WO2011160652A1 - Acoustical machine condition monitoring - Google Patents
Acoustical machine condition monitoring Download PDFInfo
- Publication number
- WO2011160652A1 WO2011160652A1 PCT/EP2010/003728 EP2010003728W WO2011160652A1 WO 2011160652 A1 WO2011160652 A1 WO 2011160652A1 EP 2010003728 W EP2010003728 W EP 2010003728W WO 2011160652 A1 WO2011160652 A1 WO 2011160652A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- alarm
- condition monitoring
- monitoring unit
- fail
- unit according
- Prior art date
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M13/00—Testing of machine parts
- G01M13/04—Bearings
- G01M13/045—Acoustic or vibration analysis
Definitions
- the invention is related to condition monitoring of machines, especially the bearings of the machines, in particular in relation to bearings of idler rollers of for example conveyor belts.
- Bearings are a very important component in rotating machinery. If a bearing fails, then the complete functionality of the machinery usually also fails. In some applications it might be very difficult or just extremely expensive to replace a failed bearing outside regular scheduled maintenance. Such applications might be continuous manufacturing lines, including conveyors of for example mines and cement plants.
- a trough type conveyor will traditionally comprise a conveyor belt that is typically supported by three idler rollers for approximately every meter. Each idler roller will usually comprise two bearings each, giving six bearings per meter of conveyor. Other types of conveyors, such as pipe conveyors, will also comprise idler rollers for support. Conveyors in mines and cement plants can easily be several kilometers long, resulting in more than ten thousand bearings being incorporated in a single conveyor system. If a bearing of an idler roller fails, then most likely the idler roller will seize. A seized idler roller can fray and cut the conveyor belt. Conveyor belt damage can result in expensive belt repair/replacement, risk of workers' safety, and/or unplanned downtime if no parallel belt or buffer stock exists.
- Condition monitoring is done in an attempt to predict when a bearing needs to be replaced before it fails, suitably enabling replacement in an orderly scheduled manner. It is difficult to overcome the logistical problem of monitoring possibly more than ten thousand bearings of a single conveyor system.
- condition monitoring has been done with a worker walking along the conveyor belt listening and looking for failing or failed conveyor idler rollers. It is hard to detect failing idlers over commonly a very high background noise level.
- a more advanced method of condition monitoring is to use a thermographic camera to detect failing idlers. Unfortunately it is very time consuming and only catches failing idlers at a very late stage of failure. Thus there seems to be room for improvement in the ways of assessing the condition of a bearing, especially a bearing of an idler roller.
- An object of the invention is to define a method and means to monitor the condition of a machine, suitably comprising at least one bearing, especially a bearing of an idler roller.
- a machine condition monitoring unit is usually analyzed by several different methods, each one looking at a different aspect of a machine/bearing. This is at least in part due to the fact that we do not know what sort of failure it will be and what caused it. Different failures have different ways of breaking down and thus give different indications that something is not completely right.
- a machine/bearing condition monitoring unit will then preferably comprise many different methods of trying to predict and detect failures. As a user it can be difficult to assess all the different alarm indications from the different methods. Therefore according to the invention, an alarm voting controller determines if an alarm should be made or not.
- the alarm voting controller analyzes the alarms from the individual methods and possibly how serious each one is and then outputs a single for example go/no go, pass/fail, or pass/warning/fail indication to a user.
- a machine condition monitoring unit comprising digital signal processing means and input means arranged to receive electrical signals from an acoustic sensor.
- the digital signal processing means is arranged to analyze the electrical signals by at least two different methods of bearing analysis, each with an individual signal processing path, each path with an individual alarm level controller creating an alarm output indicating at least either pass or fail.
- the machine condition monitoring unit further comprises an alarm voting logic controller arranged to create one alarm condition indication of at least either pass or fail based on the outcome of all of individual alarm outputs of the individual alarm level controllers.
- the alarm voting logic controller is arranged to create an alarm condition indication of fail when two or more alarm outputs indicate fail.
- at least two of the individual alarm level controllers are arranged to create at least three different alarm outputs, indicating pass, fail and warning, and then preferably the alarm voting logic controller is arranged to create an alarm condition indication of warning if one alarm output is warning and the rest of the alarm outputs are pass, and in that the alarm voting logic controller is arranged to create an alarm condition indication of fail if two or more alarm outputs are warning or at least one alarm output is fail.
- One signal path can be arranged to analyze the level of crackling, rumbling and hissing by means of a Kurtosis algorithm.
- One signal path can be arranged to analyze the level of harmonic energy there is by means of a harmonic activity index algorithm.
- One signal path can be arranged to analyze the level of screeching sounds by means of an analysis in the frequency domain.
- One signal path can be arranged to analyze the amount of possible damage by means of determining the level of total energy in different frequency spectrums.
- Fig. 1A shows a view of a condition monitoring unit according to the invention in a typical application monitoring a conveyor
- Fig. 1 B a cross section of a conveyor
- Fig. 2 shows a functional block diagram of a condition monitoring unit according to the invention
- Fig. 3 shows a functional block diagram of the signal processing and alarm generation according to the invention
- Fig. 4 shows a block diagram of an embodiment with four different signal processing units feeding an alarm voting controller according to the invention
- FIG 1A shows a view of a condition monitoring unit 100 according to the invention in a typical application monitoring a conveyor 1 10.
- a conveyor 1 10 will convey goods, such as gravel, on a conveyor belt 1 19 from a first place to a second place. The conveyed distance can be in the range of kilometers.
- the illustrated conveyor section 110 is of a trough conveyor and will typically require three idler rollers 112, 114, 116 approximately every meter along the length of the conveyor 110.
- One idler roller 114 is located directly underneath the belt 119 and one idler roller 112, 116 is located on each side, bending up the belt, to thereby create a trough.
- Figure 1B illustrates a cross section of such a conveyor.
- Each idler roller 112, 114, 116 along the whole conveyor will typically comprise two bearings, one on each end of an idler roller 112, 114, 116.
- a complete conveyor will then easily comprise more than a thousand bearings. It is these bearings that the condition monitoring unit 100 according to the invention is monitoring the condition of.
- a worker will walk along the length of the conveyor 110 and point the condition monitoring unit 100, or at least an acoustic sensor of the condition monitoring unit, towards the idler rollers 112, 114, 116 carrying the bearings of interest. It has been found that by analyzing sound pressure signals within the acoustic frequency range, an assessment of the condition of a bearing can be made.
- Acoustics cover frequencies from zero Hertz up to several mega Hertz and is usually subdivided into infrasound covering 0 Hz to about 20 Hz, sound covering about 20 Hz to 20 kHz and ultra sound being above 20 kHz up to several mega Hz.
- FIG. 2 illustrates a functional block diagram of a typical condition monitoring unit according to the invention.
- an acoustic sensor 250 typically a microphone, with a wide enough bandwidth to be able to sense sound pressure in the frequency range of interest.
- an acoustic sensor 250 for the purposes of the invention should have a bandwidth of 20 Hz to at least 40 kHz.
- the acoustic sensor 250 is coupled to a pre-processing block 272, which will suitably comprise amplification, analog to digital conversion, filtering and possibly automatic normalization. It is possible to implement the invention completely in the analog domain, but normally it is a desire to do as much processing as possible in the digital domain, thus putting the analog to digital conversion as close as possible to any analog input.
- FIG. 3 illustrates a functional block diagram of the signal processing 332, 334, 336, 338 and alarm generation 342, 344, 346, 348, 390 according to the invention.
- the invention relates to the use of more than one analysis of acoustic signals and then making an assessment of the outcome of the plurality of analyses to determine if there should be an alarm and if so, possibly also what kind of alarm.
- the illustrated example uses four different types of analysis/diagnosis of a captured acoustic signal and is thus illustrated with four different types of signal processing 332, 334, 336, 338.
- An acoustic sensor 350 will capture a sound pressure and convert this sound pressure into an electrical signal that is then pre-processed 372 into a format and range suitable for further processing.
- the pre-processing 372 may include amplification, automatic normalization, analog to digital conversion, preliminary filtering. It is to be noted that all signal processing can be in the analog domain, in the digital domain or a mixture of the two. Normally it is desirable to enter the digital domain as soon as possible and do all signal processing as digital signal processing.
- the signal processing 332, 334, 336, 338 is illustrated as being performed in parallel; it might just as well be performed sequentially.
- the signal processing might for example be directed to detect hissing, crackling and rumbling sounds due to, for example, abrasion, to detect the loudness of sounds associated with the amount of damage, to detect screeching sounds due to metal to metal contact, to detect whether a fault is associated with a bearing or not, and/or to detect ticking sounds due to bearing damage or debris on an idler roller.
- Each of the signal processing 232, 334, 336, 338 paths leads to a respective alarm level controller 342, 344, 346, 348 and for display of the value of each diagnosis also to a display 378.
- Each respective alarm level controller 342, 344, 346, 348 can be set by, for example, a keyboard input 376.
- the alarm level controllers 342, 344, 346, 348 can also have pre-set values, selected, for example, by selecting a specific conveyor in question or a conveyor type.
- the alarm levels can be of a binary type or of a multi-level type. A binary type alarm only has two levels: on or off, good or bad.
- a multi-level type alarm can give three or more levels, such as no alarm, alert alarm, and critical alarm.
- the alarm level outputs of the alarm level controllers 342, 344, 346, 348 go to an alarm voting controller 390.
- the alarm voting controller 390 can be preset and/or programmable in which way it is to deliver an alarm output to a user, suitably both via the display 378 and an audio output 380, 382.
- the preset can for example be by specific conveyor in question or by conveyor type.
- the voting controller 390 determines by how many and possibly by how serious the alarms are, if the user should be alerted or not.
- the voting controller 390 may require that at least two of the diagnoses gives an alarm, for it to alert a user, or if the alarm level controllers 342, 344, 346, 348 are set for multi level alarms, then require at least two alert alarms or one critical alarm to alert a user.
- the voting controller 390 might also react on one alarm input on its own, while another alarm input is only an indicator that must have another alarm input for the voting controller 390 to react and output an alarm.
- the voting controller 390 might also in turn have a binary output: no alarm or alarm, or a multi level output such as no alarm, alert alarm and a critical alarm output.
- These alarm outputs may be represented by different colors on the display 378, in the form of e.g. a traffic light, with green for no alarm, orange for an alert alarm and red for a critical alarm.
- the audio output with, for example, nothing for no alarm, intermediate level pulsating for an alert alarm and a high level continuous sound for a critical alarm.
- Figure 4 illustrates a block diagram of an embodiment with four different signal processing units 433, 435, 437, 439 feeding an alarm voting controller 491 according to the invention.
- a microphone/acoustic sensor 450 will detect any sound pressure and convert this pressure into an electrical signal that is fed into a pre-processing unit 472 which can comprise an amplifier, analog to digital converter, high pass filtering or band pass filtering. It has been determined that for diagnosing acoustic signals according to the invention, it is favorable to high pass filter the signals with a cut-off frequency somewhere between 3.5 kHz and 5 kHz. There are also advantages to low pass filter the signals with a cut off frequency about 40 kHz.
- Three of the illustrated signal processing paths utilize an enveloped 473 signal, two of these three do further processing in the frequency domain and therefore a Fourier transform 431 of the signal is done.
- the transformation from the time domain into the frequency domain is most commonly done by a Fast Fourier Transformation (FFT) 431. Since it is desirable to do as much as possible in the digital domain, it is possible that a re-sampling of the signal is performed before the Fourier transformation 431.
- the first of the signal processing paths determines how spiky the signal is by doing a Kurtosis 433. Kurtosis is a statistical analysis of the time data.
- the second of the signal processing paths determines how harmonic the spectrum of the signal is by a Harmonic Activity Index algorithm 435.
- this signal processing path uses the Fourier transformed 431 spectrum.
- the third signal processing path determines the total energy of the spectrum, that is in the frequency domain, by doing a Root Sum Squares (RSS) 437.
- RSS Root Sum Squares
- a Root Sum Squares 437 sums all of the data blocks vertically in the spectrum and squares the value of each, sums the results horizontally before finally taking the square root. This is the other signal processing path that works in the frequency domain and thus uses the Fourier transformed 431 spectrum.
- the fourth and final signal processing path in this example works in the time domain and is a peak detector 439.
- a peak detection algorithm determines the largest positive and negative peaks and returns the difference between them.
- An alarm signal will commonly always be coupled to an audio output with a user being able to switch between listening to the raw signal from the acoustic sensor 450 and the enveloped signal 473.
- FIGURE 1A shows a view of a condition monitoring unit according to the invention in a typical application monitoring a conveyor
- FIGURE 1 B shows a cross section of a conveyor
- FIGURE 2 shows a functional block diagram of a condition monitoring unit according to the invention
- FIGURE 3 shows a functional block diagram of the signal processing and alarm generation according to the invention
- FIGURE 4 shows a block diagram of an embodiment with four different signal processing units feeding an alarm voting controller according to the invention.
- HAI Harmonic Activity Index
Landscapes
- Physics & Mathematics (AREA)
- Acoustics & Sound (AREA)
- General Physics & Mathematics (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
A machine and/or a bearing is usually analyzed by several different methods, each one looking at a different aspect of a machine/bearing. This is at least in part due to the fact that we do not know what sort of failure it will be and what caused it. Different failures have different ways of breaking down and thus give different indications that something is not completely right. A machine/bearing condition monitoring unit will then preferably comprise many different analysis methods of trying to predict and detect failures. As a user it can be difficult to assess all the different alarm indications from the different methods. Therefore according to the invention, an alarm voting controller determines if an alarm should be made or not. The alarm voting controller analyzes the alarms from the individual methods and possibly how serious each one is and then outputs a single, for example, go/no go, pass/fail, or pass/warning/fail indication to a user.
Description
ACOUSTICAL MACHINE CONDITION MONITORING
TECHNICAL FIELD
The invention is related to condition monitoring of machines, especially the bearings of the machines, in particular in relation to bearings of idler rollers of for example conveyor belts.
BACKGROUND
Bearings are a very important component in rotating machinery. If a bearing fails, then the complete functionality of the machinery usually also fails. In some applications it might be very difficult or just extremely expensive to replace a failed bearing outside regular scheduled maintenance. Such applications might be continuous manufacturing lines, including conveyors of for example mines and cement plants.
A trough type conveyor will traditionally comprise a conveyor belt that is typically supported by three idler rollers for approximately every meter. Each idler roller will usually comprise two bearings each, giving six bearings per meter of conveyor. Other types of conveyors, such as pipe conveyors, will also comprise idler rollers for support. Conveyors in mines and cement plants can easily be several kilometers long, resulting in more than ten thousand bearings being incorporated in a single conveyor system. If a bearing of an idler roller fails, then most likely the idler roller will seize. A seized idler roller can fray and cut the conveyor belt. Conveyor belt damage can result in expensive belt repair/replacement, risk of workers' safety, and/or unplanned downtime if no parallel belt or buffer stock exists. Some plants can have as many as 5 to 8% of the conveyor idlers failing in a month.
Condition monitoring is done in an attempt to predict when a bearing needs to be replaced before it fails, suitably enabling replacement in an orderly scheduled manner. It is difficult to overcome the logistical problem of monitoring possibly more than ten thousand bearings of a single conveyor system. Traditionally condition monitoring has been done with a worker walking along the conveyor belt listening and looking for failing or failed conveyor idler rollers. It is hard to detect failing idlers over commonly a very high background noise level. A more advanced method of condition monitoring is to use a thermographic camera to detect failing idlers. Unfortunately it is very time consuming and only catches failing idlers at a very late stage of failure. Thus there seems to be room for improvement in the ways of assessing the condition of a bearing, especially a bearing of an idler roller.
SUMMARY
An object of the invention is to define a method and means to monitor the condition of a machine, suitably comprising at least one bearing, especially a bearing of an idler roller.
The aforementioned object is achieved according to the invention by the use of a machine condition monitoring unit according to the invention. A machine and/or a bearing is usually analyzed by several different methods, each one looking at a different aspect of a machine/bearing. This is at least in part due to the fact that we do not know what sort of failure it will be and what caused it. Different failures have different ways of breaking down and thus give different indications that something is not completely right. A machine/bearing condition monitoring unit will then preferably comprise many different methods of trying to predict and detect failures. As a user it can be difficult to assess all the different alarm indications from the different methods. Therefore according to the invention, an alarm voting controller determines if an alarm should be made or not. The alarm voting controller
analyzes the alarms from the individual methods and possibly how serious each one is and then outputs a single for example go/no go, pass/fail, or pass/warning/fail indication to a user. The aforementioned object is further achieved according to the invention by a machine condition monitoring unit comprising digital signal processing means and input means arranged to receive electrical signals from an acoustic sensor. According to the invention the digital signal processing means is arranged to analyze the electrical signals by at least two different methods of bearing analysis, each with an individual signal processing path, each path with an individual alarm level controller creating an alarm output indicating at least either pass or fail. Suitably the machine condition monitoring unit further comprises an alarm voting logic controller arranged to create one alarm condition indication of at least either pass or fail based on the outcome of all of individual alarm outputs of the individual alarm level controllers. In some embodiments the alarm voting logic controller is arranged to create an alarm condition indication of fail when two or more alarm outputs indicate fail. In other embodiments at least two of the individual alarm level controllers are arranged to create at least three different alarm outputs, indicating pass, fail and warning, and then preferably the alarm voting logic controller is arranged to create an alarm condition indication of warning if one alarm output is warning and the rest of the alarm outputs are pass, and in that the alarm voting logic controller is arranged to create an alarm condition indication of fail if two or more alarm outputs are warning or at least one alarm output is fail.
One signal path can be arranged to analyze the level of crackling, rumbling and hissing by means of a Kurtosis algorithm. One signal path can be arranged to analyze the level of harmonic energy there is by means of a harmonic activity index algorithm. One signal path can be arranged to analyze the level of screeching sounds by means of an analysis in the frequency domain. One signal path can be arranged to analyze the amount
of possible damage by means of determining the level of total energy in different frequency spectrums.
Other advantages of the invention will become apparent from the detailed description below.
BRIEF DESCRIPTION OF THE DRAWINGS
The invention will now be described in more detail for explanatory, and in no sense limiting, purposes, with reference to the following figures, in which:
Fig. 1A shows a view of a condition monitoring unit according to the invention in a typical application monitoring a conveyor,
Fig. 1 B a cross section of a conveyor,
Fig. 2 shows a functional block diagram of a condition monitoring unit according to the invention,
Fig. 3 shows a functional block diagram of the signal processing and alarm generation according to the invention,
Fig. 4 shows a block diagram of an embodiment with four different signal processing units feeding an alarm voting controller according to the invention,
DETAILED DESCRIPTION
In order to clarify the inventions, some examples of its use will now be described in connection with Figures 1A to 4
Figure 1A shows a view of a condition monitoring unit 100 according to the invention in a typical application monitoring a conveyor 1 10. A conveyor 1 10 will convey goods, such as gravel, on a conveyor belt 1 19 from a first place to a
second place. The conveyed distance can be in the range of kilometers. The illustrated conveyor section 110 is of a trough conveyor and will typically require three idler rollers 112, 114, 116 approximately every meter along the length of the conveyor 110. One idler roller 114 is located directly underneath the belt 119 and one idler roller 112, 116 is located on each side, bending up the belt, to thereby create a trough. Figure 1B illustrates a cross section of such a conveyor.
Each idler roller 112, 114, 116 along the whole conveyor will typically comprise two bearings, one on each end of an idler roller 112, 114, 116. A complete conveyor will then easily comprise more than a thousand bearings. It is these bearings that the condition monitoring unit 100 according to the invention is monitoring the condition of. Typically a worker will walk along the length of the conveyor 110 and point the condition monitoring unit 100, or at least an acoustic sensor of the condition monitoring unit, towards the idler rollers 112, 114, 116 carrying the bearings of interest. It has been found that by analyzing sound pressure signals within the acoustic frequency range, an assessment of the condition of a bearing can be made. Acoustics cover frequencies from zero Hertz up to several mega Hertz and is usually subdivided into infrasound covering 0 Hz to about 20 Hz, sound covering about 20 Hz to 20 kHz and ultra sound being above 20 kHz up to several mega Hz.
Figure 2 illustrates a functional block diagram of a typical condition monitoring unit according to the invention. There will be an acoustic sensor 250, typically a microphone, with a wide enough bandwidth to be able to sense sound pressure in the frequency range of interest. Typically an acoustic sensor 250 for the purposes of the invention should have a bandwidth of 20 Hz to at least 40 kHz. The acoustic sensor 250 is coupled to a pre-processing block 272, which will suitably comprise amplification, analog to digital conversion, filtering and possibly automatic normalization. It is possible to implement the invention completely in the analog domain, but normally it is a desire to do as much
processing as possible in the digital domain, thus putting the analog to digital conversion as close as possible to any analog input. For most further digital signal processing 274 it is advantageous that the signal is enveloped 273. The core digital signal processing 274 will have some user interface input 276, such as a keyboard, for changing parameters and limits. There will also be an output 278, such as a screen, to show the results of the analysis. In some applications it is desirable to have an audio output 280, for example in the form of headphones. Figure 3 illustrates a functional block diagram of the signal processing 332, 334, 336, 338 and alarm generation 342, 344, 346, 348, 390 according to the invention. As mentioned, the invention relates to the use of more than one analysis of acoustic signals and then making an assessment of the outcome of the plurality of analyses to determine if there should be an alarm and if so, possibly also what kind of alarm. The illustrated example uses four different types of analysis/diagnosis of a captured acoustic signal and is thus illustrated with four different types of signal processing 332, 334, 336, 338.
An acoustic sensor 350 will capture a sound pressure and convert this sound pressure into an electrical signal that is then pre-processed 372 into a format and range suitable for further processing. The pre-processing 372 may include amplification, automatic normalization, analog to digital conversion, preliminary filtering. It is to be noted that all signal processing can be in the analog domain, in the digital domain or a mixture of the two. Normally it is desirable to enter the digital domain as soon as possible and do all signal processing as digital signal processing.
After the pre-processing 372 the signal will enter the illustrated four different signal processing 332, 334, 336, 338 paths for diagnosis of the condition of relevant bearings. The signal processing 332, 334, 336, 338 is illustrated as being performed in parallel; it might just as well be performed sequentially. The signal processing might for example be directed to detect hissing, crackling and
rumbling sounds due to, for example, abrasion, to detect the loudness of sounds associated with the amount of damage, to detect screeching sounds due to metal to metal contact, to detect whether a fault is associated with a bearing or not, and/or to detect ticking sounds due to bearing damage or debris on an idler roller. Each of the signal processing 232, 334, 336, 338 paths leads to a respective alarm level controller 342, 344, 346, 348 and for display of the value of each diagnosis also to a display 378. Each respective alarm level controller 342, 344, 346, 348 can be set by, for example, a keyboard input 376. The alarm level controllers 342, 344, 346, 348 can also have pre-set values, selected, for example, by selecting a specific conveyor in question or a conveyor type. The alarm levels can be of a binary type or of a multi-level type. A binary type alarm only has two levels: on or off, good or bad. A multi-level type alarm can give three or more levels, such as no alarm, alert alarm, and critical alarm.
The alarm level outputs of the alarm level controllers 342, 344, 346, 348 go to an alarm voting controller 390. The alarm voting controller 390 can be preset and/or programmable in which way it is to deliver an alarm output to a user, suitably both via the display 378 and an audio output 380, 382. The preset can for example be by specific conveyor in question or by conveyor type. The voting controller 390 determines by how many and possibly by how serious the alarms are, if the user should be alerted or not. The voting controller 390 may require that at least two of the diagnoses gives an alarm, for it to alert a user, or if the alarm level controllers 342, 344, 346, 348 are set for multi level alarms, then require at least two alert alarms or one critical alarm to alert a user. The voting controller 390 might also react on one alarm input on its own, while another alarm input is only an indicator that must have another alarm input for the voting controller 390 to react and output an alarm. The voting controller 390 might also in turn have a binary output: no alarm or alarm, or a multi level output such as no alarm, alert alarm and a critical alarm output. These alarm outputs may be represented by different colors on
the display 378, in the form of e.g. a traffic light, with green for no alarm, orange for an alert alarm and red for a critical alarm. Likewise for the audio output with, for example, nothing for no alarm, intermediate level pulsating for an alert alarm and a high level continuous sound for a critical alarm.
Figure 4 illustrates a block diagram of an embodiment with four different signal processing units 433, 435, 437, 439 feeding an alarm voting controller 491 according to the invention. A microphone/acoustic sensor 450 will detect any sound pressure and convert this pressure into an electrical signal that is fed into a pre-processing unit 472 which can comprise an amplifier, analog to digital converter, high pass filtering or band pass filtering. It has been determined that for diagnosing acoustic signals according to the invention, it is favorable to high pass filter the signals with a cut-off frequency somewhere between 3.5 kHz and 5 kHz. There are also advantages to low pass filter the signals with a cut off frequency about 40 kHz.
Three of the illustrated signal processing paths utilize an enveloped 473 signal, two of these three do further processing in the frequency domain and therefore a Fourier transform 431 of the signal is done. The transformation from the time domain into the frequency domain is most commonly done by a Fast Fourier Transformation (FFT) 431. Since it is desirable to do as much as possible in the digital domain, it is possible that a re-sampling of the signal is performed before the Fourier transformation 431. The first of the signal processing paths determines how spiky the signal is by doing a Kurtosis 433. Kurtosis is a statistical analysis of the time data. The second of the signal processing paths determines how harmonic the spectrum of the signal is by a Harmonic Activity Index algorithm 435. Since this algorithm works on the spectrum of the signal, that is in the frequency domain, this signal processing path uses the Fourier transformed 431 spectrum.
The third signal processing path determines the total energy of the spectrum, that is in the frequency domain, by doing a Root Sum Squares (RSS) 437. A Root Sum Squares 437 sums all of the data blocks vertically in the spectrum and squares the value of each, sums the results horizontally before finally taking the square root. This is the other signal processing path that works in the frequency domain and thus uses the Fourier transformed 431 spectrum.
The fourth and final signal processing path in this example works in the time domain and is a peak detector 439. A peak detection algorithm determines the largest positive and negative peaks and returns the difference between them.
The outputs of each of these signal processing paths 433, 435, 437, 439 are then fed to alarm level controllers and an alarm voting controller for further processing 491 as has been previously described. Input and output devices have not been illustrated.
It is advantageous for a user to be able to listen to the signals that the condition monitoring unit according to the invention gets subjected to. An alarm signal will commonly always be coupled to an audio output with a user being able to switch between listening to the raw signal from the acoustic sensor 450 and the enveloped signal 473.
The invention is not restricted to the above-described embodiments, but may be varied within the scope of the following claims.
FIGURE 1A shows a view of a condition monitoring unit according to the invention in a typical application monitoring a conveyor,
100 A condition monitoring unit according to the invention
110 Conveyor
112 Right side idler roller,
114 Bottom idler roller,
116 Left side roller,
119 Conveyor belt.
FIGURE 1 B shows a cross section of a conveyor,
112 Right side idler roller,
114 Bottom idler roller,
116 Left side roller,
119 Conveyor belt.
FIGURE 2 shows a functional block diagram of a condition monitoring unit according to the invention,
250 Acoustic sensor unit/microphone,
272 Amplification, A/D conversion, Filtering,
273 Enveloping,
274 Further digital signal processing,
276 Input/keyboard,
278 Output/Screen,
280 Optional output/headphones,
FIGURE 3 shows a functional block diagram of the signal processing and alarm generation according to the invention,
332 First signal processing,
334 Second signal processing,
336 Third signal processing,
338 Fourth signal processing,
342 First alarm level controller,
344 Second alarm level controller,
346 Third alarm level controller,
348 Fourth alarm level controller,
350 Microphone/acoustic sensor unit,
372 Amplification and preprocessing,
376 Input device, keyboard,
378 Output device, display,
380 Headphones,
382 Headphone amplifier,
390 Alarm voting controller.
FIGURE 4 shows a block diagram of an embodiment with four different signal processing units feeding an alarm voting controller according to the invention.
431 Fast Fourier Transform (FFT)
433 Kurtosis
435 Harmonic Activity Index (HAI)
437 Root Sum Squares (RSS)
439 Peak Detection
450 Microphone/acoustic sensor unit,
472 High Pass Filter, Band Pass Filter,
473 Enveloping,
480 Headphones,
482 Headphone amplifier,
491 Alarm Voting Controller, Output/Display, Input/Keyboard
Claims
A machine condition monitoring unit comprising digital signal processing means and input means arranged to receive electrical signals from an acoustic sensor, characterized in that the digital signal processing means is arranged to analyze the electrical signals by at least two different methods of bearing analysis, each with an individual signal processing path, each path with an individual alarm level controller creating an alarm output indicating at least either pass or fail.
A machine condition monitoring unit according to claim 1 characterized in that the machine condition monitoring unit further comprises an alarm voting logic controller arranged to create one alarm condition indication of at least either pass or fail based on the outcome of all of individual alarm outputs of the individual alarm level controllers.
A machine condition monitoring unit according to claim 2 characterized in that the alarm voting logic controller is arranged to create an alarm condition indication of fail when two or more alarm outputs indicate fail.
4. A machine condition monitoring unit according to claim 1 or 2 characterized in that at least two of the individual alarm level controllers are arranged to create at least three different alarm outputs, indicating pass, fail and warning.
A machine condition monitoring unit according to claim 4 characterized in that the alarm voting logic controller is arranged to create an alarm condition indication of warning if one alarm output is warning and the rest of the alarm outputs are pass, and in that the alarm voting logic controller is arranged to create an alarm condition indication of fail if two or more alarm outputs are warning or at least one alarm output is fail.
A machine condition monitoring unit according to any one of claims 1 to 5 characterized in that one signal path is arranged to analyze the level of crackling, rumbling and hissing by means of a Kurtosis algorithm.
A machine condition monitoring unit according to any one of claims 1 to 6 characterized in that one signal path is arranged to analyze the level of harmonic energy there is by means of a harmonic activity index algorithm.
A machine condition monitoring unit according to any one of claims 1 to 7 characterized in that one signal path is arranged to analyze the level of screeching sounds by means of an analysis in the frequency domain.
A machine condition monitoring unit according to any one of claims 1 to 8 characterized in that one signal path is arranged to analyze the amount of possible damage by means of determining the level of total energy in different frequency spectrums.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/EP2010/003728 WO2011160652A1 (en) | 2010-06-21 | 2010-06-21 | Acoustical machine condition monitoring |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
PCT/EP2010/003728 WO2011160652A1 (en) | 2010-06-21 | 2010-06-21 | Acoustical machine condition monitoring |
Publications (2)
Publication Number | Publication Date |
---|---|
WO2011160652A1 true WO2011160652A1 (en) | 2011-12-29 |
WO2011160652A8 WO2011160652A8 (en) | 2012-04-19 |
Family
ID=43598483
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/EP2010/003728 WO2011160652A1 (en) | 2010-06-21 | 2010-06-21 | Acoustical machine condition monitoring |
Country Status (1)
Country | Link |
---|---|
WO (1) | WO2011160652A1 (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
DE102017209310A1 (en) * | 2017-06-01 | 2018-12-06 | Takraf Gmbh | Conveyor belt system and system for monitoring rollers of the conveyor belt system |
CN112897262A (en) * | 2021-02-26 | 2021-06-04 | 浙江理工大学 | Elevator running state evaluation system and method based on sound feature extraction |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4493042A (en) * | 1979-04-16 | 1985-01-08 | Mitsubishi Denki Kabushiki Kaisha | Bearing failure judging apparatus |
EP1431727A2 (en) * | 1999-03-13 | 2004-06-23 | Textron Systems Corporation | Method and apparatus for monitoring rotating machinery and estimating torque therein |
WO2004059399A2 (en) * | 2002-12-30 | 2004-07-15 | Rsl Electronics Ltd. | Method and system for diagnostics and prognostics of a mechanical system |
WO2007147611A1 (en) * | 2006-06-23 | 2007-12-27 | Ab Skf | Intrinsically safe vibration and condition monitoring system and the parts thereof |
-
2010
- 2010-06-21 WO PCT/EP2010/003728 patent/WO2011160652A1/en active Application Filing
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4493042A (en) * | 1979-04-16 | 1985-01-08 | Mitsubishi Denki Kabushiki Kaisha | Bearing failure judging apparatus |
EP1431727A2 (en) * | 1999-03-13 | 2004-06-23 | Textron Systems Corporation | Method and apparatus for monitoring rotating machinery and estimating torque therein |
WO2004059399A2 (en) * | 2002-12-30 | 2004-07-15 | Rsl Electronics Ltd. | Method and system for diagnostics and prognostics of a mechanical system |
WO2007147611A1 (en) * | 2006-06-23 | 2007-12-27 | Ab Skf | Intrinsically safe vibration and condition monitoring system and the parts thereof |
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
DE102017209310A1 (en) * | 2017-06-01 | 2018-12-06 | Takraf Gmbh | Conveyor belt system and system for monitoring rollers of the conveyor belt system |
AU2018203717B2 (en) * | 2017-06-01 | 2019-08-15 | Takraf Gmbh | Conveyor belt system and system for monitoring the rolls of the conveyor belt system |
US10486910B2 (en) | 2017-06-01 | 2019-11-26 | Takraf Gmbh | Conveyor belt system and system for monitoring the rolls of the conveyor belt system |
CN112897262A (en) * | 2021-02-26 | 2021-06-04 | 浙江理工大学 | Elevator running state evaluation system and method based on sound feature extraction |
Also Published As
Publication number | Publication date |
---|---|
WO2011160652A8 (en) | 2012-04-19 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US10175070B2 (en) | Selective decimation and analysis of oversampled data | |
KR101491017B1 (en) | System and method for diagnosing sound of power device | |
KR101608275B1 (en) | Power equipment acoustic diagnostic system | |
KR20090010430A (en) | Apparatus for detecting mechanical trouble | |
US10311703B1 (en) | Detection of spikes and faults in vibration trend data | |
KR101663820B1 (en) | Method and apparatus of machine diagnosis using sound signal | |
WO2011160652A1 (en) | Acoustical machine condition monitoring | |
JP2015114294A (en) | Inspection apparatus of acoustic device and inspection method of acoustic device and inspection program of acoustic device | |
DE102016208002B4 (en) | Method and device for detecting errors on rotating knives | |
CN113376457B (en) | Equipment operation state detection method, system, device, equipment and medium | |
KR101641269B1 (en) | Partial discharge diagnostic system using emf sensor and acoustic sensor | |
CN102759905B (en) | Method and apparatus for characterizing process control facility integrality | |
KR101180687B1 (en) | System and method for quality management using sound detection | |
JP6158439B2 (en) | Detection of cleaning processes in plants with filters arranged spatially offset from each other | |
CN103604605A (en) | Portable swivel bearing fault diagnostic device based on ultrasonic information | |
JP4658743B2 (en) | Vibration monitoring system | |
JP3713860B2 (en) | Conveyor abnormality detection method and apparatus | |
CN201843767U (en) | Failure audio frequency analysis and diagnosis device for turbine type compressor | |
KR102230463B1 (en) | Diagnosis system and method of defect of equipment component | |
WO2007020767A1 (en) | Acoustic sensor device, acoustic analysis diagnosis device, and acoustic analysis diagnosis device manufacturing method | |
JPH11118593A (en) | Plant failure monitoring device | |
CN110307899A (en) | Sound anomaly detection system based on deep learning | |
CN112268960A (en) | Defect analysis and early warning method for strain clamp of power transmission line | |
WO2011125233A1 (en) | Abnormality detecting device for equipment | |
KR102684336B1 (en) | Smart safety management system |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 10726443 Country of ref document: EP Kind code of ref document: A1 |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 10726443 Country of ref document: EP Kind code of ref document: A1 |