Method for detecting damage to a part made of conductive material

Abstract

The invention relates to a method (300) for detecting damage (201) to a part (200) made of conductive material, the damage (201) being a burn or a crack, the part (200) being heated (310) by induction by injecting an energy pulse. The method comprises receiving (330) a temporal sequence of thermal images of the part after heating the part (200), and the temporal signal of the intensity of each pixel of the thermal image is used to construct (370) a phase image of the part by means of a discrete Fourier transformation of the temporal signal.

Description

[0001]DESCRIPTION TITLE: Method for detecting damage to a part made of conductive material TECHNICAL FIELD OF THE INVENTION The technical field of the invention is that of non-destructive testing of mechanical parts. The present invention relates to a method for detecting damage to a part made of conductive material, the damage being a burn or a crack, and a system for implementing the method. TECHNOLOGICAL BACKGROUND OF THE INVENTION In aeronautics and in particular in the sector of production and maintenance of metal parts, certain parts such as the sliding rods of landing gear attached to the wheels of the aircraft are subjected to significant mechanical forces during difficult landings for example. To withstand these mechanical conditions, the base materials forming the mechanical parts must be sufficiently ductile. A ductile material is a material capable of deforming without breaking.This is particularly the case for low-carbon, low-alloy steels, such as 300M steel, for example. When the materials used are sensitive to corrosion, they are generally covered with an anti-corrosion coating on their sensitive surfaces, for example, a chromium-based conductive coating. Health checks on the coating and the steel are important to ensure good resistance to mechanical fatigue and corrosive environments, such as humidity. These tests, preferably non-destructive, are carried out during the part production phase, and in several stages depending on the production cycle and the type of defects to be detected, for example, cracks or burns. Conventional non-destructive testing by penetrant testing, magnetic particle testing, and observation after Nital chemical etching are carried out in production.Penetrant testing reveals cracks, whether on the base steel without coating or cracks (crazing) on ​​the chrome coating. Magnetic particle testing reveals cracks on the base ferromagnetic steel without coating as well as non-metallic inclusions. Tests such as chemical etching with Nital or Barkhausen noise analysis (micro-magnetic method) are generally used to detect burns generated by grinding material. The purpose of grinding material is to adjust the dimensional profile of parts and coating thicknesses to a few microns of tolerance using a grinder.However, grinding can cause burns of the material and / or its coating if it is locally too intense, which occurs for example when the grinder is off-center, or when the grinding conditions are inappropriate (for example, lubrication problem, depth of cut, etc.). Consequently, locally in the steel, two phenomena can occur depending on the temperature reached during the burn. Either the temperature reached is lower than the phase transformation temperature of the steel, and in this case the temperature field releases the residual stresses by thermal expansion (state of local tension instead of compression), thus generating a so-called "black" burn (or over-tempering) because it is conventionally detected after Nital attack by operators in the form of a black colorimetry gradient.Either the temperature reached locally is higher than the phase transformation temperature of the steel, and a hard and brittle microstructure is produced. In this case, this burn is called "white" because it is detected by Nital etching in the form of a white color gradient. When testing by Nital etching, white burns are often more difficult to detect than black burns, but they are often surrounded on their edges by a black stress relaxation zone, which facilitates their detection. For these two types of burns, the thermally affected zone is generally a few tens of micrometers below the surface of the steel (30-100 µm for light burns, 200-300 µm for intense burns), their depths can be measured a posteriori and conventionally by metallographic analysis in optical microscopy after destructive cutting of the parts.Burns significantly degrade the mechanical properties of materials, particularly their local mechanical hardness, which promotes the subsequent appearance of cracks during the part's life cycle. These burns can be caused either directly on the uncoated base steel or on the coated steel, as chromium is a good thermal conductor that transmits heat to the steel by conduction during grinding. The methods mentioned above have several drawbacks. On the one hand, numerical controls by Barkhausen Noise Analysis are carried out by sensors whose spatial resolution is greater than the dimension of submillimeter and millimeter burns, which often makes them undetectable. In addition, the sensors are designed specifically for each part geometry, which requires the design of different sensors for each type of part.On the other hand, penetrant testing, magnetic particle testing, and Nital etching use chemicals that are harmful to operators and the environment. Furthermore, Nital etching is a non-digital method subject to human judgment by operators, making automating burn detection difficult, if not impossible. Finally, to detect different types of defects, specific methods are used, making part health monitoring complex and time-consuming. Thus, there is a need for a method to effectively detect several types of defects on aeronautical part materials with different geometry types, without the use of chemicals.SUMMARY OF THE INVENTION The invention provides a solution to the problems mentioned above, by making it possible to detect crack or burn type damage to a part without chemicals and without manual action by an operator, taking into account the inhomogeneities of the thermal field resulting from the deflection of the electric currents induced in the part, which makes the thermography detection method particularly suitable for this type of damage. A first aspect of the invention relates to a method for detecting damage to a part made of conductive material, the damage being a burn or a crack, the part being heated by induction by injecting an energy pulse, the method comprising: a.Receiving a time sequence of thermal images of the room following heating of the room, wherein each image of the time sequence of thermal images comprises a plurality of pixels, each pixel of each thermal image having a respective intensity, each pixel of the plurality of pixels being associated with a time intensity signal comprising the intensities of said pixel in the thermal images of the time sequence of thermal images; b.For each analysis frequency of a set of analysis frequencies: For each pixel of the plurality of pixels, calculating a discrete Fourier transform of the intensity time signal of said pixel, evaluated at said analysis frequency; Constructing a phase image from the discrete Fourier transforms calculated for each pixel of the plurality of pixels; and Determining, in the phase image, a presence or absence of a damage pattern, a damage pattern being present when the phase image comprises a group of neighboring pixels having intensities in a first interval of values, for which each neighboring pixel of the group of pixels has an intensity in a second interval of values, the first interval and the second interval of values ​​being disjoint; c.Detecting the presence of crack or burn damage when the presence of a damage pattern has been determined for at least one analysis frequency of the set of analysis frequencies. "Conductive material" means a material capable of allowing an electric current to pass through and of diffusing heat. In the context of the present invention, a conductive material has, for example, a thermal conductivity greater than 1 Wm -1 .K -1 and an electrical conductivity greater than 1 MS.m -1. "Energy pulse" means an energy whose value suddenly changes from a zero value to a non-zero value and is then maintained at this value for a short time, the short time being of the order of a few tens of milliseconds. "Thermal image" (or "thermogram") means an image acquired by thermography, i.e. by a thermal camera. The image thus obtained comprises a plurality of pixels each having an intensity corresponding to a heat flux or a temperature of a corresponding zone of the part. "Analysis frequency" means a frequency at which a phase image is calculated, in order to determine the presence or absence of a pattern revealing particular damage. "Group of neighboring pixels" means a group in which each pixel is connected to at least one pixel, the connection being defined by a neighborhood relationship, for example a 4-neighborhood.The determination of such groups of pixels corresponds to a segmentation of the image to detect a pattern therein. By "following heating of the part", it is understood that the images of the sequence of thermal images are acquired during a period integrating the entire heating time, at least part of the heating time (including the end of the heating time) and / or after the heating time. In other words, the acquisition of the images can begin before, during or just at the end of the heating time (i.e. the energy pulse). The invention advantageously makes it possible to detect damage such as cracks or burns on the part made of conductive material without using harmful chemicals. The invention exploits the thermal properties of the material of the part which, following induction heating, produces heat radiation by the Joule effect.In the event of damage, overheating areas can be detected and recorded using an infrared thermal camera as a sequence of thermal images. In addition, obtaining a phase image of a discrete Fourier transform, for a chosen analysis frequency, from the sequence of thermal images makes it possible to represent thermal effects such as cooling time differences and therefore potential damage. In addition to the features just mentioned in the previous paragraph, the method according to the first aspect of the invention may have one or more additional features from the following, considered individually or in all technically possible combinations.According to one embodiment, the method further comprises: when the presence of damage is detected, characterizing the damage from the at least one analysis frequency of the set of analysis frequencies for which the presence of a damage pattern has been determined and / or a shape of the damage pattern, the characterization of the damage comprising a determination of a type of damage from a burn and a crack, information on the depth of the damage relative to a surface of the part and / or information on the length of the damage. In one embodiment, the phase image is constructed from a complex argument of the discrete Fourier transform of the pixel intensity time signal.In one embodiment, the method further comprises: receiving a discrete time signal representing a time evolution of an electric field associated with the energy pulse, and the phase image is further constructed from a complex argument of the discrete Fourier transform of the time signal representing the time evolution of the electric field associated with the energy pulse. According to one embodiment, the method further comprises: modifying the sequence of thermal images by subtracting, pixel by pixel, a first thermal image from the sequence of thermal images from each image of the sequence of images, wherein steps b and c are implemented on the modified sequence of thermal images. This operation makes it possible to reduce the contribution of the spatial variation of emissivity at the surface of the part as well as potential reflections of the environment on the part.In particular, the emissivity corresponds to a radiative flux of the thermal radiation emitted by the surface of the part at a given temperature, related to a reference value corresponding to the flux emitted by a black body at this same temperature. According to one embodiment, each thermal image of the time sequence of thermal images is associated with a respective acquisition instant, in which the first image is associated with a minimum acquisition instant among the acquisition instants of the thermal images of the sequence of thermal images. According to one embodiment, the discrete intensity time signal comprises a plurality of values, each value being associated with a corresponding respective instant among the set of instants. In one embodiment, the part is heated for a heating time of between 10 ms and 100 ms. According to one embodiment, the conductive material is a conductive metal or a material comprising carbon fibers.According to one embodiment, the conductive material is a conductive metal, the conductive metal being a ferromagnetic metal or a paramagnetic metal. According to one embodiment, the material is at least partially covered with a conductive metal coating or a layer of paint. According to one embodiment, the material is at least partially covered with a conductive metal coating, the conductive metal coating of the part being made of chromium, zinc, nickel or cadmium. According to one embodiment, the part is an aeronautical part. For example, the part may be an aircraft landing gear. A second aspect of the invention relates to a device for detecting damage to a part made of conductive material, the damage being a burn or a crack, the part being heated by induction by injecting an energy pulse, the device comprising a circuit configured to implement the steps of the above method.The device is for example a computer. A third aspect of the invention relates to a system comprising: - an excitation chain configured to heat a part made of conductive material by injecting an energy pulse; - a thermal camera configured to acquire a time sequence of thermal images of the heated part; and - a device as defined above. The invention and its various applications will be better understood upon reading the description which follows and examining the figures which accompany it. BRIEF DESCRIPTION OF THE FIGURES The figures are presented for information purposes only and in no way limit the invention. Figure 1 is a representation of a system for detecting damage to a part made of conductive material according to one embodiment of the invention.Figure 2 is a representation of a computer configured to implement a method for detecting damage to a part made of conductive material according to an embodiment of the invention. Figures 3a and 3b represent the temperature curves during induction heating of two parts to be checked, one not having a crack (Figure 3a), and the other having a crack (Figure 3b). Figure 4 represents the system of Figure 1 during a method for detecting damage to a part comprising a crack, according to an embodiment of the invention. Figure 5 is a block diagram of a method for detecting damage to a part made of conductive material according to an embodiment of the invention.Figure 6 is a representation of a first signal representing an energy pulse emitted by a power supply and a second signal representing the evolution of the pulse after its passage through an excitation chain for heating the part. Figure 7 is a representation of a sequence of thermal images from which a time signal corresponding to a pixel of the pixel grid is constructed. Figure 8 is an example of a phase image, in which a pattern, determined according to a step of the method, representing potential damage is found. DETAILED DESCRIPTION Unless otherwise specified, the same element appearing in different figures has a single reference. A first aspect of the invention relates to a method for detecting damage to a part made of conductive material, the damage being a burn or a crack, the part being heated by induction by injecting an energy pulse.The part is, for example, an aeronautical part, such as an airplane landing gear. The conductive material is, for example, a conductive metal or a carbon fiber material. The conductive metal can be ferromagnetic or paramagnetic. Steel is an example of a ferromagnetic metal. Paramagnetic materials include, for example, moderately conductive materials such as nickel or titanium and highly conductive materials such as aluminum or copper. The part material may be coated with an anti-corrosion coating, for example, in a conductive metal such as chromium, or with a layer of paint thin enough for the assembly formed by the material and the paint layer to be conductive. For example, the paint layer may be less than 200 µm thick.Regardless of the type of coating used, it is assumed that the assembly formed by the material and the coating is sufficiently conductive. Subsequently, unless otherwise specified, the term "material" is used to designate the assembly formed by the material and the coating. Figure 1 is a representation of a system 100 for detecting damage to a part made of conductive material according to one embodiment of the invention. The detection system 100 comprises an excitation chain 101, a thermal camera 103 and a computer 102. The excitation chain 101 comprises a heating head 1011, an inductor 1012 (comprising a copper coil and a capacitor) and a power supply not shown. The power supply is configured to generate currents in the heating head 1011 then in the inductor 1012. The heating head 1011 comprises a current step-up transformer.The thermal camera 103 is configured to measure and record heat waves and infrared radiation emitted by a body or object 200 (hereinafter called a “room”), to generate so-called infrared image sequences. The thermal camera 103 may be, for example, a FLIR camera (for example from the X6580sc range), a NoxCam camera (for example from the 640M range) or any other type of thermal camera. The images are recorded in the form of a temporal sequence of thermal images ^^. ^ ^ ^∈^^,^^ , each thermal image ^^ of the sequence being associated with an instant respective ^^, ^ ^1, … , ^, with ^ a non-zero natural integer. The thermal camera 103 has, for example, a spectral band between 1.5 micrometers and 5.5 micrometers. Only thermal radiation having respective wavelengths within the spectral band can be recorded by the thermal camera 103. Each image ^ ^of an image sequence generated by the thermal camera 103 comprises a set of pixels ^^ ^,^ ^with 1 ^ ^ ^ ^ and 1 ^ ^ ^ ^,^ and ^ being two non-zero natural integers. acquired by the thermal camera 103 have for example sizes of 640 x 512 pixels, each pixel being for example a square with an edge of 15 micrometers. Of course, these values ​​are given as an example and do not constitute a limitation of the invention. The computer 102, shown in more detail in Figure 2, comprises a memory 1024 for storing instructions allowing the implementation of a method for detecting damage to a part made of conductive material according to the invention, predefined values ​​of parameters relating to manufacturing constraints, and temporary data for carrying out different steps of the method for detecting damage. The computer 102 further comprises a circuit 1022.This circuit may be, for example, a processor capable of interpreting instructions in the form of a computer program, an electronic card whose steps of the method of the invention are described in the silicon, or a programmable electronic chip such as an FPGA chip (for "Field-Programmable Gate Array" in English). The computer 102 comprises an input interface 1021 for receiving the time sequence of thermal images, and an output interface 1023 for providing information relating to the presence or absence of damage, for example. Finally, the computer may comprise, to allow easy interaction with a user, a screen 1026 and a keyboard 1025. Of course, the keyboard is optional, in particular in the context of a computer having the form of a touch pad, for example.Referring again to Figure 1, the inductor 1012 is placed close to (for example above) the surface of the part to be tested to generate currents, for example alternating currents, transmitted from the heating head 1011 to the inductor 1012. Consequently, according to the Lenz-Faraday law, eddy currents are generated within the part 200 to be tested. The currents are generated in a thickness δ of the part, called the skin thickness, which depends on the frequency of the currents generated in the inductor. Due to the resistivity value of the material of the part (or of the material and coating assembly), the propagation of the eddy currents is accompanied by a release of heat by the Joule effect. As detailed below with reference to Figures 3a and 3b, in the absence of damage in the part, the heating is homogeneous and the surface temperature does not exhibit any sudden local variation.Conversely, the presence of damage deflects the eddy currents: their local density increases, thus creating an overheating zone that can be detected and recorded using the infrared thermal camera as a sequence of images. The thermal camera is configured to transmit the recorded sequence of images to the computer configured to perform specific processing on the sequence of images in order to detect potential damage such as a burn or a crack. Figures 3a and 3b represent the temperature curves during induction heating of two parts to be tested, one without a crack (Figure 3a), and the other with a crack (Figure 3b). More precisely, Figures 3a and 3b represent the temperature curves along an x ​​direction of the part (for example a longitudinal direction). The part to be tested is referenced 200 in both figures.In the case of Figure 3b, the part 200 comprises a crack 201. The broken lines (dashed lines) represent the eddy currents generated by the inductor 1012 and modified (in the case of Figure 3b) by the presence of a crack. δ corresponds to the skin thickness mentioned above. In Figure 3a, T0 represents the initial temperature of the part during induction heating, and TF represents the so-called “final” temperature of the part after a certain heating time. When the part does not have any damage, the final temperature is homogeneous at all points of the part considered. In Figure 3b, representing the temperature of a part comprising a crack during induction heating, the final temperature of the part is not homogeneous. The area of ​​the part 200 located around the crack 201 has a final temperature Tfis higher than the temperature TF at the rest of the part.For a paramagnetic material, for example, hot spots correspond to a higher concentration of eddy currents, while colder spots correspond to a lower concentration of the latter. The bottom of the damage is the site of a high concentration of eddy currents and therefore of a local temperature maximum. Thus, observing the temperature change following induction heating makes it possible to reveal areas of damage in the part. Figure 4 represents the system of Figure 1 during a method for detecting damage to a part 200 comprising a crack 201, according to one embodiment of the invention. It is noted that the invention also applies to detecting burn-type damage. Figure 4 represents eddy currents (in broken lines) generated by induction and deflected by the damage 201.In the case of artificially generated damage such as notches by electrical discharge machining (EDM), the eddy currents plunge below the defect (z axis) while moving laterally away from the defect (y axis). This results in a characteristic “butterfly wing” thermal signature as shown in Figure 4. Figure 5 is a block diagram of a method 300 for detecting damage to a part made of conductive material according to one embodiment of the invention. The method 300 comprises a step 310 of induction heating of the part 200. The induction heating can be carried out by the inductor 1012 shown in Figure 1 by injecting an energy pulse through the inductor 1012 into the part 200, for a duration tc called the heating duration or heating time. The heating duration tc influences the level of energy injected into the part and the thermal contrast.Indeed, it is preferable to inject sufficient energy without going as far as thermal saturation of the part, by maximizing the energy injected in a short duration. For example, the heating time tc can be between 10 ms and 300 ms, or between 10 ms and 100 ms. More particularly, the heating time tc can be between 75 ms and 100 ms. Thus, as detailed above, the induction heating step 310 makes it possible to create, according to the Lenz-Faraday law, eddy currents in a skin thickness δ of the part 200, which are accompanied by a release of heat by Joule effect in the part 200 and therefore an increase in temperature in this part 200.Although in theory an energy pulse corresponding to a step is injected onto the part, storage inductances in the power supply, self and parasitic inductances in the voltage or current transformers for impedance matching in the heating head and self and leakage inductances for the inductor impact the energy pulse, which no longer corresponds to a step. This impact on the energy pulse is thus intrinsic to the inductive nature of the excitation chain and can therefore be difficult to completely cancel in practice. In certain embodiments, the invention advantageously proposes to take this effect into account in the simulation of the thermal response of the material of the part to a pulse.In these embodiments, in order to know the impact of the inductances on the energy pulse emitted by the power supply, the method 300 comprises a step 320 of receiving a discrete time signal representing the electric field received by the part during the induction heating step 310. In particular, Figure 6 represents a first signal (1) of an electric field corresponding to the theoretical energy pulse emitted by the power supply, and a second time signal (2) corresponding to the energy actually received by the part after passing through the heating head 1011 and the inductor 1012. The first signal (1) thus represents an energy pulse of amplitude E1 emitted by the power supply, which has the form of a perfect square wave over a duration (called heating duration) tc. The time signal (2) represents the evolution of the first signal (1) after its passage through the power supply, the heating head and the inductor.The rise rate of the second signal (2) is finite, unlike the first signal (1). The second signal thus reaches its maximum at a non-zero attachment time (or duration) t1 (in practice, of the order of several milliseconds). The rise rate is called parasitic speed, and results from the self and leakage inductances, and more particularly from a transient regime due to the sudden switching on and off of the inductances present in the excitation chain. However, for this attachment duration, a thermal diffusion length for moderately diffusive materials is a few tens of microns, i.e. of the same order of magnitude as the depth and width of the damage to be detected. Thus, in one embodiment, it is possible to take them into account to detect this type of damage reliably and efficiently.Referring again to Figure 5, the method 300 further comprises a step 330 of receiving a sequence of thermal images of the part following the heating of the part 200. The sequence of thermal images is recorded by the thermal camera 103 shown in Figure 1 during an acquisition duration tacq. For example, the acquisition of the sequence of thermal images can begin at the start of heating of the part, or just after the start of heating of the part (between times 0 and tc), or even at time tc. In the case where the sequence of thermal images begins to be recorded from the start of heating of the part, the acquisition duration tacq thus comprises the heating duration tc and a so-called relaxation duration during which the part reacts to the energy pulse. The sequence of images comprises a plurality of K images, acquired with an acquisition frequency facq. Thus, the sequence of images corresponds to the thermal evolution of the part.The heating time tc can be chosen such that the energy pulse injected into the part does not saturate the camera, which would degrade the detection performance by reducing the thermal contrast. Each image of the image sequence is recorded at a respective time ti, called acquisition time, during the acquisition duration tacq. Each image of the image sequence comprises a plurality of pixels associated with a plurality of positions on a pixel grid and each pixel has a respective intensity. The intensity of each pixel is for example a gray level, corresponding to a temperature at said position and at time ti. By "pixel position" is meant a pair {i,j} of values ​​representing the coordinates of a pixel on the pixel grid. A gray level is a luminance or intensity value that can be measured on a scale of shades ranging from black (zero intensity, equal to 0) to white (intensity equal to 255).According to one embodiment, the method 300 further comprises an optional step 340 of obtaining a sequence of modified thermal images from the sequence of images. In this embodiment, each image of the sequence of thermal images can be modified by subtracting from it, pixel by pixel, a first image of the sequence of thermal images. This step makes it possible to dispense with the raw values ​​of the intensities of each pixel, to consider only differences in intensities. For example, the first image corresponds to the image having a minimal acquisition time among the entire sequence of thermal images. The method 300 further comprises steps 350 and 360 implemented for each pixel {i,j} of the grid of pixels which constitutes the thermal images. In step 350, a discrete time signal representing the evolution of the intensity of the pixel {i,j} over time is obtained.This step is implemented for each pixel {i,j} of the thermal image pixel grid. For each pixel {i,j}, the associated discrete time signal ^^,^ ^ ^^. ^,^ ^ includes the values ​​^ ^,^ of the pixel intensities ^ ^,^ for all ^ ^ e laWe therefore have: ∀^ ^ 1, … , ^, ^ ^,^ ^ ^ ^ ^,^ ^ . In other words, for each pixel we obtain a time signal which represents the thermal behavior of this pixel in response to the energy pulse. Figure 7 is a representation of step 350. In particular, Figure 7 represents a sequence Seq of thermal images, said sequence Seq being obtained at the end of step 330 or at the end of step 340 when the latter is implemented. The discrete signal ^ ^,^representing the intensity of pixel {i,j} as a function of time includes the intensity values ​​^ ^,^ ^ at pixel {i,j} for the different images of the sequence Seq. Step 360 is a determination step, from the discrete time signal ^ ^,^ corresponding to the pixel {i,j} obtained in step 350 - and in certain embodiments from the time signal of the electric field received in step 320, from a discrete frequency signal, called the phase signal, and, optionally, from a second discrete frequency signal, called the amplitude signal. The phase signal is defined as the signal associated with the complex argument (or "phase") of a complex signal, and the amplitude signal is defined as the signal associated with the modulus (or "amplitude") of the complex signal. For example, for a complex signal ^ !" ^ # !"$%^& '" ^ |^ !"|$%^& '" function of a frequency !, the phase signal is ) !" !". In step 360, a discrete Fourier transform is applied to all or part of the time signal determined in step 350 and corresponding to a given pixel. In embodiments, a discrete Fourier transform is also applied to the electric field signal calculated in step 320. A first discrete Fourier transform * ^,^^,^ !" ^ +, ^ " of the time signal ^ is thus obtained. ^,^ of the evolution of the intensity of the pixel {i,j}, and, in certain embodiments, a second discrete Fourier transform -!" ^ +, $" of the temporal signal of the electric field $ received by the part during the heating step and determined in step 320. The notations * ^,^ !" and - !" here indicate that the calculated discrete Fourier transforms * ^,^and - depend on the frequency!, but it is understood that these signals are discrete signals, evaluated for a plurality of frequencies!^ ^ 1 / ^^, ^ ^ 1, … , ^ (also called subsequently “set of frequencies”). discrete Fourier function is characterized by a complex argument (which corresponds to the phase) and an amplitude for each frequency for a set of frequencies. Thus, Si,j(f) = |Si,j(f)|e- jφi,j(f) , with |Si,j(f)| representing the amplitude of the Fourier transform given f and φ i,j representing the argument (or phase) of the discrete Fourier transform Si,j at a given frequency f. Thus, E(f)=|E(f)|e- jα(f) . with |E(f)| representing the amplitude of the discrete Fourier transform E at a given frequency f and αrepresenting the argument (or phase) of the signal Ei,j at a given frequency f. The first discrete Fourier transform corresponds to a response, at a given pixel, to thermal heating and is in particular a heat transfer function associated with the area of ​​the material represented by the pixel and the electromagnetic field signal received by this area. The first discrete Fourier transform * ^,^!) can be proportional or equal to the product: Hi,j(f).E(f), f being a frequency among the set of frequencies, Hi,j corresponding to the heat transfer function for pixel {i,j} in response to a dirac and E(f) being the discrete Fourier transform of the time signal of the electric field received at pixel {i,j}. The time signal of the electric field e received by the part does not correspond to a pulse, its discrete Fourier transform E(f) is not a cardinal sine as expected but a cardinal sine whose secondary lobes are attenuated, because of the limitation of the speed of rise and fall of the electric field signal. An example allowing to free oneself from at least a large part of the effects of the inductances and therefore of the time signal of the electric field received is to calculate the phase signal and the amplitude signal from the following ratio: 1 '" 51 738 2,3 9" ^ 2,3 ^ 2,3 '"56 '" => " associated with the signal Ri,j(f) can be calculated from the discrete Fourier transforms * ^,^ !" And - ! " for the different values ​​of !. The (discrete) signal of amplitude 5 / !"5 ^ 5* !"5 / |- !"| can be calculated from the amplitudes of the transforms and -!" for different values ​​of !. The phase signal and the amplitude signal are defined for example for respective frequencies !^ ^ 1 / ^^, ^ ^ 1, … , ^ as defined previously. In other words, only the transfer function associated with the material is analyzed and not that associated with the material / signal pair of the electromagnetic field received by the material. It is noted that step 360 is, like step 350, implemented for each pixel {i,j} of the pixel grid. Thus, at the end of step 360, a plurality of phase signals and possibly amplitude signals each corresponding to a pixel of the grid is obtained. During a step 370, a phase image is obtained for a set of frequencies called "analysis frequencies" among the frequencies ! ^associated with the samples of the phase signals and the amplitude signals (if any) determined. By "analysis frequency" is meant a frequency among the frequencies! ^associated with the signal samples for which the phase (and possibly amplitude) signal samples are analyzed to determine the presence of particular damage. Indeed, certain types of damage (for example, burn or crack) and certain associated characteristics (for example, the depth of the damage or the length of the damage) are particularly visible at certain frequencies. This or these analysis frequencies can be determined in advance, based on a certain number of factors as detailed below. Each analysis frequency can thus be selected based on several factors including, for example: the types of damage expected such as burns and cracks, their size (for example, the length of the damage), their depth in the part the thermal diffusivity of the assembly formed by the material and its coating, if applicable.Some analysis frequencies thus make it possible to discriminate between surface damage and others make it possible to discriminate between subsurface damage. For example, the smaller the damage, the more the analysis frequency increases and the larger the damage, the more the analysis frequency decreases. It is noted that the analysis frequency depends inversely on the heating time tc. For example, the analysis frequency, noted !. ABA , can be such that F ^ ! ABAF It is thus possible to determine the heating time depending on what one wishes to highlight. For example, to detect a small defect (30 to 100 µm for example), the analysis frequency is quite high (for example 15 Hz), and therefore the pulse is quite short (which corresponds to a low tc value, for example between 0.05 s and 0.1 s). Conversely, to detect a large defect (200 to 300 µm for example), the analysis frequency is quite low (for example 5 Hz), and therefore the pulse is longer (which corresponds to a higher tc value, for example between 0.1 s and 0.3 s but a short pulse can also be deliberately chosen to detect small damages). In addition, the higher the thermal diffusivity of the material (and the coating), the higher the analysis frequency.For example, for burn-type damage, there are several classes of burns, including so-called black burns and so-called white burns. For these two classes of burns, the thermally affected area of ​​the part generally has a depth of about ten micrometers below the surface of the material forming the part, for example a depth of between 30 µm and 100 µm for light burns and between 200 µm and 300 µm for intense burns. From the expected burn and / or crack dimensions and the heating time tc, simulations can be carried out to theoretically determine the analysis frequency(ies). Thus, in certain embodiments, the method according to the invention can comprise a preliminary step of determining the analysis frequency(ies), from a simulation, for example by finite elements, of the thermal behavior of a physical model of the part comprising theoretical damage.The known dimensions for intense or light burns for example make it possible to simulate the thermal response of the physical model of the part to an energy pulse for example. The physical model of the part can be obtained from a thermal diffusion equation in the part for example. In particular, the finite element simulation is carried out in the frequency domain and makes it possible to obtain one or more analysis frequencies from which the theoretical damage is detectable. During step 370, a phase image is obtained for each analysis frequency, the phase image comprising a plurality of pixels each corresponding to an intensity equal to a function of a value of the sample of the phase signal associated with said analysis frequency and determined for the pixel {i,j} of the pixel grid. In other words, a phase image at an analysis frequency !. ABAis an image of the same size as the images in the thermal image sequence, and each pixel {i,j} of this image is associated with a value equal to the phase of the signal / ^,^ ! ABA " , ie = ^,^ ! ABA " . In a particular, when an amplitude signal has been calculated in step 360, step 370 may comprise obtaining an amplitude image comprising a plurality of pixels each corresponding to an intensity equal to a function of a value of the sample of the amplitude signal associated with said analysis frequency and determined for the corresponding pixel of the pixel grid. In other words, the amplitude image at an analysis frequency! ABA is an image of the same size as the images in the thermal image sequence, and each pixel {i,j} of this image is associated with a value equal to the amplitude of the signal / ^,^ ! ABA " , ie! ABA "5. amplitude image is more representative of the effects of the surface emission of the material while a phase image is more representative of thermal effects such as cooling time differences. The method 300 further comprises a step 380 of determining, in the phase image obtained in the previous step, the presence or absence of a group of neighboring pixels of the phase image having intensities in a first interval I1 of values ​​for which each neighboring pixel of the group of pixels has an intensity in a second interval I2 of values, the first interval and the second interval of values ​​being disjoint. This step makes it possible to detect a “damage pattern” in the phase image (and possibly in the amplitude image) obtained, revealing the presence of damage. Figure 8 is an example of a phase image of a part of an aeronautical part.The phase image makes it possible to highlight a pattern M determined according to step 380. Finally, the method 300 may comprise a step 390 of characterizing damage detected in step 380. This characterization of the damage may comprise a determination of the type of damage among: burning and crack, information on the depth of the damage relative to a surface of the part and / or information on the length of the damage. This characterization is carried out from the shape of the damage pattern and / or the analysis frequency(ies) for which the damage pattern was detected. Indeed, the shape of the pattern makes it possible to determine the type of damage (for example, a butterfly wing pattern is indicative of a crack in the part).The analysis frequency at which the pattern is observed allows the damage to be characterized, in particular to determine information relating to the length or depth of the damage.

Claims

CLAIMS

1. A method (300) for detecting damage (201) to a part (200) made of conductive material, the damage (201) being a burn or a crack, the part (200) being heated (310) by induction by injecting an energy pulse, the method comprising: a. Receiving (320) a discrete time signal representing a temporal evolution of an electric field associated with the energy pulse and receiving (330) a time sequence of thermal images of the part following heating of the part (200), each image of the time sequence of thermal images comprising a plurality of pixels, each pixel of each thermal image having a respective intensity, each pixel of the plurality of pixels being associated with a time intensity signal comprising the intensities of said pixel in the thermal images of the time sequence of thermal images; b.For each analysis frequency of a set of analysis frequencies: o For each pixel of the plurality of pixels, calculating (360) a discrete Fourier transform of the intensity time signal of said pixel, evaluated at said analysis frequency and calculating a discrete Fourier transform of the discrete time signal representing the time evolution of the electric field associated with the energy pulse; o Constructing (370) a phase image from the discrete Fourier transforms calculated for each pixel of the plurality of pixels and from a complex argument of the discrete Fourier transform of the time signal representing the time evolution of the electric field associated with the energy pulse; and o Determining (380), in the phase image, a presence or absence of a damage pattern, a damage pattern being present when the phase image comprises a group of neighboring pixels having intensities in a first range of values,.for which each neighboring pixel of the group of pixels has an intensity in a second interval of values, the first interval and the second interval of values being disjoint; c. Detecting a presence of damage (201) of the crack or burn type when the presence of a damage pattern has been determined for at least one analysis frequency of the set of analysis frequencies.

2. The method (300) of the preceding claim, further comprising: when the presence of damage (201) is detected, characterizing (390) the damage (201) from the at least one analysis frequency of the set of analysis frequencies for which the presence of a damage pattern has been determined and / or a shape of the damage pattern, the characterization of the damage (201) comprising a determination of a type of damage from a burn and a crack, information on the depth of the damage relative to a surface of the part and / or information on the length of the damage.

3. The method (300) of the preceding claim, wherein the phase image is constructed (370) from a complex argument of the discrete Fourier transform of the pixel intensity time signal.

4. The method (300) of any preceding claim further comprising: modifying (340) the sequence of thermal images by subtracting, pixel by pixel, a first thermal image of the sequence of thermal images from each image of the sequence of images, wherein steps b and c are performed on the modified sequence of thermal images.

5. The method (300) of the preceding claim, wherein each thermal image of the temporal sequence of thermal images is associated with a respective acquisition instant, wherein the first image is associated with a minimum acquisition instant among the acquisition instants of the thermal images of the sequence of thermal images.

6. Method (300) according to any one of the preceding claims, wherein the material is at least partially covered with a conductive metal coating or a layer of paint.

7. Method (300) according to any one of the preceding claims, wherein the part (200) is an aeronautical part.

8. Device (102) for detecting damage (201) to a part (200) made of conductive material, the damage (201) being a burn or a crack, the part (200) being heated (310) by induction by injecting an energy pulse, the device comprising a circuit configured to implement the steps of the method according to any one of the preceding claims.

9. System (100) comprising: - An excitation chain (101) configured to heat a part (200) made of conductive material by injecting an energy pulse; - A thermal camera (103) configured to acquire a time sequence of thermal images of the heated part (200); and - A device (102) according to the preceding claim.

10. Computer program product comprising instructions for implementing the method according to one of claims 1 to 7 when this program is executed by a processor.

Images