System and method for predictive maintenance of a machine

Abstract

A sensor (1; 1') for the predictive maintenance of a machine (M), comprising a support base (2), two vibration transducers (3a, 3b, 3c) constrained to corresponding mutually different points of the support base (2) by means of constraint means (4) to receive the vibrations thereof, joining means (5) for rigidly associating the support base (2) to the machine (M).

Description

[0001] SYSTEM AND METHOD FOR PREDICTIVE MAINTENANCE OF A MACHINE DESCRIPTION

[0002] FIELD OF THE INVENTION

[0003] The present invention generally relates to the technical field of systems for the predictive maintenance of an industrial machine. In particular, the invention relates to a sensor for the predictive maintenance of a machine, a device for the predictive maintenance of the machine comprising the sensor mentioned above and a method for the predictive maintenance thereof.

[0004] STATE OF THE ART

[0005] As known, there are systems capable of predicting the onset of a failure in a machine. This allows to schedule the maintenance of the machine before a malfunction occurs, so as to avoid expensive downtime.

[0006] The systems mentioned above provide for continuously measuring some specific parameters related to the operation of the machine and identify the variances thereof with respect to an excellent condition, even through automatic learning algorithms (so-called "machine learning").

[0007] The systems mentioned above of the known type reveal the drawback of not precisely allowing to identify the component that will lead to the future malfunction.

[0008] Therefore, there arises the need to provide a predictive maintenance system and / or method that is more precise with respect to systems of the known type.

[0009] SUMMARY OF THE INVENTION

[0010] The present invention sets out to at least partially overcome the drawbacks of the prior art mentioned above.

[0011] In particular, an object of the invention is to provide a system for the predictive maintenance of a machine which enables to precisely identify the point in which the onset of a malfunction is predicted.

[0012] The object mentioned above is achieved by a sensor for the predictive maintenance and / or by a device comprising a such sensor and / or by a method for the predictive maintenance.

[0013] Further detailed characteristics of the invention are outlined in the relative dependent claims. Variants of the invention may provide for that the device detects the vibration spectrum of the machine through vibration sensors.

[0014] Further variants may provide for that the vibration sensors mentioned above comprise accelerometers.

[0015] Further variants may provide for that the data acquired by the sensors be made available by an open integrated platform of the " sensor-to-cloud” type.

[0016] Further variants may provide for a pre-processing step in which the data are processed so as to be adapted to the creation of a model by the learning device.

[0017] Advantageously, the possibility of precisely knowing the point where there is predicted the malfunction and / or the greater processing efficiency allowed by the method of the invention allow to identify the component of the machine on which the maintenance is to be carried out, so as to programme it in advance.

[0018] The object and advantage mentioned above, as well as others which will be mentioned below, will be more apparent from the following description of a preferred embodiment of the invention, illustrated by way of non-limiting example, with the aid of the attached drawings.

[0019] BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Fig. 1 shows the sensor of the invention, in plan view.

[0021] Fig. 2 shows a variant of the sensor of Fig. 1 , in plan view.

[0022] Fig. 3 schematically shows the device of the invention.

[0023] Fig. 4 shows a block flow diagram which shows the operation of the device of Fig. 3 that is the method of the invention.

[0024] DETAILED DESCRIPTION OF A PREFERRED EMBODIMENT

[0025] The sensor of the invention, schematically shown in Fig. 1 with reference numeral 1 , is particularly adapted to be used for the predictive maintenance of a machine M.

[0026] The sensor 1 mentioned above comprises a support base 2 which can be rigidly associated with the machine M by means of joining means 5 which, preferably but not necessarily, may comprise screw means.

[0027] The sensor 1 further comprises two vibration transducers 3a, 3b, constrained to corresponding mutually different points of the support base 2 by means of constraint means 4 so as to receive the vibrations of the support base 2 and, therefore, of the machine M.

[0028] Each vibration transducer 3a, 3b is configured to detect the vibration that reaches the corresponding point of the support base 2 and, as a result, generate a signal which can be processed through per se known algorithms to identify unusual vibrations indicating future malfunctions.

[0029] The fact that there are present two vibration transducers 3a, 3b arranged in two corresponding different points allows to compare the respective signals through per se known triangulation algorithms so as to also identify the source direction of unusual vibrations.

[0030] This allows to identify the component from which the vibration comes, therefore attaining the object of the invention.

[0031] Preferably, the signal propagation time detected among various vibration transducers 3a, 3b of the sensor 1 are mainly measured in order to detect the vibration source direction. The detected propagation time allows to calculate the source angle of the frequency component of the detected signal and / or estimate the distance of the emission point.

[0032] Preferably, there is also present a third vibration transducer 3c. Advantageously, the use of a larger number of transducers facilitates the discrimination between the detecting vibrations and the non-detecting ones such as for example noises.

[0033] Still preferably, the three transducers 3a, 3b, 3c are arranged according to the vertices of a triangle. Advantageously, the triangular configuration increases the precision with which the source of the vibration can be determined.

[0034] Preferably, the triangle mentioned above is scalene. Advantageously, the asymmetry of the scalene triangle configuration increases the precision of the sensor further.

[0035] Still preferably, the sensor 1 also comprises a fourth vibration transducer, not shown in the drawings, which advantageously allows to obtain a greater precision.

[0036] Preferably, the distance between any two vibration transducers 3a, 3b, 3c is comprised between 1 cm and 6 cm. The party filing the present invention observed that the range mentioned above allows to carry out accurate predictions in the use in industrial machines.

[0037] Furthermore, positioning the transducers 3a, 3b, 3c diagonally on the support base 2 allows to run a propagation time "software” self-calibration.

[0038] Using the arctangent function to calculate the source angle allows to inherently normalise the measurement and, therefore, a precise reference measurement is no longer indispensable.

[0039] Preferably, each vibration transducer 3a, 3b, 3c is an accelerometer or a microphone.

[0040] With regard to the constraint means 4, they are preferably configured to allow to each vibration transducer 3a, 3b, 3c a relative movement with respect to at least one portion 2’ of the support base 2. The relative movement mentioned above increases the sensitivity of the sensor 1 , given that the transducers are free to perceive the differences in the propagation of the vibrations on the machine.

[0041] Preferably, the constraint means 4 comprise through cuts 6a, 6b in the support base 2 that partially surround each vibration transducer 3a, 3b, 3c so as to circumscribe corresponding areas 2a, 2b, 2c of the support base 2 to which the transducers are constrained, segregating them from remaining part 2’ of the support base 2. In this manner, the segregated areas 2a, 2b, 2c may move relatively to the remaining portion 2’ of the support base 2 due to the elasticity of material of the latter.

[0042] Therefore, the segregated areas 2a, 2b, 2c and the respective transducers 3a, 3b, 3c may move with respect to the portion 2’ mentioned above.

[0043] Such portion 2’ may therefore be identified as the part of the support base 2 which excludes the segregated areas 2a, 2b, 2c.

[0044] Preferably, the through cuts 6a, 6b mentioned above extend around each segregated area 2a, 2b, 2c for an overall angle, around the centre of each segregated area, equal to at least 300°, even more preferably equal to at least 325°.

[0045] Still preferably, one or more first through cuts 6a which surround each segregated area 2a, 2b, 2c extend according to respective circumference arcs. This configuration allows to eliminate any reflections of the vibrations on the means.

[0046] Preferably, each segregated area 2a, 2b, 2c is surrounded by a first through cut 6a which extends without interruption on the angle mentioned above, maintaining the area 2a, 2b, 2c connected to the rest of the support base 2 through a connection portion.

[0047] Still preferably, the through cuts comprise a second through cut 6b separate from the first through cut 6a, obtained in the connection portion mentioned above and interposed between the ends of each first through cut 6a. Preferably, the second through cut 6b has the shape of a circular hole.

[0048] A variant embodiment of the sensor, indicated in Fig. 2 in its entirety with 1', differs from the one above given that each segregated area 2a, 2b, 2c is surrounded by two first mutually separate through cuts 6c respectively arranged on two opposite sides with respect to the corresponding transducer 3a, 3b, 3c, which define two of the connection portions mentioned above. Preferably, the two first through cuts 6c are symmetrical so that the two connection portions are aligned according to an axis passing through the centre of the respective segregated area 2a, 2b, 2c.

[0049] According to a variant embodiment not shown in the drawings, the constraint means 4 comprise the support base 2, or part thereof, made of kapton, a flexible material which allows a certain freedom of movement of the transducers.

[0050] The present invention also comprises a device for the predictive maintenance of a machine M, schematically shown in Fig. 3 and therein indicated in its entirety with 7. In figure, the device 7 was enclosed by a dash and dot line, for the sake of simplicity of representation.

[0051] The device 7 comprises at least one of the sensors 1 , 1' of the type described above, for example the sensor 1. It is clear that, should the sensor 1 be replaced or used along the variant 1', the description outlined below also similarly applies thereto.

[0052] The device 7 further comprises a processing and control unit 8 which receives the signals coming from the vibration transducers 3a, 3b, 3c of the sensor 1 through connection means 9, shown in Fig. 3 with dashed lines.

[0053] In particular, the processing and control unit 8 is configured to process the signals mentioned above so as to determine a corresponding frequency spectrum.

[0054] Advantageously, the analysis of the frequency spectrum mentioned above allows to predict a potential malfunction with greater precision. Preferably, the frequency spectrum mentioned above extends between 0.1 kHz and 100 kHz.

[0055] Preferably, the processing and control unit 8 comprises a local processing unit 11 in each sensor 1.

[0056] Still preferably, the processing and control unit 8 also comprises a remote computer 10, for example a "cloud server1', to which the signals coming from the sensors 1 , possibly processed by the respective local processing unit 11 , can be transmitted through connection means 9 of any known type, for example by cable and / or by means of electromagnetic waves.

[0057] Advantageously, the use of a remote computer 10 to process the signals allows, on the one hand, to limit the complexity of the device 7 and, on the other hand, to use high computing capacity and, therefore, more complex processing algorithms, for example comprising automatic learning algorithms ("machine learning").

[0058] Preferably, the device 7 comprises two or more sensors 1 , with the advantage of being able to obtain more precise data on the operation of the machine M and, therefore, allow a more effective learning algorithm so as to prevent malfunctions more effectively. For the sake of simplicity, Fig. 3 shows two of the sensors 1 mentioned above.

[0059] The connection means 9 mentioned above further comprise a synchronisation line, not shown in the drawings, connected to each sensor 1 for mutually synchronising them. For example, the sensors 1 can be synchronised so as to be able to compare the propagation times with high precision, even among sensors 1 that are different and far from each other. Preferably, the synchronisation occurs through a quick dedicated signal, which is transmitted on an I / O line which directly connects the pins of the microprocessors of the various sensors.

[0060] Preferably, one or more of the sensors 1 comprises an antenna, not shown in the drawings but per se known, which detects the changes in the electromagnetic field and provides a corresponding signal whose frequency analysis, combined with that of the vibration transducers, increases the predictive capacity of the device 7.

[0061] Still preferably, the device 7 is configured to also consider the harmonic distortion of the current absorbed by the machine M in the analysis, to the further advantage of the predictive effectiveness.

[0062] With regard to the connection means 9, they preferably comprise a field bus not shown in the drawings but per se known, which connects the sensors 1 to each other to the advantage of ease of manufacture.

[0063] Preferably but not necessarily, the field bus mentioned above is of the RS485 type with modbus protocol. However, it is clear that variant embodiments of the invention can use a bus of any known type.

[0064] Fig. 4 shows a method, herein indicated in its entirety with 100, for the predictive maintenance of the machine M by means of one or more sensors 1. The sensors 1 are configured to detect the vibrations of the machine M in operating conditions and generate corresponding signals representing the vibrations mentioned above. It is clear that, should one or more of the sensors 1 be replaced or used along the variant 1', the description outlined below remains equally applicable.

[0065] The method 100 provides for an operation 101 in which the sensors 1 detect the vibrations mentioned above.

[0066] In the subsequent operation 102, the signals generated by the sensors 1 are transmitted to the processing and control unit 8, preferably by means of the field bus. The processing and control unit 8 is configured to identify, in the signals, indications on the potential onset of a malfunction of the machine M.

[0067] In particular, during the learning period, there is provided for an operation 103 for processing the signals through a learning algorithm resident in the processing and control unit 8.

[0068] The learning algorithm mentioned above is configured to define a selection model adapted to identify, among the signals mentioned above, those corresponding to a malfunction that could lead to a failure in the machine M.

[0069] In addition, the processing operation 103 further allows to determine the detected signal propagation time to calculate the source angle of the frequency component of the detected signal and / or estimate the distance of the emission point.

[0070] For example, the source angle is calculated using the arctangent function.

[0071] Preferably, the learning period extends for a few months during which the learning algorithm is preferably carried out on the remote computer 10. At the end of the learning, after the selection model mentioned above has been defined, in an operation 104 the sensors 1 are configured so that they apply a filter adapted to implement the selection model mentioned above. The filter mentioned above is aimed at limiting the operation 102 for transmitting to the signals indicating a malfunction only, excluding the other signals, which are not transmitted instead.

[0072] Advantageously, the filter mentioned above limits the amount of data to be transmitted to the processing and control unit 8 and to be processed, so as to save band and avoid occupying memory in the processing and control unit 8.

[0073] For example, the operation 104 for configuring the sensors 1 is carried out by applying the filters configured to analyse the frequencies of the vibrations coming from a determined source angle.

[0074] The description outlined above is preferably implemented by transmitting the selection model to the sensors 1 using connection means 9. Each sensor 1 is provided with a local processing unit 11 , in which the selection model is stored.

[0075] Preferably, there is provided for a further synchronisation operation 105 so that the detection 101 among the various sensors 1 is synchronised.

[0076] Still preferably, the synchronisation operation 105 comprises sending 106 a synchronisation signal to the sensors 1 so as to be able to compare propagation times.

[0077] Still preferably, there is provided for an operation 107 for measuring the harmonic distortion of the current absorbed by the machine M. The processing 103 is carried out also including the datum thus measured.

[0078] It is clear that the sensor 1 , 1’, the device 7 and the method 100 may be different aspects of the same invention or the sensor 1 , 1’ and the device 7 may define an invention, while the method 100 may define a separate invention.

[0079] The invention is susceptible to modifications and variants all falling within the inventive concept outlined in the attached claims. In particular, the elements of the invention can be replaced by other technically equivalent elements.

[0080] Furthermore, the materials may be selected depending on the needs, without departing from the can scope of protection of the invention.

[0081] Furthermore, one or more elements of a specific embodiment of the invention technically compatible with another specific embodiment may be introduced into the latter additionally to or to replace elements of the latter.

[0082] Should the technical elements specified in the claims be followed by reference signs, such reference signs are included with the sole purpose of improving the intelligibility of the invention and, therefore, they do not entail any limitation to the claimed scope of protection.

Claims

CLAIMS1 . A sensor (1 ; 1 ') for the predictive maintenance of a machine (M), comprising:- a support base (2);- at least two vibration transducers (3a, 3b, 3c) constrained to corresponding mutually different points of said support base (2) through constraint means (4) so as to receive the vibrations thereof;- joining means (5) for rigidly associating said support base (2) to said machine (M).

2. Sensor (1 ; 1 ') according to claim 1 , characterised in that it comprises at least three of said vibration transducers (3a, 3b, 3c), that are arranged according to the vertices of a triangle.

3. Sensor (1 ; 1 ') according to claim 2, characterised in that said triangle is scalene.

4. Sensor according to any one of the preceding claims, characterised in that it comprises at least four of said vibration transducers.

5. Sensor (1 ; 1 ') according to any one of the preceding claims, characterised in that the distance between any two of said vibration transducers (3a, 3b, 3c) is comprised between 1 cm and 6 cm.

6. Sensor (1 ; 1 ’) according to any one of the preceding claims, characterised in that said vibration transducers (3a, 3b, 3c) are positioned diagonally on said support base (2).

7. Sensor (1 ; 1 ') according to any one of the preceding claims, characterised in that said constraint means (4) are configured so as to allow each of said vibration transducers (3a, 3b, 3c) a movement with respect to at least one portion (2’) of said support base (2).

8. Sensor (1 ; 1 ') according to the preceding claim, characterised in that said constraint means (4) comprise through cuts (6a, 6b; 6c) that partially surround each vibration transducer (3a, 3b, 3c) so as to define respective segregated areas (2a, 2b, 2c) of said support base (2), the segregated areas (2a, 2b, 2c) being movable with respect to said at least one portion (2’) of the latter, each of said vibration transducers (3a, 3b, 3c) being constrained to a corresponding of said segregated areas (2a, 2b,2c) to allow the movement thereof with respect to said at least one portion (2’) of said support base (2).

9. Sensor according to the preceding claim, wherein said relative movement of each of said segregated areas (2a, 2b, 2c) with respect to said at least one portion (2’) occurs due to the elasticity of the material of said support base (2).

10. Device (7) for the predictive maintenance of a machine (M), comprising: one or more sensors (1 ; 1 '), at least one of said sensors (1 ; 1 ') being according to any one of the preceding claims; a processing and control unit (8); connection means (9) for transmitting the signals coming from said vibration transducers (3a, 3b, 3c) to said processing and control unit (8), said processing and control unit (8) being configured to process said signals so as to determine a corresponding frequency spectrum.1 1. Device (7) according to the preceding claim, characterised in that said processing and control unit (8) comprises a remote computer (10).

12. Device (7) according to any one of claims 10 or 1 1 , characterised in that it comprises at least two of said sensors (1 ; 1 '), said connection means (9) comprising a synchronisation line connected to each of said sensors (1 ; 1 ') to synchronize them.

13. Device (7) according to any one of claims 10 to the preceding, characterised in that at least one of said sensors (1 ; 1 ') comprises an antenna for detecting the changes in the electromagnetic field.

14. Device (7) according to any one of claims 10 to the preceding, characterised in that said processing and control unit (8) is configured to compare said signals coming from said vibration transducers (3a, 3b, 3c) through at least one triangulation algorithm to identify the vibration source direction.

15. Device (7) according to any one of claims 10 to the preceding, characterised in that said processing and control unit (8) is configured to process said signals coming from said vibration transducers (3a, 3b, 3c) so as to further determine the propagation time of the signal detected by said vibration transducers (3a, 3b, 3c) to calculate the source angle of the frequency component of the detected signal and / or estimate the distance of the emission point.

16. Device (7) according to the preceding claim, wherein said source angle is calculated using an arctangent function.

17. Method (100) for the predictive maintenance of a machine (M) by means of one or more sensors (1 ; 1 ') according to one or more of claims 1 to 9 configured to detect the vibrations of said machine (M) in operating conditions and generate corresponding signals representing said vibrations, said method (100) comprising: detecting (101 ) said vibrations by means of said one or more sensors (1 ; 1 '); transmitting (102) said signals from said one or more sensors (1 ; 1 ') to a processing and control unit (8) configured to identify, depending on said signals, conditions of potential onset of malfunction in said machine (M); during a learning period, processing (103) said signals through a learning algorithm configured to define a selection model adapted to identify, among said signals, those indicating said potential onset of a malfunction in said machine (M); after said learning period, configuring (104) said one or more sensors (1 ; 1 ') to apply a filter adapted to implement said selection model so as to carry out said transmission (102) only for said identified signals, excluding the other said signals.

18. Method (100) according to the preceding claim, characterised in that said processing and control unit (8) comprises a remote computer (10).

19. Method (100) according to any one of claims 17 or 18, characterised in that there are present at least two of said sensors (1 ; 1 '), said method comprising a synchronisation operation (105) so as to synchronise said detection operation (101 ) in said at least two sensors (1 ; 1 ').

20. Method (100) according to the preceding claim, characterised in that said synchronisation operation (105) comprises sending (106) a synchronisation signal to said sensors (1 ; 1 ').

21. Method (100) according to any one of claims 17 to the preceding, characterised in that said transmission (102) of said signals occurs by means of a field bus.

22. Method (100) according to any one of claims 17 to the preceding, characterised in that it comprises an operation (107) to measure the harmonic distortion of the current absorbed by said machine (M), said processing (103) alsooccurring on the datum thus measured.

23. Method (100) according to any one of claims 17 to the preceding, characterised in that said operation (103) for processing the signals detected by said sensors (1 ; 1 ’) further allows to determine the propagation time of the detected signal to calculate the source angle of the frequency component of the detected signal and / or estimate the distance of the emission point.

24. Method according to the preceding claim, characterised in that said source angle is calculated by means of the arctangent function.

25. Method according to claim 23 or 24, characterised in that said operation (104) for configuring said one or more sensors (1 ; 1 ') is carried out by applying filters configured to analyse the frequencies of the vibrations coming from a determined source angle.

26. Method according to any one of claims 17 to the preceding, wherein said processing and control unit (8) is a component of a device (7) according to one or more of claims 10 to 16.

27. A method (100) for the predictive maintenance of a machine (M) by means of one or more sensors (1 ; 1 ') configured to detect the vibrations of said machine (M) in operating conditions and generate corresponding signals representing said vibrations, said method (100) comprising:- detecting (101 ) said vibrations by means of said one or more sensors (1 ; 1 ');- transmitting (102) said signals from said one or more sensors (1 ; 1 ') to a processing and control unit (8) configured to identify, depending on said signals, conditions of potential onset of malfunction in said machine (M);- during a learning period, processing (103) said signals through a learning algorithm configured to define a selection model adapted to identify, among said signals, those indicating said potential onset of a malfunction in said machine (M);- after said learning period, configuring (104) said one or more sensors (1 ; 1 ') to apply a filter adapted to implement said selection model so as to carry out said transmission (102) only for said identified signals, excluding the other said signals.

28. Method (100) according to claim 27, characterised in that said processing and control unit (8) comprises a remote computer (10).

29. Method (100) according to any one of claims 27 or 28, characterised in that there are present at least two of said sensors (1 ; 1 '), said method comprising a synchronisation operation (105) so as to synchronise said detection operation (101 ) in said at least two sensors (1 ; 1 ').

30. Method (100) according to the preceding claim, characterised in that said synchronisation operation (105) comprises sending (106) a synchronisation signal to said sensors (1 ; 1 ').

31. Method (100) according to any one of claims 27 to the preceding, characterised in that said transmission (102) of said signals occurs by means of a field bus.

32. Method (100) according to any one of claims 27 to the preceding, characterised in that each of said sensors (1 ; 1 ') comprises a vibration transducers (3a, 3b, 3c) and / or an electromagnetic field transducer.

33. Method (100) according to any one of claims 27 to the preceding, characterised in that it comprises an operation (107) to measure the harmonic distortion of the current absorbed by said machine (M), said processing (103) also occurring on the datum thus measured.

34. Method (100) according to any one of claims 27 to the preceding, characterised in that said operation (103) for processing the signals detected by said sensors (1 ; 1 ’) further allows to determine the propagation time of the detected signal to calculate the source angle of the frequency component of the detected signal and / or estimate the distance of the emission point.

35. Method according to the preceding claim, characterised in that said source angle is calculated by means of the arctangent function.

36. Method according to claim 34 or 35, characterised in that said operation (104) for configuring said one or more sensors (1 ; 1 ') is carried out by applying filters configured to analyse the frequencies of the vibrations coming from a determined source angle.

37. Sensor (1 ; 1 ') for the predictive maintenance of a machine (M) which can be used for a method (100) for the predictive maintenance of a machine (M) according to one or more of claims 27 to the preceding, comprising:- a support base (2);- at least two vibration transducers (3a, 3b, 3c) constrained to corresponding mutually different points of said support base (2) through constraint means (4) so as to receive the vibrations thereof;- joining means (5) for rigidly associating said support base (2) to said machine (M).

38. Sensor (1 ; 1 ') according to the preceding claim, characterised in that it comprises at least three of said vibration transducers (3a, 3b, 3c), that are arranged according to the vertices of a triangle.

39. Sensor (1 ; 1 ') according to the preceding claim, characterised in that said triangle is scalene.

40. Sensor according to any one of claims 37 to the preceding, characterised in that it comprises at least four of said vibration transducers.

41. Sensor (1 ; 1 ') according to any one of claims 37 to the preceding, characterised in that the distance between any two of said vibration transducers (3a, 3b, 3c) is comprised between 1 cm and 6 cm.

42. Sensor (1 ; 1 ’) according to any one of claims 37 to the preceding, characterised in that said vibration transducers (3a, 3b, 3c) are positioned diagonally on said support base (2).

43. Sensor (1 ; 1 ') according to any one of claims 37 to the preceding, characterised in that said constraint means (4) are configured so as to allow each of said vibration transducers (3a, 3b, 3c) a movement with respect to at least one portion (2’) of said support base (2).

44. Sensor (1 ; 1 ') according to claim 43, characterised in that said constraint means (4) comprise through cuts (6a, 6b; 6c) that partially surround each vibration transducer (3a, 3b, 3c) so as to define respective segregated areas (2a, 2b, 2c) of said support base (2), the segregated areas (2a, 2b, 2c) being movable with respect to said at least one portion (2’) of the latter, each of said vibration transducers (3a, 3b, 3c) being constrained to a corresponding of said segregated areas (2a, 2b, 2c) to allow the movement thereof with respect to said at least one portion (2’) of said support base (2).

45. Device (7) for the predictive maintenance of a machine (M), comprising:- one or more sensors (1 ; 1 '), at least one of said sensors (1 ; 1 ') being according to any one of claims 37 to the preceding;- a processing and control unit (8);- connection means (9) for transmitting the signals coming from said vibration transducers (3a, 3b, 3c) to said processing and control unit (8), said processing and control unit (8) being configured to process said signals so as to determine a corresponding frequency spectrum and carry out a method according to any one of claims 27 to 36.

46. Device (7) according to the preceding claim, characterised in that said processing and control unit (8) comprises a remote computer (10).

47. Device (7) according to any one of claims 45 or 46, characterised in that it comprises at least two of said sensors (1 ; 1 '), said connection means (9) comprising a synchronisation line connected to each of said sensors (1 ; 1 ') to synchronize them.

48. Device (7) according to any one of claims 44 to the preceding, characterised in that at least one of said sensors (1 ; 1 ') comprises an antenna for detecting the changes in the electromagnetic field.

49. Device (7) according to any one of claims 45 to the preceding, characterised in that said processing and control unit (8) is configured to compare said signals coming from said vibration transducers (3a, 3b, 3c) through at least one triangulation algorithm to identify the vibration source direction.

50. Device (7) according to any one of claims 45 to the preceding, characterised in that said processing and control unit (8) is configured to process said signals coming from said vibration transducers (3a, 3b, 3c) so as to further determine the propagation time of the signal detected by said vibration transducers (3a, 3b, 3c) to calculate the source angle of the frequency component of the detected signal and / or estimate the distance of the emission point.

51. Device (7) according to the preceding claim, wherein said source angle is calculated using an arctangent function.

Images