therefore, a large eigenvalue λ k The information contained in the K eigenvectors with V is that of the dominant spurious noise (typically surface waves), while the information contained in the other eigenvectors with smaller eigenvalues ​​corresponds to any defects / artifacts present in the signal. k and the eigenvalue λ k The decomposition property is that the K largest eigenvalues ​​λ k K eigenvectors V corresponding to k corresponds to the dominant information, and a small eigenvalue λ k The eigenvector V associated with k corresponds to the information embedded in this dominant information. Therefore, the number K is the sum of the sampling matrix A Δ The covariance matrix C corresponding to Δ is the number of components to retain from the decomposition of into eigenvectors of

[0063] In a seventh step E7 following the sixth step E6, the computer CAL calculates the sampling matrix A Δ Row X i,j From each of these, this sampling matrix A Δ The covariance matrix C corresponding to Δ The K largest eigenvalues ​​λ of k K eigenvectors V corresponding to k For this row X i,j Projection X for k from 1 to Ki,j projk The result of this operation is the sampling matrix A Δ Following the same format as above, the residual defect detection row X* formed by the residual defect detection measurement signal x*(n,i,j) for n from 1 to N (hereinafter referred to as signal x*) i,j and index i is the residual defect detection row X* i,j and the index j is the same for each residual defect detection row X* i,j and the pair i,j is the residual defect detection row X* i,j This removes a certain amount of noise from the signal, thereby carrying over the defect information of the material MS. Therefore, each row vector X i,j From the K largest eigenvalues ​​λ k K eigenvectors V corresponding to k Projection of self onto X i,j proj " is subtracted. Therefore, the number K is the number of removed components. Therefore, the computer CAL calculates the residual defect detection row X* corresponding to the residual defect detection measurement signal x*(n,i,j) for n from 1 to N. i,j Calculate the value of , which is defined by the following formula:

[0064]

number

[0065]

number

[0066] Therefore, X* i,j =X i,j -X i,j proj is obtained.

[0067] By convention, vectors are written in uppercase.

[0068] Thus, the residual defect detection measurement signal x*(n,i,j) is freed from a certain amount of noise to reveal small variations that represent defects in the material MS.

[0069] According to one embodiment of the present invention, the computer CAL calculates the residual defect detection measurement signal x*(n,i,j) for several predetermined gaps Δ that are different from one another.

[0070] According to one embodiment of the present invention, the computer CAL calculates the residual defect detection measurement signal x*(n,i,j) for a given gap Δ corresponding to all combinations of M transmitting ultrasonic transducers 14 with index i and M receiving ultrasonic transducers 15 with index j.

[0071] According to one embodiment of the present invention, K is calculated for each predetermined gap Δ, and thus the sampling matrix A Δ can be made equal for each

[0072] According to another embodiment of the present invention, K is calculated for each predetermined gap Δ, and thus the sampling matrix A Δ It can be different for each.

[0073] According to one embodiment of the present invention, the computer CAL performs post-processing in an eighth step E8 following the seventh step E7 to detect defects in the material MS from the residual defect detection measurement signal x*(n,i,j).

[0074] According to one embodiment of the present invention, the computer CAL executes, in an eighth step E8, a focusing algorithm on the residual defect detection measurement signal x*(n,i,j) as a defect detection post-processing. Of course, the defect detection post-processing of the material MS may be other than a focusing algorithm or may not even exist, and the eighth step E8 is optional.

[0075] An embodiment of this focusing algorithm of the type of Total Point Focusing (TFM) in step E8 is described below.

[0076] According to one embodiment of the present invention, a focusing algorithm is executed in step E8 by the computer CAL on the residual defect detection measurement signal x*(n,i,j), which calculates an image, each pixel of which represents a probed point w of the material MS and to which a focused amplitude for this probed point w is associated.

[0077] As a non-limiting example, during an ultrasound shot at position z of the multi-element probe 10, one (or several) emitting transducers 14 emit an ultrasonic wave S1, which enters the material MS at the level of the surface S of the material MS and then propagates within the material MS before being received by the receiving transducer 15. To illustrate the propagation of the ultrasonic waves S1 and S2 within the material, FIG. 4 represents a first progression T1 constituting a short path of the ultrasonic waves S1 and S2 that are diffracted by the defect DEF in the direction of the receiving transducer 15, and a second progression T2 constituting a longer path of the ultrasonic waves S1 and S2 that are reflected by another surface S' of the material M away from the surface S in the direction of the defect DEF and then meet up with the receiving transducer 15.

[0078] In one approach, the different configurations are distinguished from one another by the oscillator or receiver functions performed by the different transducers 14,15.

[0079] For example, in a first transmit-receive configuration, a first transducer 14 (or a first set of transducers 14) is individually excited with a pulsed electrical signal to transmit ultrasonic waves S1. These ultrasonic waves propagate through the material and are then received by all of the transducers 14, 15 (or a second set of transducers 15). Then, at the same probe position z, another transducer 14 (or another first set of transducers 14) is individually excited with a pulsed electrical signal to transmit ultrasonic waves S1. These ultrasonic waves S1 propagate through the material and are then received by all of the transducers 14, 15 (or another second set of transducers 15). Preferably, each of the transducers 14, 15 transmits ultrasonic waves in at least one probe configuration at position z. Typically, each of the transducers 14, 15 alternately serves as the only transmitting transducer, while all of the transducers 14, 15 receive ultrasonic waves. There are then as many ultrasound shots in the first transmit-receive configuration as there are transducers 14, 15. Of course, this first transmit-receive configuration is not limiting and other transmit-receive configurations of the transducers 14 and 15 can also be provided.

[0080] According to one embodiment of the present invention, the all points focusing algorithm comprises, in step E8:

[0081]

number

[0082] Therefore, t(w,i,j) is Transmitted by the transducer i, reflected by a point of assumed defect DEF at position w, ·Captured by transducer j It can be the travel time of the ultrasonic signal. But t(w,i,j) is also Transmitted by the transducer i, Reflected by the back surface S' of the material MS, reflected by a point of assumed defect DEF located at position w, ·Captured by transducer j It can also be the travel time of the signal.

[0083] These travel times t(w,i,j) are calculated by the computer CAL from the wave speed, which depends on the type of propagation (shear, longitudinal). Mode conversions can also be assumed between the different reflections.

[0084] The index I*(w) is calculated for the cross section of the material MS as a function of the position z of the probe 10, and the computer CAL finally obtains an image I*, which will be referred to in the following as the TFM imaging or image.

[0085] The TFM equation can be generalized in step E8 to the following form:

[0086]

number

[0087] Depending on the choice of g, several variations of the TFM in step E8 are possible. The most common choices of g are: g(x)=x g(x) is different from x g depends on the probed position w and / or the wave propagation velocity ·g can also be the absolute value of the signal or the modulus of the analytical signal.

[0088] The method and device 100 for detecting defects according to the present invention allows a gain in decibels, which facilitates the detection of defects, especially in the vicinity of the surface S of the material MS.

[0089] In fact, Figure 7 represents a TFM image obtained from the signal x(n,i,j) alone, without implementing the method and device for detecting defects 100 according to the present invention. This TFM image in Figure 7 contains a defect near the surface S, but it is difficult to identify due to the phenomenon explained above.

[0090] By applying the method and device 100 for detecting defects according to the present invention, first the parameter K is set to 4 and the TFM image of FIG. 8 is obtained from the above signal x*(n,i,j).

[0091] FIG. 8 highlights a defect approximately 2.5 mm deep that was embedded in the noise that was just present near the surface S in FIG.

[0092] In one embodiment of the present invention, a further step E9 of post-processing the TFM image can be applied by the computer CAL after the eighth step E8. This further step E9 of post-processing the TFM image can comprise bilateral filtering applied to the image. Figure 9 shows an image obtained by applying bilateral filtering to the image in Figure 8. This filtering improves the visual quality of the image and its SNR (signal-to-noise ratio). SNR is a quantitative measure of the quality of such an image. SNR is expressed in decibels (dB) and is calculated, for example, as follows:

[0093]

number

[0094] The example shown in Figures 7 and 8 made it possible to obtain an SNR gain of 18.1 dB (by changing the SNR from 4.7 in Figure 7 to 22.8 in Figure 8). The image in Figure 9 has an SNR of 33.2.

[0095] A summary of the performance for different transducer configurations and different defect depths is given below.

[0096] [Table 1] The graph in FIG. 10 shows the SNR as a function of K for the same configuration. Curve C1 shows the SNR obtained from signal x (without x*). Curve C2 shows the SNR obtained from signal x* without applying bilateral filtering of the final image. Curve C3 shows the SNR obtained from signal x* with applying bilateral filtering of the final image. The SNR of the raw signal represented by curve C1 formed by a horizontal line is −1.0. This means that initially, the maximum intensity of the noise is greater than the maximum intensity of the defect. By applying the method and device 100 for detecting defects according to the present invention, more or fewer components are removed (according to K), and the SNR is clearly improved. The SNR gain can be increased by adding bilateral filtering in further post-processing E9 of the TFM image.

[0097] In another embodiment of the invention, there is no further step E9 of post-processing the TFM image.

[0098] The larger K is, the more components are removed, and removing more information is removing that which is related to possible defects DEF, which is why curves C2 and C3 increase up to a certain threshold and then decrease.

[0099] The optimum value of K is not the same for all configurations, as it depends on the properties of the material MS and therefore on the value obtained.

[0100] According to one embodiment of the present invention, the number K is defined in the calculator CAL.

[0101] According to one embodiment of the present invention, K can be determined by the computer CAL by using one of the methods for automatically selecting the number of components known from the literature from the results obtained during steps E3 to E5. It is therefore ensured that the choice of K is relevant and dependent on the particular case under consideration. In other words, the nature of the material MS to be examined influences the choice of K.

[0102] According to one embodiment of the present invention, the computer CAL calculates the eigenvalues ​​(λ k ) and / or eigenvectors (V k ) and / or covariance matrix C Δ and / or matrix A Δ and / or a method for calculating K from the obtained values ​​x(n,i,j) may be implemented.

[0103] According to one embodiment of the present invention, the computer CAL calculates the covariance matrix C Δ The N eigenvalues ​​(λ k ) 1≦k≦N , we keep only those greater than m+2s, where m is the mean of the N eigenvalues ​​and s is the standard deviation of the N eigenvalues. k ) 1≦k≦N are arranged in descending order, and λ1 ≥ λ2 ≥ ... ≥ λN The integer K held by the computer CAL is λ K >m+2s and λ K+1 ≦m+2s.

[0104] According to one embodiment of the present invention, when p=N, all eigenvalues ​​are accessible, so the mean m and standard deviation s are calculated by the computer CAL as follows:

[0105]

number

[0106] According to another embodiment of the present invention, the computer CAL calculates the covariance matrix C Δ The mean m and standard deviation s of all eigenvalues ​​are calculated directly from the eigenvalues ​​(λ k ) 1≦k≦N and covariance matrix C Δ The property that links the

[0107]

number

[0108] l is the covariance matrix C Δ is raised to a power on the left side, and the eigenvalue (scalar) λ k is an integer greater than or equal to 1 corresponding to the power to be raised on the right-hand side.

[0109] Using this property with l=1, we obtain the following result for the mean m:

[0110]

number

[0111]

number

[0112] Figure 11 shows an image with the same configuration as in Figure 7, using a selection criterion for the number of components K and adding bilateral filtering. The SNR is 35.7.

[0113] FIG. 12 shows a TFM image of the configuration used for the graph in FIG. 10 obtained from the signal x(n,i,j) alone, without implementing the method and device 100 for detecting defects according to the present invention (SNR=−1.0).

[0114] Figure 13 shows the TFM image obtained for the same example as before by using the transformed acquisition values ​​x*(n,i,j) by selecting the number of components K for each Δ and adding bilateral filtering. The method used for selecting K in this example is the m+2s method described above. The SNR is 46.6, which is slightly higher than the best case of the graph shown in Figure 10. This improvement comes from the fact that for the graph in Figure 10, K is the same for all Δ.

[0115] According to one embodiment of the present invention, the calculator CAL recalculates the eigenvectors and the number K of components to be removed for each new inspection of the material MS.

[0116] According to another embodiment of the invention, the computer CAL records the eigenvectors and the number K of components removed from a given material MS so that they can be reused next time for this given material MS, and the calculation time is automatically significantly reduced next time.

[0117] According to another embodiment of the invention, instead of considering all samples x(n,i,j) acquired for n from 1 to N, it may be interesting to limit to a given time domain, for example x([n0,n0+1,...,n1],i,j), where n0 (or n1) is the first (or last) sample considered for each signal x(n,i,j). This allows to focus on the time domain of interest corresponding to the determined defect depth and reduces the calculation time.

[0118] According to one embodiment of the present invention, the defect detection method is performed for several positions z (different from each other) of the multi-element probe 10 on the surface S of the material MS to be inspected.

[0119] Of course, the above embodiments, features, possibilities, variations and examples can be combined with each other or selected independently of each other.

Claims

1. 1. A method for detecting defects in a material by ultrasound, comprising: a) continuously transmitting ultrasonic waves to the surface of the material by M transmitting ultrasonic transducers with index i of a multi-element probe, where i is a first natural number from 1 to M, and M is a second predetermined natural number equal to or greater than 3; Sampling time nT e receiving a measurement signal x(n,i,j) representing the amplitude of the ultrasonic wave propagating in the material by M receiving ultrasonic transducers with index j of the multi-element probe, where n is a third natural number from 1 to N, and N is a fourth predetermined natural number equal to or greater than 2; e is a predetermined sampling period, and j is the fifth natural number from 1 to M; b) By computer, for at least one predetermined gap Δ that is positive or zero, N columns Y n forming a sampling matrix having The N columns Y for n from 1 to N n is formed by all measurement signals x(n,i,j) and N sampling instants nT e Corresponding to, Each column Y n is the sampling time nT e , the distance between the receiving ultrasonic transducer with index j and the transmitting ultrasonic transducer with index i is expressed as the distance between the receiving ultrasonic transducer with index j and the transmitting ultrasonic transducer with index i in the N columns Y n and all measurement signals x(n,i,j) equal to said predetermined gap Δ are identical for The sampling matrix is ​​a matrix of rows X formed by all the measurement signals x(n,i,j). i,j and the index i is for each row X i,j and the index j is the same for each row X i,j are identical in the pair i,j in row X i,j Each step is different, c) calculating, by the computer, a covariance matrix from the sampling matrix, the covariance matrix being a square symmetric matrix of dimensions N×N; d) calculating, by the computer, p eigenvectors and p eigenvalues ​​associated with the eigenvectors for the covariance matrix, where p is a sixth predetermined natural number greater than or equal to 2 and a predetermined maximum number of calculated eigenvectors and calculated eigenvalues ​​less than or equal to N; e) calculating by the computer the row X of the sampling matrix for the K eigenvectors corresponding to the K largest eigenvalues; i,j Projection of X i,j proj,k where K is a selected number less than the maximum number p of computed eigenvectors and computed eigenvalues; f) calculating by the computer the row X of the sampling matrix; i,j From each of the K eigenvectors, ij The projection of X i,j proj,k to obtain the residual defect detection row X* formed by the set of residual defect detection measurement signals x*(n,i,j). i,j and the index i is obtained for each residual defect detection row X* i,j and the index j is the same for each residual defect detection row X* i,j and the pair i,j is the residual defect detection row X* i,j Each step is different, g) performing post-processing by the computer to detect defects in the material from the residual defect detection measurement signal x*(n,i,j); Including, the post-processing for detecting defects includes an algorithm for focusing and generating an image of the residual defect detection measurement signal x*(n,i,j); A method according to claim 1, wherein said algorithm for focusing is a focusing algorithm for focusing on all points.

2. The focusing algorithm for focusing on all points comprises: By the computer, [Equation 1] and calculating an index I*(w) of a probed position w in the cross section of the material such that t(w,i,j) corresponds to the travel time of the signal transmitted by the transmitting ultrasonic transducer with index i, reflected at the probed position w, and received by the receiving ultrasonic transducer with index j, and t(w,i,j) corresponds to the sampling instant nT e where n is calculated between 1 and N; forming, by the computer, the image where the probed location w corresponds to a location of a pixel in the image, the value of the pixel in the image at the location w being equal to the index I*(w); 2. The method of claim 1, comprising:

3. The focusing algorithm is a focusing algorithm for focusing on all points, and the focusing algorithm for focusing on all points comprises: By the computer, [Equation 2] and calculating an index I*(w) of a probed position w in the cross section of the material such that t(w,i,j) corresponds to the travel time of the signal transmitted by the transmitting ultrasonic transducer with index i, reflected at the probed position w, and received by the receiving ultrasonic transducer with index j, and t(w,i,j) corresponds to the sampling instant nT e where n is calculated between 1 and N; generating, by the computer, the image where the probed location w corresponds to a pixel location in the image, the value of the pixel in the image at the location w equals the index I*(w), and g is a predetermined function; 3. The method of claim 2, comprising:

4. h) performing, by said computer, bilateral filtering of said image.

4. The method of claim 2 or 3, further comprising:

5. The computer calculates λ K >m+2s and λ K+1 Calculate an integer K such that m+2s (λ k ) 1≦k≦N represents the eigenvalue for the seventh natural number k between 1 and N, and m is the number of N eigenvalues ​​(λ k ) 1≦k≦N is the average of the N eigenvalues ​​(λ k ) 1≦k≦N The standard deviation of the step 5. The method according to claim 1, further comprising:

6. 6. The method according to claim 1, wherein the M transmitting ultrasonic transducers with index i and the M receiving ultrasonic transducers with index j are uniformly distributed over the surface of the material.

7. 7. The method according to claim 1, wherein the M transmitting ultrasonic transducers with index i form part of each of M ultrasonic transmitting-receiving units respectively arranged at M different predetermined positions in the multi-element probe, and the M receiving ultrasonic transducers with index j form part of each of the M ultrasonic transmitting-receiving units.

8. 8. The method according to claim 1, wherein the M transmitting ultrasonic transducers with index i and the M receiving ultrasonic transducers with index j are distributed in a plane.

9. 9. The method according to claim 1, wherein the M transmitting ultrasonic transducers with index i and the M receiving ultrasonic transducers with index j are distributed along at least one linear axis.

10. 10. The method according to claim 1, further comprising the step of calculating, by the computer, the residual defect detection measurement signal x*(n,i,j) for several predetermined gaps Δ different from one another.

11. 11. The method of claim 10, further comprising the step of calculating, by the computer, the residual defect detection measurement signal x*(n,i,j) for the predetermined gap Δ corresponding to all combinations of the M transmitting ultrasonic transducers with index i and the M receiving ultrasonic transducers with index j.

12. 12. The method according to claim 1, wherein the M transmitting ultrasonic transducers with index i of the multi-element probe correspond to the M receiving ultrasonic transducers with index j of the multi-element probe.

13. 12. The method according to claim 1, wherein the M transmitting ultrasonic transducers of index i of the multi-element probe are distinct from the M receiving ultrasonic transducers of index j of the multi-element probe.

14. 1. A device for detecting defects in materials by means of ultrasound, comprising: a multi-element probe including M transmitting ultrasonic transducers with index i, capable of continuously transmitting ultrasonic waves to a surface of the material, where i is a first natural number from 1 to M, and M is a second predetermined natural number equal to or greater than 3; The multi-element probe is e The method includes: receiving M ultrasonic transducers with index j that can receive a measurement signal x(n,i,j) representing the amplitude of the ultrasonic wave propagating in the material; n is a third natural number from 1 to N; and N is a fourth predetermined natural number equal to or greater than 2; and T e is a predetermined sampling period, and j is the fifth natural number from 1 to M. Including, The device is For at least one given gap Δ that is positive or zero, n configured to form a sampling matrix having where, for n from 1 to N, the N columns Y n is formed by all measurement signals x(n,i,j) and N sampling instants nT e Corresponding to, Each column Y n is the sampling time nT e , the distance between the receiving ultrasonic transducer with index j and the transmitting ultrasonic transducer with index i is expressed as the distance between the receiving ultrasonic transducer with index j and the transmitting ultrasonic transducer with index i in the N columns Y n having all measurement signals x(n,i,j) equal to a given gap Δ that is the same for The sampling matrix is ​​a matrix of rows X formed by all measurement signals x(n,i,j). i,j and the index i is for each row X i,j and the index j is the same for each row X i,j are identical in the pair i,j in row X i,j Each time is different, configured to calculate a covariance matrix from the sampling matrix, wherein the covariance matrix is ​​a square symmetric matrix of dimensions N×N; configured to calculate p eigenvectors and p eigenvalues ​​associated with the eigenvectors for the covariance matrix, where p is a sixth predetermined natural number greater than or equal to 2 and is a predetermined maximum number of calculated eigenvectors and calculated eigenvalues ​​less than or equal to N; The rows X of the sampling matrix for the K eigenvectors corresponding to the K largest eigenvalues i,j Projection of X i,j proj,k where K is a selected number less than the maximum number p of computed eigenvectors and computed eigenvalues; The row X of the sampling matrix i,j From each of the K eigenvectors, ij The projection of X i,j proj,k to obtain the residual defect detection row X* formed by the set of residual defect detection measurement signals x*(n,i,j). i,j where the index i is the residual defect detection row X* i,j and the index j is the same for each residual defect detection row X* i,j and the pair i,j is the residual defect detection row X* i,j Each time is different, configured to perform post-processing to detect defects in the material from the residual defect detection measurement signal x*(n,i,j). Including a calculator, the post-processing for detecting defects includes an algorithm for focusing and generating an image of the residual defect detection measurement signal x*(n,i,j); A device characterized in that the algorithm for focusing is a focusing algorithm for focusing on all points.

15. 15. The device of claim 14, wherein the M transmitting ultrasonic transducers of index i of the multi-element probe coincide with the M receiving ultrasonic transducers of index j of the multi-element probe.

16. 15. The device of claim 14, wherein the M transmitting ultrasonic transducers of index i of the multi-element probe are distinct from the M receiving ultrasonic transducers of index j of the multi-element probe.

17. 1. A computer program for detecting defects by ultrasound, comprising: A computer program comprising code instructions which, when executed by a computer, implements the method for detecting defects by ultrasound according to any one of claims 1 to 13.

Images