FR3003951A1 - DEFECT DETECTION SYSTEM IN AN AIRCRAFT ENGINE ROTATING ELEMENT - Google Patents

Abstract

The invention relates to a system for automatically detecting defects on at least one rotating element of an aircraft engine, comprising: embedded heating means (9) for heating said rotating element (3) of the engine (5) by a thermal stress, - on-board thermographic means (11) for acquiring at least one infrared image (23) translating a transient phase thermal field of said rotating element, and - processing means (15) for calculating differentials relating to a component of the thermal field between different subdivisions of said image to detect variations of said component of the thermal field indicative of defects on said rotating element.

Description

The present invention relates to the field of rotary element monitoring systems of an aircraft engine and more particularly to the detection of defects or anomalies. BACKGROUND OF THE INVENTION in a rotating element of the blade type or blade of a bladed wheel. STATE OF THE PRIOR ART An aircraft engine has several bladed wheels which are potentially exposed to heavy stresses that may in the worst case result in the loss of a blade. More particularly, in a mass reduction objective, the composite materials progressively replace the metal in the manufacture of the engines. Nevertheless, the rotating elements in composite materials can undergo fatigue which manifests itself for example by delamination. These defects are not necessarily on the surface and are therefore difficult to detect. Currently, during production tests or inspections of the blades of an engine, we apply different techniques of non-destructive testing based on the use of thermal imaging cameras. These techniques involve using a mobile heat emitter to heat the blade and a moving thermal camera to take an infrared image of the blade. Image analysis can detect defects in the blade.

In particular, these techniques are not very suitable for a propeller or fan "not careened" also called "Open Rotor". Indeed, the fan / propeller is fixed on the power turbine outside the nacelle and has very large blades. Thus, for an inspection of the blades in situ, it is necessary to use forklifts. These inspection operations are complex, costly and lead to a more or less long immobilization of the aircraft and therefore are not necessarily performed on each flight. The object of the present invention is therefore to provide a detection system that is simple to implement and capable of accurately and reliably detecting defects on a rotating element of an engine without having the aforementioned drawbacks. PRESENTATION OF THE INVENTION The present invention is defined by an automatic system for detecting defects on at least one rotating element of an aircraft engine, comprising: embedded heating means for heating said rotating element of the engine by a thermal stress on-board thermographic means for acquiring at least one infrared image reflecting a transient phase thermal field of said rotating element, and processing means for calculating differentials relating to a component of the thermal field between different subdivisions of said image in order to detect variations of said component of the thermal field indicative of defects on said rotating element. Thus, one can monitor each flight and automatically rotating parts of the engine to detect the first signs of fatigue. This allows for predictive maintenance and not just preventative maintenance as rotational elements can be changed when they really experience damage, increasing profitability (fewer parts changed) and safety (less risk of loss). blade). The analysis is carried out according to differential measurements that make it possible to get rid of the context. In particular, the fact of making comparisons between spatially close zones makes it possible to avoid problems due to the distance of the source of heat or the illumination by the sun. Advantageously, when the differential corresponding to a current subdivision is indicative of anomaly, the processing means are configured to calculate other differentials by rearranging the subdivisions and / or refining the current comparison subdivision in order to locate the locations of the defects. This makes it possible to reduce the number of subdivisions to be studied and consequently to reduce the calculation time and the solicitation of a calculator. Advantageously, the processing means are configured to record at each flight said differentials relating to the thermal fields of the different subdivisions and to analyze the evolution of said flight differentials in flight. This makes it possible to consolidate the result of the detection and to systematically follow the health of the rotating elements of flight in flight. Advantageously, the detection system comprises a database of degradative signatures representative of different forms of degradations and their progress, and the processing means are configured to compare the differentials relative to the thermal fields of the subdivisions exhibiting audit defects. damage signatures. This makes it possible to determine the most probable type of defect.

According to an advantageous embodiment of the present invention, the heating means consist of at least one anti-icing heating element already existing in the engine. This reduces the onboard weight and also allows monitoring the heating means itself.

According to a first embodiment, the heating means are intended to heat said element by thermal pulsations. Thus, the rotating element can be heated in a sufficiently short time so that the material of the rotating element does not reach a constant temperature. According to the first embodiment, the processing means are configured to calculate differentials between an amplitude of the thermal field of a current subdivision and amplitudes of the thermal fields of the neighboring subdivisions. According to a second embodiment, the heating means are intended to heat said element by periodic thermal waves. According to the second embodiment, the processing means are configured to calculate phase shifts between the thermal field of a current subdivision and the thermal fields of the neighboring subdivisions.

Detection according to the phase shift has the advantage of being little influenced by the distance of the heat source or the illumination of the sun, because the temperature is not measured but the phase shift. Advantageously, the rotating element is a blade of a bladed wheel of said motor and the system further comprises identification means for identifying the blade having defects. BRIEF DESCRIPTION OF THE DRAWINGS Other characteristics and advantages of the invention will appear on reading preferred embodiments of the invention with reference to the appended figures in which: FIG. 1 schematically illustrates a system for detecting failure of at least one rotating element of an aircraft engine, according to the invention; Fig. 2 schematically illustrates the effects of the context on a rotating element of an aircraft engine; Fig. 3 schematically illustrates an example of a detection system comprising several heating and thermographic means, according to the invention; Fig. 4 is an algorithm illustrating various data processing steps implemented by acquisition and processing means, according to a preferred embodiment of the invention; Figs. 5A-5E are grids of an image schematically illustrating the steps of the flowchart of FIG. 4; Figs. 6A-6D illustrate the detection of point and progressive defects on different types of grids; Fig. 7 is a detection algorithm comprising a confirmation phase according to an embodiment of the invention; Fig. 8 is a detection algorithm comprising a confirmation phase according to another embodiment of the invention; Fig. 9 is a block diagram illustrating a method for detecting failure of a bladed wheel of the motor, according to the invention; Fig. 10 schematically illustrates a first-of-failure detection system of a bladed wheel of the engine according to a first embodiment of the invention; Fig. 11 schematically illustrates a system for detecting failure of a bladed wheel of the engine according to a second embodiment of the invention; and FIG. 12 schematically illustrates a system for detecting failure of a bladed wheel of the motor according to a third embodiment of the invention. DETAILED DESCRIPTION OF PARTICULAR EMBODIMENTS The concept underlying the invention is based on the use of an on-board detection system performing differential thermal measurements on a rotating element visible from the outside, enabling automatic and precise detection of defects. or anomalies on the rotating element. Fig. 1 schematically illustrates a fault detection system 1 on at least one rotating element 3 of an aircraft engine 5, according to the invention.

The rotating element 3 is visible from the outside and corresponds for example to a blade or blade of a bladed wheel 7 or to a revolving hood of the engine 5. The rotating hood or the bladed wheel 7 can belong to a compressor of the engine 5 and may for example correspond to a streamlined or non-ducted fan of the aircraft engine. It will be noted that the rotating elements may be metallic or composite. In one case, the dreaded defects are break primers and cracks, in the other case they are delamination not always visible from the outside. The detection system 1 comprises on-board heating means 9, embedded thermographic means 11, acquisition means 13 and data processing means 15.

The heating means 9 are intended to heat at least one rotating element 3 of the engine 5 by a thermal load 19a. Of course, the heat enters the material of the rotating element 3. Thus, the local temperatures will vary from one region to another because the heat will more or less penetrate depending on the presence or absence of defects. For example, the heating means 9 may consist of one or more thermal transmitter (s) fixed (s) on the engine 5 or the aircraft 21 with respect to the rotating element 3. Thus, each heat emitter 9 remains permanently on the aircraft 21 and can be set to heat the rotating member 3 periodically or pulsationally. Moreover, the thermographic means 11 are intended to acquire at least one infrared image 23 of the rotating element 3 reflecting, following the thermal bias 19a of the heating, a thermal field in transient phase. It will be noted that the thermographic means 11 may consist of one or more thermal camera (s) fixed (s) on the engine 5 or the aircraft 21 with respect to the rotating element 3.

Advantageously, the rotating elements 3 are filmed during a rotation at very low speed (that is to say, at the beginning of the starting phase, at the end of the stopping phase or during windmill operation milling "on the ground). This makes it possible to have a complete view of the rotating elements 3 without disturbing the acquisition of the images 23. The advantage of filming at low speed is the use of a single camera to detect defects on a whole set of blades, one after the other. In addition, by filming at low speed, one can also advantageously monitor two sets of blades of a bladed wheel double helix. One can of course film while stopped but in this case, it is necessary a plurality of cameras to detect the defects on all the blades. It should be noted that the heating and the acquisition of the images have the advantage of being able to be done without contact, which allows on the one hand not to damage the tested material and, on the other hand, not to have sensors to put on the blades 3 or very close to the blades 3, which could disturb their aerodynamics. In addition, the fact that the heating means 9 and thermographic 11 are embedded allows to acquire images 23 at each flight, automatically, and without human intervention and expensive. The acquisition means 13 and processing means 15 are configured to acquire the infrared image 23 from the thermographic means 11 and to calculate differentials relating to a component (for example, the amplitude or the phase) of the thermal field between different zones. comparisons or subdivisions 25a-25d of the infrared image 23. The detection of the variations of the component of the thermal field is indicative of defects or premices of rupture of the rotating element 3. Thus, it is possible to monitor at each flight and automatically the rotating elements 3 of the engine 5 to detect the first signs of fatigue before a dawn loss occurs. In particular, the detection system is well suited to monitor the rotating elements 3 of composite materials that can undergo fatigue causing non-visible defects on their surfaces. It will be noted that performing the analysis of the data according to differential measurements on transient thermal phases makes it possible to dispense with the context such as the outside temperature or the illumination by the sun S (see FIG. The external conditions act in the same way on a current comparison zone at two successive instants. Indeed, FIG. 2 illustrates the effects of the context on a rotating element 3 of an aircraft engine. According to this example, a part 3a of the rotating element 3 is illuminated by the sun S while the other part 3b is in the shadow hidden by another element 4 of the engine 5. Following a warming by a thermal stress 19a, the rotating element 3 emits infrared waves 20 which are translated by the curves C1-C4. More particularly, the curves C1-C4 reflect the variation of the transient phase temperature T as a function of time t on different zones of the rotating element. Curves C1 and C2 represent the shape of the thermal wave in illuminated areas while curves C3 and C4 reflect the shape of the thermal wave in shaded areas. These curves show that in the transient phase (heating or cooling), the shape of the thermal wave and the interval between two instants t1 and t2 do not affect the measurement of the differentials between neighboring subdivisions. Thus, the calculation of the differentials by the processing means 15 makes it possible to detect the presence of the defects (see curve C2) while avoiding illumination by the sun S or any other extrinsic phenomenon. Advantageously, it is possible to exploit the acquisition means 13 and the processing means 15 of a computer 27 embedded in the aircraft 21 or in a computer 27 integrated in the engine 5 of the EMU (Engine Monitoring Unit) type 5 engine to exploit the detection system 1 according to the invention. In particular, the computer 27 may be used to execute a computer program stored in storage means 29 of the computer and comprising code instructions for implementing the detection method according to the invention.

It will be noted that the images acquired can be directly processed during the flight of the aircraft. Alternatively, the image processing 23 can be performed after the landing of the aircraft 21 so as not to overload the computer 27 during the flight, especially if we watch the propellers in stop phase at the end of the flight. flight. According to yet another variant, the data of the images 23 acquired can be transmitted to the ground for processing by a computing station. Fig. 3 schematically illustrates an example of a detection system 1 comprising several heating and thermographic means, according to the invention. Indeed, in the case where the motor 5 comprises a bladed wheel 7 double helix, especially in the case of an open rotor, a camera and a heat emitter can be placed on either side of the two rows of propellers . According to the example of FIG. 3, a first thermal camera 11a and a first heat transmitter 9a are fixed on the fuselage 21a downstream of the bladed wheel 7 so as to aim the downstream propeller 31a. Similarly, a second thermal camera 11b and a second heat transmitter 9b are fixed on the fuselage 21b upstream of the bladed wheel 7 so as to aim the upstream propeller 31b. Moreover, even in the case of a single propeller, a second pair of camera and transmitter can monitor the opposite side of the propeller that is not seen with the first camera.

Fig. 4 is an algorithm illustrating various data processing steps implemented by the acquisition and processing means according to a preferred embodiment of the invention. In addition, Figs. 5A-5E are grids of an image schematically illustrating the steps of the flowchart of FIG. 4.

In step E1, the acquisition means 13 are configured to acquire the image 23 representative of the rotating element 3 picked up by the thermographic means 11. In the step E2, the processing means 15 are configured to criss-cross the image 23 into a plurality of common subdivisions and to define a determined resolution corresponding to a minimum size of defects. Fig. 5A shows a grid 25 of a portion of the image 3 in nine subdivisions 125a-125i in the form of large tiles of the same sizes. Small tiles 100 represent the selected resolution. Moreover, it should be noted that the subdivisions may also be of hexagonal or triangular shape or of any other geometrical shape. In step E3, the processing means 15 are configured to calculate first current differentials relative to a component of the thermal field between each current subdivision 125a and the neighboring current subdivisions 125b-125i. More particularly, the processing means 7 calculate a component (amplitude or phase) of the thermal field relative to each subdivision and then compare the component of each subdivision with those of its neighbors. According to the example of FIG. 5A each tile is compared to its eight neighbors by calculating the differential between firstly the component of the thermal field in a tile 125a and secondly the component relative to each of the eight tiles 125b-125i 'o adjacent. Step E4 is a test where the processing means 15 are configured to check whether or not there is a current subdivision for which first current differentials with at least a first predetermined number of neighboring subdivisions are indicative of anomaly. An anomaly indicator may for example be the comparison of the differential with a predetermined threshold. Alternatively, the abnormality indicator may be defined by the difference between the observed differential and a healthy differential measured during a learning phase and the comparison of this difference with a predetermined level. Note that the threshold or predetermined level may depend on several factors such as the number of neighbors, the size of the subdivision 125a, the material of the rotating element, the desired accuracy, etc. If the result of the test of step E4 is negative, then it is considered in step E5 that the rotating element 3 is valid.

On the other hand, if a subdivision is found for which the first current differentials, with at least a first predetermined number of neighboring subdivisions are indicative of anomaly, then this subdivision is considered potentially invalid and step E6 is taken. It will be noted that if the differentials indicate an anomaly with only another neighboring subdivision, it can probably be considered that it is an imprecision or measurement error. In other words, for the subdivision to be declared potentially invalid, there must be at least a threshold number of neighboring subdivisions with which the differentials are indicative of anomaly. This threshold number may also depend on the number of neighbors, the size of the subdivision, and the desired accuracy.

In the example of FIG. 5A, this threshold number is chosen equal to four and this figure shows that the subdivision 125a in the center has, with respect to at least four of its neighbors 125b-125i, a differential indicative of anomaly. Thus, when the test of step E4 is confirmed, the processing means 15 are configured to compare in step E6, the potentially invalid subdivision with remote subdivisions. In particular, the processing means 15 calculate second current differentials between the potentially invalid current subdivision and remote current subdivisions. For example one can compare the incriminated central tile 125a of FIG. 5A with only eight remote neighbors (not shown) to limit the computational load. Step E7 is a test where the processing means 15 are configured to check whether the current subdivision has, with at least a second predetermined number of distant current subdivisions, second current differentials indicative of anomaly. We take neighbors far enough apart to leave the potentially invalid area. If the result of the test of step E7 is negative, then it is considered in step E8 that the offending subdivision is valid. On the other hand, if the result of the test of step E7 is positive, then it is considered in step E9 that the incriminated subdivision is invalid.

As before, an anomaly is detected when the differential is greater than a predetermined threshold. In addition, in order for the offending subdivision to be declared invalid, there must be at least a second specified number of neighboring subdivisions with which the differentials are indicative of anomaly. Fig. 5A also shows that the subdivision 125a in the center presents with respect to at least four of its remote neighbors (not shown) a differential indicative of anomaly. Comparing a given subdivision with its close neighbors and then with distant neighbors makes it possible to confirm the invalidity of the subdivision and to adapt the resolution of the subdivisions. Indeed, if the differentials between the given subdivision and the nearby neighbors are indicative of an anomaly and if the anomaly results from a real defect, then the differentials with the distant neighbors must also indicate an anomaly since away from the defective area. In particular, if the anomaly is due to a progressive defect, then the differentials with distant neighbors are inevitably greater than with close neighbors. On the other hand, if the fault is very punctual then the differentials with distant neighbors are at least as important as with nearby neighbors. Note that in order to avoid false alarms, we do not take the same thresholds for near and far comparisons. In fact, the remote subdivisions are normally sufficiently distant from the incriminated zone and therefore have relatively large differentials relative to the latter. However, the context in the remote areas may be different and therefore the values of the physical parameter between the two areas may be significantly different without necessarily having a fault. Thus, to avoid false alarms, it is advantageous to choose a larger threshold for a comparison between two distant subdivisions than for a comparison between two close subdivisions. Then, the processing means 15 are configured to calculate other differentials by rearranging the subdivisions and / or refining their sizes. Indeed, in step E10, the processing means 15 are configured to survey an area 225 covering the subdivision 125a declared invalid (see Fig. 5B).

Thus, new subdivisions are formed that overlap the invalid subdivision. The new zone 225 is homothetic to the invalid subdivision, for example with a ratio strictly between 1 and 2. The example of FIG. 5B schematically illustrates a reorganization of the subdivisions with a simple shift of half a tile horizontally and half a tile vertically. Thus, this example shows that four new current tiles 225a-225d overlap the previous invalid current tile 125a (shown in dashed line). Each of these four new tiles 225a-225d covers a portion of the previous tile 125a plus part of the immediate vicinity of the latter. This makes it possible to scan the entire neighborhood of the zone declared invalid.

Again, the processing means 15 calculate new differentials relating to the new division of the comparison areas. In fact, the new subdivisions 225a-225d are considered as the current current subdivisions and for each of these new subdivisions, the steps E11-E17 which are equivalent to the steps E3-E9 respectively are carried out. Thus, in step E11, the processing means 15 are configured to calculate first current differentials between each new current subdivision 225a-225d and neighboring current subdivisions.

Step E12 is a test where the processing means 15 are configured to check whether there is a new current subdivision for which first current differentials with at least a first predetermined number of neighboring subdivisions are indicative of anomaly. If the result of the test of step E12 is negative, then it is considered in step E13 that the subdivision is valid, otherwise, it is considered to be potentially invalid and proceed to step E14. In step E14, the processing means 15 are configured to compare the new potentially invalid subdivision with remote subdivisions. Step E15 is a test where the processing means 15 are configured to check whether the new current subdivision has, with at least a second predetermined number of current subdivisions remote from the second current differentials indicative of anomaly. If the result of the test of step E15 is negative, then it is considered in step E16 that the incriminated subdivision is valid (is that correct?). On the other hand, if the result of the test of step E15 is confirmed, then it is considered in step E17 that the new incriminated subdivision is invalid.

Thus, at the end of step E17, there is at least one new invalid subdivision and one previous invalid subdivision. The example of FIG. 5B shows a new invalid current tile 225a and a previous invalid tile 125a. The overlap between the current and previous invalid tiles gives more precision on the location of the defect.

In fact, in step E18, the processing means 15 are configured to make a mask according to an AND logic operation between the previous invalid subdivisions 125a and the new subdivisions 225a-225d in the overlap zone 225. This forms subdivisions 325a-325d of reduced size comprising at least one invalid subdivision 325a of reduced size (see Fig. 5C). These new subdivisions 325a-325d of reduced size are considered current current subdivisions. In step E19 the processing means 15 are configured to check whether the size of the present current subdivision 325a-325d is greater than the predetermined resolution 100. If yes, the processing means 15 are configured to repeat the previous steps E3-E18 for each current current subdivision, and otherwise, in step E20, the at least one current subdivision (s) is declared as an abnormal subdivision (s). Fig. 5C shows that the AND mask refines the area by decreasing the length and the width of the tile by a factor of 2. However, the size of the invalid tile 325a remains greater than the size of the small tile 100 corresponding to the resolution and consequently, repeat the same steps as shown in FIG. 5D. Finally, FIG. 5E shows that the resolution 100 is reached and the small invalid tiles 100a-100d are located. The example of Figs. 5A-5E shows that the detection method according to the invention makes it possible to greatly reduce the number of calculation steps. Indeed, the part of the image according to the example of Figs. 5A-5E has 18 x 18 = 324 small boxes 100. Thus, ignoring the edge effects, if we compared each box 100 with its eight neighboring cells, we would have 2592 comparisons and the technique would be less effective because we would only detect very localized anomalies on a box 100. With the above technique and always ignoring the effects of edges, it is realized, in the step of FIG. 5A, 9x8 = 72 comparisons, at the step of FIG. 5B, 4 x 8 = 32 comparisons, in the step of FIG. 5C, zero comparison, and finally at the step of FIG. 5D, 9x8 = 72 comparisons, for a total of only 176 comparisons. This reduces the calculation time and the solicitation of the computer. More generally, for an image of an object of 100cm x 20cm and resolution of 1mm, if we took independently each zone of 1mm and compared it with its eight neighbors, we would have, without counting the effects of edge 1600000 comparisons and we do not can detect only 1mm defects or very marked defects.

However, by applying the detection method according to the invention, taking an initial grid of 1 cm and assuming that there is only one defect, the total number of comparisons is about 16,000. method according to the invention considerably reduces the number of calculations by optimizing the number of comparisons. In addition, it detects defects whose size is between the dimensions of an initial subdivision 125a and the resolution 100 selected. Indeed, Figs. 6A-6D illustrate the detection of point and progressive defects on different grids.

Figs. 6B and 6D show that if we looked at each small box individually, we would detect intense or point defects 17a (FIG 6B), but not progressive defects 17b (FIG 6D) for which the comparison with the close neighbors is weak, but from neighbor to neighbor there is more and more lag. Thus, Figs. 6A and 6B show that a point defect 17a can be detected on a large tile 425 or a small tile 525. However, Figs. 6C and 6D show that a progressive defect 17b can be detected on a large tile 425 because it is characterized by a not insignificant difference between the large tile 425 and its neighbors (FIG 6C), but not on a small tile 525 because the The difference between a small tile 525 and its neighbors is very small (Fig. 6D). Thus, one would not detect a progressive defect with a conventional method which considers only small tiles. Fig. 7 is a detection algorithm comprising a confirmation phase according to one embodiment of the invention. The confirmation phase comprises a comparison of the differentials relating to an abnormal subdivision belonging to an image taken during a last flight with differentials relating to the same abnormal subdivision belonging to each of a given number of images taken during the previous flights. The images of the last flights being for example recorded in the storage means 29 associated with the processing means 15.

Step E21 concerns the acquisition of an image 23 of the rotating element 3. In step E22, the image 23 is transmitted to the processing means 15. In the step E23, the processing means 15 are configured to process image 23 according to the flowchart of FIG. 4.

In particular, in steps E4 and E12 (FIG.4), it is checked whether the first current differentials of a component of the thermal field are greater than a first predetermined threshold. Similarly, in steps E7 and E15 (FIG 4), it is checked whether the second current differentials of a component of the thermal field are greater than a second predetermined threshold. It should be noted that the values of the first and second thresholds can be modified according to the size of the subdivision and therefore the rank of the iteration. At the end of step E23, if no fault is found, then in step E24, the images of the last incriminated flights are reset. On the other hand, if at the end of the step E23, one or more abnormal subdivision (s) is (are) detected, then in the step E25, the storage means 29 record the information concerning the last abnormal subdivision (s) before proceeding to step E26. In step E26, the processing means 15 are configured to compare the differentials of a component of the thermal field relating to the abnormal subdivision (s) belonging to the last image with differentials relating to the ( the same abnormal subdivision (s) belonging to each of the previous images of the rotating element 3. If it is found that the image 23 presents an anomaly for the first time, then no alert n ' is generated (step E27). On the other hand, if it is found that the differentials have increased during the last images, then an alert of high importance is generated (step E28). Finally, if it is found that the differentials remain constant during the last images, then an alert of average importance is generated (step E29).

Fig. 8 is a detection algorithm comprising a confirmation phase according to another embodiment. The steps of the algorithm of FIG. 8 are identical to those of FIG. 7 except for steps E32 and E33.

As previously, step E31 concerns the formation of an image 23 of the rotating element 3. If it is a first image, then step E32 and if not, step E33. Step E32 is a learning phase in which a learning database is constructed by comparing the subdivision differentials of the first image of the healthy rotating element. This can be done according to the steps of comparisons between neighboring subdivisions of the flowchart of FIG. 4. Thus, in step E32, a learning database is created recording healthy differentials between different subdivisions of the healthy original image, knowing that the latter is not necessarily uniform because of intrinsic but normal differences. of the rotating element 3. For example, on a blade can have a leading edge of a different material and therefore react to a thermal stress always in a different way. If the image of the rotating element 3 is not a first image, then proceed to step E33 where the processing means 7 are configured to process the data according to the flowchart of FIG. 4. However, in steps E4 and E12 (Fig. 4), the differences between the first current differentials and corresponding healthy differentials are calculated to check whether they are greater than a predetermined level. On the other hand, in steps E7 and E15 (FIG.4), it is always possible to check whether the second current differentials are greater than a second predetermined threshold. Fig. 9 is a block diagram illustrating an example of defect detection on a bladed wheel of the engine, according to the invention. According to this example, the rotating element 3 corresponds to each of the blades of the bladed wheel 7. At the block B1, the processing means 15 receive data from the engine 5 (represented by the block B2), concerning the rotation speed of the bladed wheel 7 to watch. The processing means 15 trigger the detection process when the bladed wheel 7 begins to rotate at a very low speed. In block B3, the heating means 9 heat up the blades 3 of the bladed wheel 7 of the engine (block B2) by a thermal load 19a generating a thermal field that evolves in the heating and cooling phases. It will be noted that the thermal bias 19a (a thermal pulsation or a periodic thermal wave) penetrates into the material of the blade 3 so that if the material has a defect 17 (at the surface or at depth), the amplitude and / or the phase of the thermal field at the surface will be different. Then, while the thermal field is in its transient phase of heating or cooling, the thermal imaging camera (s) 11 films (s) the blades 3 of the bladed wheel 7 of the engine (block B2) to acquire at least one Thus, at block B4, at least one infrared image 23 is generated. It should be noted that each thermal camera 11 can be configured to acquire an image by blade or a single image for all the blades of the bladed wheel 7.

In block B5, identification means 33 of the blades 3 are used to distinguish the different blades 3 of the bladed wheel 7. This makes it possible to track over time the different blades 3 and to identify which one (s) present (s) defects. These identification means are, for example, optical form recognition means. One can for example use the thermal camera 11 itself coupled to a shape recognition algorithm to identify the blades. Alternatively, the identification means are means of individualization by a marking 33 or labeling on one or two blades 3 of the bladed wheel 7. It is possible to individualize the blades 3 by numbering them with paint or any other inserted material in the blade 3 or disposed on its surface.

In block B6, the processing means 15 perform, for example, a Fourier analysis to calculate a component (amplitude or phase) of the thermal field of each subdivision of the infrared image 23 for each of the blades 3. The component (amplitude or phase) of the thermal field relative to each subdivision is then compared with those of the neighbors of the same rotating element by calculating differentials between the different components according to the steps of the flow chart of Figs. 4, 7 or 8. If the material of the blade is uniform, its entire surface responds identically to the thermal load 19a and therefore the component of the thermal field is constant on all areas. On the other hand, if the material has a surface or depth roughness 17, the component of the thermal field at the surface following the thermal bias 19a will be different. Thus, by comparing the components on different subdivisions, the defects can be detected. Note that when a thermal pulse is used to heat the blades 3 of the bladed wheel 7, the component of the thermal field then corresponds to the amplitude of the thermal field (that is to say, the temperature). On the other hand, when a periodic thermal wave is used to heat the blades 3, then the component of the thermal field corresponds to the phase of the thermal field. In block B7, when an anomaly is detected, the processing means 15 compare the results on several flights, to see if the anomaly is still detected. Indeed, the processing means 15 are configured to record at each flight the differentials relating to the thermal fields of the different subdivisions in order to analyze the evolution of these flight differentials in flight. Thus, it is possible to quantify the evolution of the defects for each blade 3 by comparing the data from the current flight with data from previous flights stored in a database 129a (block B8). In block B9, the processing means 15 perform a follow-up over the time of the evolution of the defect found by comparing this evolution to a predetermined critical threshold in order to send an alert (block B10) to the user or maintenance operator if the critical threshold is exceeded.

Advantageously, the detection system 1 comprises a library of anomalies or a database 129b (block B8) of damage signatures representative of different forms of degradation and their states of progress. This allows the processing means 15 to compare the differentials relative to the thermal fields of the subdivisions presenting the first signs of failure to the characteristic signatures of deteriorations and thus to decide on the type of degradation and on the state of progress thereof. Fig. 10 schematically illustrates a detection system according to a first embodiment of the invention.

According to this embodiment, the heating means 9 are intended to heat the rotating element 3 by thermal pulsations 19b or transient thermal phases. The example of FIG. 10 illustrates a heat emitter 9 (for example a heat lamp) fixed directly on the engine 5 or the aircraft 21, facing the rotating element 3 to heat the latter pulsationally. The rotating element 3 is then heated in a sufficiently short time (a few milliseconds) so that the material of the rotating element does not reach a constant temperature. The heat emitter 9 is fixed at a predetermined distance from the rotating element 3 which may vary from a few millimeters to a few meters.

The thermal camera 11 is installed near the rotating element 3 for example, between a few centimeters and a few meters and acquires the images during heating. In this case, the processing means 15 are configured to calculate differentials between an amplitude 37b of the thermal field (i.e., the temperature) of a current subdivision 25b and amplitudes 37a, 37c, 37d (c that is, temperatures) of the thermal fields of the neighboring subdivisions 25a, 25b, 25c. Thus, if the material of the rotating element 3 has a defect 17 at the surface or at depth, the temperature at the surface following the pulsation thermal load 19b will be different. The comparison of the temperatures between the different subdivisions makes it possible to detect the defects. Of course, the data processing (comparison with close subdivisions, comparison with remote subdivisions, comparison with an anomaly library, monitoring the temporal evolution, comparison with a threshold, etc.) according to the flowchart of FIG. . 4, 7, 8 or 9 applies to this first embodiment. Fig. 11 schematically illustrates a detection system according to a second embodiment of the invention. According to this second embodiment, the heating means 9 are intended to heat the rotating element 3 by periodic heat waves 19c for a predetermined time, for example of the order of a few seconds. The example of FIG. 11 illustrates a heat emitter 9 fixed on the engine 5 (or the aircraft 21) facing the rotating element 3 at a predetermined distance which may vary from a few millimeters to a few meters. The heat emitter 9 corresponds for example to a flash-type heat lamp sending a periodic heat wave 19c of a predetermined frequency to heat the rotating element 3 periodically. The thermal camera 11 is installed near the rotating element 3 for example, between a few centimeters and a few meters and acquires the images during heating.

The heat emitter 9 and the thermal camera 11 can be placed directly on the fuselage or the wing of the aircraft 21. In this second embodiment, the processing means 15 are configured to perform, for example, a Fourier analysis for determine the phase variation 39 between the different subdivisions 25a-25d of the infrared image 23 of the rotating element 3. If the material is uniform, the thermal energy is distributed identically and there is no phase difference between the different areas. On the other hand, if the material of the rotating element has a defect, the thermal energy will not propagate identically and the thermal wave will either be accelerated or slowed down in the defect, which will result in a phase shift.

Thus, the processing means 15 calculate the phase shifts between the thermal field of a current subdivision and the thermal fields of neighboring subdivisions in order to detect the defects. The data processing according to the flowchart of FIG. 4, 7, 8 or 9 also applies to this second embodiment. It will be noted that the second embodiment has the advantage of being little influenced by the distance of the heat source or the illumination of the sun, because the temperature is not measured, but the phase shift. In order to increase the accuracy of the measurements, it is preferable that the heat emitter 9 not be too far from the rotating element 3. FIG. 12 schematically illustrates a detection system according to a third embodiment of the invention. According to this third embodiment, the heating means 9 consist of at least one anti-icing heating element 91 already existing in the motor 5.

Indeed, if the rotating elements 3 already have heating means for preventing frost, the detection system 1 of the present invention can cleverly use this heat source and can therefore omit the installation of the additional heating means and by therefore, reduce the on-board weight. In this case, the anti-icing heating element 91 is set, for example during start-up tests, to provide heat for predetermined periods of time. If the heating element 91 is not integrated in the blade 3 but fixed to the outside, then the detection process is strictly identical to that of the first or second embodiment of FIGS. 11 and 12. Heating element 91 is sufficiently powerful and is integrated in blade 3, a relatively short heating time of a few seconds followed by a cooling time of a few seconds can be used. More particularly, if the heating element 91 is constituted for example by heating wires distributed over the surface of the blade 3, the heating element 91 is fed with a constant current intensity during a determined heating time and then it is stopped. feed in order to decrease the temperature. After a determined waiting time (always identical flight in flight), it is then in transient phase of the thermal field and the processing means 15 trigger the camera 11 to take an infrared photo. In case of anomaly of the material of the rotating element 3, the cooling will be different and we can then compare each subdivision to its near and far neighbors and this flight in flight. On the other hand, in this case, defects can not be detected under the wires because their temperature will distort the thermal response of the material at this location.

On the other hand, if the heating wires are not at the surface but integrated inside the material of the blade 3, the situation is more favorable than before, since the wires do not hide any surface of the blade 3 and one accesses directly the thickness response and can then detect internal defects and the entire surface of the blade 3. The data processing is the same as that detailed above in Figs. 4 to 9. In addition, the processing means 15 are advantageously configured to verify the proper operation of the heating element 91 anti-icing by monitoring the amplitude differential of the rotating elements 3. Thus, if the amplitude response is more in addition to low or zero or increasingly high flight in flight even considering the aging effect of the blades 3 on the thermal responses, the processing means 15 may incriminate the heating element 91. The present invention thus allows to monitor rotating elements made of metal or composite materials to detect the first signs of fatigue using the means fixed on the engine or the aircraft, each flight, automatically, and individually. The present invention is advantageously applied to the tracking of the turbofan fan blades, to the propellers of a turboprop or open rotor as well as to the rotating cowlings of the latter.

Claims (10)

REVENDICATIONS1. Automatic defect detection system on at least one rotating element of an aircraft engine, characterized in that it comprises: - heating means (9) on board for heating said rotating element (3) of the engine (5) by a thermal load, - on-board thermographic means (11) for acquiring at least one infrared image (23) reflecting a transient phase thermal field of said rotating element, and - processing means (15) for calculating component-related differentials the thermal field between different subdivisions of said image in order to detect variations of said component of the thermal field indicative of defects on said rotating element. 2. System according to claim 1, characterized in that when the differential corresponding to a current subdivision is indicative of anomaly, the processing means (15) are configured to calculate other differentials by rearranging the subdivisions and / or refining the current subdivision to locate the locations of the defects. 3. System according to claim 1 or 2, characterized in that the processing means (15) are configured to record each flight said differential relative to the thermal fields of different subdivisions and to analyze the evolution of said flight differentials in flight. 4. System according to any one of the preceding claims, characterized in that it comprises a database (129b) of damage signatures representative of different forms of damage and their states of progress, and in that the means processors (15) are configured to compare the thermal field differential differentials of the subdivisions exhibiting auditory defects of degradations. 5. System according to any one of the preceding claims, characterized in that the heating means (9) consist of at least one anti-icing heating element already existing in the engine. 6. System according to any one of the preceding claims, characterized in that the heating means (9) are intended to heat said element by thermal pulsations (19b). 7. System according to claim 6, characterized in that the processing means (15) are configured to calculate differentials between an amplitude of the thermal field of a current subdivision and the amplitudes of the thermal fields of neighboring subdivisions. 8. System according to any one of claims 1 to 5, characterized in that the heating means (9) are intended to heat said element by periodic thermal waves (19c). 9. System according to claim 8, characterized in that the processing means are configured to calculate phase shifts between the thermal field of a current subdivision and the thermal fields of neighboring subdivisions. 10. System according to any one of the preceding claims, characterized in that the rotating element is a blade of a bladed wheel of said engine, and in that the system further comprises means for identifying the blade having defaults.