FR2970329A1 - Method for determining defect of squeal generated by idler-roller in e.g. 1.6 liter direct injection four valves type diesel engine of car, involves measuring vibro-acoustic signal by sensor that is arranged perpendicular to window - Google Patents

Abstract

The method involves measuring a vibro-acoustic signal by an ultrasound sensor i.e. test probe, which is arranged perpendicular to a window. The vibro-acoustic signal is processed by a Fourier transform to obtain an ultrasonic spectrum, where the Fourier transform utilizes certain parameters such as sampling rate of 100 kilohertz (KHz), spectral resolution of 12.5 Hz, 4096 spectral lines and 500 averages. An absence of noise is detected in a presence of non-resonance peaks in spectra that is obtained for winding rollers (22). An independent claim is also included for a measuring stand for implementing a method for determining lack of noise generated by roller in an engine of a vehicle.

Description

[0001] The invention relates to a method for determining the noise fault generated by the tensioner roller and to the vehicles and measuring bench associated with it [1]. 0002] In the passenger compartment of an automobile, one of the aims pursued is to reduce more and more the level of noise defect heard by the driver and his passengers. This improves comfort and safety. By following these objectives, new sounds specific to the engine components appear. These noise defects were previously masked by the overall noise. The squeaking of the tensioning roller which is a new noise fault identified by this request, is an example. It is therefore desirable to have experimental tools, diagnostic methodologies and antidefault solutions for noise to determine the most quickly the origin of a defect, especially if it had not been previously identified. These tools provide technical and economic benefits. Solutions are known from the prior art for detecting noise defects. This is particularly the case of WO-A1-03 / 038426, US-B1-6 182 018, EP-A-1 087 214, US-B1-6 332 362 and US-A1-2007 / 0137284. However, none of the previous solutions do not correctly detect periodic impulse type noise defects which includes the squeal of the tensioner roller. [000s] There is therefore a need for a method for detecting new noises, and in particular the squeaking of the tensioning roller. For this, the invention provides a method for determining the noise fault generated by the tensioning roller of a vehicle engine, the roller comprising a setting window directly connected to the tensioning mechanism, the method comprising the measurement steps. a vibro-acoustic signal by an ultrasonic sensor, the ultrasonic sensor being perpendicular to the window. According to the embodiments, the method may comprise one or more of the following features: the method comprises a signal processing step by a Fourier transform to obtain an ultrasonic spectrum. the parameters used for the Fourier transform are the following: a sampling frequency of 100 kHz, a spectral resolution of 12.5 Hz, a number of spectral lines of 4096, a number of averages of 500, a temporal weighting of Hanning and an overlap of 80 ° / o. - the sensor is a touch point. the method further comprises a step of comparing the ultrasound spectrum with ultrasonic spectra obtained by implementing the method described previously for rollers that do not have the noise defect. the method comprises a step of detecting the noise fault in the presence of resonance peaks not present in the spectra obtained for rollers that do not exhibit the noise fault. The method comprises a step of determining the noise fault by the presence of peaks at 380.Hz and 780 Hz in the sound spectrum after a heterodyning operation on the ultrasonic spectrum. - the method is implemented in the part of an experience plan. [0008] The invention also relates to the vehicle and the associated measurement bench adapted to the implementation of the method previously described. Other features and advantages of the invention will appear on reading the detailed description which follows of the embodiments of the invention, given by way of example only and with reference to the drawings which show: FIGS. 4, a histogram of the distribution of the number of return files of 25 pieces under contractual warranty respectively according to the suppliers, the engine, the family of parts and the type of customer effect. - Figure 5, a schematic representation of a distribution system; - Figure 6, a representation of the back of a timing belt portion; - Figure 7, a time-frequency representation of the sound pressure signal; Figure 8 is a representation of sound pressure spectra with or without a noise defect; FIG. 9, a time-frequency representation of the sound pressure signal with the temporal evolution of Kurtosis; - Figure 10, a representation of the digital file; FIG. 11, a screen view of the image processing software after separation and synchronization of the sound and video bands; FIGS. 12 to 15, images of the film at several moments; FIG. 16, a view of an experiment; - Figure 17, a representation of the observation without strobe; - Figure 18, a representation of the observation with strobe; figure 19, a view of another experience; - Figure 20, a graph representing the idle spectrum obtained; FIG. 21, a test matrix; FIG. 22, a view of the measurement protocol; - Figures 23 and 24, a graph showing the evolution of Kurtosis respectively in CO3 and C04 configurations; - Figure 25, the temporal evolution of acoustic Kurtosis in logarithmic scale; - Figure 26, a schematic view illustrating the positioning of an accelerometer; - Figures 27 and 28, the temporal evolution of acoustic Kurtosis for two tests with a different sensor; FIG. 29, a view of an example of an ultrasonic sensor; - Figure 30, a view of an example of a dispensing system with a contact probe; FIGS. 31 to 33, 35, 37 and 38, representations of the ultrasonic spectral evolution for various tensioning rollers; Figure 34, a view of another example of a dispensing system with a contact probe; - Figure 36, a view of a tensioner roller screw; - Figure 39, a table giving access to the measuring time to detect the noise of the roller; - Figures 40 to 42, schematic views of another experiment; FIG. 43, representations of the ultrasonic spectral evolution for various tensioning rollers in the context of the experiment of FIGS. 40 to 42; - Figure 45, a schematic representation of the principle of the test bench; - Figures 46, 47 and 78 views of an example of a test bench; FIG. 48, a force representation delivered by the vibratory pot for various tensioning rollers, and FIGS. 49 and 50, representations of the force-displacement laws for various tensioning rollers. The invention relates to a method for determining the noise failure generated by the tensioning roller of a vehicle engine, the roller comprising a window directly connected to the tension mechanism, the method comprising the steps of measuring a vibro-acoustic signal by an ultrasonic sensor, the ultrasonic sensor being perpendicular to the window. This makes it possible to easily detect new vehicle noise defects, namely periodic impulse type noises, such as tensioner roller squeaks. A noise fault on the diesel engine distribution system type DV6TED4 (the abbreviation "DV6TED4" meaning 1.6L diesel engine with turbocharger, exchanger, direct injection and 4 valves per cylinder) appeared in concession early 2008 on vehicles with thousands of kilometers and less than one year of use. The term "noise fault" in the following description means an abnormal noise fault, a sound signal that is perceived as inconvenient by the driver and / or passengers of the vehicle. Figures 1 to 4 show the importance of the distribution noise fault which is the main contributor on the parts of contractual guarantees. These different figures come from analyzes of the return of parts in contractual warranty. Figure 1 is a histogram showing the distribution of the number of return of parts in contractual warranty according to suppliers. For reasons of confidentiality, the suppliers are anonymized and simply numbered from 1 to 9. The number of files are indicated for each supplier in the bar of the histogram. Suppliers 1 and 2 account for 800/0 of parts returned for parts under contractual warranty, supplier 1 representing 64.5% and supplier 2, 14.5%. The diesel engine of the type DV6TED4 is equipped with a distributor belt 12 of the supplier 1 and a tensioner roller 20 of the supplier 2. For these two suppliers 1 and 2, FIG. 2 is a histogram representing the distribution of the number of return of parts in contractual warranty depending on the engine. The engines considered are the diesel engines (DV6, DV4 and DW10) and petrol (EP, TU and EW). The bars consist of two parts, the lower part represents the number of parts returned by the supplier under contractual warranty 1 and the upper part the number of supplier files 2. The number of files is specified for each supplier in both parts of the supplier. the bar of the histogram. The histogram of Figure 2 shows that the diesel engine DV6TED4 type has by far the highest volume of parts return parts as a contractual warranty. Figure 3 is a bar graph representing the distribution in number of incident months returns parts of parts as a contractual warranty depending on the family of parts. The families of parts considered are five in number: the friction wheel, the water pump pulley (also known as PAE), the engine, the distribution drive and the drive accessories. The bars have several parts from bottom to top: a first part corresponding to the friction wheel, a second part the water pump pulley, a third part the engine, a fourth part to drive the distribution and a fifth part in training accessories.

The numbers of records are specified for each room family in both parts of the histogram bar. The more precise study of the histogram of Figure 3 shows that most of the depositions occur between 8 and 15 months of driving with a peak at 11 months of driving. During this period, it is the constituent elements of the distribution system that are mainly deposited. On the other hand, during the first three months of driving, the components of the accessories training are mainly deposited. Figure 4 is a histogram showing the distribution of the number of return files of parts under contractual warranty depending on the type of customer effect. Among the customer effects listed, there is the lack of noise, glare, breakage or other damage. The numbers of records are specified for each vendor in both parts of the histogram bar. The lack of noise is the main customer effect on parts under contractual warranty. [0o17] The noise fault is noticeable inside and outside the passenger compartment, especially when the engine is idling or at low speeds. It concerns all vehicles equipped with diesel or diesel hybrid engines. Subjectively using a noise-vibration diagnostic tool, it seems possible to locate the environment in which this noise fault is generated. This is the vehicle engine distribution system 10 as shown schematically in Figure 5. The dispensing system 10 shown is a belt 12 distribution. The transmission belt 12 could be replaced by a transmission chain. The motor drives in rotation the transmission belt 12 on which different rotary members are meshing. The belt 12 thus makes it possible to make a connection between the camshaft and the crankshaft. The crankshaft pinion and the pulley 30 of the camshaft are indeed driven by the belt 12. A damping pulley 18 limits the amplitude of the crankshaft modes. This pulley is often referred to as AVT. The dispensing system 10 further comprises a tensioner roller 20 and a retractor roller 22 which prevents the strand beat of the belt 12. A strand of the belt 12 is a part of the belt 12 which connects two elements. The more specific function of the tensioning roller 20 is to ensure a sufficient and almost constant tension in the soft strands of the dispensing system 10, in order to avoid the appearance of slipping or more serious phenomena of topping. A soft strand is a part of the timing belt that is pushed by the crankshaft. In addition, the belt 12 meshes with a pinion 26 of a pump 14. The pump 14 pumps a heat transfer liquid to circulate in a cooling circuit of the engine. The circuit is therefore a hydraulic circuit connected to a heater to heat the interior of the vehicle livery with the heat transfer liquid warmed by the engine. The coolant is usually brine so that pump 14 is also known as a "water pump". The timing belt 12 also allows to drive synchronously to the rotation of the crankshaft various elements, including the water pump 14 and the pump 16 at high pressure via a pulley 28. The high pressure pump 16 is used to inject the fuel. More precisely, the noise-vibration diagnostic tool makes it possible to locate the noise fault in the environment of the distribution system 10 close to the tensioner rollers 20 and the winder 22 (part framed by dashed lines in FIG. 5 ). In addition, a visual and auditory analysis seems to indicate that the duration of emission of the noise fault corresponds to the duration of passage of an identification marking 32 placed on the back of the belt 12 on the tensioner roller 20. This marking 32 is obtained by screen printing and is visible in FIG. 6, which represents a portion of the back of the belt 12. The acoustic analyzes also show that the noise fault studied is a broadband noise fault having a frequency of between 2 and 8 kHz. In addition, it is periodic since it appears each time the timing belt 12 performs one revolution. For an engine running at 750 rpm, the repetition frequency of the noise fault is about 0.5 seconds. In addition, this noise defect temporarily disappears by depositing, cleaning the components of the distribution system 10 (belt 12, rollers, pulleys, ..). The same phenomenon occurs by brushing, moistening or applying a solvent on the back of the timing belt 12. The noise defect disappears more sustainably by replacing the distribution kit. This operation includes the change of the timing belt 12 and the tensioning rollers 20 and the reel 22. The various indices mentioned below seem to indicate that the timing belt 12 is the cause of the noise failure, especially since that its appearance coincides with the introduction in series of a belt 12 teflon replacing the original belt 12, it aramid, for reasons of technical savings. More specifically, the identification marking 32 affixed to the back of the distribution belt 12 is suspected to be at the origin of the noise fault. Therefore, in case of customer complaint, the replacement operation of the distribution kit is carried out in concession. But, this operation is expensive and entirely the responsibility of the manufacturer for vehicles under warranty. It should therefore be ensured that this operation is a permanent solution to this problem of noise failure. For this, it is desirable to verify that it is the distribution kit, and in particular the marking 32 present on the belt 12 which is at the origin of this noise fault. [0o3o] However, since the noise problem to be studied may temporarily disappear, for example by cleaning the components of the distribution system 10, the conclusions relating to the changes in the configurations of the distribution system 10 may be controversial because the noise fault appears and disappears without reliable tools being able to attest to the absence, presence, reinforcement or attenuation of the noise fault in question. Indeed, to build the acoustic analysis, the only instruments available are our ears, diagnostic tools and listening such as noise-vibration diagnostic tool or standard vibroacoustic measurements and analysis. These various tools, despite their powers, fail to highlight this phenomenon of noise failure. FIG. 7 illustrates this inefficiency in the case of a time-frequency representation (also designated by the acronym RTF) of the acoustic pressure signal in the form of wavelets. Listening to the headphones of this sound sample makes it possible to position the noise fault between times t1 and t2 with t1 = 19.42 s and t2 = 19.52 s. These two instants are materialized by the two vertical bands at times t1 and t2 in FIG. 7. A range between t1 and t2 is thus highlighted. With the observation of FIG. 7, it is not possible to distinguish differences in level or frequency bands for the sound pressure signal in the particular range from acoustic pressure signals outside the range. . The noise fault can not be detected in this way. [0033] Figure 8 also shows the inefficiency for detecting the noise defect of the study of spectra in fine bands of sound pressure with or without noise failure. Curve 34 corresponds to the acoustic spectrum in the presence of the noise defect while curve 36 corresponds to the acoustic spectrum in the absence of a noise fault. The noise-free spectrum is obtained by recording the sound pressure just after the component cleaning of the dispensing system. In this figure, the acoustic pressure differences are of the order of measurement uncertainty. As a result, it is impossible to distinguish between a configuration with and without a noise fault. The two examples of Figures 7 and 8 therefore illustrate that the methods mentioned fail to highlight this phenomenon of noise fault. Therefore, given the technical elements available after the after-sales missions, it is envisaged to remove the marking 32 located on the back of the belts 12 distribution, especially since it is There are no reliable tools to position yourself or decide, if this removal is really the right solution to eliminate this noise defect. It is therefore proposed to determine by the combination of experimental methods and digital tools the presence or absence of the previously mentioned noise defect, acoustic source hitherto unknown technical reference. Detection can be done in two life situations: when the identified noise fault is audible and when it is not audible. In the case of an audible noise defect, a method for detecting noise is provided. periodic pulse-type noise in a vehicle engine comprising measuring a vibro-acoustic signal by a single sensor and signal processing by the Kurtosis function. An acoustic measurement combined with the calculation of a statistical function called Kurtosis is thus implemented. Such a technique and the resulting indicator perform well in a noisy environment. They can therefore be used on a vehicle, on an engine bench or on a subassembly bench of the engine. Such tools can therefore be used in a car dealership or test bench to detect the noise fault in a noisy environment. Kurtosis is a statistical function denoted K. It corresponds to the measurement of the flattening of a distribution of a real random variable. Applied to a physical signal, its formulation is as follows: [0039] - 4 x x-x K = J f 64 P (x) dx-x [0040] where: x is the instantaneous amplitude of the signal; - x is the average value of the signal; - P (x) is the probability of appearance of the value x, and - 6 is the standard deviation (variance) of the value x. Kurtosis K is 1.5 for a harmonic signal and 3 for a random signal. It has a high sensitivity to periodic shocks or not. The analysis can also be performed in the spectral domain in different frequency bands related to the resonances of the structure. Among the applications of this statistical function, there is the detection of termites in wood or the field of preventive maintenance conditional or coupled to a vibratory measurement for industrial rotating machinery. In the latter case, the Kurtosis can detect at an early stage, the wear of the shaft bearings rotating at low rotational speeds (less than 600 rpm). When K is 3, it means that the bearing is in good condition whereas for K between 4 and 7, it is a risk area in which the bearing can be simply deteriorated to completely deteriorated.

The use of Kurtosis thus makes it possible to prevent mechanical damage. It is proposed here to use this same statistical function to quantitatively highlight a perfectly audible noise defect in a noisy environment whereas this noise defect remains undetectable by standard vibroacoustic analysis tools. For this, a microphone is used to pick up the emitted sound signal. The microphone transforms the audible signal picked up into an electrical signal proportional to the acoustic signal received. For the processing of this signal, the microphone is either associated with a conventional measurement system or directly integrated into a digital device. By way of illustration, the digital apparatus may be a video-camera, a dictaphone or a mobile phone. The audible signal is recorded. The recording time can be short insofar as the squealing noise defect is a periodic impulse phenomenon very little damped whose appearance is related to the engine speed. In the example taken previously, the noise fault appears every 0.5 seconds for a motor running at idle, that is to say at a speed of approximately 750 rpm. Thus to have several noise fault patterns, the recording time can be set to 10 s. This duration remains valid even if the appearance of the noise fault occurs on low speeds because in this case the frequency of occurrence is lower than that of the idle. Such recording times limit the number of data to be recorded. This has the advantage of facilitating the processing, the computation and memory capacities solicited by the processing of these data being relatively small. The signal is sampled so as to capture all of the acoustic energy radiated by the noise fault. This corresponds to an extended frequency band of up to 10 kHz. The amplitude of acoustic kurtosis depends on this parameter. As the sampling frequency is not always a parameter accessible on digital devices, the noise fault indicator based on the calculation of Kurtosis can be used in this case only to perform detection is to know if the noise fault is present or absent. The signal is then filtered. The time signal can advantageously be inputted to a high-pass filter with a cut-off frequency of 1000 Hz. Such a filter in fact limits the noise induced by the primary acoustic sources of the motor. The calculation of the Kurtosis can be performed on filtered time samples of 0.05 s due to the duration of emission of the noise defect. This avoids a signal averaging operation. [oo5o] To illustrate the effectiveness of the detection method, three examples are given. The first example is "mission to Loison sous Lens", the second "Hazebrouck mission" and the third "Armentières mission". All three examples presented below are part of an anechoic engine test bench trial. This demonstrated the effectiveness of the proposed detection method for detecting noise and the source of squealing noise defect criticized by vehicle users. It will be shown effectively that the noise fault comes only from the tensioning roller 20. In the first example, the engine noise fault was registered after-sales by means of a digital camera in front of the vehicle hood in concession. It is therefore an audio-video recording or film whose visualization is compatible with standard computer tools as shown in Figure 10. This media just identifies the listening to the presence of a defect of periodic noise, abnormal that is similar to a friction or a squeak. As it stands, the raw film of the mission can only provide subjective impressions. Listening, the squeak appears periodically and its intensity seems subjectively vary very slightly from one pattern to another. The selected temporal extract of the film comprises two complete noise failure patterns which are located between the instants t1 = 19.5 s and t2 = 20.06 s. The time-frequency representation corresponds to that of FIG. 8 and has already been studied. As explained above, it does not show any difference in time as well in frequency as in amplitude. It is then impossible to distinguish the squealing source from the surrounding noise, that is to say the noise of operation of the engine and the ambient noise of the concession. FIG. 9 makes it possible to make a comparison between the time-frequency representation method and the method proposed with Kurtosis. In fact, an enlarged view of FIG. 5 is compared with a graph showing the temporal evolution of the acoustic Kurtosis (see ordinate on the right) and the standard acoustic kurtosis (see ordinate on the left). Normalized Kurtosis is the difference between the Kurtosis value and the Kurtosis value for a random distribution, which is mathematically translated by Knorm = K-3 noting Knorm the normalized Kurtosis. It reflects the excess flattening or a contrario of punchiness with respect to the value given for a pure normal distribution (which corresponds to K = 3). The observation of the temporal evolution of acoustic Kurtosis normalized or not highlights two peaks, one at t1 = 19.5s and the other at t2 = 20.06s. These two peaks correspond to the emission of squealing noise defects on the film of the first example. In the case of standardized acoustic Kurtosis, the amplitude of the squeak is between 1.5 and 2 at the two peaks. Elsewhere, she approaches zero. In addition, the value of Kurtosis does not remain constant on the peaks. This indicates that the intensity of the squealing noise defect is also different from one pattern to another. This corresponds to what is felt by the subjective listening of the soundtrack. In addition, the impulse aspect of the squeaks results in an appearance density curve different from the random appearance of the acoustic signals radiated by the motor. This first test also shows that the noise defect is not produced by the marking 32 placed on the back of the belt 12 distribution. To separate the sound of the image from the original movie while synchronizing them on two independent channels, an image processing software is used. In this way, it becomes possible to make simultaneous or dissociated observations of the three major parameters: time, sound and image. This is shown in FIG. 11 which is a view of the screen of the image processing software after separation and synchronization of the sound and video bands. This view shows that the user can arbitrarily fix a moment of the film and then observe the other parameters. The time is indicated in the window by the reference 42 and corresponds to the clock of the film. Figure 11 illustrates that the user can also arbitrarily set an image or set of images of the film and then observe the other parameters. Thus, as shown by the different images 46, it is possible to perform a decomposition of the image-by-image film. The view also shows that the user can arbitrarily set sound events with cursors and then observe the other parameters. Element 48 is a temporal representation of the soundtrack. To confirm the acoustic signature of the customer criticized noise fault, a time range containing several reasons noise defects is isolated in the temporal representation of the mission film. This insulation is achieved by means of listening in which the speed of scrolling is slowed by about 90%. By musical analogy, this device is like listening to a 78 rpm record on a turntable set to 33 rpm. In this way, by positioning cursors where the noise defects occur, the user can accurately record the times corresponding to the appearance of each of them, the periodicity (s) between them. appearance of two noise fault patterns and the average duration of emission of a noise fault pattern. From the knowledge of the moments when the noise fault patterns appear, it becomes possible to determine the frequency band characteristic of the noise fault by processing the time signal using time-frequency representation in wavelets ( Figure 7). The acoustic signature of the noise fault recorded during subjective listening can then be determined. The periodicity of appearance is equal to 0.46 s; this time corresponds exactly to the time required for the timing belt 12 to make a turn, at idle speed. The average emission time is of the order of 0.1s. Finally, the frequency band is between 2 and 4 kHz; the high value of the frequency band is lower than the value found during the acoustic analyzes because of the metrological limits of the microphone integrated into the digital camera. All the signal processing carried out on the extract of the temporal representation of the mission film makes it possible to affirm that the noise defect noted during the subjective listenings corresponds to the lack of noise criticized by the customers (acoustic signatures identical). With this conclusion and now having the appearance time, the transmission time of a noisy pattern, it becomes easy to view the corresponding film images. The noise fault pattern appears at time t = 19.43 s and ends 0.1 s later at t = 19.56 s. The images of the film related chronologically are Figures 12 to 15. It is observed that the emission of the noise defect begins slightly after the image before the passage of the marking 32 on the camshaft and ends on the image just after passing the marking 32 on the camshaft. As the marking 32 is in contact with nothing during the emission of the noise fault, it can not be considered as causing the noise fault. Removing it from the belt 12 will therefore have no impact on the existence of the noise fault. The identification marking 32 affixed to the back of the timing belt 12 is therefore not responsible for the noise fault criticized on the diesel engine of the DV6TED4 type. In the second example, similar results are found because the measuring means and the shooting are the same. Again, a digital camera has indeed been used. The third example shows that the amplitude of the standardized acoustic Kurtosis increases near the source of the noise fault. In addition, the amplitude of this quantity quickly becomes zero in the near field by deviating laterally from the source of the noise fault. As the timing belt 12 is not the component responsible for the noise fault, the mystery remains intact on the origin and the component related to the noise fault. Based on the elements of the previous analyzes, other components that have similarities to the characteristics of the noise defect should be found.

For this, attention is focused on the frequency of appearance of the noise fault: T = about 0.5 s at idle. As shown in Figure 16 which is a view of the experiment, the facade 10 distribution is illuminated by means of a strobe 50 calibrated at 2 Hz (corresponding to a periodicity of 0.5 s) to search of immobile components despite the rotation of the engine.

By comparison of Figures 17 and 18 which correspond respectively to the movement of the tensioner roller 20 with and without stroboscope, it is observed that if the voltage control index 52 of the tensioner roller 20 moves from bottom to top in the absence of strobe, the tensioning mechanism of the tensioning roller 20 remains stationary if it is observed with the aid of the stroboscope. This result is confirmed by placing a displacement sensor 54 in front of the index 52 of the tensioner roller 20 as shown in FIG. 19. The spectrum at idle corresponds to the spectrum of FIG. this spectrum confirms the presence of a periodicity equal to 0.5 s corresponding to the time taken by the timing belt 12 to make a turn. Peak 56 has a frequency of 2 Hz which is the frequency associated with a periodicity of 0.5 s. These preliminary results suggest that the tensioner roller 20 would be responsible for the squeak criticized by customers on vehicles equipped with diesel engines type DV6TED4. In order to rule on this presumption, the test matrix of the table of FIG. 21 corresponding to the first, second and third examples mentioned previously is studied. It involves combining a diesel engine type DV6TED4 all coming from the components of noisy distribution drives (deposited in concession following customer feedback) and non-noisy distribution drives (new and honed). The goal is not to analyze the test campaign conducted between December 2008 and January 2009 as a whole but rather to use some strong results to complete the demonstration. The CO2, CO3 and CO4 configurations of the test matrix will be studied more particularly. In the three configurations CO2, CO3 and C04, components deposited in concession are used. The CO2 test has the particularity of using a new Teflon dispensing belt 12 and lapped while the C04 test uses a new tensioning roller and lapped. A subjective perception of the noises in the three configurations gives an idea of the final result. Indeed, in the CO2 and CO3 configurations of squeak appearing periodically is perceived. The regularity in the appearance of the lack of noise is less than that found in after-sales (mission film). This squeak is not perceived in the configuration C04 either after change of the tensioning roller 20. These observations confirm the responsibility of the roller in the squeak. These experimental results are confirmed by the implementation of the detection method with acoustic Kurtosis. Figure 23 shows the value of the Kurtosis calculation for the CO3 configuration. In this configuration CO3, a tensioning roller 20 comes from a removal of a noisy customer vehicle. Curve 62 corresponds to the time evolution of the Kurtosis value for the tensioner roller 20 while curve 64 gives that for the retractor roller 22. The measurement protocol is shown in FIG. 22 in which the two zones 58 and 60 for the tensioner roller 20 and the roller roller 22, respectively, are highlighted. For the same protocol but applied to the C04 configuration, Figure 24 shows the value of the Kurtosis calculation. The curve 66 corresponds to the temporal evolution of the Kurtosis value for the tensioning roller 20 whereas the curve 68 gives that for the retractor roller 22. Comparing FIGS. 23 and 24, it is found that the curves 64 and 68 are similar and nothing remarkable happens. The value of the standardized acoustic kurtosis, calculated from the retractable roller microphone signal 22, remains in effect equal to zero, reflecting a pure random distribution and the absence of a squeak or impulse signal fault. The roller 22, although separated from the tensioner roller 20 by only a few centimeters, is therefore not responsible for the squealing noise defect. Curve 62 has a contrario resonances that does not present the curve 66. These resonances are highlighted by the ellipses in Figure 23. Faithful to the subjective feeling, peaks of different amplitudes when squeak noise defects more or Less audible occur occur in the acoustic Kurtosis calculated from the tensioner microphone 20 tensioner signal. This again confirms that the tensioner roller 20 is responsible for the squealing noise defect and that the detection method with the acoustic Kurtosis allows obtain information on the localization of the noise fault when the noise failure measurements are made closer to the components. Because the tensioner roller 20 generates the noise fault criticized by users, the noise fault will be referred to in the following description "squirting of the tensioner roller." These three examples therefore show the power of the indicator associated with the implementation of the detection method to identify the squeaking of the tensioner roller 20 on a heat engine and with the adaptation of signal processing parameters the detection of any impulse sound phenomenon in a noisy environment. It has also been found that the higher the value of the acoustic kurtosis, the higher the intensity or the subjective feeling of the squeak is strong. It then becomes possible to build a quantitative scale for the Knorm Standard Acoustic Kurtosis to objectively quantify this source of noise defect. For a Knorm value between 0 and 1, there is no squeak. For a Knorm value between 1 and 1.5, there is a risk of squealing. In addition, for a Knorm value greater than 1.5, squeak is present. This scale for the standardized acoustic Kurtosis is valid for any position of the microphone. Therefore, inside or outside the vehicle, the value of the acoustic Kurtosis makes it possible to quantify the intensity of the squeaking of the tensioning roller 20. It then becomes possible to compare this intensity with a requirement or a level of desired performance. In addition, in the near field, the use of the quantitative scale exposed below makes it possible to locate the source of the noise fault as explained above. The noise fault indicator (acoustic kurtosis or standardized acoustic kurtosis) thus makes it possible to quantify the intensity of the noise fault. It also allows to approach or estimate the duration of emission of the squeak. For this, it is assumed that the transmission duration corresponds to an attenuation of 3 dB. This amounts to calculating the halfway width of the curve on a peak of the noise fault indicator. Figure 25 graphically illustrates how this width can be determined on a peak. FIG. 25 represents the temporal evolution of the acoustic kurtosis in logarithmic scale between the instants t1 = 7.85 s and t2 = 8.15 s. The maximum value of Kurtosis is 12.70 dB. For the instants of 7.92s and 8.00 s, the Kurtosis value is 9.69 dB. Thus, the transmission time calculated from the half-height width of the acoustic Kurtosis curve is approximately 0.1 s. This is the same result that is obtained by listening.

The noise fault indicator makes it possible to estimate the duration of squeal emission. [oo8li The approach that led to the design of a noise fault indicator based on the combination of an acoustic measurement and the Kurtosis statistical function finds its explanation in the analysis of the acoustic signature of the noise fault and in the identification of the engine component actually responsible for the noise fault. The indicator associated with the detection method is thus an aid to the localization of noise defects. The detection method thus makes it possible to perform a diagnosis on a distribution system 10 from the point of view of these noise performances. This diagnostic work can be done either by a specialist in the field using conventional vibroacoustic acquisition equipment and programming the calculation of the indicator either by a neophyte (automotive mechanic for example) using a noise-vibration diagnostic tool in a version where he would have in his electronics the algorithm that deals with this noise fault and a man-machine interface that signals him. The detection method applies to the vehicle in general for the detection of any periodic pulsed sound phenomenon in a noisy environment. As the implementation of this detection method has made it possible to find which component of the vehicle is involved in the generation of the identified noise fault (namely the tensioning roller 20), it is then possible to envisage other methods / tools. of determination specific to this noise failure. Another detection method similar to that considered so far can be applied to detect the squeaking of the tensioner roller 20 while it is not audible. In this case, the vibro-acoustic signal sensor is not a microphone but an accelerometer 70 placed on the moving part of the tensioning roller 20. This particular positioning of the accelerometer 70 on the roller 20 is of importance. It is illustrated by FIG. 26 which shows the accelerometer 70 placed on a tensioning roller 20. The detection method comprises a processing by the Kurtosis statistical function of the signal emitted by the accelerometer 70. The efficiency of this detection process is highlighted with the help of an example on an intervention carried out after-sales on a vehicle of a user bought by the concession. Characterization and identification of the noise fault is done by means of the detection method. The curve 72 of Figure 26, figure showing the spectra obtained, allows undoubtedly to decide on the origin of the complaint. This is the squeaking of the tensioner roller 20. The curve 74 of Figure 26 is obtained in turn after replacement of the tensioner roller 20 by a new tensioner roller 20. The lack of squealing noise disappears as well in the ear as on the representation: absence of periodic peaks. [0087] However, to reassure the diagnosis, the original tensioner roller 20 is raised in the hope of perceiving the squeak again. However, as shown in curve 76 of FIG. 27, the squealing fault has disappeared. This is also confirmed by our ears. Is this the reality? This is where the alternative with the accelerometer 70 which gives the spectrum of FIG. 28 comes into play. On the curve 78 of FIG. 28, it is not observed periodicity that makes it possible to rule in fine on the signature of the squeaking of the tensioner roller 20. This result joins the previous and the fact that the roller 20 is a new roller. Conversely, the periodicity noted on the signal from the original roller after reassembly clearly highlights the lack of squealing noise, although it is not yet audible. The detection method using the accelerometer as a sensor thus makes it possible to detect the squealing noise. A method for determining the squeal of the tensioning roller comprises the detection of an ultrasonic signal in contact with the tensioner roller 20 by means of an ultrasonic sensor. We will see in the study of the implementation of the determination process that the contact between the sensor and the roller is at the window of the roller for a motor at idle. The ultrasonic sensor can be any type of sensor. According to the example of Figure 29, it is a touch point, that is to say a contact probe for receiving information integral that is beyond the audible frequencies (f> 20kHz). The tip is a metal point to place in contact with the object to listen. The probe also has a handle for holding and hosting the ultrasonic sensor. This ultrasonic sensor has a resonance frequency equal to about 40 kHz and has a narrow bandwidth around its resonance which is ± 1 kHz. Beyond that, without ultrasonic excitation, the signal drops by 6dB / octave. The conditions of occurrence of squealing noise on the vehicle include the fact that the engine is idling. When such conditions are satisfied, the contact probe is to be applied perpendicularly to the window of the tensioning roller 20. This is illustrated by FIG. 30 in which a distribution element with a contact probe is shown. The distribution comprises in particular a tensioning roller 20, the latter being provided with a window for holding the roller around a centering pin and an index. The zone of contact between the ultrasonic sensor and the tensioner roller 20, that is to say the tip-roller zone is highlighted by the dashed circle. In addition, this clearly demonstrates the perpendicularity between the probe and the tensioning roller 20. The determination method also comprises the calculation of the Fourier transform of the ultrasonic signal received by the ultrasonic sensor. To perform this step, it is possible to calculate the Fast Fourier Transform or FFT (Fast Fourier Transform) of the time signal. The following parameters are preferably used: sampling frequency of 102.4 kHz, spectral resolution of 12.5 Hz, number of spectral lines of 4096, number of averages of 500, temporal weighting of Hanning and recovery 800/0. For after-sales applications from the noise-vibration diagnostic tool, the control box of the device must integrate an electronic card for transforming ultrasonic information into sound information. This operation is easily achievable using heterodyne technology which makes it possible to subtract from the incoming signal a frequency slightly lower than the resonance frequency of the listening sensor. Thus, the signal can be heard because it is included in the bandwidth of the sensor which is at audible frequencies. This is the subject of the patent application No. 1054664 filed on 11/06/2010. For applications in the Technical Center by those skilled in the art, this device is not necessary given the sampling frequency of acquisition front ends which is generally equal to about 104 kHz. All graphic representations included in this document are constructed using this technique. However, for more readability and more precision (non-linearities introduced by a piezoelectric ultrasonic sensor), the set of results is presented between 200 and 1000Hz using a signal processing device similar to the heterodyne technique. The calculations are thus consciously limited to a bandwidth of 800 Hz, ie ± 400 Hz around the resonant frequency of the sensor, now almost equal to 600 Hz. That's about half of the initial sensor bandwidth. To show that it makes it possible to distinguish the noise fault, the determination method is implemented in the conditions of appearance of the squealing noise defect by performing the acquisition of the signal for 30 s. The conditions of appearance of squealing noise defect are hot engine idling and no load (low torque). The result of the implementation of this determination method is illustrated in FIG. 31. [0098] This figure thus shows the ultrasonic spectral evolution of a noiseless new tensioner roller 20 and a noisy tensioner roller 20. Curve 94 is the ultrasonic spectrum corresponding to the noiseless roller while curve 92 is the ultrasonic spectrum corresponding to the noisy roller. The difference in acoustic signature appears clearly between the two configurations. In particular, a significant increase in the ultrasonic vibratory energy is observed over the entire spectrum between the curve 92 and the curve 94. This increase is symbolized by an arrow 96. In addition, two energy peaks that are little damped towards 380 Hz and 780 Hz appear in the case of the tensioner 20 noisy. These two peaks are relatively far from the resonant frequency of the ultrasonic sensor present at 600Hz. The ultrasound peaks are highlighted by the dotted ellipses numbered 98 and 100. The resonant frequency of the ultrasonic sensor is symbolized by a dashed arrow. The increase in ultrasonic vibratory energy over the entire spectrum comes from the age difference of the components. This phenomenon is not surprising in that a worn tensioning roller 20 has more mechanical clearances than a new tensioner 20, thus producing higher vibratory energy. This is evidenced by a comparative experiment conducted on a non-noising new tensioner roller and a non-noisy worn tensioner roller. The acoustic spectra obtained by the implementation of the determination method are illustrated in FIG. 32. The curve 106 corresponds to the spectrum obtained for the new non-noisy roller while the curve 104 is the spectrum obtained for the non-noisy worn tensioning roller. . As in the case of FIG. 31, an increase in the vibratory sound energy is observed since the curve 104 is above the curve 106. This characteristic is therefore not specific to the squealing noise defect. [oo1 oo] On the other hand, the energy peaks with little damping around 380 Hz and 780 Hz are not present in the curve 104. This is clearly shown in FIG. 33 in which the curves 92, 94 and 104 are superimposed on the same graph 33 The dotted ellipses numbered 100 and 102 highlight the peaks at 380 Hz and 780 Hz. This means that these energy peaks are an ultrasonic acoustic signature of a tensioner roller 20 having, having presented or going to present the noise failure of squealing. . [oo1o1] The determination method may further comprise a comparison step as has been explained above. This process is powerful regardless of the application environment: test bench, vehicle in a quiet or noisy environment. In a conventional ultrasonic application, the force with which the sensor is pressed on the measured organ influences the result. This is not the case for the determination method described here. An experiment makes it possible to show this advantage for this determination method. The experiment is a test carried out on the fixing screw of the tensioner 20. This particular positioning of the ultrasonic sensor for this test is shown in FIG. 34. The spectra of FIG. 35 are then obtained. FIG. 35 illustrates indeed the evolution of the acoustic signature as a function of the bearing force of the sensor on the tensioner roller 20. The curve 110 is obtained for a stronger support of the ultrasonic sensor than the curve 112. It is possible to note that, on both sides of the resonance at about 600 Hz, the appearance of the two curves 110 and 112 is the same. This means that no matter how strong the touch point is, the ultrasonic response is the same. The conclusions are unchanged for tests on the window of the tension roller 20. [00104] To this advantage over the independence of the support force, is added the fact that the measuring position has little influence on the measurement . Thus, the contact area between the sensor and the roller does not have to be defined very precisely. An experiment makes it possible to show this advantage for this determination method. The experiment is a test performed on the fastening screw of the tensioner 20 as for the support force. The different positions of the sensor on the screw used for the experiment are the positions numbered 114, 116, 118, 120, 122 and 124 in FIG. 36. The spectra obtained are those of FIG. 37. The numbers of the spectra 126, 128 , 130, 132, 134 and 136 respectively correspond to the numbers 114, 116, 118, 120, 122 and 124 of the positions of FIG. 36. FIG. 37 thus illustrates the evolution of the ultrasonic spectrum as a function of the contact position of the sensor on the element. keeps the same pace for several positions of the touch point. The observation of the results obtained shows that the ultrasonic spectrum keeps the same pace for several positions of the touch point. The differences between the different configurations are visible only on the amplitude. These differences are all the more marked at the resonance frequency of the sensor (around 600Hz). Therefore, it is preferable to keep an identical measurement position from one test to another, but to deviate from it has little influence on the result because the appearance of the variation of the spectrum is maintained even with position changes. The conclusions are unchanged for similar tests on the window of the tensioning roller 20. [0105] This determination method thus has the advantages of being independent of the position of the sensor as well as the support force imposed on it. This makes the implementation of this determination process easier. However, if the determination method is relatively independent of the position of the sensor, an experiment can show the importance of the contact zone between the tensioner roller 20 and the ultrasonic tip. In this experiment, the test that led to the results of FIG. 31 is reproduced by applying the contact probe perpendicularly to the fixing screw to the motor as is the case in FIG. 34 and no longer perpendicular to the window of FIG. tensioner roller 20 as in the case of Figure 30. The results obtained are the spectra of Figure 38 which illustrates the evolution of the ultrasonic spectrum for noisy and noisy rollers. Curve 140 represents the spectrum of the noiseless roller and curve 142 represents the spectrum of the noisy roller. To the uncertainties of measurements, it is not possible to distinguish the curves 140 and 142. This method of determination also has the advantage that it makes it possible to avoid waiting after each configuration change several hours of the engine before it can judge the presence or absence of squealing noise. FIG. 39 is a table showing the operating time of the engine to rule on the presence or absence of squeak in the case of the configurations C01 to C10 already presented above with reference to FIG. several hours are thus gained by the implementation of the method of determining the noise failure of the tensioner roller with an ultrasonic sensor. The determination method therefore allows the detection and characterization of the squeak, that this noise defect is audible or not. [00109] Studies previously described, it seems that the squeaking of the tensioner roller 20 has its source on the tension mechanism of the roller. It then becomes interesting to test this hypothesis by artificially modifying the mechanical characteristics of the spring 146 of the tensioning roller 20. For this, at the end of the turn of the spring 146, an elastomer tube 144 is inserted as shown diagrammatically in the figure 40. Figures 41 and 42 show respectively the tensioner roller 20 provided with the tube 144 and an enlarged view of Figure 42 centered on the tube 144. [oo11o] The tensioner roller 20 thus modified is reassembled on the engine on which the method of determination of the noise of the roller is implemented since no squeaking is audible. The results obtained are the spectra of curves 148, 150 and 152 which correspond to the spectrum calculated after respectively 1 hour, 1 hour and 2 hours of operation. For comparison purposes, the curves 92 (tensioner roller 20 exhibiting the noise defect) and 94 (tensioner roller 20 not exhibiting the noise defect) have also been reproduced. The absence of resonance peaks at the frequency of 380 Hz and 780 Hz clearly shows that in this situation, the noise is absent, which confirms the hypothesis, the local modification of the stiffness and damping characteristics of the spring tensioner roller 20 causing well the disappearance of the squeak as expected. [00112] In application of the above observation, it is proposed a test method for tensioning rollers for revealing squealing phenomena. The method comprises measuring the displacement of the index as a function of the tension to which the roller is subjected. By comparison of FIGS. 49 and 50 which respectively correspond to the force-displacement laws for tensioning rollers presenting the noise defect and not presenting it, it is actually observed that the force-displacement law of a tensioning roller 20 " non-noisy "has a characteristic look. The force is proportional to the displacement when the timing belt 12 is stretched or relaxed, which makes it possible to estimate the stiffness of the tensioning roller 20. The shape of the force-displacement law of a tensioning roller 20 noisy present in but a very different look. It is characteristic of an infinite friction because the increase of the effort has no effect on the displacement which remains null. Therefore, the tensioning roller is determined to have a noise defect if the displacement is zero as a function of the voltage, whereas it is determined to have no noise defect if the displacement is proportional to the voltage. This determination has the advantage of being easy to implement once the force-displacement law is known. [00115] To determine it, the block diagram of Figure 44 made in practice in Figures 45, 46 and 47 which respectively show the assembly in its entirety and views of some parts of the assembly that can be used. According to this example, the tensioner roller 20 is placed on a bracket 158, which is fixed to a marble 160. The tightening torque applied to the screw of the tensioner roller 20 is 25N.m, as on the motor. Instead of directly exciting the tensioning roller 20 via a jack for example, the roller 20 drives a belt 12, the voltage variation being generated by injection of a sinusoidal force on the belt 12. For this, one end of a timing belt 12 is cut and is attached to the marble by means of a jaw 156. [00116] The other end of the belt 12 is connected to an electrodynamic exciter 154 so as to wind the roller tensioner 20 as on the engine. The sinusoidal force is thus created by the electrodynamic exciter 154 shown more specifically in FIG. 46. The latter can in particular be a vibrator or vibratory pot. FIG. 48 shows that for a vibratory pot creating a sinusoidal force of amplitude 3 V at the frequency of 6 Hz, the force seen by the roller 20 whether it is squeak or not is the same since the sinusoidal curves 162 and 164 recorded by a sensor attached directly to the belt 12 have the same amplitude. According to the example of FIG. 47, the movement of the index of the roller 20 is measured by a laser sensor 166. Such a sensor has the advantage of being easy to implement while ensuring that relatively accurate results without being intrusive. This method can thus be implemented on a test bench or measurement to reveal in the laboratory if the tensioner roller 20 is likely to squeak. This is particularly advantageous in the development phases of this component. The noise fault identified in the present application was not known in the technical references. As a result, the source was unknown and there was no method to detect it, nor a solution to remove or mitigate it. [00120] The device and the diagnostic tools described in this application not only allow to remove the squeak of the tensioner roller 20 but also allow to monitor its appearance before it is audible or in any case critical. The determination and detection methods can advantageously be implemented in the context of an experimental design, in a vehicle or on a test bench because they make it possible to have a tool for detecting and / or characterizing previously undetectable noises such as squeaking of the tensioning roller 20.

Claims (10)

REVENDICATIONS1. A method for determining the noise fault generated by a tensioning roller (20) of a vehicle engine, the roller comprising a window connected to the tensioning mechanism, characterized in that the method comprises the steps of: measuring a vibro-acoustic signal by an ultrasonic sensor, the ultrasonic sensor being perpendicular to the window. 2. The method according to claim 1, characterized in that the method comprises a step of: - processing the signal by a Fourier transform to obtain an ultrasonic spectrum. 3. The method according to claim 1 or 2, characterized in that the parameters used for the Fourier transform are as follows: a sampling frequency of 100 kHz, a spectral resolution of 12.5 Hz, a number spectral lines of 4096, a number of averages of 500, a temporal weighting of Hanning and an overlap of 80 ° / 0. 4. The method according to one of claims 1 to 3, characterized in that the sensor is a touch tip. 5. The method according to one of claims 2 to 4, characterized in that the method further comprises a step of comparing the ultrasonic spectrum with ultrasonic spectra obtained by the implementation of the method according to claim 2 for rollers not not presenting the noise fault. 6. The method according to claim 5, characterized in that the method comprises a step of detecting the noise fault in the presence of resonance peaks not present in the spectra obtained for rollers that do not have the noise defect. 7. The method according to one of claims 2 to 6, characterized in that the method comprises a step of determining the noise fault by the presence of peaks at 380.Hz and 780 Hz in the sound spectrum after an operation. heterodyning on the ultrasound spectrum for example. 8. The method according to one of claims 1 to 7, characterized in that the method is implemented in the context of an experimental plan. 9. Vehicle adapted to the implementation of the method according to one of claims 1 to 8. 10. Measuring bench adapted to the implementation of the method according to one of claims 1 to 8.