Non-destructive testing method for a plurality of elements
The method addresses the inefficiencies of existing non-destructive testing by using dynamic time deformation and statistical analysis to reliably detect defects in multiple elements, enhancing detection accuracy and reducing human error.
Patent Information
- Application Number
- FR2022011287
- Authority / Receiving Office
- FR · FR
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2022-10-28
- Publication Date
- 2024-11-01
- Estimated Expiration
- 2042-10-28
AI Technical Summary
Existing non-destructive testing methods for elements are complex, require long development periods, and are not suitable for detecting unanticipated defects or inspecting elements other than those designed for, especially when the number of elements to inspect is large and safety is a concern.
A method using dynamic time deformation to calculate distances between signal portions from measurement signals, followed by statistical analysis of these distances to determine the health of elements, allowing for reliable detection of defects across multiple similar elements.
The method enables efficient and reliable detection of defects in a large number of similar elements, reducing human error and improving detection accuracy by using statistical indicators and dynamic time warping to analyze signal variations.
Smart Images

Figure 00000014_0000 
Figure 00000015_0000 
Figure 00000015_0001
Abstract
Description
Title of the invention: Method for non-destructive testing of a plurality of elements Technical field
[0001] The invention belongs to the field of non-destructive testing of a plurality of elements, in order to determine whether the element is healthy or unhealthy. More specifically, the invention applies to the testing of similar elements from measurement signals, each of the measurement signals corresponding to a different determined element. Technological background
[0002] Non-destructive testing methods are based on the search for anomalies in measurement signals, likely to correspond to a defect in an unhealthy element. The measurement signals can be of different natures, and for example are signals resulting from ultrasonic acquisitions, or electrical responses to stimulations such as the generation of eddy currents, the detection of radio waves, or even images or videos.
[0003] In these measurement signals, the search for anomalies can be made difficult. Indeed, multiple factors are likely to alter the measurement signals, such as for example a complex geometry of the inspected element or a lack of cleanliness thereof, and the presence of an anomaly can be hidden among these alterations. Furthermore, if the detection must be carried out by a human operator analyzing the measurement signals, the control can be very long and tedious, and its quality depends on the skill of this operator. These problems can become very troublesome in fields where the number of elements to be inspected is large and / or when it is essential to be able to detect all the defects for safety reasons.
[0004] Various methods have been proposed for searching for anomalies that may correspond to defects in measurement signals. Most of these methods are based on comparing values derived from the signals with references. However, each method is then specific to a particular application, and the sensitivity of the method depends greatly on the relevance of the calculation of the derived values and the definition of the reference used. These methods therefore generally require a long development time and may not be suitable for detecting unanticipated defects or for inspecting elements other than those for which they were designed.
[0005] For example, patent application FR3029288 proposes an inspection of tank bottom penetrations; for each time sample, statistical values such as the mean or standard deviation of the values taken by the signal at that instant are calculated on the P available signals. These statistics are then used to construct a temporally homogenized signal, with, for each time sample, a constant mean equal to zero and a constant standard deviation equal to 1. If the application of this method to tank bottom penetrations gives satisfactory results, it is not as effective when applied to the inspection of other elements such as partitioning screws. Indeed, the defects do not systematically generate increases in amplitudes, but often variations in the shape of the signals. On the other hand, the signals are not exactly synchronized with each other, local expansions / contractions / shifts may appear. Presentation of the invention
[0006] The invention therefore aims to propose a method for non-destructive testing of a plurality of elements from measurement signals, which is simple to implement, applicable to the testing of a large number of similar elements, and which makes it possible to reliably detect defects.
[0007] A method is proposed for non-destructive testing of a plurality of elements from measurement signals, each of the measurement signals corresponding to a determined different element, comprising the following steps:
[0008] - obtaining a sequence of at least one portion of signal from each signal of measure,
[0009] - for each pair associating two portions of signal presenting the same position in their respective sequence, a distance between these signal portions of said pair is calculated, said distance being calculated using dynamic time warping,
[0010] - for each portion of signal, determination of at least one statistical indicator to from the distances of pairs involving said portion of signal, the statistical indicator being associated with the element of said portion of signal,
[0011] - for each element, comparison of at least one indicator of said element with indicators of other elements, and based on the comparison, determination of the healthy or unhealthy nature of the element.
[0012] Dynamic time deformation is applied here in an original way, since a statistical indicator is then used from the distances of couples in order to detect the portions of the signal that are distinguished from the others.
[0013] This method is advantageously supplemented by the following characteristics, taken alone or in any of their technically possible combinations: - two portions of signal which follow one another in a sequence obtained from the same measurement signal partially overlap, - the plurality of elements comprises P elements, P natural integer greater than 10, and each portion of signal is involved in Pl couples - a statistical indicator is an indicator of central tendency, such as the mean or the median, - a statistical indicator is a dispersion indicator that measures the variability of distances, such as the standard deviation or the variance, - at least two statistical indicators are determined, including a central tendency indicator and a dispersion indicator, - the comparison of at least one indicator of said element with indicators of other elements comprises a determination of a difference between said statistical indicator and a closer other statistical indicator of another element, - at least two statistical indicators are determined, and the comparison of the pair of indicators of one element with pairs of indicators of other elements includes a statistical modeling of a probability density of the values taken by the pair of two statistical indicators of said element, - each portion of signal corresponds to a measurement distance at least equivalent to a typical size of defects sought.
[0014] The invention also relates to a computer program product comprising program code instructions for executing the steps of the method according to the invention, when said program is executed on a computer. The computer program product may take the form of a non-volatile medium on which the instructions are stored. Presentation of figures
[0015] Other characteristics, aims and advantages of the invention will emerge from the following description, which is purely illustrative and non-limiting, and which must be read in conjunction with the appended drawings in which:
[0016] - [Fig.l] is a diagram showing schematically the steps of implementation of a method according to a possible embodiment of the invention;
[0017] - [Fig.2] shows an example of alignment by dynamic time warping between two portions of healthy element signal;
[0018] - [Fig.3] shows an example of alignment by dynamic time warping between a portion of signal from a healthy element and a portion of signal from an unhealthy element;
[0019] - [Fig.4] shows an example of a histogram of couple distances for a healthy element;
[0020] - [Fig.5] shows an example of a histogram of couple distances for a unhealthy element;
[0021] - [Fig.6] shows an example of distribution of values for two sta indicators tistics. Detailed description
[0022] The invention applies to the non-destructive testing of a plurality of similar elements, a majority of which are considered to be sound. By unsound is meant an element having a defect, for example mechanical (crack, corrosion, etc.), which potentially makes it unfit to perform its function. For example, a fastening member such as a screw is no longer sound if cracks are detected in its body. The elements can be of various types, provided that measurement signals are available, each corresponding to a different determined element. If the method can be implemented with a limited number of elements, such as for example at least 10 elements, it is more reliable and more advantageous to implement it on a large number of elements. Consequently, the plurality of elements preferably comprises at least 50 elements, and more preferably at least 100 elements.
[0023] The measurement signals may be of various natures, since the measurement signals are likely to present variations linked to the healthy or unhealthy nature of an inspected element, and in particular variations depending on the presence of mechanical defects on the inspected elements. For example, the measurement signals may result from ultrasonic acquisitions, or electrical responses to stimulations such as the generation of eddy currents, the detection of radio waves, or even images or videos. By way of non-limiting example, the method is described in an application to the control of mechanical fastening elements such as partition screws, and with measurement signals resulting from ultrasonic acquisitions. Even in this application, the measurement signals could be different.It should be noted that the method may comprise the acquisition of the measurement signals by means of a probe or a sensor near each of the elements, for example successively from one element to another.
[0024] The method aims to determine the healthy or unhealthy nature of each element. It is then possible to continue the method with a more in-depth inspection of the elements considered unhealthy.
[0025] The method is implemented on a set of measurement signals, each of the measurement signals corresponding to a determined different element. For example, each measurement signal can be acquired specifically at a respective element, typically by placing a probe close to the element to be inspected: the measurement signal is then naturally associated with a particular element. In our example, an ultrasonic probe is placed against a partition screw, then emits and receives ultrasonic waves from this position against the partition screw. It is It is also possible that the initial acquisition includes measurement signals from several elements. This is the case, for example, when a probe is moved along a trajectory where it successively encounters several elements to be inspected. A breakdown of the acquisition result makes it possible to recover measurement signals, each corresponding to a different specific element.
[0026] Preferably, the measurement signals are of similar sizes, in temporal and / or spatial terms, which facilitates their common manipulation in the rest of the method.
[0027] A sequence of at least one signal portion is obtained from each measurement signal (step S1). Preferably, several signal portions forming a sequence are extracted from each measurement signal. If the measurement signal comes from the propagation of a wave, for example ultrasonic, or from the movement of a probe, each signal portion is a time window corresponding to a distance. Thus, each signal portion can correspond to a measurement distance, which is chosen to be equivalent to at least one typical size of defects sought, and more advantageously to approximately twice a typical size of defects sought. Typically, the defects sought are of the order of millimeters (cracks, cracks, etc.), so that each signal portion corresponds to a measurement distance of between 1 mm and 4 mm. In the example which follows, each signal portion corresponds to a measurement distance which is a depth of approximately 2 mm.
[0028] Preferably, at least 2 signal portions are extracted from the same measurement signal, and preferably at least 4 signal portions are extracted from the same measurement signal. The number of signal portions, however, obviously depends on the size of the measurement signal and the size of the defects sought. It should be noted that if the measurement signal is sufficiently small, it may only be necessary to extract one signal portion, and in this case the sequence of signal portions may comprise only one signal portion. Preferably, the signal portions of two different elements, having the same position in their respective sequence, are of a comparable size, for example in terms of measurement depth. Thus, the extraction of signal portions can also be used to ensure that the comparable signal portions relate to similar areas of the inspected elements, and therefore that these signal portions of different elements are comparable.
[0029] Therefore, even when the measurement signals are small in size so that only one signal portion is extracted, the extraction can serve at least to ensure that signal portions of different measurement signals are similar (not necessarily identical) in terms of physical representativeness of the measurement (e.g., measurement or penetration distance). If the measurement signals are already similar to each other, obtaining the sequence of at least one signal portion can simply consist of making the measurement signal available.
[0030] Preferably, two successive signal portions in a sequence of the same measurement signal partially overlap. In other words, some data of the measurement signal are found in these two successive signal portions, or part of a signal portion is also in another signal portion. The overlap with another signal portion is preferably between 0.1% and 25% of a signal portion, and preferably between 0.5% and 10% of a signal portion.
[0031] The signal portions form a sequence in the sense that they are ordered among themselves: there is a first signal portion, followed by a second signal portion, followed by a third signal portion, etc. We can denote by ai>n the nth signal portion of the ith measurement signal. It should be noted that not all the portions obtained from a measurement signal are necessarily part of the sequence of measurement portions, but that two signal portions of the sequence retain the same order as in the measurement signal: the first signal portion is in the measurement signal before the second signal portion, the third signal portion is in the measurement signal before the second signal portion, etc.
[0032] Then, for each pair associating two signal portions having the same position in their respective sequence, a distance between these signal portions of said pair is calculated: let dtwn(ai>n, aj>n) be the distance between the nth portion of the ith measurement signal and the nth portion of the jth measurement signal. As mentioned above, the portions obtained from a measurement signal are not necessarily part of the sequence of measurement portions, and for example the sequence of measurement portions may only start from a certain extracted portion, the preceding signal portions then not being used in the comparison and therefore not being part of the sequence of measurement portions. Similarly, it is possible that the signal portions of the sequence are not immediately consecutive in the measurement signal.However, it is constant that the order of the signal portions is respected: the (n+l)th portion of the ith signal is compared with a measurement portion of the jth measurement signal which is, in the measurement signal, after the portion of the jth measurement signal to which the nth portion of the ith measurement signal is compared.
[0033] Since each measurement signal is associated with a particular element, we can also define this distance dtwn(ai>n, aj>n) as the distance between the nth portion of the ith element and the nth portion of the jth element. The distance is understood here in the sense of a criterion of similarity or rather of dissimilarity (the greater the distance, the greater the dissimilarity). This is in particular the distance taken into account in the dynamic time warping.
[0034] For this purpose, the distance is calculated using dynamic time warping, or DTW, which is a method already used in other applications, allowing to measure the similarity between two sequences which can vary. In general, DTW is a method which seeks an optimal matching between two time series, under certain restrictions. The time series are deformed by non-linear transformation of the time variable, to determine a measure of their similarity, independently of certain non-linear transformations of time.
[0035] DTW defines a distance that allows the similarity between two portions Si and s2 to be measured in a more relevant way than by comparing the signals point by point, as is the case for example with the Euclidean distance. With DTW, a point of Si is associated with one or more points of s2 (and vice versa), based on a minimization of a cost function. This makes the measure particularly suitable for the analysis of time series because it is robust to expansions, compressions and time shifts between the two signal portions. [Fig.2] shows an example of alignment by dynamic time warping between two signal portions of healthy elements. The pairings are illustrated by the lines connecting the points of the curves corresponding to the two signal portions.Although most of the features are globally vertical, approaching a Euclidean distance between synchronous points, some matching features are strongly inclined, showing the influence of the transformations undergone which, thanks to the DTW, do not prevent highlighting a great similarity.
[0036] [Fig. 3] shows an example of alignment by dynamic time warping between a signal portion of a healthy element and a signal portion of an unhealthy element. Compared to [Fig. 2], we see greater inclinations of the matching lines, and therefore greater lengths, i.e. greater matching distances, indicating a greater warping to be applied to match the signal portions. In addition, we see that certain points of one signal portion are matched with points of the other signal arranged at various locations of this other signal. For example, several peaks of the signal portion at the top of [Fig. 3] are matched with the same point at the end of the signal portion at the bottom of [Fig. 3].Even after alignment, the result will be a final distance, a measure of dissimilarity obtained by calculating the differences between the amplitudes of the points after matching, which will remain significant, since not all of these peaks can be aligned with the same point on the other portion of the signal.
[0037] It is however difficult to formulate hypotheses on the quality of the alignment by DTW between two portions of signal of unhealthy elements. It may happen that two portions of signal of unhealthy elements have similarities, and in this case the DTW measurement may be of the same order of magnitude as the DTW between two respective portions of signal of two healthy elements. It may also happen that the alignment between two respective signal portions of unhealthy elements is very complicated, and in this case the DTW measurement could be of the same order of magnitude as the DTW between a signal portion of a healthy element and a signal portion of an unhealthy element. The preliminary step of dividing the signal into a series of signal portions makes it possible to have local and comparable orders of magnitude of DTW. Indeed, if the DTW measurement were carried out on the entire measurement signal, the impact of a deformation at a given point in time would be too discrete to be detected.
[0038] The DTW is applied to each pair associating two signal portions having the same position in their respective sequence, so that if the plurality of elements comprises P elements, P being a natural integer greater than 10, each signal portion is involved in Pl pairs and this would result in Pl distances involving this portion of the signal and therefore a particular element. Preferably, the distances are calculated for several portions of the signal, and preferably for all the sequences of signal portions.
[0039] This approach can be translated by means of a matrix Mn of dimension P x P with P the number of elements and therefore of measurement signals. The components Mn^ of the matrix are: Mn;j= dtwn(ai>n, aj>n), with dtwn(ai>n, aj>n) the distance between the nth portion portion of the jth element, noted dtwn(a;, aj) for more dtw n (has b «2) ' ' ' dtw^a^ a P ) dtw n (a2, æ2) ... dtw^a2, a p ) dtw n (has P ,a2) ... dtw^ap, a P ) /
[0041] This matrix Mn is symmetrical, with zero diagonal, and groups together all the possible couple distances for the same position in the sequences obtained from the measurement signals. It should be noted that the matrix notation is not essential, but it allows the distances to be organized. Thus, column j of the matrix Mn groups together the distance dtwn(aj, aj) between aj the nth signal portion of element j and itself, which is 0, and the set of distances dtwn(ak, aj) between aj the nth signal portion of element j and the Pl other nth portions ak of the other elements, k^j.
[0042] However, having these torque distances does not allow one to immediately conclude whether an element is healthy or unhealthy, since there is no reference to compare them to. It is therefore preferable to adopt a statistical approach aimed at identifying elements having signal portions whose characteristics deviate from the others. As an illustration, in application of the inspection of partition screws, [Fig.4] shows an example of a histogram of torque distances for a healthy element, while [Fig.5] shows an example of the ieme elerr simplicity: jyeme
[0040] M n dtWflm dtw^ \dtw n (has N , ad of a histogram of couple distances for an unhealthy element. On the abscissa are the couple distances for this element, and on the ordinate the number of distances in classes of distance values. Of course, the values indicated are only indicative, these depending on many parameters such as the amplitudes of the measurement signals. We note that the histogram of couple distances for an unhealthy element is generally translated and dilated compared to the histogram of couple distances for a healthy element. Thus, a statistical approach can make it possible to identify such deviations.
[0043] Consequently, for each signal portion, at least one statistical indicator is determined from the distances of pairs involving said signal portion (step S03), the statistical indicator being associated with the element of said signal portion. Using the matrix notation, the distances of pairs involving an element], and therefore involving the signal portion aj, correspond to column j of the matrix Mn.
[0044] Preferably, a statistical indicator is a central tendency indicator, such as the mean or the median, or a statistical indicator is a dispersion indicator measuring the variability of the distances, such as the standard deviation or the variance. Preferably, at least two statistical indicators are determined, comprising a central tendency indicator and a dispersion indicator. Typically, the statistical indicators are the mean p and the standard deviation o. It should be noted that the statistical indicator(s) can be determined directly from the pair distances, or from a histogram such as that illustrated in Figures 4 and 5.
[0045] For each element, a comparison is then made of at least one indicator of said element with indicators of other elements (step S04). Preferably, at least two statistical indicators are determined, and the comparison is carried out by pairs of indicators, a pair of indicators of an element being compared with pairs of indicators of other elements. Depending on the comparison, the healthy or unhealthy nature of an element is determined (step S05), preferably on the basis of a deviation of the statistical indicators compared to the others.
[0046] The comparison aims to identify the elements for which the statistical indicator(s) are far from those of the healthy elements, assumed to be in the majority. Several approaches can be adopted. The comparison of the at least one indicator of said element with indicators of other elements can include a determination of a difference between said statistical indicator and a closest other statistical indicator of another element, and even between said statistical indicator and several closest other statistical indicators of other elements. For example, the “K nearest neighbors” method consists, for each element, in calculating the Euclidean distance to its K-th nearest neighbor. A threshold is then chosen on this distance to decide whether the element is potentially unhealthy. Other variants can be used, such as the local outlier factor method.
[0047] [Fig.6] shows an example of a two-dimensional spatial representation of pairs {mean, standard deviation] for distances between partition screws. The ordinates show the means and the abscissas show the standard deviations. By applying the "K nearest neighbors" method, elements whose statistical indicators deviated from the majority were identified, and appear in black on [Fig.6], while the majority of elements whose statistical indicators were close to each other appear in white, and are therefore considered healthy.
[0048] First of all, we note that healthy elements have means and standard deviations higher than those of unhealthy elements, and that some unhealthy elements have means much higher than those of healthy elements, notably higher than 70. Thus, a simple comparison of values with a fixed threshold would make it possible to detect certain unhealthy elements, but would also determine a healthy character for many unhealthy elements. Such an approach is therefore not entirely satisfactory. On the contrary, identifying the elements whose indicators deviate from those of the majority of the others makes it possible to detect potentially unhealthy elements even when these have means and standard deviations lower than the means and standard deviations of healthy elements.
[0049] Alternatively, the comparison of the pair of indicators of an element with pairs of indicators of other elements comprises a statistical modeling of a probability density of the values taken from the pair of two statistical indicators of said element. Parametric methods (Gaussian, mixture of Gaussians) or non-parametric methods, such as a kernel method, can be used, for example.
[0050] It should be noted that there is at least one statistical indicator per signal portion, so that the comparison can be made between several statistical indicators of several signal portions of the same element. It is then, for example, possible to determine an unhealthy character if it appears that the statistical indicator differs from the others for a single position in the sequence, or on the contrary, require that the statistical indicator must differ from the others in several positions in the sequence for the element to be considered healthy. In order not to miss a potentially defective element, it is preferable to consider as unhealthy an element whose statistical indicator(s) differ from the others for at least one position in the sequence of signal portions.
[0051] Once the healthy or unhealthy nature of each of the elements has been determined, it is possible to implement a thorough inspection of the unhealthy elements, or directly a physical intervention on them, such as their replacement or their scrapping. In this regard, it should be noted that the method is configured to be able to detect all unhealthy elements, and that it is less harmful to consider a healthy element as unhealthy rather than taking the risk of considering an element that is not healthy as healthy.
[0052] The invention is not limited to the embodiment described and shown in the attached figures. Modifications remain possible, in particular from the point of view of the constitution of the various technical characteristics or by substitution of technical equivalents, without departing from the scope of protection of the invention.
Claims
Claims
1. Method for non-destructive testing of a plurality of elements from measurement signals, each of the measurement signals corresponding to a determined different element, comprising the following steps: - obtaining (SOI) a sequence of at least one signal portion from each measurement signal, - for each pair associating two signal portions having the same position in their respective sequence, a distance (S02) between these signal portions of said pair is calculated, said distance being calculated using dynamic time warping, - for each signal portion, determining (S03) at least one statistical indicator from the distances of pairs involving said signal portion, the statistical indicator being associated with the element of said signal portion, - for each element, comparing (S04) at least one indicator of said element with indicators of other elements, and depending on the comparison,determination (S05) of a healthy or unhealthy character of the element.,
2. A method according to claim 1, wherein two signal portions which follow one another in a sequence obtained from the same measurement signal partially overlap.
3. A method according to any preceding claim, wherein the plurality of elements comprises P elements, P being a natural integer greater than 10, and each signal portion is involved in Pl pairs.
4. A method according to any preceding claim, wherein a statistical indicator is an indicator of central tendency, such as the mean or the median.
5. A method according to any preceding claim, wherein a statistical indicator is a dispersion indicator measuring the variability of distances, such as standard deviation or variance.
6. A method according to any preceding claim, wherein at least two statistical indicators are determined, comprising a central tendency indicator and a dispersion indicator.
7. A method according to any preceding claim, wherein comparing at least one indicator of said element with indicators of other elements includes a determination of a difference between said statistical indicator and a closer other statistical indicator of another element.
8. Method according to any one of claims 1 to 7, in which at least two statistical indicators are determined, and the comparison of the pair of indicators of one element with pairs of indicators of other elements comprises a statistical modeling of a probability density of the values taken by the pair of two statistical indicators of said element.
9. Method according to any one of the preceding claims, in which each signal portion corresponds to a measurement distance at least equivalent to a typical size of defects sought.
10. A computer program product comprising program code instructions for carrying out the steps of the method according to any one of the preceding claims, when said program is executed on a computer.
