CN100516789C - Abnormality diagnosis system for machinery - Google Patents

Abnormality diagnosis system for machinery Download PDF

Info

Publication number
CN100516789C
CN100516789C CNB2005800062882A CN200580006288A CN100516789C CN 100516789 C CN100516789 C CN 100516789C CN B2005800062882 A CNB2005800062882 A CN B2005800062882A CN 200580006288 A CN200580006288 A CN 200580006288A CN 100516789 C CN100516789 C CN 100516789C
Authority
CN
China
Prior art keywords
signal
frequency
vibration
diagnosis
unit
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CNB2005800062882A
Other languages
Chinese (zh)
Other versions
CN1926413A (en
Inventor
佐原淳太郎
武藤泰之
宫坂孝范
山添正信
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
NSK Ltd
Original Assignee
NSK Ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Family has litigation
First worldwide family litigation filed litigation Critical https://patents.darts-ip.com/?family=36381618&utm_source=google_patent&utm_medium=platform_link&utm_campaign=public_patent_search&patent=CN100516789(C) "Global patent litigation dataset” by Darts-ip is licensed under a Creative Commons Attribution 4.0 International License.
Application filed by NSK Ltd filed Critical NSK Ltd
Publication of CN1926413A publication Critical patent/CN1926413A/en
Application granted granted Critical
Publication of CN100516789C publication Critical patent/CN100516789C/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Abstract

An abnormality diagnosis system for machinery comprises an envelope processing section (103) for determining the envelope of a detection signal, an FFT section (104) for converting the envelope into a frequency spectrum, a peak detecting unit (105) in which the frequency spectrum is smoothed by the moving average method, the resultant spectrum is subjected to a smoothed differentiation, frequency points at which the sign of differential coefficient changes from positive to negative are detected as peaks, peaks higher a predetermined threshold are extracted and sorted, and the higher ones are detected as peaks, and a diagnosis section (T) for performing abnormality diagnosis on the basis of the detected peaks.

Description

The abnormity diagnostic system of plant equipment
Technical field
The present invention relates to the abnormity diagnosis technology that rail truck, aviation machine, wind power generation plant, lathe, automobile, system iron machinery, papermaking equipment, rotating machinery etc. comprise the plant equipment of bearing, in more detail, relate to by analyzing the sound that takes place from plant equipment or vibration and diagnose the abnormity diagnosis technology of the unusual plant equipment of bearing in this plant equipment or the related member of bearing.
Background technology
In the past, as this abnormity diagnosis technology, known following technology, promptly detect slide unit or the sound of slide unit related member or the signal of vibration of expression from plant equipment, ask the frequency spectrum of detected signal or its envelope signal, from this frequency spectrum, only extract the frequency component that causes unusually, judge that according to the size of the frequency component that extracts no abnormal (with reference to patent documentation 1) arranged in the employed slide unit of plant equipment by the related member of slide unit of the slide unit of plant equipment or plant equipment.
In addition, known following technology, promptly detect from the sound or the vibration of rotor or the related member generation of rotor, from detected signal, take out the signal of the necessary frequency band of diagnosis, and then ask the envelope of the signal that has taken out, the envelope of obtaining is carried out frequency resolution, ask size by frequency resolution by the frequency component of the size of the fundamental component of the frequency that causes unusually of the related member of rotary body or rotary body and its natural several times, size to the frequency component of the size of the fundamental component obtained and its natural several times compares, to this comparative result of major general as the unusual benchmark (with reference to patent documentation 2) of judging plant equipment.
In addition, known following technology, the simulating signal that is about to the sound that takes place from plant equipment or vibration by A/D (analog/digital) thus conversion is transformed to digital signal to be generated and surveys numerical data, thereby this actual measurement numerical data is carried out suitably dissection process generation actual measurement frequency spectrum data such as frequency analysis and envelope analysis, according to having or not, carry out for the N/R judgement of having of plant equipment (with reference to patent documentation 3) for peak value by the actual measurement frequency spectrum data of 1 rank of the frequency component that causes unusually of plant equipment, 2 rank, 4 rank values.
In addition, known following technology, with the envelope waveform transformation of vibration acceleration is digital signal, ask digitizing the rumble spectrum of each time of vibration data distribute, the rotating speed of the rolling bearing when constantly asking vibration measurement simultaneously, time variation diagram shape in the frequency of the time variation diagram shape of rotating speed and peaks spectrum during rumble spectrum distributes is consistent, and arbitrarily the time the frequency of the peaks spectrum situation consistent with the characteristic frequency of the rolling bearing damage of obtaining according to the geometry size of the rotating speed of rolling bearing and rolling bearing under, damage (with reference to patent documentation 4) has taken place in the privileged site that is judged to be at rolling bearing.
Clearly put down in writing the method for the peak value that detects the unusual frequency of expression in these patent documentations, but under the unusual situation of the rotating shaft eccentric etc. of peeling off life-span or machinery that bearing has taken place, represent that the peak value of the frequency of these anomalous signals (abnormal signal) can on average come easily to obtain according to the accumulative total of frequency spectrum.Accumulative total on average is owing to the method in the field that removing of random noise effectively is widely used in the analysis of fast Fourier transform (FFT) parsing equifrequent.
In addition, in these prior aries, the processing (envelope processing) of asking envelope signal is simulation process or digital processing, but frequency resolution is handled the fast Fourier transform (FFT) processing of using digital processing.For carrying out the FFT computing, before or after handling, envelope carries out the A/D conversion.And, in any prior art, all after handling, envelope carries out the FFT computing immediately.
In the mode of carrying out the envelope processing by simulation process, need the envelope processing unit.Thereby in the cost reduction and miniaturization of the system of realization, the mode of carrying out the envelope processing with digital processing is favourable.
In the mode of carrying out the envelope processing with digital processing,, consider the method for the efficient of raising FFT computing as the method that improves abnormity diagnosis efficient.The efficient of FFT computing improves and can realize by reducing counting of FFT computing.
And then known utilization vibration (comprising the vibration of sound equipment) detects the apparatus for diagnosis of abnormality of the axletree of rail truck with the damage of bearing or wheel.Existing this apparatus for diagnosis of abnormality is provided with vibration transducer respectively in each axle box, the damage of each bearing or wheel is detected (with reference to patent documentation 5, patent documentation 6 etc.).
In the past, the rotary part of rail truck used certain during after, checked termly that for vehicle bearing and other rotary part not damaged and wearing and tearing etc. are unusual.This regular inspection is decomposed by the mechanical hook-up that will assemble rotary part and is carried out, and finds the damage and the wearing and tearing that take place on the rotary part by the visual inspection of operator.And, as found main defective in checking, under the situation of bearing, tired the peeling off and other wearing and tearing etc. of causing of impression, the rotation of generations such as foreign matter is engaging-in arranged, under the situation of gear, the damaged of tooth portion and wearing and tearing etc. are arranged, under the situation of wheel, the wearing and tearing of flat grade are arranged, under any situation, if new product just is replaced by in unexistent concavo-convex or wearing and tearing etc. on the discovery new product.
But, plant equipment integral body is decomposed and method by operator's visual examination in, taking off the assembling operation that the apportioned effort of rotary body and slide unit, the rotary body that will check and slide unit re-assembly the device once more from device needs very big labour, has the problem that significantly increases of the maintenance cost that causes device.
In addition, exist when re-assemblying rotary body and slide unit are produced the possibility that the percussion mark that do not have before checking etc., inspection itself become the defective of rotary body or slide unit.In addition since in the limited time by a plurality of bearings of visual examination, therefore also have the problem of the possibility of leaking the discovery defective.And then also there is individual differences in the judgement of the degree of this defective, even there is not defective to carry out part replacement in fact yet, and therefore consuming cost in vain.
Therefore, propose a kind of mechanical hook-up of having assembled rotary part the decomposition and under the real-world operation state, be rotated the abnormality diagnostic method (for example, patent documentation 1,7 and 8) of parts.As the most general method, put down in writing as patent documentation 1, known bearing portions is provided with accelerometer, the vibration acceleration of metering bearing portions, and then this signal carried out FFT (fast fourier transform) thus handle and extract the method that the signal of vibration occurrence frequency component is diagnosed.
In addition, as on the plane of rotation of the wheel of rail truck, the breech lock of the wheel that causes by maloperation of detent etc. or slide that cause and friction track, wearing and tearing and the detection method of the flat of being known as of producing flat (flat), various schemes (for example, with reference to patent documentation 6,9 and 10) have also been proposed.Especially in patent documentation 6, the device that detects for the defect state of the circuit that railroad carriage wheel and train is passed through by vibration transducer and rotation determinator etc. has carried out motion.
Patent documentation 1: the spy opens the 2003-202276 communique
Patent documentation 2: the spy opens the 2003-232674 communique
Patent documentation 3: the spy opens the 2003-130763 communique
Patent documentation 4: the spy opens flat 09-113416 communique
Patent documentation 5: the spy opens flat 4-235327 communique
Patent documentation 6: the spy opens flat 9-500452 communique
Patent documentation 7: the spy opens the 2002-22617 communique
Patent documentation 8: the spy opens the 2004-257836 communique
Patent documentation 9: the spy opens flat 4-148839 communique
Patent documentation 10: special table 2003-535755 communique
But, in vibration transducer and the acoustic sensor, because under the situation from the impulsive sound of outside or grating, vibrating mass, the acceleration effect that revolution causes is therefore because these non-stable interference are more by the situation that flase drop is surveyed unusually.Therefore by accumulative total on average to the detection method of the peak value of frequency when cumulative number increases, owing to be subjected to the variation of speed easily or from the influence of the impulsive sound of outside etc., so also do not imitate sometimes.
In addition, the little wound before arriving the life-span, peel off, under the unusual situation that rust etc. causes, little under a lot of situations of power from the signal of vibration transducer or acoustic sensor to the degree that is buried in mechanical noise or electrical noise.Therefore the predicting abnormality the life-span before is in the stage, much do not use under the situations threshold value is set only extracts the method that is worth high-power signal than this.Carrying out in the unusual prediction, the problem of trouble is, abnormal signal like this or represent the signal (unusual omen signal) of unusual omen and the S/N of noise signal than little situation under, the noise signal mistake is judged as abnormal signal or omen signal unusually.It is favourable on the accuracy of the predicting abnormality that improves bearing etc. that minimum abnormal signal or unusual omen signal are also leaked situation about finding, but its result, if the noise signal erroneous judgement is decided to be abnormal signal or unusual omen signal, then continually plant equipment is shut down and check, therefore cause the increase of operating cost.
In addition, reduce counting when improving counting yield of FFT computing, there is the problem that causes that abnormality diagnostic precision reduces in the frequency resolution variation.
And then, in the mode of carrying out the envelope processing with digital processing,, consider the method for the efficient of raising FFT computing as the method that improves abnormity diagnosis efficient.The efficient of FFT computing improves and can realize by reducing counting of FFT computing.But, reduce counting when improving counting yield of FFT computing, there is the problem that causes that abnormality diagnostic precision reduces in the frequency resolution variation.
In rotating machinery, be used to diagnose that its size of unusual arithmetic unit that bearing defect etc. causes or consumed power are more little then is suitable for assembling more and uses.In addition, from the aspect of computational accuracy or the aspect of memory span, require to count and carry out FFT with few computing.But, on the other hand, as mentioned above, if frequency resolution is not high to a certain degree then cause that abnormality diagnostic precision reduces.Even the frequency that can recover original waveform need be made as below the 10kHz (sample frequency is more than the 20kHz), the upper limit of the defect frequency of bearing also is below the 1kHz as a result.
But, in apparatus for diagnosis of abnormality in the past, because vibration transducer need be set respectively in each axle box, therefore the number that is provided with of the sensor of each vehicle becomes many, the number that is used for the input circuit of signal processing unit of processes sensor signal and wiring is huge, has the circuit structure complicated problems.
But in the pick-up unit of defect state of record, it is that the flat of vehicle causes that existence can not be discerned abnormal vibrations in patent documentation 6, or axle bearing causes, or circuit or other cause unusually.
Summary of the invention
The present invention In view of the foregoing finishes; its purpose is to provide a kind of abnormity diagnostic system of plant equipment; under the little condition of the S/N of abnormal signal or unusual omen signal and noise signal, also can implement abnormity diagnosis accurately and the noise signal flase drop can not surveyed is unusual or unusual omen signal.
The present invention In view of the foregoing finishes, its purpose is to provide a kind of abnormity diagnostic system of plant equipment, take into account from the raising of the frequency resolution of the signal of plant equipment and the efficient of FFT computing and improve, can high precision and implement abnormity diagnosis expeditiously.
The present invention In view of the foregoing finishes, and its purpose is to provide a kind of abnormity diagnostic system of plant equipment, thereby can implement abnormity diagnosis with frequency resolution arbitrarily accurately to carrying out FFT from the detected signal of diagnosis object.
The present invention In view of the foregoing finishes, its purpose is to provide a kind of abnormity diagnostic system, every vehicle only is provided with a vibration transducer, can be based on waveform signal from this vibration transducer, detect the flat etc. unusual of the peeling off of bearing in this vehicle, wheel.
The present invention In view of the foregoing finishes, its purpose is to provide a kind of abnormity diagnostic system, according to coming the abnormal vibrations of inspection vehicle axle bearing and wheel, thereby determine that this abnormal vibrations is that flat that cause or the axle bearing of wheel causes from output signal to the vibration transducer of the vibration of axle bearing or wheel.
To achieve these goals, the abnormity diagnostic system of plant equipment of the present invention is feature with following (1) to (3).
(1) a kind of abnormity diagnostic system by detecting sound or the vibration that takes place from plant equipment, and is analyzed its detection signal, thereby bearing in the diagnosis plant equipment or the related member of bearing is unusual, and this system comprises:
The envelope processing unit is asked the envelope of described detection signal;
The FFT unit will be transformed to frequency spectrum by the envelope that this envelope processing unit obtains;
Peak detection unit carries out smoothing by the frequency spectrum that is obtained by this FFT unit is carried out the moving average processing, thereby detects its peak value; And
Diagnosis unit comes diagnosing abnormal based on the peak value by the detected frequency spectrum of described peak detection unit.
(2) in the abnormity diagnostic system of the structure of above-mentioned (1), described peak detection unit comprises smoothing differential peak extraction unit, this unit is implemented the smoothing differential to the frequency spectrum that obtained by described FFT unit and is handled, and the Frequency point of the sign change of the differential value that obtains is extracted peak value as frequency spectrum.
(3) in the abnormity diagnostic system of the structure of above-mentioned (1) or (2), the weighting coefficient in the described moving average processing is left-right symmetric (with the current time is benchmark, the front and back object).
(4) in the abnormity diagnostic system of the structure of above-mentioned (2) or (3), described peak detection unit comprises first module of selection, the peak value in the peak value that this unit selection is extracted by described smoothing differential peak extraction unit more than the threshold value.
(5) in the abnormity diagnostic system of the structure of above-mentioned (4), described peak detection unit comprises second module of selection, and the peak value up to the regulation number is selected in this unit the big peak value from amplitude levels in the peak value of being selected by described first module of selection.
(6) in the abnormity diagnostic system of any structure of above-mentioned (1)~(5), described diagnosis unit is by asking the consistent degree by principal component and the pairing peak value of high order component and the unusual frequency of expression diagnosis object of pairing peak value of principal component that vibrates in the detected peak value of described peak detection unit or vibration, and the accumulated result repeatedly of this unanimity degree estimated, thereby diagnosing abnormal.
(7) a kind of abnormity diagnostic system of plant equipment, by detecting sound or the vibration that takes place from plant equipment, and its detection signal analyzed, thereby diagnose the unusual of interior bearing of this plant equipment or the related member of bearing, this system comprises: filter processing unit, from described detected signal, take out the signal of diagnosing required frequency band; The envelope processing unit is asked the envelope signal by the signal of this filter processing unit taking-up; Select processing unit, the envelope signal that is obtained by this envelope processing unit is extracted processing;
The FFT arithmetic element is carried out frequency resolution to selected extraction that processing unit the carries out envelope signal after handling by this; And diagnosis unit, come diagnosing abnormal based on the analysis result of this FFT arithmetic element.
(8) a kind of abnormity diagnostic system of plant equipment, by detecting sound or the vibration that takes place from plant equipment, and its detection signal analyzed, thereby diagnose the unusual of interior bearing of this plant equipment or the related member of bearing, this system comprises: sample processing unit, with the sample frequency higher described detected signal is sampled in advance than the sample frequency of necessity; Filter processing unit is taken out the signal of diagnosing required frequency band from the signal of being sampled out by this sample processing unit; Select processing unit, the signal that is taken out by this filter processing unit is extracted processing; The envelope processing unit is asked by this and is selected the envelope signal that processing unit has carried out extracting the signal of handling; The FFT arithmetic element is carried out frequency resolution to the envelope signal that is obtained by this envelope processing unit; And diagnosis unit, come diagnosing abnormal based on the analysis result of this FFT arithmetic element.
(9) in the abnormity diagnostic system of the structure of above-mentioned (7) or (8), this system also comprises the digital filtering processing unit of the frequency band low-frequency bandization that makes described envelope signal.
(10) in the abnormity diagnostic system of the structure of above-mentioned (7), (8) or (9), realize described FFT arithmetic element by DSP, the data number that will import described FFT arithmetic element simultaneously is made as the data number in the storer that can be contained in this DSP.
(11) a kind of abnormity diagnostic system of plant equipment, by detecting sound or the vibration that takes place from plant equipment, and this signal analyzed, thereby the related member of bearing in the diagnosis plant equipment or bearing is unusual, it is characterized in that, this system comprises: the A/D converter unit is a digital signal with described signal transformation; The digital filtering processing unit is from the signal by the required frequency band of taking-up diagnosis the digital signal of this A/D converter unit conversion; The envelope processing unit is asked the envelope by the signal of this digital filtering processing unit taking-up; Interpolation processing unit carries out 0 cover interpolation, so that with frequency resolution arbitrarily the envelope of being obtained by this envelope processing unit is carried out fast fourier transform; The FFT unit carries out fast fourier transform to the signal that has been carried out 0 cover interpolation by this interpolation processing unit; And diagnosis unit, come diagnosing abnormal based on the frequency spectrum that obtains by this FFT unit.
(12) in the abnormity diagnostic system of the structure of above-mentioned (11), described interpolation processing unit carries out 0 cover interpolation, so that the sample frequency in the described FFT unit becomes the multiple hertz of the Nth power of 2 Nth power hertz or 2.
(13) in the abnormity diagnostic system of the structure of above-mentioned (11) or (12), this system also comprises the peak detection unit that the peak value to the frequency spectrum that is obtained by described FFT unit detects, described diagnosis unit is by asking the consistent degree by principal component and the pairing peak value of high order component and the unusual frequency of expression diagnosis object of pairing peak value of principal component that vibrates in the detected peak value of described peak detection unit or vibration, and the accumulated result repeatedly of this unanimity degree estimated, thereby diagnosing abnormal.
(14) a kind of apparatus for diagnosis of abnormality, in the middle diagnosing abnormal of travelling of vehicle, this device comprises: vibration transducer, detect the vibration of vehicle; The parameter value testing circuit, based on the waveform signal of described vibration transducer output, asking crest factor, impacting one of them dimension in index, form factor and the kurtosis is 1 parameter value; And comparator circuit, output expression is first voltage signal of 1 the parameter value situation that surpasses certain benchmark or represents that described dimension is that 1 parameter value is second voltage signal of the situation below certain benchmark from the dimension of described parameter value testing circuit output, detects unusually based on the output of described comparator circuit.
(15) a kind of apparatus for diagnosis of abnormality, in the middle diagnosing abnormal of travelling of vehicle, this device comprises: vibration transducer, detect the vibration of vehicle; Computing circuit based on the waveform signal of described vibration transducer output, is asked the parameter value of average one of them of RMS (square root of Fang Jun) and absolute value; Peak detection circuit is asked the peak value of described waveform signal; And comparator circuit, certain times value of described parameter value and the peak value of exporting from described peak detection circuit are compared, according to its comparative result, the dimension that expression is obtained as the ratio of described peak value and described parameter value is first voltage signal of 1 the parameter value situation that surpasses certain benchmark or represents that described dimension is that 1 parameter value is second voltage signal output of the situation below certain benchmark, detects unusually based on the output of described comparator circuit.
(16) in the abnormity diagnostic system of the structure of above-mentioned (15), this device also comprises peak value-reference point comparator circuit that peak value and predefined reference point from described peak detection circuit output are compared, the comparative result of described peak value-reference point comparator circuit, under the situation of described peak value, make the output of described comparator circuit invalid greater than described reference point.
(17) in the abnormity diagnostic system of the arbitrary structures of above-mentioned (14)~(16), recently detect unusually based on the duty of described first voltage signal.
(18) in the abnormity diagnostic system of the arbitrary structures of above-mentioned (15)~(17), this device also comprises filtering circuit, be used for the output signal of described vibration transducer only the signal of allocated frequency band be input to described parameter value testing circuit and described peak detection circuit.
(19) in the abnormity diagnostic system of the arbitrary structures of above-mentioned (14)~(18), described vehicle is a rail truck.
(20) a kind of apparatus for diagnosis of abnormality, it is apparatus for diagnosis of abnormality with mechanical hook-up of the different a plurality of parts of vibration characteristics, it is characterized in that, this apparatus for diagnosis of abnormality comprises: the sensor signal processing unit, the output signal of the vibration transducer of the vibration that detects described mechanical hook-up is sampled; And diagnostic process unit, carry out abnormity diagnosis based on vibration data by described sensor signal processing unit sampling, described diagnostic process unit is taken into the vibration data from described sensor signal processing element continuously, be divided into the interval of each fixed cycle simultaneously, the vibration data in 1 interval is handled as the vibration data of the ponent design that is used for first vibration characteristics, and the data of data that will connect the last stipulated time in its previous interval simultaneously at the front end of the vibration data in 1 interval are handled as the vibration data that is used for the ponent design of second vibration characteristics.
(21) a kind of apparatus for diagnosis of abnormality, be the vehicle bearing of rail truck and the apparatus for diagnosis of abnormality of wheel, it is characterized in that this apparatus for diagnosis of abnormality comprises: the sensor signal processing unit, the output signal of the vibration transducer of the vibration of inspection vehicle axle bearing and wheel is sampled; And diagnostic process unit, carry out the abnormity diagnosis of axle bearing and wheel based on vibration data by described sensor signal processing unit sampling, described diagnostic process unit is taken into the vibration data from described sensor signal processing element continuously, be divided into the interval of each fixed cycle simultaneously, the vibration data in 1 interval is handled as the vibration data that is used for bearing diagnosis, and the data of data that will connect the last stipulated time in its previous interval simultaneously at the front end of the vibration data in 1 interval are handled as the vibration data that is used for the wheel diagnosis.
(22) in the abnormity diagnostic system of above-mentioned (21), it is characterized in that, described diagnostic process unit is handled and the frequency peak that obtains is come the unusual of inspection vehicle axle bearing based on the rotating speed of axle bearing with to the envelope waveform of vibration, the frequency that surpasses threshold value based on the grade of the vibration that produces synchronously with the rotation of wheel detects the unusual of wheel, carries out abnormity diagnosis based on each unusual testing result.
In the abnormity diagnostic system arbitrarily of above-mentioned (21)~(22), it is characterized in that (23) described Signal Processing Element respectively switches a channel with the output signal of a plurality of vibration transducers samples.
(24) in the abnormity diagnostic system of above-mentioned (21) or (23), it is characterized in that, based on synchronously the output signal of vibration transducer being sampled with the rotation of wheel and carrying out summation averaging to handle and the vibration data that obtains carries out the abnormity diagnosis of axle bearing and wheel.
(25) a kind of apparatus for diagnosis of abnormality, it is apparatus for diagnosis of abnormality with mechanical hook-up of the different a plurality of parts of vibration characteristics, it is characterized in that, this apparatus for diagnosis of abnormality comprises: the sensor signal processing unit, the output signal of the vibration transducer of the vibration that detects described mechanical hook-up is sampled; And diagnostic process unit, carry out abnormity diagnosis based on vibration data by described sensor signal processing unit sampling, described diagnostic process unit is taken into the vibration data from described sensor signal processing element continuously, and the sample frequency or the two kinds of different data of sampling length that simultaneously it are transformed to ponent design that is used for first vibration characteristics and the ponent design that is used for second vibration characteristics are handled.
(26) a kind of apparatus for diagnosis of abnormality, be the axle bearing of rail truck and the apparatus for diagnosis of abnormality of wheel, it is characterized in that this apparatus for diagnosis of abnormality comprises: the sensor signal processing unit, the output signal of the vibration transducer of the vibration that detects described mechanical hook-up is sampled; And diagnostic process unit, carry out abnormity diagnosis based on vibration data by described sensor signal processing unit sampling, described diagnostic process unit is taken into the vibration data from described sensor signal processing element continuously, simultaneously it is transformed to be used for axle bearing diagnosis and to handle with the sample frequency or the two kinds of different data of sampling length that are used for the wheel diagnosis.
(27) apparatus for diagnosis of abnormality arbitrarily of above-mentioned (22), (23), (24), (26) is characterized in that, implements repeatedly abnormality detection respectively for axle bearing and wheel, according to separately aggregate-value statistical ground repeatedly carrying out abnormity diagnosis.
(28) apparatus for diagnosis of abnormality arbitrarily of above-mentioned (20)~(27) is characterized in that, this device has and is kept at the function that detects employed data when unusual.
(29) a kind of apparatus for diagnosis of abnormality, it is apparatus for diagnosis of abnormality with mechanical hook-up of the parts that rotate or slide, it is characterized in that this device comprises: the AD transducer will be transformed to digital signal from the simulating signal of the vibration transducer of the vibration that detects described mechanical hook-up; And diagnostic process unit, to carry out Fourier transform processing from the digital signal of this AD transducer, carry out abnormity diagnosis based on its result, described diagnostic process unit than the resolution of described AD transducer also the growth data width digital signal from described AD transducer is carried out Fourier transform processing.
(30) a kind of apparatus for diagnosis of abnormality, it is apparatus for diagnosis of abnormality with mechanical hook-up of the parts that rotate or slide, it is characterized in that this device comprises: the AD transducer will be transformed to digital signal from the simulating signal of the vibration transducer of the vibration that detects described mechanical hook-up; And diagnostic process unit, to carry out Fourier transform processing from the digital signal of this AD transducer, carry out abnormity diagnosis based on its result, it is 1 that the diagnostic process unit makes the resolution of described AD transducer, and its predetermined data width that expands to more than 2 is carried out Fourier transform processing.
(31) a kind of apparatus for diagnosis of abnormality, it is apparatus for diagnosis of abnormality with mechanical hook-up of the parts that rotate or slide, it is characterized in that, this device comprises comparer, the voltage and the reference voltage of the simulating signal of the vibration transducer that will detect from the vibration to described mechanical hook-up compare, it still is the signal of low two-value that thereby output is used to represent the voltage ratio reference voltage height of this simulating signal, and described diagnostic process unit will carry out Fourier transform processing for the predetermined data width from the signal extension of described comparer.
Abnormity diagnostic system according to the structure of above-mentioned (1); owing to detect the sound or the vibration that take place from plant equipment; ask the envelope of this detection signal; this envelope is transformed to frequency spectrum; by the frequency spectrum gliding smoothing that will obtain on the basis of smoothing; detect its peak value; come diagnosing abnormal based on detected peak value; so under the little condition of the S/N of abnormal signal or unusual omen signal and noise signal, also can implement abnormity diagnosis accurately and the noise signal flase drop can not surveyed is unusual or unusual omen signal.
Abnormity diagnostic system according to the structure of above-mentioned (2), because frequency spectrum is implemented the smoothing differential to be handled (promptly, with identical point is the center, the difference in a plurality of intervals and interval long long-pending and), the Frequency point of the sign change of this differential value is extracted peak value as frequency spectrum, detect so can carry out being buried the peak value of the frequency spectrum in noise accurately.
According to the abnormity diagnostic system of the structure of above-mentioned (3), because the weighting coefficient left-right symmetric in the moving average processing is abnormal signal or unusual omen signal so can prevent from noise signal is detected mistakenly.
According to the abnormity diagnostic system of the structure of above-mentioned (4), amplitude levels is the above peak value of threshold value in the peak value that is extracted out owing to select, and detects so can carry out being buried the peak value of the frequency spectrum in noise accurately.
Abnormity diagnostic system according to the structure of above-mentioned (5), because in amplitude levels is peak value more than the threshold value, from the big peak value of the square mean square root of amplitude levels, select the peak value till the regulation number, carry out on the abnormity diagnosis effectively peak value so be limited in, can high precision and carry out abnormity diagnosis expeditiously.
Abnormity diagnostic system according to the structure of above-mentioned (6), because principal component by pairing peak value of the principal component of vibrating in the peak value of asking detected frequency spectrum or vibration and the pairing peak value of high order component and represent the consistent degree of the unusual frequency of diagnosis object, and the accumulated result repeatedly of this unanimity degree estimated, thereby diagnosing abnormal, so can implement abnormity diagnosis accurately.Abnormity diagnostic system according to the structure of above-mentioned (7), owing to handle the extraction processing that signal is carried out in displacement at envelope, reduce counting of the FFT computing be used for the envelope wave analysis, improve so take into account the raising of frequency resolution of detected signal and the efficient of FFT computing, can high precision and implement the abnormity diagnosis of bearing expeditiously.
Abnormity diagnostic system according to the structure of above-mentioned (8), because the sampling rate during with the A/D conversion of detected signal is set to such an extent that carry out frequency band limits after high and is extracted and handle, therefore can omit antialiasing (anti-aliasing) wave filter, carrying out the extraction of signal after envelope is handled handles, reduce counting of the FFT computing be used for the envelope wave analysis, improve so take into account the raising of frequency resolution of the signal that is detected and the efficient of FFT computing, can high precision and implement the abnormity diagnosis of bearing expeditiously.
According to the abnormity diagnostic system of the structure of above-mentioned (9), because the digital filtering of the frequency band low-frequency bandization by making envelope signal handles, thus the influence that can suppress to obscure etc. and carry out the FFT calculation process of low-frequency band reliably.
According to the abnormity diagnostic system of the structure of above-mentioned (10), can carry out handling by the high speed FFT of DSP.Abnormity diagnostic system according to the structure of above-mentioned (11), owing to detect the sound or the vibration that take place from plant equipment, with this signal transformation is digital signal, from this digital signal, take out the signal of the required frequency band of diagnosis and ask its envelope, for with frequency resolution arbitrarily to carrying out FFT by this envelope and carrying out on the basis of 0 cover interpolation, come diagnosing abnormal based on the frequency spectrum that obtains by FFT, so can implement abnormity diagnosis accurately.
Abnormity diagnostic system according to the structure of above-mentioned (12), owing to carry out 0 cover interpolation so that the sample frequency in the FFT unit becomes the multiple hertz of the Nth power of 2 Nth power (for example N=8~12) hertz or 2, so the frequency resolution in the time of can be with the FFT computing is made as the 1.0Hz benchmark, and can be set at resolution arbitrarily.
Abnormity diagnostic system according to the structure of above-mentioned (13), owing to ask the pairing peak value of principal component or the principal component of vibration and the consistent degree of the pairing peak value of high order component and the unusual frequency of expression diagnosis object that vibrate in the peak value by detected frequency spectrum, and the accumulated result repeatedly of this unanimity degree estimated, thereby diagnosing abnormal, so can implement abnormity diagnosis accurately.Apparatus for diagnosis of abnormality according to the structure of above-mentioned (14), comprise waveform signal based on vibration transducer output, output expression crest factor, impact index, the dimension of one of them in form factor and the kurtosis is first voltage signal that 1 parameter value surpasses certain benchmark, or represent that described dimension is that 1 parameter value is the comparator circuit of the second following voltage signal of certain benchmark, can detect based on the output of this comparator circuit unusually, so only every vehicle is provided with a vibration transducer, can be based on waveform signal from this vibration transducer, detect peeling off of bearing in this vehicle, the flat grade of wheel is unusual.
According to the apparatus for diagnosis of abnormality of the structure of above-mentioned (15), owing to comprise: computing circuit, based on the waveform signal of described vibration transducer output, ask the parameter value of average one of them of RMS (square root of Fang Jun) and absolute value; Peak detection circuit is asked the peak value of described waveform signal; And comparator circuit, the certain of described parameter value doubly (determined magnification by mimic channel in [embodiment] for example described later, so generally be not integer but certain doubly or constant times) value and compare from the peak value of described peak detection circuit output, according to its comparative result, the ratio of output described peak value of expression and described parameter value (promptly, dimension is 1 parameter value) surpass first voltage signal of the situation of certain benchmark, or represent that described dimension is that 1 parameter value is second voltage signal of the following situation of certain benchmark, can detect based on the output of described comparator circuit unusually, so only every vehicle is provided with a vibration transducer, can be based on waveform signal from this vibration transducer, detect peeling off of bearing in this vehicle, the flat grade of wheel is unusual.In addition, according to the apparatus for diagnosis of abnormality of the structure of above-mentioned (15),
Apparatus for diagnosis of abnormality according to the structure of above-mentioned (16), because peak value and predefined reference point are compared, under the situation of peak value greater than reference point, it is invalid to be used in the first and second unusual voltage signal of detection, so can prevent because the very large signal that noise causes causes the output of sensor unit saturated.
According to the apparatus for diagnosis of abnormality of the structure of above-mentioned (17), recently detect unusually by constituting the duty that surpasses first voltage signal of certain benchmark based on the expression parameter value, can avoid The noise to carry out abnormity diagnosis simultaneously.
According to the apparatus for diagnosis of abnormality of the structure of above-mentioned (18), the signal of only catching allocated frequency band in the output signal of vibration transducer just can carry out abnormity diagnosis.
According to the apparatus for diagnosis of abnormality of the structure of above-mentioned (19), owing to can detect the unusual of rail truck, so can improve the reliability of rail truck.
And then, according to above-mentioned (20) apparatus for diagnosis of abnormality, obtain the effect of following (I)~(IV) to the structure of (28).
(I) owing to be taken into the interval that vibration data is divided into each fixed cycle simultaneously continuously, the vibration data in 1 interval is handled as the vibration data of the parts that are used to diagnose first vibration characteristics, the data of data that will connect the last stipulated time in its previous interval simultaneously at the front end of the vibration data in 1 interval are handled as the vibration data that is used for the ponent design of second vibration characteristics, so detect the abnormal vibrations of the parts of two vibration characteristics in real time according to the output signal of the vibration transducer of the vibration of the parts that detect two vibration characteristics, can determine unusually cause still unusually the causing by the parts of second vibration characteristics of this abnormal vibrations by the parts of first vibration characteristics.
(II) owing to be taken into the interval that vibration data is divided into each fixed cycle simultaneously continuously, diagnose the bearing vibration data to handle as being used to the vibration data in 1 interval, the data of data of last stipulated time that will connect its previous interval simultaneously at the front end of the vibration data in 1 interval are as being used to diagnose the vibration data of wheel to handle, so come the abnormal vibrations of inspection vehicle axle bearing in real time and wheel according to the output signal of the vibration transducer of the vibration of inspection vehicle axle bearing and wheel, can determine that this abnormal vibrations is caused or caused by axle bearing by the flat of wheel.
(III) being taken into sample frequency or the two kinds of different data of sampling length that vibration data is transformed to it ponent design that is used for first vibration characteristics and the ponent design that is used for second vibration characteristics simultaneously continuously handles, so detect the abnormal vibrations of the parts of two vibration characteristics in real time according to the output signal of the vibration transducer of the vibration of the parts that detect two vibration characteristics, can determine unusually cause still unusually the causing by the parts of second vibration characteristics of this abnormal vibrations by the parts of first vibration characteristics.
(IV) being taken into vibration data continuously is transformed to it simultaneously and is used for axle bearing diagnosis and handles with the sample frequency or the two kinds of different data of sampling length that are used for the axletree diagnosis, so come the abnormal vibrations of inspection vehicle axle bearing in real time and wheel according to the output signal of the vibration vibration transducer of inspection vehicle axle bearing and wheel, can determine that this abnormal vibrations is caused or caused by axle bearing by the flat of wheel.
<invention effect〉according to the present invention, even the S/N of abnormal signal or unusual omen signal and noise signal than little condition under, also can implement abnormity diagnosis accurately and the noise signal flase drop can not surveyed and be unusual or unusual omen signal.According to abnormity diagnostic system of the present invention, can high precision and implement the bearing in the plant equipment or the abnormity diagnosis of the related member of bearing expeditiously.According to the present invention, because each vehicle only is provided with a vibration transducer, so, can detect the flat etc. unusual of the peeling off of bearing in this vehicle, wheel, so can construct abnormity diagnostic system with low cost based on waveform signal from this vibration transducer.
According to apparatus for diagnosis of abnormality of the present invention, can use the AD transducer of low resolution or cost degradation and the save spaceization that simple comparer is realized circuit, and can carry out abnormity diagnosis and do not cause that precision reduces.
Description of drawings
Fig. 1 is the block scheme of the embodiment of expression abnormity diagnostic system of the present invention.
Fig. 2 is the oscillogram of illustration frequency spectrum and moving average result thereof.
Fig. 3 is the oscillogram of illustration frequency spectrum and moving average result thereof.
Fig. 4 is the oscillogram of illustration frequency spectrum and moving average result thereof.
The example of the vibrational waveform when Fig. 5 represents that the noise of impact enters.
Fig. 6 is the process flow diagram of the abnormity diagnosis action example of expression abnormity diagnostic system shown in Figure 1.
Fig. 7 is the oscillogram of illustration frequency spectrum and moving average result thereof.
Fig. 8 is the microlesion product of expression bearing and the abnormity diagnosis result's of normal product figure.
Fig. 9 is the block scheme of the 2nd embodiment of expression abnormity diagnostic system of the present invention.
Figure 10 is the block scheme that expression constitutes the embodiment of the microcomputer of abnormity diagnostic system of the present invention and peripheral circuit thereof.
Figure 11 is the oscillogram of the frequency-gain characteristic of first wave digital lowpass filter among illustration Fig. 9.
Figure 12 is the oscillogram of the frequency-gain characteristic of first wave digital lowpass filter among illustration Fig. 9.
Figure 13 (a) is the oscillogram that the FFT spectrum waveform under the situation of processing is extracted in expression, and Figure 13 (b) is the oscillogram that the FFT frequency spectrum under the situation that extracts magnetic force has been omitted in expression.
Figure 14 be on curve map expression by the figure of the effect that reduces counting of FFT computing and cut down the FFT calculation process time.
Figure 15 is a curve map of comparing the result of a plurality of diagnosis of carrying out than change condition with S/N.
Figure 16 is the block scheme of the 3rd embodiment of expression abnormity diagnostic system of the present invention.
Figure 17 is the block scheme of the 4th embodiment of expression abnormity diagnostic system of the present invention.
Figure 18 is the process flow diagram of flow process of a series of processing in the abnormity diagnostic system of expression the 4th embodiment.
To be expression extract the key diagram of the situation of processing with respect to the envelope waveform of the vibration phase place that staggers for Figure 19 (a) and Figure 19 (b).
Figure 20 is the figure of diagnostic result of the situation of expression the 4th embodiment.
Figure 21 be expression about time of FFT calculation process, contrast is used the situation of DSP and is only used the figure of the situation that CPU carries out.
Figure 22 is the process flow diagram of flow process of a series of processing in the abnormity diagnostic system of expression the 5th embodiment.
Figure 23 is the figure that hinders diagnostic result of the situation of expression the 5th embodiment.
Figure 24 is the functional-block diagram of the embodiment of expression abnormity diagnostic system of the present invention.
Figure 25 is the key diagram that 0 interpolation of passing through 0 interpolation unit among Figure 24 is handled.
Figure 26 is the abnormity diagnosis result's of expression defective and normal product figure.
Figure 27 is that being taken into of expression data handled and the timing of operation of data processing and the timing diagram of used time.
Figure 28 is the summary construction diagram of rail truck that comprises the apparatus for diagnosis of abnormality of the 7th embodiment.
Figure 29 is the block scheme of the 7th embodiment of expression sensor unit.
Figure 30 is the figure of output waveform of the sensor unit of expression Figure 29.
Figure 31 is the block scheme of the 8th embodiment of expression sensor unit.
Figure 32 is the figure of output waveform of the sensor unit of expression Figure 31.
Figure 33 is the block scheme of the 9th embodiment of expression sensor unit.
Figure 34 is the block scheme of the 10th embodiment of expression sensor unit.
Figure 35 is the block scheme of the 11st embodiment of expression sensor unit.
Figure 36 is the block scheme of the 12nd embodiment of expression sensor unit.
Figure 37 is the block scheme of the 13rd embodiment of expression sensor unit.
Figure 38 is the oscillogram of the crest factor (Peak/RMS) of one of expression parameter of deterioration such as peeling off, and the expression parameter value increases owing to peeling off.
Figure 39 is the vibrational waveform figure of rail truck, sneaks into the situation of the impact shock (noise) that the seam of track causes in the vibrational waveform that expression detects.
Figure 40 (a) is the general view of rail truck that has carried the apparatus for diagnosis of abnormality of the 14th embodiment, and Figure 40 (b) represents the summary side elevation of this rail truck.
Figure 41 is the skeleton diagram of the position relation of illustration axle bearing and vibration transducer.
Figure 42 is the block scheme of the 14th embodiment of apparatus for diagnosis of abnormality of the present invention.
Figure 43 is being taken into and the sequential chart of data parsing of vibration data of 4 channels of apparatus for diagnosis of abnormality.
Figure 44 is the process flow diagram of movement content of the diagnostic process unit of expression Figure 42.
Figure 45 be the expression axle bearing wound the position and by the figure of the relation of the vibration occurrence frequency that hinder to cause.
Figure 46 is the block scheme of the 15th embodiment of apparatus for diagnosis of abnormality of the present invention.
Figure 47 is the block scheme of the 16th embodiment of apparatus for diagnosis of abnormality of the present invention.
Figure 48 is the process flow diagram of movement content of the diagnostic process unit of expression Figure 47.
Figure 49 is the block scheme of the 17th embodiment of apparatus for diagnosis of abnormality of the present invention.
Figure 50 is the block scheme of the 18th embodiment of apparatus for diagnosis of abnormality of the present invention.
Figure 51 is the block scheme of the 19th embodiment of apparatus for diagnosis of abnormality of the present invention.
Figure 52 is the process flow diagram of vibration content of diagnostic process unit of the 19th embodiment of apparatus for diagnosis of abnormality of the present invention.
Figure 53 (a) is the concept map that the expression bearing is peeled off the relation on the T/F plane of diagnostic data and the flat diagnostic data of wheel, and Figure 53 (b) is the concept map of relation of the frequency range of expression bearing and wheel.
Figure 54 is the process flow diagram of vibration content of diagnostic process unit of the 20th embodiment of apparatus for diagnosis of abnormality of the present invention.
Figure 55 (a) and (b) be the part block scheme of the diagnostic process unit of the 21st embodiment.
Figure 56 is the process flow diagram of vibration content of diagnostic process unit of the 21st embodiment of apparatus for diagnosis of abnormality of the present invention.
Figure 57 is the process flow diagram of movement content of the diagnostic process unit of expression Figure 51.
Figure 58 (a) is that Figure 58 (b) is the key diagram of expression from the example of the simple sign extended of the digital signal of AD transducer about the key diagram of the processing that will also expand than its resolution from the digital signal of AD transducer.
Figure 59 is the major part block scheme of the 23rd embodiment of apparatus for diagnosis of abnormality of the present invention.
Figure 60 (a) is that the simulating signal of self-excited oscillation sensor in the future is transformed to the key diagram of the Signal Processing of two-value by comparer, and Figure 60 (b) has carried out digital filtering with the signal from comparer with the microcomputer in the diagnostic process unit to handle oscillogram afterwards.
Figure 61 is the process flow diagram of the movement content of the diagnostic process unit among expression the 23rd embodiment.
Figure 62 is the major part block scheme of the 24th embodiment of apparatus for diagnosis of abnormality of the present invention.
Embodiment
Below, about being used to implement best mode of the present invention, be object with the plant equipment that comprises rolling bearing, be that example describes to the unusual situations about judging that have or not such as wound of the rolling bearing in the plant equipment.
Fig. 1 is the block scheme of the embodiment of expression abnormity diagnostic system of the present invention.As shown in Figure 1, abnormity diagnostic system of the present invention comprises: amplifilter (filter processing unit) 101, A/D transducer 102, envelope processing unit 103, FFT unit 104, peak detection unit 105, diagnosis unit 106 and diagnostic result output unit 107.
Amplifilter 101 is transfused to the detected signal of sensor (vibration transducer, acoustic sensor etc.) that is detected by sound that the plant equipment of diagnosis object is taken place or vibration.Amplifilter 101 amplifies the signal of importing with the gain of regulation, the above signal of the frequency of truncation specification simultaneously.
A/D transducer 102 to the analog signal sampling by amplifilter 101, and is transformed to digital signal with the sample frequency of regulation.
Envelope processing unit 103, ask the FFT unit 104 of the envelope (envelope waveform) of the digital signal that generates by A/D transducer 102, the envelope that envelope processing unit 103 is obtained carries out frequency resolution and is transformed to the peak detection unit 105 of frequency spectrum, 106 pairs of characteristic frequencies of the line feed diagnosis unit of the peak value of the frequency spectrum that detection is obtained by FFT unit 104 by each element decision of inside of detected rotating speed of not shown rotation sensor that is provided with on the rolling bearing and bearing, the peak value that obtains with peak detection unit 105 compares, and by its consistent degree evaluation is come diagnosing abnormal.The diagnostic result of diagnostic result output unit 107 output diagnosis units 106.Peak detection unit 105 comprises moving average processing unit 105a, smoothing differential peak extraction unit 105b, the first module of selection 105c, the second module of selection 105d.Moving average processing unit 105a will be weighted moving averageization by frequency spectrum (discrete data of the frequency domain) left and right symmetrically that FFT unit 104 obtains.For example, in 5 moving average, by the frequency spectrum that obtained by FFT unit 104 being implemented the computing of following formula,
[formula 1]
Figure C20058000628800201
a>b,a>c
Generally, by the computing of following formula (1),
[formula 2]
Figure C20058000628800202
Frequency spectrum is carried out smoothing carry out alleviating of noise.Smoothing differential peak extraction unit 105b is after the moving average processing by moving average processing unit 105a, with moving average frequency spectrum further level and smooth and obtain differential value, the Frequency point of the sign change of differential coefficient is extracted as the peak value of frequency spectrum.That is, smoothing differential peak extraction unit 105b is considered as the value (smoothing differential coefficient yj) of following formula (2) candidate of the peak value of frequency spectrum from the negative Frequency point that changes of forward.
[formula 3]
y i = Σ i = 1 m ( x i + j - x j - i ) i - - - ( 2 )
From this formula (2) as can be known, the data that connect near away from point between inclination person can be considered as weight big.As seen from formula (2), peak detection unit 105 comprises smoothing differential peak extraction unit 105b, for the frequency spectrum that obtains by FFT unit 104 implement with the j point be the difference in a plurality of intervals of carrying out, center and its interval long long-pending and the smoothing differential handle, with the Frequency point of the sign change of the differential value that obtains peak extraction as frequency spectrum.
Thereby,, do not use formula (1) can be embedded in the detection of the peak value of noise yet, but can and use formula (1) yet according to formula (2).
It is peak value more than the threshold value that the first module of selection 105c selects amplitude levels in the peak value that is extracted by smoothing differential peak extraction unit 105b.Threshold value is used the relative value according to the square mean square root decision of the power average value of the peak value that is extracted by smoothing differential peak extraction unit 105b or whole (overall) signals.Absolute threshold value is effective under the low situation of relative noise level, but not necessarily effective under the big situation of noise level.
The second module of selection 105d selects the peak value till the regulation number from the big peak value of amplitude levels in the peak value of being selected by the first module of selection 105c.As its simplest method, for example, can enumerate and use known sorting algorithm, a plurality of peak values are made as about grade after ascending order or the descending class, from the peak value of a high position, promptly be worth the method that big peak value is selected successively.
Fig. 2 represents the example of spectrum waveform.The vibration data that this routine expression will be diagnosed as wound carries out the frequency spectrum of envelope processing and the frequency spectrum after the moving average processing thereof.The moving average here is 7 the moving average that is shown below.
[formula 4]
Figure C20058000628800212
w 0=4,w 1=w -1=3,w 2=w -2=2,w 3=w -3=1
Weighting coefficient w is not limited to above-mentioned value, but does not preferably remove about j=0 symmetry and make the condition of weight maximum of the point of j=0.In the example of Fig. 2 as can be known, because S/N is better frequently, so the fundametal component f1 that the wound of bearing foreign steamer causes and higher harmonic components f2, f3, f4 obviously can see in the front and back of moving average processing, but after the moving average processing, the pseudo-peak value that noise causes is considerably less.
As shown in Figure 2, carry out the smoothing differential by the frequency spectrum of moving average processing unit 105a after to the moving average processing, smoothing differential peak extraction unit 105b with the symbol of differential coefficient after the negative Frequency point that changes of forward has detected as peak value, by the peak value more than the first module of selection 105c extraction threshold value, with them by after the second module of selection 105d classification, will be wherein till high-order 5 as peak extraction, thereby peaking frequency f 1, f2, f3, f4.When discrete spectrum was made as xi, the smoothing differential coefficient yi of this moment was expressed from the next.
[formula 5]
y i = Σ j = 1 m ( x i + j - x j - i ) j
Different with common numerical differentiation, in this formula, carry out in order to make it have the effect of smoothing than further from point between the big weighting of difference, so can only differentiate, do not need division by integer arithmetic.Thereby Float Point Unit (FPU) or the microcomputer that does not have a division order also can carry out computing easily.
The data of the peak value of the above-mentioned frequency spectrum that is obtained by the second module of selection 105d (envelope frequency distribution) are transfused to diagnosis unit 106.
Diagnosis unit 106 will compare with the unusual frequency of representing diagnosis object corresponding to the peak value of the principal component of vibrating or corresponding to the principal component of vibration and the peak value of higher hamonic wave in the peak value of the frequency spectrum of input, and asks its consistent degree.Then, the consistent degree of obtaining given count and add up, thereby carry out the high diagnosis of reliability.For example, carry out principal component, 2 rank, 4 times three components and represent the comparison of unusual frequency, if principal component and other component are detected, then are judged as the possibility of hindering take place, add counting accordingly in the predefined some numerical table.The example of some numerical table is shown in following table 1.In the example of Fig. 2, because three components on principal component, 2 rank, 4 rank all are detected, so add 4 points.
[table 1]
Detected components Count
Fundametal compoment, 2 rank, 4 rank 4
Fundametal compoment, 2 rank 2
Fundametal compoment, 4 rank 1
In the example of spectrum waveform shown in Figure 3, although receive the noise that external impact causes, the peak value of the frequency that the wound of bearing foreign steamer causes also is detected.Same with the situation of Fig. 2, carry out the smoothing differential again and carry out after peak value detects, with will be up to the result after detecting as peak value till high-order 5 after the classification of the peak value more than the threshold value, principal component and 2 order components be detected.It is 2 minutes that addition under this situation is counted.
In the example of spectrum waveform shown in Figure 4, because the noise that external impact causes is excessive, peak value is not detected.It is 0 point that addition under this situation is counted.
The example of the vibrational waveform when Fig. 5 represents that the noise of impact enters.Like this, the frequency analysis result of the envelope of the vibrational waveform that the amplitude big like this and noise of impact burst has entered increases near the lower frequency side of DC (direct current) component, and as the example of Fig. 4, the peak value of the vibration that small wound causes is hidden.Under these circumstances, needn't be used to detect the processing of hindering the component of signal that causes forcibly.
As shown in Figure 6, a series of processing till this abnormity diagnostic system will detect abnormal point numerical and judge from above-mentioned vibration signal repeats stipulated number N (for example 30 times) and comes above-mentioned the counting of accumulative total, judges unusually according to this cumulative points.In Fig. 6, n is current number of times, and PA is represented among Fig. 2, Fig. 3 of diagnostic points in measuring of frequency spectrum, aggregate-value that PACC represents PA and Fig. 4 that illustrative spectrum waveform is once sampled and carried out the required time of abnormity diagnosis respectively is about 1 second.Thereby,, then repeat about 40~60 times diagnosis with the above-mentioned accumulative total of counting, and can carry out abnormity diagnosis by this cumulative points if the time that allows in order to obtain diagnostic result is about 40~60 seconds.By in once the abnormity diagnosis of sampling only, as Fig. 2~Fig. 4, do not know to obtain which type of frequency spectrum, but detect and add at every turn that at it diagnosis counts by the repetition frequency peak value, and the aggregate-value of counting estimated, thereby thereby abnormity diagnosis is carried out in the influence that can alleviate the deviation of frequency spectrum accurately.
Fig. 7 is the diagnosis frequency spectrum of the normal bearing of not hindering, and the result that peak value detects has been carried out in expression, the measured result that the frequency component of the vibration that is caused by wound is not detected.Which kind of feature sees among the frequency analysis result of moving averageization just has as can be known, but the result who selects processing of threshold value and classification, and it doesn't matter with frequency component that bearing is unusual, so the abnormity diagnosis point of table 1 is not coupled with.
Fig. 8 is the abnormity diagnosis of the microlesion product of No. 40 bearings of repetition and normal product and the figure that represents the accumulative total of its diagnostic points in histogram.In microlesion product and the normal product, there is big gap in cumulative points, therefore as can be known by will totally putting accumulative total about 40 times, can carry out the abnormity diagnosis of bearing exactly.In addition, although be small wound because and produce between the normal product big poor, therefore as shown in Figure 8 owing to the scope of threshold value can be got greatly, therefore can be with this scope classification and the alarm that send stage.
As described above; in the abnormity diagnostic system of this embodiment; sound or vibration that detection takes place from plant equipment; ask the envelope of this detection signal; this envelope is transformed to frequency spectrum; the frequency spectrum that obtains is carried out the moving average processing; and then this frequency spectrum carried out the smoothing differential; thereby with the symbol of differential coefficient after the negative Frequency point that changes of forward detects as peak value; extract the above peak value of defined threshold; to after their classification a high position wherein be stipulated that number is as peak extraction; ask in these peak values corresponding to the peak value of the principal component of vibration or corresponding to the peak value of the principal component of vibration and higher hamonic wave and the consistent degree of the unusual frequency of representing diagnosis object; and this unanimity degree given count and repeatedly add up; come diagnosing abnormal by estimating this aggregate-value; even so the S/N of abnormal signal or unusual omen signal and noise signal than little condition under, also unusual high precision and implement abnormity diagnosis expeditiously and the noise signal flase drop can not surveyed and be unusual or unusual omen signal.
In addition, the invention is not restricted to the foregoing description.For example, shown in dashed rectangle among Fig. 1, digital filter (LPF/HPF) 108 is set between A/D transducer ADC102 and envelope processing unit 103, preferably removes the noise component of high frequency and remove the DC side-play amount simultaneously.In addition, between FFT unit 104, be provided with and select unit 109, extract processing (decimation selects) according to the frequency of necessity.Handle by the extraction of after envelope is handled, carrying out signal, and reduce counting of the FFT computing be used for the envelope wave analysis, improve thereby take into account the raising of frequency resolution of detected signal and the efficient of FFT computing, can high precision and implement the abnormity diagnosis of bearing expeditiously.
[the 2nd embodiment]
Fig. 9 is the block scheme of the 2nd embodiment of expression abnormity diagnostic system of the present invention, and Figure 10 is the block scheme of the embodiment of the microcomputer (MPU) of concrete inscape of expression abnormity diagnostic system of the present invention and peripheral circuit thereof.
As shown in Figure 9, abnormity diagnostic system of the present invention comprises: amplifilter (filter processing unit) 201, A/D transducer 202, first wave digital lowpass filter 203, first are selected unit (extracting unit) 204, envelope processing unit 205, second wave digital lowpass filter 206, second and are selected unit (extracting unit) 207, FFT arithmetic element 208, diagnosis unit 209, transformation of speed processing unit 210 and diagnostic result output unit 211.
Amplifilter 101 is transfused to the detected signal of sensor (vibration transducer, acoustic sensor etc.) that is detected by sound that the plant equipment of diagnosis object is taken place or vibration.Amplifilter 201 amplifies the signal of importing with the gain of regulation, simultaneously the above signal of truncation specification frequency (being 80kHz here).
A/D transducer 202 to passing through the analog signal sampling of amplifilter 201, and is transformed to digital signal with the sample frequency (be 250kHz) of regulation here.Counting of once sampling is about 200,000.Data length is 16.As shown in figure 10, this system uses microcomputer 220 as signal processing circuit, but owing to comprise external RAM221, is easy to so guarantee the variable zone of this degree.Microcomputer 220 comprises floating point unit (FPU).
First wave digital lowpass filter 203 only makes that the following signal of assigned frequency (being 10kHz here) passes through in the digital signal that A/D transducer 220 generates, and for example has the FIR wave filter on 55 rank to constitute.The frequency characteristic of Figure 11 illustration first wave digital lowpass filter 203.This wave filter 203 is unattenuated fully below 10kHz (fp), and till 10kHz (fp) arrived 25kHz (fq), attenuation rate increased, and more than 25kHz, becomes the attenuation rate of 60dB.For the waveform that reaches the frequency band of stopband at 25kHz (fq), as long as sample frequency is at least 50kHz just enough.
First select unit (extracting unit) 204 by with the sample frequency (being 50kHz here) of regulation to sampling by the signal of first wave digital lowpass filter 203, thereby extract processing.Because the sample frequency of A/D transducer 202 is 250kHz, so sampling number (data number) is extracted with 1/5 ground.Thus, there are 200,000 data to be reduced to 40960 data.Envelope processing unit 205 is asked the envelope signal (envelope waveform signal) of selecting the signal that takes out unit 204 by first.
Second wave digital lowpass filter 206 is only to make assigned frequency in the envelope signal that is obtained by envelope processing unit 205 (the being 1kHz here) wave filter that following signal passes through, and for example the FIR wave filter by 110 rank constitutes.The characteristic waveforms of Figure 12 illustration second wave digital lowpass filter 206.This wave filter 206 cooperates the unusual characteristic frequency of expression bearings to carry out Filtering Processing, and is unattenuated fully below 1kHz (fp), and till the 2.5kHz, attenuation rate increases, and more than 25kHz (fq), becomes the attenuation rate of 60dB at 1kHz.
Second select unit (extracting unit) 207 by with the sample frequency (being 5kHz here) of regulation to sampling by the signal of second wave digital lowpass filter 206, thereby extract processing.Because first sample frequency (fs) of selecting unit 204 is 50kHz, so by with 1/10 sampled point that extracts.For the waveform that reaches the frequency of stopband at 2.5kHz (fq), as long as sample frequency is at least 5kHz just enough.Extract to handle by this, have 40960 data be reduced to 4096 data.
208 pairs of FFT arithmetic elements are selected the envelope signal that has carried out extracting after handling in unit 207 by second and are carried out frequency resolution.Under the situation of this example, use 4096 data to carry out the frequency analysis of the envelope of detected signal.Thus, carry out frequency analysis with the resolution of 5000/4096=1.22Hz.
Diagnosis unit 209 is in the peak value that the result who has carried out frequency resolution by FFT arithmetic element 208 obtains, the fundamental component of the frequency that will cause by rolling bearing and the size of high fdrequency component, the unusual frequency of the bearing that obtains from each element of bearing with the determinating reference data (rotating speed) that provided by transformation of speed processing unit 210 and expression compares, and diagnoses the unusual of rolling bearing based on its result.
Transformation of speed processing unit 210 generate with from the corresponding determinating reference data of the rotating signal of the not shown rotation sensor that is provided with on the rolling bearing, and these data are offered diagnosis unit 209.
The diagnostic result of diagnostic result output unit 211 output diagnosis units 209.
The spectrum waveform of the envelope that the operation result of Figure 13 (a) expression FFT arithmetic element 208 obtains.This is the waveform that the foreign steamer of having caught rolling bearing is hindered component, has conclusivelyed show fundamental component (f1) and high fdrequency component (f2~f6 etc.).In this case, diagnosis unit 209 calculates the unusual frequency of the bearing that the rotating speed that obtains from transformation of speed processing unit 210 and expression obtain from each element of bearing, the result that fundamental frequency among Figure 13 (a) and the higher hamonic wave till 6 rank are compared, because the frequency component that causes with the foreign steamer defective is consistent, therefore exporting foreign steamer exists unusual diagnostic result.
Here, as a comparative example, the spectrum waveform that Figure 13 (b) expression obtains carrying out carrying out the FFT computing after envelope is handled with the waveform that sample frequency is made as 25kHz, cutoff frequency fc is made as the identical condition of the above-mentioned example of 10kHz.Because counting of FFT computing is 16384, so the frequency resolution in this comparative example is 25000/16384=1.526Hz.In embodiments of the invention (Figure 13 (a)), for comparative example (Figure 13 (b)), the FFT computing count from 16384 reduce to its 1/4 4096, and resolution is brought up to 1.22Hz from about 1.53Hz.This is to have carried out extracting the effect that (selecting) handles in the front and back that envelope is handled.
Figure 14 represents to reduce the effect that FFT calculation process time that counting of FFT computing cause cuts down.Under the situation of this embodiment,, as shown in figure 10, use the microcomputer 220 that has high-speed RAM 220c in inside as the hardware of carrying out the FFT calculation process.Can hold the FFT data till 4096 o'clock among the high-speed RAM 220c of these microcomputer 220 inside.Its result compared with the computing time under having held the FFT data conditions more than 8192, can calculate fast overwhelmingly.In the system that does not have such high-speed RAM 220c, obtain the reduction effect of the execution cycle number (FFT calculation process time) shown in the dotted line among Figure 14.FFT need be by the calculating of counting of 2 index, thus sample in this example and extract processing, finally becoming 4096 points, even but 4096 of hypothesis have too much and not enough, omit counting or 0 data supplementing being got final product to front and back of this part.
Figure 15 is the curve map of absolute comparison of S/N ratio of the foreign steamer diagnosis of expression rolling bearing.First-harmonic and represent with the S/N ratio up to 6 times higher hamonic wave and the ratio of having removed the component till 1kHz of these components.A is corresponding to above-mentioned comparative example among Figure 15.C is corresponding to the foregoing description.B is the S/N ratio that has omitted under the situation of second wave digital lowpass filter 206.In the comparison of A and C, both S/N do not have so big difference than we can say, but C omits.Although counting of FFT computing is that C lacks than A, S/N is than having improved, and this is the effect by the frequency band limits of second wave digital lowpass filter 206.
As described above, in the abnormity diagnostic system of this variation, carry out the extraction of signal handles in the front and back that envelope is handled, the counting of FFT computing that is used to resolve the envelope waveform of the signal that goes out by sensor by minimizing, take into account the raising of frequency resolution of signal and the efficient of FFT computing and improve, and can high precision and implement the abnormity diagnosis of bearing expeditiously.
In addition, in this abnormity diagnostic system, the sampling rate during the A/D conversion of the signal that will be gone out by sensor is set after the height, carries out frequency band limits and extracts and handle, and therefore can omit antialiasing filter.Promptly, signal owing to the above frequency of 1/2 (being the frequency f s/2 of Qwest) of the sample frequency dispersion (250kHz) of blocking A/D transducer 202 by first wave digital lowpass filter 203, therefore need to insert antialiasing filter usually, but here with respect to the frequency band of amplifilter 201 less than 80kHz, and the sample frequency of A/D transducer 202 is 250kHz, the needs antialiasing filter not so satisfy sampling thheorem.Thus, can realize the low cost of abnormity diagnostic system.
[the 3rd embodiment]
Figure 16 is the block scheme of the 3rd embodiment of expression abnormity diagnostic system of the present invention.In the 3rd embodiment, omitted in the 2nd embodiment and to have selected processing unit 204 before the wave digital lowpass filter 203,206 that is provided with in envelope processing unit 205 front and back and the envelope processing unit 205.Even this structure can be applied to S/N than reducing a little, as long as can improve the situation that the frequency resolution of envelope wave analysis gets final product with few FFT computing of counting.
Do not use the extraction of wave digital lowpass filter to handle the influence of being obscured, himself play the effect that low-pass filter is handled on the contrary.And envelope is handled the also effect handled of double as low-pass filter of himself result, so think that to omit the situation of selecting the wave digital lowpass filter 206 before the processing unit 207 more.Do not cause under the situation about obscuring learning, do not use digital filter and extract processing without any influence according to the frequency characteristic of the amplifier of prime or transmission path yet.
In addition, the situation of the operation efficiency of digital filter and FFT computing also has property difference a little.FFT since be unified calculation process so data number arithmetic speed is fast more more at least, and digital filter is owing to handle successively basically, so the exponent number of wave filter becomes problem.But, in the above-described embodiments, estimate that also the wave filter of 100~200 exponent numbers gets final product in second wave digital lowpass filter 206 after envelope processing unit 205.If the filter order of this degree, the processing among the high-speed memory 210a in the then general microcomputer 210 is without any problem.
In addition, in the 2nd and the 3rd embodiment, because microcomputer 220 does not have FPU (Float Point Unit), therefore used the FIR wave filter that is suitable for fixed-point arithmetic, but under the situation of system with FPU, if wave digital lowpass filter is used iir filter, just can reduce filter order and do not reduce computational accuracy.
[the 4th embodiment]
Figure 17 is the block scheme (hardware structure diagram) of the 4th embodiment of expression abnormity diagnostic system of the present invention.Figure 18 is the process flow diagram of flow process of a series of processing in the abnormity diagnostic system of expression the 4th embodiment.Be connected with synchronous dram (SDRAM) 221a, flash memory 222, amplifilter (filter processing unit) 223 and LCD (LCD) 224 on the microcomputer 220.
Microcomputer 220 also comprises DSP220b and flash RAM220c except CPU220a.
DSP220b is built-in with the X/Y-RAM220e that X-RAM and Y-RAM constituted that is connected by private bus respectively, so that can the long-pending and computing with the command execution of special use in one-period.The capacity of X-RAM and Y-RAM respectively is 8 kilobyte.DSP220b adds that command line simultaneously can visit three buses, and can carry out a plurality of orders simultaneously.X/Y-RAM is known as binary channels RAM, dual access RAM, hyperchannel RAM etc.
Synchronous dram 221a, flash memory 222 and amplifilter 223 are connected to the external bus of CPU220a.Synchronous dram 221a is the storer of the capacity of the 32MB (megabyte) that works as main memory.Flash memory 222 is storeies of the capacity of the 4MB that works as program storage area.Storage is used to implement the program of a series of processing shown in Figure 180 in the flash memory 222.Amplifilter 223 comprise amplifier 223a that the signal of autobiography sensor in the future amplifies and with the sample frequency (being 250kHz here) of regulation to sampling by amplifier 223a amplifying signal and being transformed to the A/D transducer 223b of 16 resolution of digital signal.
The responsiveness of synchronous dram 221a and flash memory 222 is slower than CPU220a, and therefore in order to produce the high speed of CPU220a, flash memory is indispensable.Therefore, be built-in with the RAM cache 220c of data/order mixed type in the microcomputer 220.
DMAC220d control DMA action is not promptly used CPU220a and will synchronous dram 221a be transmitted by the data that A/D transducer 223b obtains.LCD 224 is the output units that are used to show diagnostic message.
That the data volume maximum of processing was FFT calculation process (S204) during the digital operation that comprises in a series of processing shown in Figure 180 was handled.In order to be carried out FFT calculation process (S204) by DSP220b, the data that FFT calculation process (S204) is used need be contained among the X/Y-RAM220e.
On the other hand, detect wound, need detect vibration, but the vibration number that passes through of rotor that is used to catch the bearing of wound is generally below the 1kHz with the frequency band till about 10kHz in order to resolve by bearing vibration.In this example, be low frequency below the 100Hz as the rotor of the bearing of diagnosis object by vibration number.
Under the rotor situation low like this, accurately diagnose the relatively more long waveform sampling of unusual needs of bearing by vibration number.
Therefore, in the sampling processing (S201) of Figure 18, to sampling, the Wave data that is made of the data more than 40000 o'clock is sampled from the signal of amplifilter 223 with the sample frequency of 48kHz.In this case, can guarantee the sampling time Tw that 800ms is above.The frequency resolution Δ f of FFT calculation process (S204) is determined by this sampling time Tw.That is, frequency resolution Δ f is the inverse (1/Tw) of sampling time Tw.
Absolute value is handled (S202) and is handled same processing with envelope, in digital processing, compares with the method by Hilbert transform, can simplify computing significantly.In this is handled, for the envelope or the absolute value waveform of the signal of having sampled, average in order to eliminate the DC component by sampling processing (S201), rebuild the line of amplitude 0.
In selecting processing (S203), by with assigned frequency (being 4.8kHz here) to handle through absolute value (S202) thus envelope or absolute value waveform signal sample and extract.In the FFT calculation process (S204), to carrying out frequency resolution by selecting the signal that processing (S203) carried out extracting after handling.
Data in the FFT calculation process (S204) are made of real part and imaginary part, are assigned to X-RAM and the Y-RAM of X/Y-RAM220e respectively.If adopt the mode of supplying with storage area in input and output, then can be to the data progress row FFT of 8kB.Because the resolution of A/D transducer 223b is 16 (2 bytes), so can be by the data of DSP220b processing till 8192/2 byte is at 4096.Otherwise the data of counting above 4096 can not be handled in DSP220b.Therefore, in this example, extracting processing by selecting processing (S203), is 4096 so that data grow up to.When sample frequency fs=48kHz is extracted with 1/10 ground, become fs=4.8kHz.Even like this, also be to guaranteeing to detect the enough sample frequency of frequency band of the required 1kHz of bearing defect.
Handle in (S205) in the frequency spectrum evaluation, the peak value of the frequency spectrum that the result of frequency resolution obtains has been carried out in detection by FFT calculation process (S204), this peak value and bearing abnormal frequency are compared, by estimating whether unusually with reference to the different parts abnormity diagnosis index corresponding with this comparative result.
In abnormity diagnosis point addition process (S206), count being evaluated as unusual number by frequency spectrum evaluation processing (S205).In number of occurrence determination processing (S207), judge whether the number of times (estimating frequency n 1) that has carried out frequency spectrum evaluation processing (S205) reaches the times N 1 of regulation.Phase-shift processing (S208) is performed under the situation that is judged to be the times N 1 (among the S207 denying) that does not reach regulation by number of occurrence determination processing (S207).Dephase and select the later processing of processing (S203) repeatedly by this processing.
Waveform is taken into number of times determination processing (S209) and is performed under the situation that is judged to be the times N 1 (among the S207 being) that has reached regulation by number of occurrence determination processing (S207).Be taken under the situation of times N 2 that frequency n 2 do not reach regulation (among the S209 not), the later processing of sampling processing (S201) repeatedly at waveform.Be taken under the situation of times N 2 that frequency n 2 reaches regulation (among the S209 being) at waveform, proceed to evaluation/determination processing (S210).
In evaluation/determination processing (S210), carry out the unusual evaluation/judgement of bearing based on counting by the anomaly evaluation of abnormity diagnosis point addition process (S206) counting.
As mentioned above, in the present embodiment, when (S205) handled in the evaluation of each enforcement frequency spectrum, phase deviation is implemented repeatedly to select processing (S203), adopt for a sample waveform and carry out repeatedly the method that FFT calculation process (S204) is come the accumulative total diagnostic points.This is because the meaning of sampling with the frequency of 48kHz in the data that will only carry out extracting are estimated with once FFT diminishes, and becomes with from the beginning of sampling identical with the frequency of 4.8kHz.Even the rotor of bearing is long by the cycle and since hinder more little then by in the shock wave that causes decaying in the short time more, so high sampling is effectively original, carry out repeatedly phase-shift processing and FFT calculation process for it is applied in a flexible way.
Figure 19 (a) and (b) represent to extract for vibration pack winding thread waveform skew phase place the situation of processing.Phase shift is equivalent to sampled point is offset 1 point.In the example of Figure 19, expression is with respect to only sampling ● the state of (a) of (stain), and state (b) is the situation of 1 of only phase shift and only sample again zero (white point).In the example of Figure 19,, therefore obtain the maximum 5 groups group of sampling again owing to extract with 1/5 ground.
Thereby, under situation about extracting, obtain the maximum 10 groups group of sampling again with 1/10 ground.In the following table illustration to all these 10 groups carry out the FFT computing, and detected frequency component pairing evaluation point has been carried out the result of accumulative total.
[table 2]
Detected components Count
Fundametal compoment, 2 rank, 4 rank 4
Fundametal compoment, 2 rank 2
Fundametal compoment, 4 rank 1
Figure 20 represents to hinder diagnostic result by the foreign steamer of the rolling bearing repeatedly of phase shift and FFT calculation process.In this example, prepared to have added people's industrial injury from sample (1,2) with less than the normal sample (3,4) of hindering on the foreign steamer orbital plane of rolling bearing, the characteristic frequency component of hindering with the foreign steamer of bearing is that principal component is tested.The accumulative total of getting this point at certain hour, with it as foreign steamer defective index.The moving average by detecting the FFT frequency spectrum and the peak value of smoothing differential carry out the detection of frequency component, and then according to the size of spectrum component and high-order 9 components of boil down to.
Figure 21 be expression for time of FFT calculation process, contrast has been used the situation of DSP220b and the curve map of the situation of only carrying out with CPU220a.In this example, owing to counting of FFT computing being cooperated being made as the 4k word length with the capacity that order is read in the high-speed memory 220e of DSP 220b with two data simultaneously, so carry out the FFT computing at high speed with DSP.
As described above, in the abnormity diagnostic system of this embodiment, because for the envelope waveform that is digitized, thereby with sample frequency drop to DSP220b in the data number of capacitance balance of X/Y-RAM220e carry out the FFT computing, so can carry out handling by the high speed FFT of DSP220b, and handle and reduce counting of the FFT computing that is used to resolve the envelope waveform by after absolute value is handled, extracting, improve thereby can take into account the raising of frequency resolution of signal and the efficient of FFT computing, and high precision and implement the abnormity diagnosis of bearing expeditiously.
[the 5th embodiment]
Figure 22 is the process flow diagram of flow process of a series of processing in the abnormity diagnostic system of expression the 5th embodiment.This flow process is being selected the Filtering Processing (S211) that processing (S203) is inserted through wave digital lowpass filter before, and this point is different with Figure 18.In addition, number of occurrence determination processing (S207) and phase-shift processing (S208) are omitted.
Like this, when selecting processing (S203), by reducing frequency band in advance by wave digital lowpass filter, thus the influence that can knownly obscure etc. and carry out the FFT calculation process of low-frequency band reliably.Compare with the mode of Figure 18, the filter factor that additionally need be used for the program code of wave digital lowpass filter and cooperate the property calculation of wave filter to go out, but reliable carry out aspect the removing of noise favourable.Figure 23 represents that the foreign steamer of the situation of the 5th embodiment hinders diagnostic result.It is identical with the situation of Figure 20 to test employed sample.
[the 6th embodiment]
The functional-block diagram of embodiment of representing the abnormity diagnostic system of the 6th embodiment during Figure 24.
As shown in figure 24, the abnormity diagnostic system of the 6th embodiment comprises: simulation amplifilter 301, A/D transducer 302, digital filtering unit 303, select unit 304, absolute value unit (envelope processing unit) 305, null fill-in unit (interpolation processing unit) 306, peaceful (hanning) window function processing unit 307 of the Chinese, FFT unit 308, peak detection unit 309, bearing defect fundamental frequency computing unit 310, comparing unit 311, accumulated unit 312, diagnosis unit 313 and diagnostic result output unit 314.
Simulation amplifilter 301 is transfused to sensor (acoustic sensor etc.) the 317 detected signals that detected by sound that the plant equipment of diagnosis object is taken place or vibration.Simulation amplifilter 301 amplifies the signal of importing with the gain of regulation, the above signal of the frequency of truncation specification simultaneously.
A/D transducer 302 to the analog signal sampling by amplifilter 301, and is transformed to digital signal with the sample frequency of regulation.
Digital filtering unit 303 only makes the signal of allocated frequency band in the digital signal that is generated by A/D converter unit 302 pass through.
Select unit 304 and extract processing to sampling by the signal of digital low-pass filtering unit 303 by sample frequency with regulation.
Absolute value unit 305 will be obtained as the discretize data by the envelope (envelope waveform) of selecting the signal that extracts unit 304.
Null fill-in unit 306 carries out fast fourier transform and carries out the interpolation of spot patch position for the discretize data of the envelope that absolute value unit 305 obtained with frequency resolution arbitrarily.Here, null fill-in is meant for the sample frequency that makes FFT unit 308 becomes 2 power and is producing the interpolation of under the not enough situation discretize data supplementing 0 of envelope being adjusted.
Hanning window function processing unit 307 has been by to having been carried out signal times after interpolation is handled with the Hanning window function of specified period by null fill-in unit 306, thereby draws the employed signal of diagnosis.
FFT unit 308 carries out frequency resolution by fft algorithm to the signal that is drawn by Hanning window function processing unit 307, and generates the spectrum waveform signal.
Peak detection unit 309 detects the peak value of the frequency spectrum that is obtained by FFT unit 308.
Bearing defect fundamental frequency computing unit 310 calculates the basic function of the defective of expression bearing based on each element of inside by the rotating speed of revolution detector 315 detected rolling bearings and the bearing read from the ROM316 of each element of storage bearing.
Peak value that 311 pairs of comparing units are obtained by peak detection unit 309 and the frequency that is calculated by bearing defect fundamental frequency computing unit 310 compare, and will export after its consistent number of degrees value.
Accumulated unit 312 will add up and export its result from the output valve of comparing unit 311.
Diagnosis unit 313 is based on the accumulated result diagnosing abnormal of accumulated unit 312.
The diagnostic result of diagnostic result output unit 314 output diagnosis units 313.
Above-mentioned peak detection unit 309 comprises moving average processing unit, smoothing differential processing unit, threshold value module of selection, classification module of selection.
The moving average processing unit will be undertaken moving averageization by frequency spectrum (discrete data of frequency domain) the left and right symmetrically weighting that FFT unit 308 obtains.
Smoothing differential processing unit carries out the numerical differentiation computing when the moving average processing by the moving average processing unit, with the Frequency point of the sign change of the differential coefficient peak extraction as frequency spectrum.
The square mean square root that the threshold value module of selection is selected amplitude levels in the peak value that is extracted by smoothing differential processing unit is the peak value more than the threshold value.Threshold value is used the power average value of the peak value that is extracted by smoothing differential processing unit or the relative value that determines corresponding to square mean square root.
The classification module of selection in the peak value of selecting by the threshold value module of selection, from the square mean square root of amplitude levels big select peak value till the regulation number.As its simplest method, for example, can enumerate and use known sorting algorithm, a plurality of peak values are made as after the descending class method of selecting from the peak value of a high position about grade.
Here, about the block scheme (hardware structure diagram) of the embodiment of the microcomputer (MPU) of the concrete textural element of the abnormity diagnostic system of representing the 6th embodiment and peripheral circuit thereof, consider and the same structure of Figure 17.In addition, about these explanations and aforementioned same, so omit explanation here.
What in addition, carry out in the functional block that digital operation handles data volume maximum that once (per 1 circulation) handle by abnormity diagnostic system shown in Figure 24 is FFT unit 308.In order to realize FFT unit 308 by the DSP220b in the MPU220, the data that the FFT calculation process is used need be contained among the X/Y-RAM220e.
On the other hand, detect wound, need observe waveform, but the frequency that becomes the feature of wound is generally below the 1kHz with the frequency band till about 10kHz in order to resolve by bearing vibration.
In this example, the characteristic frequency of wound of supposing the bearing of diagnosis object is the following low frequency of 100Hz, with the frequency resolution of FFT unit 308 be made as 1Hz (± 0.5Hz), sample frequency is made as 32.768kHz, (Tw) is made as 750ms with the sampling time.Thereby the number of samples of original waveform is 32768 * 0.75=24576.The final stage of asking frequency spectrum in FFT unit 308, by carrying out null fill-in so that the sampling time becomes 1s, frequency resolution become 1Hz (± 0.5Hz).
The bandwidth of passing through of digital filtering unit 303 cooperates by the maximum frequency band of S/N ratio of vibration that causes unusually and noise chosen.For example, under knowing that in advance the S/N that peels off defective is than the situation in the frequency band maximum of 1kHz~4kHz, the bandwidth of passing through of digital filtering unit 303 is chosen to be 1kHz~4kHz.This digital filter can be by the formations such as wave filter of having used FIR wave filter, IIR filtering, FFT and contrary FFT (IFFT), but under the situation of the risc type microcomputer of the DSP of built-in fixed-point arithmetic mode, the FIR wave filter is suitable for.
Absolute value processing unit 305 is in order to eliminate the DC component and to average the line of reconstruction amplitude zero envelope or absolute value waveform.Handle (envelope processing) by this absolute value, the low frequency signal less than 1kHz that bearing defect causes is significantly changed.At this constantly, owing to also comprise high-frequency signal, carry out adding before the FFT FIR low-pass filter that the low frequency signal less than 1kHz is passed through so be preferably in, but owing to original waveform has been implemented to handle by the bandpass filtering treatment of digital filtering unit 303 with by the absolute value (envelope extraction) of absolute value processing unit 5, even so omit the previous low-pass filtering treatment in FFT unit 308, also very little to the influence of the diagnostic accuracy of bearing defect.
Selecting the interpolation rate of the extraction yield (extraction amount) of the FFT computing point of unit 304 and null fill-in unit 306 or interpolation figure place counts etc. according to frequency band, frequency resolution, the FFT computing that will analyze and determines.In this example, owing to realize the hypervelocity FFT calculation process of the DSP220b in the MPU220, so the FFT computing is counted nature by the capacity limit that can pass through the X/Y-RAM220e of parallel private bus visit from DSP220b.
Data by FFT unit 308 calculation process are made of real part and imaginary part, are assigned to X-RAM and the Y-RAM of X/Y-RAM220e respectively.If adopt the mode of supplying with storage area in input and output, then can be to the data progress row FFT of 8kB.If the resolution of A/D transducer 223b is 16 (2 bytes), then also be made as 2 byte longs in advance by the computing variable, can be by the data of DSP220b high speed processing till 8192/2 byte is at 4096.
With sample frequency (fs, fft) the interval length of the FFT when the 1.0Hz is made as Tw, during fft, necessary frequency resolution Δ fw is expressed as Δ fw=1/Tw, fft.Thereby as long as Tw, fft=1 just satisfies essential condition.
In this example, because with sampling time Tw, fft is made as 0.75s, so 0.25s sampling time deficiency.This insufficient section carries out interpolation by null fill-in unit 306, but has only carried out null fill-in, and the data number also reaches 32786.
Here, originally number of samples (32768) is extracted at 4096 o'clock that count for by the FFT computing of the upper limit of the byte long decision of the capacity of the X/Y-RAM220e of DSP220b and computing, the data number become originally 1/8, the sample frequency of FFT unit 308 is also cut down to 32768/8=4096.Thereby the upper limit (being Qwest's frequency) of the frequency that can be analyzed by FFT unit 308 becomes its half 2.048kHz, even but also covered the frequency (less than 1kHz) of the defective of expression bearing so fully.
In this example, carry out the sampling (24576 point) of 0.75s at 32.768kHz as benchmark, handle in the extraction of carrying out 1/8 by digital filtering unit 303 and absolute value processing unit 305 after with the frequency band low frequencyization, hits and sample frequency are reduced to and 4.096Hz respectively at 3072, for the point of less than 4096, cover 1024 zero (0) after 3072 and be made as 4096 sample waveform data (with reference to Figure 25).This Wave data is transfused to FFT unit 308 via Hanning window function unit 307.This Hanning window function unit 307 is by multiply by the Hanning window function to the incoming wave graphic data, thereby alleviates the influence at the Wave data two ends that are transfused to FFT unit 308.By carrying out FFT, obtain frequency spectrum with the resolution of 1Hz by 308 pairs of these Wave datas in FFT unit.The frequency spectrum data that obtains is transfused to peak detection unit 309.
The frequency spectrum left and right symmetrically weighting that will obtain by FFT unit 308 in the peak detection unit 309 and moving averageization.Thus, smoothedization of frequency spectrum, noise are alleviated.
And then, when the moving average processing, carry out the numerical differentiation computing.Then, with the Frequency point of the sign change of differential coefficient as the peak value of frequency spectrum and extract.Then, selecting the square mean square root of amplitude levels in the peak value that extracts is the above peak value of threshold value, selects the peak value till the number (for example 10) of regulation from the big peak value of the square mean square root of amplitude levels.
On the other hand, bearing defect fundamental frequency computing unit 310 calculates the basic function of the defective of expression bearing based on each element of inside by the rotating speed of revolution detector 315 detected rolling bearings and the bearing read from ROM316.The detection of the rotating speed by 315 pairs of bearings of revolution detector repeats a plurality of cycles with vibration detection synchronous (for example 0.75s once) by vibration transducer 317.
Then, be transfused to comparing unit 311 by the frequency of peak detection unit 309 detected peak values with by fundamental frequency and each cycle synchronisation that bearing defect fundamental frequency computing unit 310 calculates.
Comparing unit 311 compares the frequency of fundamental frequency and higher hamonic wave and peak value when the frequency of each peak value and fundamental frequency are transfused to, and gives corresponding the counting of the degree consistent with both (quantizing), and this value is outputed to accumulated unit 312.The example of the adding method of counting here is as shown in the table.
[table 3]
Detected components Abnormity diagnosis is counted
Basic wave, 2 rank, 3 rank, 4 rank 16
Basic wave, 2 rank, 3 rank 8
Basic wave, 2 rank, 4 rank 8
Beyond above-mentioned 0
In this case, the diagnostic process that is taken into for once waveform is equivalent to give the degree corresponding diagnosis consistent with detected peak value to the fundamental frequency of the vibration that is caused by bearing defect that calculates with the higher harmonic components till 4 times and counts.
Then, detect, add to diagnose each time and count, and estimate the aggregate-value of counting, thereby thereby abnormity diagnosis is carried out in the influence that can alleviate the deviation of frequency spectrum accurately by the repetition frequency peak value.
The aggregate-value that Figure 26 has imported the bearing (defective) of defective with histogram graph representation and do not had the diagnosis of the bearing of defective (normal product) to count.This example internally bearing of wheel rotation type uses above-mentioned abnormity diagnostic system to test.In to produce the situation of damage in the bearing of wheel rotation type outside on the wheel track more, so in this example also externally wheel track import to hinder and test.4 groups of bearings are used for diagnosis object, 2 groups of importing foreign steamers are wherein hindered.Under this test condition, the fundamental frequency under the defective situation of outer wheel track is below the 100Hz.About first-harmonic, if the frequency of peak value fundamental frequency ± scope of 1.0Hz in, then be judged as consistent with the frequency of peak value, about 2 rank, each higher hamonic wave of 3 times, 4 times, if the frequency of peak value each higher hamonic wave frequency ± scope of 2.0Hz in, then be judged as consistently, add counting of above-mentioned table 1 with the frequency of peak value.Test duration was made as 60 seconds.The sampling time in 1 cycle is 0.75 second, is used for output abnormality and diagnoses the used time cost of the computing of counting about 0.15 second.Data are taken into by DMA is undertaken, and calculation process was carried out with 0.15 second during the data that are taken into next time, so as shown in figure 27, all in all, by making the native system running 60 (+0.15) second continuously, can carry out the inferior diagnostic process in 80 (=60/0.75).
As shown in figure 26, even normal product, noise a little also is counted, but with the difference of defective clearly.Since produce very big difference between defective and the normal product, thus very big owing to the scope of threshold value being got, therefore this scope can be sent interim alarm as gray area.
Figure 21 be expression about time of FFT calculation process, contrast is used the situation of DSP220b and is only used the curve map of the situation that CPU220a carries out.Because for consistent with the capacity of X/Y-RAM220e in the DSP220b and counting of FFT computing reduced after the frequency bands by selecting unit 304 extractions, so can carry out the FFT computing very at high speed by DSP220b by digital filtering unit 303.
As described above, abnormity diagnostic system kind at present embodiment, for detection signal is transformed to digital signal, take out the signal of the required frequency band of diagnosis, ask the envelope that it has been carried out extracting the signal of handling, and this envelope is carried out FFT with frequency resolution arbitrarily, and carry out the interpolation of spot patch position, and then obtain diagnosing after the employed signal by the Hanning window function, carry out frequency analysis by FFT, come diagnosing abnormal based on the frequency spectrum that obtains, so thereby with sample frequency consistent and frequency resolution detection signal is carried out FFT and can implement abnormity diagnosis accurately with the employed arithmetic unit of FFT computing.
In addition; to carry out the moving average processing by the frequency spectrum that FFT obtains; and then this frequency spectrum carried out the smoothing differential; thereby with the symbol of differential coefficient after the negative Frequency point that changes of forward detects as peak value; extract the above peak value of threshold value of regulation; to after their classification a high position wherein be stipulated that several are as peak extraction; ask these peak value kinds corresponding to the peak value of the principal component of vibration or with the consistent degree of the unusual frequency of the principal component of vibration and corresponding peak value of higher hamonic wave and expression diagnosis object; to this unanimity degree give count and accumulative total repeatedly; judge unusually by estimating this aggregate-value; even so the S/N of abnormal signal or unusual omen signal and noise signal than little condition under, also unusual high precision and implement abnormity diagnosis expeditiously and the noise signal flase drop can not surveyed and be unusual or unusual omen signal.
[the 7th embodiment]
Figure 28 is the summary construction diagram of rail truck that comprises the apparatus for diagnosis of abnormality of the 7th to the 13rd embodiment.Rail truck 401 comprises 4 groups of wheels (totally 8) 402-1~402-4, they is remained on 4 bearing 404-1~404-4 under the chassis 403 with rotating freely, and chassis 403 is provided with apparatus for diagnosis of abnormality 410.
Apparatus for diagnosis of abnormality 410 sensor packet unit 420 and console panel 430.Sensor unit 420 is the unit that detect the vibration on chassis 403.Console panel 430 comprises diagnostic circuit 431, and based on the output signal of sensor unit 420, diagnosis has or not that peeling off of bearing 404-1~404-4 or wheel 402-1~402-4's is flat etc. unusual.It is the diagnosis content (alarm signal) of diagnostic circuit 431 is sent to pilothouse or instruction place by the communication lines in the vehicle 401 system.
In the 8th to the 12nd following embodiment, as Figure 38,, utilize Peak/RMS (Root Mean Square) as the parameter of deteriorations such as peeling off of expression bearing 404-1~404-4.Here, Peak is the absolute value of the peak swing in certain interval, and RMS is the square root of the Fang Jun of the vibration voltage in certain interval.Here, be the waveform of the vibration on the chassis 403 of expression as Figure 39 as the waveform of object, comprise the deterioration noise that it doesn't matter of impulsive sound that the seam of track causes or grating etc. and the mechanical organ of formation vehicle 401.This noise and bearing 404-1~404-4 peel off or the vibration that causes unusually such as flat of wheel 402-1~402-4 is compared, have very large amplitude.
Figure 29 is the block scheme of the 7th configuration example of expression sensor unit 420.Sensor unit 420 involving vibrations sensors (Sens) 421 shown in Figure 29, as the square root calculation circuit (RMS-DC of the Fang Jun of parameter value testing circuit and analog operational circuit; After, record and narrate computing circuit into RMS) 422, peak detection circuit (Peak) 423, first comparer (CMP1) 424, second comparer (CMP2) 425, the reference voltage output circuit (Vref.) 426 of circuit as a comparison as peak value-reference point comparator circuit.In addition, RMS computing circuit 422 carries out the square root calculation of Fang Jun based on following calculating formula (3).
[formula 6]
Σ x i 2 n - - - ( 3 )
X i: i measured value (value of I constantly)
N: hits (interval long) vibration transducer 421 is the sensors that detected the vibration of vertical direction by piezoelectric ceramics, detects the vibration of the frequency band of 50Hz~10kHz, and this vibrational waveform is exported as electric signal.The output signal of vibration transducer 421 (vibration signal) is imported RMS computing circuit 422 and peak detection circuit 423 after being amplified by amplifying circuit 427 simultaneously.
The vibration signal that RMS computing circuit 422 is transfused to by processing, thereby the direct current signal of the output voltage (following note make RMS voltage) suitable with the voltage RMS of this vibration signal.This RMS computing circuit 422 for example uses buffer amplifier built-in, absolute value circuit, square/division circuit, output be with the RMStoDC converter IC of wave filter amplifier circuit etc.As the object lesson of this RMStoDC converter IC, enumerate commodity signal ' AD637 ': Analog Devices, Inc (ア Na ロ グ デ バ イ セ ズ) system etc.
The time constant of RMS computing circuit 422 can be by external capacitor decision.In this example, be made as 100ms.In addition, RMS computing circuit 422 comprises the circuit that the RMS voltage amplification is exported for certain multiplying power.In this example, suppose the voltage of 4 times of outputs.
The crest voltage of the vibration signal that peak detection circuit 423 outputs are transfused to.The time constant of peak detection circuit 423 equates that with RMS computing circuit 422 the amplification degree of the voltage level during output is 1.
The output signal of RMS computing circuit 422 is transfused to first input end of first comparer 424.The output signal of peak detection circuit 423 is transfused to second input terminal of first comparer 424 and first input end of second comparer 425.Second input terminal of second comparer 425 is transfused to the reference voltage from reference voltage output circuit 426.
424 pairs of signal voltages from RMS computing circuit 422 of first comparer are 4 times voltage of RMS voltage and are that crest voltage compares from the signal voltage of peak detection circuit 423.Then, voltage (first voltage) signal of output+5V if crest voltage is big, the voltage of output-5V if crest voltage is little (second voltage) signal.Promptly export crest factor (Peak/RMS) and whether surpass 4.
425 pairs of signal voltages from peak detection circuit 423 of second comparer are crest voltage and compare from the reference voltage of reference voltage output circuit 426.Then, the voltage signal of output+5V if crest voltage is bigger than reference voltage, the voltage signal of output-5V if crest voltage is little.Reference voltage is picked as the high level of voltage of signals level that causes unusually than bearing etc.
The output signal of first comparer 424 is transfused to the detection signal input terminal of gate circuit 428.The output signal of second comparer 425 is transfused to signal input end of gate circuit 428.Gate circuit 428 providing from second comparer 425-still export under the situation of the voltage signal of 5V signal from first comparer 424 (+5V or-5V), but providing from second comparer 425+situation of the voltage signal of 5V under, export the voltage signal of 0V usually.
The relation of the output of comparative result and gate circuit 428 in first and second comparer 424,425 of following table 4 expression sensor unit 420.In table 1,4 * RMS is the input voltage of first comparer 424, and peak is the input voltage (crest voltage) of second comparer 425, and Vref is the output voltage (reference voltage) of reference voltage output circuit 426, and Output is the output voltage of gate circuit 428.In addition, the output voltage that should be noted that first and second comparer 424,425 is respectively 2 values.True and false according to from the signal of second comparer 425 of gate circuit 428, control make from the signal former state of first comparer 424 by (5V or+5V), or make its invalid (0V).The output of second comparer 425 is open and close controlling signals, and the output of gate circuit 28 is an output, and the output of first comparer 424 is signal sources.
[table 4]
peak≥4×RMS peak<Vref Output
T T +5V
F T +5V
T F 0
F F 0
T:True F:False
As shown in table 4, only under the situation below the voltage of signals level that causes unusually that by the crest voltage (peak) of vibration transducer 421 detected vibration signals is bearing etc. in reference voltage (Vref), the signal of the magnitude relationship of 4 times voltage of sensor unit 420 output expression crest voltage (peak) and RMS voltage (+5V or-5V).Thus, the very large signal that the deterioration noise that it doesn't matter of impulsive sound that the seam by track causes or grating etc. and the mechanical organ that constitutes vehicle 401 causes prevents that the output of sensor unit 420 is saturated.Represent that for+5V Peak/RMS surpasses certain benchmark at the output voltage of sensor unit 420, otherwise the output voltage of sensor unit 420 represents that for-5V Peak/RMS does not satisfy certain benchmark.
The output signal of the diagnostic circuit 431 common monitoring sensor unit 420 of console panel 430, with in the unit interval (being 60 seconds here)+ratio of the output time of the voltage signal of 5V recently calculates (with reference to Figure 30) as the duty that Peak/RMS surpasses benchmark.Then, send the alarm signal of the alert level corresponding with the dutycycle that calculates.The alarm signal per second is updated, and the received signal till the calculating of dutycycle was played before 60 seconds during usually based on the up-to-date signal that receives from sensor unit 420 is carried out.This alarm signal is sent to pilothouse or instruction place by the communication line in the vehicle 401.Be provided with the different a plurality of warning lights of shades of colour in pilothouse or instruction place, according to the alert level of the alarm signal that receives from sensor unit 420, the warning light of specified color is lighted or is glimmered.
The corresponding relation of above-mentioned dutycycle of following table 5 illustration and alert level.In addition, the also corresponding relation of the color of illustration alert level and warning light in the table 5.
[table 5]
Peak/RMS surpasses the dutycycle of benchmark Alert level The color of warning light
Less than 20% Normally
More than or equal to 20% less than 40% Note Yellow
More than or equal to 40% less than 80% Alert level I Orange
More than or equal to 80% Alert level II Red
This apparatus for diagnosis of abnormality 410 is used for monitoring that having travelled of rail truck is no abnormal, and the alarm shown in the table 5 and derailing precognition etc. are different, does not suppose to stop train detecting under the unusual situation.Even the alert level II of the highest alert level in the table 5 for example is that a week is with the interior alarm that requires the degree of visual examination.Even under the flat situation that the peeling off of bearing, wheel have taken place, this train neither can not move immediately, so the standard that expression is checked can be described as the fundamental purpose of using apparatus for diagnosis of abnormality 410.But the high more urgency of then checking of the travel speed of train is high more.Surpass in the such hypervelocity railway of 200km at F-Zero, also implement to check if the alarm of above-mentioned alert level II takes place then preferably stop train fast.
[the 8th embodiment]
Figure 31 is the block scheme of the 8th embodiment of expression sensor unit 420.Sensor unit 420 involving vibrations sensors (Sens) 421, amplifier 427 shown in Figure 31, be transfused to from different three bandpass filter (BPF) 441-1~441-3 of the frequency band of the signal of amplifier 427, handle three signal processing unit 442-1~442-3 respectively by the signal of each bandpass filter 441-1~441-3.Bandpass filter 441-1~441-3 has the centre frequency of 500Hz, 1.5kHz, 3kHz in this example respectively.
Each signal processing unit 442-1~442-3 comprises RMS computing circuit (RMS-DC) 422, peak detection circuit (Peak) 423, comparer (CMP) 424 respectively.
Vibration transducer 421 is same with the example of Figure 29, detects the vibration of the frequency band of 50Hz~10kHz, and its vibrational waveform is exported as electric signal.The output signal of vibration transducer 421 (vibration signal) is transfused to three bandpass filter 441-1~441-3 after being amplified by amplifying circuit 427 simultaneously.Be transfused to RMS computing circuit 422 and peak detection circuit 423 in signal processing unit 442-1~442-3 respectively by the different signal of each frequency band of each bandpass filter 441-1~441-3.That is, in this example, be transfused to the vibration signal of the low-frequency band (centre frequency 500Hz) of having passed through the first bandpass filter 441-1 in RMS computing circuit 422 in the first signal processing unit 422-1 and the peak detection circuit 423.Be transfused to the vibration signal of the intermediate frequency band (centre frequency 1.5kHz) of having passed through the second bandpass filter 441-2 in RMS computing circuit 422 in the secondary signal processing unit 442-2 and the peak detection circuit 423.Be transfused to the vibration signal of the high frequency band (centre frequency 3kHz) that has passed through the 3rd bandpass filter 441-3 in RMS computing circuit 422 in the secondary signal processing unit 442-2 and the peak detection circuit 423.
RMS computing circuit 422 in each signal processing unit 442-1~442-3 is by handling the vibration signal of input, thereby exports 4 times voltage of the RMS voltage of this vibration signal.
The crest voltage of the vibration signal that peak detection circuit 423 outputs in each signal processing unit 442-1~442-3 are transfused to.The enlargement factor of the voltage level during peak detection circuit 423 outputs is 1.The output signal of the RMS computing circuit 422 in each signal processing unit 442-1~442-3 is transfused to first input end of comparer 424.The output signal of peak detection circuit 423 is transfused to second input terminal of comparer 424.
424 pairs of signal voltages from RMS computing circuit 422 of comparer in each signal processing unit 442-1~442-3 are 4 times voltage of RMS voltage and are that crest voltage compares from the signal voltage of peak detection circuit 423.Then, the voltage signal of output+5V if crest voltage is big, the voltage signal of output-5V if crest voltage is little.
The diagnostic circuit 431 of console panel 430 is the output signal of each signal processing unit 442-1~442-3 of monitoring sensor unit 420 all the time, with in the unit interval in each signal (being made as for 60 seconds here)+ratio of the output time of the voltage signal of 5V recently calculates (with reference to Figure 32) as the duty that Peak/RMS surpasses benchmark.Then, send the alarm signal of the alert level corresponding with the dutycycle that calculates.
In this example, different with the example of Figure 29, the employed crest voltage of the calculating of dutycycle is not established the upper limit, so, in the benchmark of dutycycle, contain noise component as the noise countermeasure.
In the example of table 5, in less than the normal dutycycle 20% of expression, should estimate the deterioration noise that it doesn't matter of impulsive sound that the seam of track causes or grating etc. and the mechanical organ that constitutes vehicle 401.
[the 9th embodiment]
Figure 33 is the block scheme of the 9th embodiment of expression sensor unit 420.Sensor unit 420 shown in Figure 33 comprises: vibration transducer (Sens) 421, amplifier 427, be transfused to low-pass filter 451 from the signal of amplifier 427, handle the signal processing unit 452 by the signal of low-pass filter 451.Low-pass filter 451 has the cutoff frequency of 1kHz degree.By blocking signal, can cut off the very large signal that the deterioration noise that it doesn't matter of impulsive sound that the seam of track causes or grating etc. and the mechanical organ that constitutes vehicle 401 causes, and only catch the vibration of mechanical organ above the 1kHz degree.
Signal processing unit 452 comprises RMS computing circuit (RMS-DC) 422, peak detection circuit (Peak) 423, comparer (CMP) 424.
Vibration transducer 421 is same with the example of Figure 29, detects the vibration of the frequency band of 50Hz~10kHz, and its vibrational waveform is exported as electric signal.The output signal of vibration transducer 421 (vibration signal) is transfused to low-pass filter 451 after being amplified by amplifying circuit 427.Then, be transfused to signal processing unit 452 by the signal below the 1kHz degree of low-pass filter 451.Processing among processing in the signal processing unit 452 and each the signal processing unit 442-1~442-3 among Figure 31 is same.
According to the 9th embodiment, owing to only catch 1kHz degree following signal and input signal processing unit 452, so the very large signal that the deterioration noise that it doesn't matter of impulsive sound that can cause by the seam of track or grating etc. and the mechanical organ that constitutes vehicle 401 causes prevents that the output of sensor unit 420 is saturated.
[the 10th embodiment]
Figure 34 is the block scheme of the 10th configuration example of expression sensor unit 420.Sensor unit 420 shown in Figure 34 comprises except the structure of Figure 33, also comprises the Hi-pass filter (HPF) 453 that is transfused to from the signal of amplifier 427, the signal processing unit 454 of handling the signal that passes through Hi-pass filter 453.The structure of signal processing unit 454 is identical with the signal processing unit 452 of low-pass filter 451 sides.Hi-pass filter 453 has the cutoff frequency of 1kHz.To the signal more than the signal processing unit 454 input 1kHz, can detect the spot corrosion of bearing by only effectively.
[the 11st embodiment]
Figure 35 is the block scheme of the 11st embodiment of expression sensor unit 420.Sensor unit 420 shown in Figure 36 comprises: vibration transducer (Sens) 421, amplifier 427, with the output transform of amplifier 427 be digital signal A/D transducer (ADC) 455, handle microprocessor (MPU) 456 from the signal of A/D transducer 455.MPU456 carries out digital processing according to being stored in its inner program to the input signal from A/D transducer 455, thereby finish the function of above-mentioned RMS arithmetic element 422, peak detection circuit 423 and comparer 424, simultaneously also carry out table 5 in illustrative diagnostic process.Thereby,, can omit the diagnostic circuit 431 of console panel 430 according to this configuration example.Also can use DSP (digital signal processor) to replace MPU456.In addition, if the microprocessor that uses the A/D transducer built-in replaces MPU456, then can omit external A/D transducer 455.
But, using MPU or DSP to carry out under the situation of calculation process, the RMS calculation process becomes quite heavy processing easily.Square the summation computing in the system that carries out the fixed-point number computing, cause saturatedly easily, generally do not have subduplicate order.
Thereby, if the RMS calculation process is quite complicated and other processing is fairly simple, then compare with using MPU or DSP, realize making the digital operational circuit of the preferential special use of the treatment effeciency of RMS computing more favourable by field programmable gate array (FPGA).About computing circuit by FPEG, according to the signal Processing of reality and difference, but wish high speed for the dominance of microcomputer or DSP by hardware, the possibility that reduces than microcomputer circuit also improves.In microcomputer initial handling totalizer and multiplier etc. write after in EPGA, needing.But, conversely, owing to can only load calculation function and peripheral function in case of necessity, so can expect the parallelization high speed of the miniaturization exclusive disjunction of device.
[the 13rd embodiment]
Here, in the 13rd embodiment shown in Figure 37, by the input signal from A/D transducer (ADC) 455 is carried out digital processing, thereby realize the function of above-mentioned RMS computing circuit 422, peak detection circuit 423 and comparer 424, comprise the digital circuit of EPGA457 simultaneously as the special use of illustrative diagnostic process in also carry out table 5.In addition, in the explanation of the foregoing description, use the parameter of Peak/RMS (crest factor), but also can replace them and use kurtosis (Kurtosis), impact index (the Peak/ absolute value is average) or form factor (the RMS/ absolute value is average) as deteriorations such as peeling off of expression bearing 404-1~404-4.
[formula 7]
1 n Σ i = 1 n ( x i σ ) 4 - - - ( 4 )
N: hits (interval long)
Xi: i measured value (value of I constantly)
σ: standard deviation
Mean value is that σ equates also to have no relations with the RMS value in zero the vibrational waveform.The biquadratic circuit can be used the squaring circuit that comprises in the RMS circuit.Thereby, use kurtosis (Kurtosis) to replace the embodiment of the crest factor (Peak/RMS) in the foregoing description also can realize.That is, use the circuit of asking kurtosis, impacting index or form factor to replace the structure of above-mentioned RMS computing circuit (RMS-DC) 422 also to be contained in the apparatus for diagnosis of abnormality of the present invention.
[the 14th embodiment]
The apparatus for diagnosis of abnormality of the 14th embodiment at first, is described with reference to Figure 40~Figure 45.
As shown in figure 40, a rail truck 500 has been installed 4 wheels 501 by former and later two chassis supportings on each chassis.The vibration transducer 511 that the on-stream vibration that takes place from rotary supporting device 510 is detected has been installed in the rotary supporting device of each wheel 501 (bearing housing) 510.
Being equipped with two whiles (roughly simultaneously) on the console panel 515 of rail truck 500 is taken into the sensor signal of 4 channels and the apparatus for diagnosis of abnormality 550 of implementing diagnostic process.That is, the output signal of 4 vibration transducers 511 that are provided with on each chassis is transfused to the different apparatus for diagnosis of abnormality of every chassis 550 via signal wire 516 respectively.In addition, also be transfused to the rotational speed pulse signal of the speed probe (omitting diagram) that detects from rotating speed in the apparatus for diagnosis of abnormality 550 to wheel 501.
As shown in figure 41, in the rotary supporting device 510 as an example, be provided with the axle bearing 530 as rotatable parts, axle bearing 530 comprises: as the interior wheel 531 that is embedded in the rotor wheel on the rotation axis (not shown) outward, as the foreign steamer 532 that is embedded in the fast pulley in the shell (not shown), the retainer (not shown) that rollably keeps as the rolling body 533 of a plurality of rotors between wheel 531 and the foreign steamer 532 in being configured in, with rolling body 533 freedom.Vibration transducer 511 is retained as the attitude of the vibration acceleration that can detect gravity direction, and is fixed near the foreign steamer 532 of shell.Vibration transducer 511 uses various sensors such as acceleration transducer, AE (acousticEmission) sensor, ultrasonic sensor, shock pulse sensor.
As shown in figure 42, apparatus for diagnosis of abnormality 550 has sensor signal processing unit 550A, diagnostic process unit (MPU) 550B.Sensor signal processing unit 550A has four amplifilters (AFILT) 551.And the output signal of 4 vibration transducers 511 is imported amplifilter 551 separately.Each amplifilter 551 has the function of analogue amplifier and the function of antialiasing concurrently.Simulating signal by these 4 amplifilters, 551 amplifications and filtered 4 channels, signal based on diagnostic process unit (MPU) 550B, switch to the signal of each channel by the Port Multiplier that works as handoff functionality (MUX), be transformed to digital signal by AD transducer (ADC) 553, and be taken into diagnostic process unit (MPU) 550B.On the other hand, by after waveform shaping circuit 555 shapings, record and narrate the umber of pulse of unit interval by time counter (omitting diagram) from the rotational speed pulse signal of speed probe, this value is used as tach signal input diagnostic process unit (MPU) 550B.Diagnostic process unit (MPU) 550B is based on by vibration transducer 511 detected vibrational waveforms with by the detected tach signal of speed probe, the execute exception diagnosis.The diagnostic result of diagnostic process unit (MPU) 550B is transfused to communication line 520 (with reference to Figure 40) via line driver (LD) 556.Communication line 520 is connected to warning horn, the alarm action of carrying out when carrying out taking place unusually in the flat grade at wheel 501.
By the detected tach signal of speed probe roughly certain fixing speed (185~370min in the present embodiment, -1) time, diagnostic process unit (MPU) 550B handles sample frequency fs and the certain waveform blocks of data of hits Ns, thereby carries out the flat detection of wheel 501.Specifically, when being made as fs=2kHz, Ns=2000, the interval length=1sec of blocks of data.By to comparing, thereby carry out flat detection at the number of times of between this second the flat vibrational waveform pulse that causes being counted with from the number of times that wheel 501 rotates between the detected speed of a motor vehicle one second by speed probe.
The vibration acceleration that wheel 501 takes place under the flat state is big, and the value of the caused vibration acceleration of vibration of common vehicle is littler than it usually.In addition, the vibration of track seam is with flat equal or than the grade of its big vibration acceleration.And then the grade of the vibration acceleration that the friction of the track at the turning of track and wheel 501 causes is also with flat or that the track seam causes is equal.Diagnostic process unit (MPU) 550B portion within it has storer (RAM) 559, can use it to carry out very at high speed FFT or digital filtering.Thus, can carry out processing (that is, comparing the calculating of the short time with suitable surplus) in real time to the vibration transducer 511 of 4 channels with the sampling time.
On the other hand, cause one-shot in flat one week of rotation, and the situation of the impact that the seam of track causes under the situation of the impact that track friction causes, takes place brokenly with more long period generation.Therefore, in the present embodiment, be conceived to surpass the systematicness that the impact (pulse) of the threshold value of flat distinctive vibration acceleration takes place, shock wave number of times to the unit interval in the certain speed is roughly counted, if this count number is roughly consistent with the rotation number of wheel, then, flat possibility height carries out abnormity diagnosis as taking place.
And then in the present embodiment, design is carried out the algorithm of diagnostic process repeatedly to identical wheel 501, compares with the statistical determination methods of the deviation of the count number of having considered umber of pulse and The noise etc., and abnormality diagnostic reliability improves.
Figure 43 represent apparatus for diagnosis of abnormality 550 4 channels vibration data be taken into sequential chart with data parsing.Vibration data constantly is taken into apparatus for diagnosis of abnormality 550, but can be divided into certain sampling interval according to diagnosis object.The diagnosis of bearing 530 (peeling off detection) is required, and to be taken into period T 1 just enough less than one second, for the influence of the contact noise that alleviates track and wheel 501, also preferably is the short time as far as possible.On the contrary, unusual for the plane of rotation that detects wheel 501 needs to detect the impacts of wheel 501 each rotations, so the period T 2 of 1 second degree that need be longer than period T 1.
The period T 1 that is taken into that will be used for the vibration data of bearing diagnosis is made as consistent for example 0.67 second of required time that is taken into the vibration data of 4 channels (channel), and when sample frequency is made as 20kHz, during 1 period T 1, be taken into 4 * 0.67 * 20000 data.Thereby, in the time of will being used for being taken into period T 2 and being made as 1 second of vibration data of vehicle diagnostics, carrying out being taken into of vibration data and became less than 0.33 second with the period T 1 that is taken into of the vibration data that is used for bearing diagnosis.Therefore, by connecting 1 interval, be the data of period T 1 and the 0.33 second last data in previous interval, thereby become the data of period T 2.Wherein, as described later since the data number can by filtered select to handle extract, so 1 channel can be made as below 2000.Its result is 0.67 second by making the required time of diagnosis of carrying out 4 channel wheels 501 and bearing 530 less than period T 1, can make the processing time of wheel, bearing diagnosis data have surplus.
In the present embodiment, diagnostic process unit (MPU) 550B walks abreast and carries out being taken into and wheel, bearing diagnosis data processing of above-mentioned vibration data.That is, carry out the real-time processing of finishing wheel, bearing diagnosis data processing in the period T 1 that is taken at the vibration data of 4 channels.Should handle in real time and by diagnostic process unit (MPU) 550B the Port Multiplier 552 of sensor signal processing unit 550A and AD transducer 553 be carried out the interrupt control line data of going forward side by side and sample and realize.In addition, also can realize by the data sampling of direct memory memory controller (DMA).
Like this, has surplus by the time processing that makes wheel, bearing diagnosis data, the parallel processing with wheel, bearing diagnosis data of being taken into of carrying out vibration data, thereby can eliminate the accident failure of data, the reliability of the diagnostic result that obtains so can improve that the data that comprise the probability process that the waving of the scrambling of track or car body, load-carrying change etc. are caused are carried out statistical treatment.
Figure 44 represents the motion flow of diagnostic process unit (MPU) 550B.Being taken into of diagnostic process unit (MPU) 550B executed in parallel vibration data, i.e. the AD conversion of the sensor signal of 4 channels and sampling (S300) and bearing, wheel diagnostic data are handled (S400).
Handle in (S400) at bearing, wheel diagnostic data, when carrying out the renewal (S401) of the vibration data of 4 channels at every turn, carry out rotating speed successively and detect the output processing (S405) that the storage of handling (S402), diagnostic process (S403), diagnostic result keeps handling (S404) and result of determination.
Rotating speed detect to be handled the processing that signal that (S402) be based on speed probe detects the rotating speed of bearing 130.
Diagnostic process (S403) handles (S410) by bearing diagnosis and wheel diagnostic process (S420) constitutes.
Bearing diagnosis is handled (S410) and is based on the rotating speed of bearing 530 and handles the envelope waveform of vibration and the frequency peak that obtains detects the abnormity processing of bearing 530.Bearing diagnosis is handled in (S410), at first carry out from the vibration data that is taken into bandpass filtering (BPF) processing (S411) of the vibration data that extracts the intermediate frequency that the component of high frequency (more than the 3kHz) and low frequency (below the 200Hz) has been decayed, the data that extract are carried out selecting processing (S412) afterwards with the extraction yield of stipulating, carrying out the low-pass filtering treatment (S414) that absolute value is handled the component of (S413), extraction low frequency (1kHz is following) successively.And, in that further having been carried out, the data that extract select processing (S415) afterwards, and handle (S416) by carrying out the null fill-in fast Fourier transform (FFT), thereby obtain the frequency data of resolution 1Hz.These frequency data are implemented the peak detection process (S417) of smoothing differential, the fundamental frequency (with reference to Figure 45) of the bearing defect that obtains to rotating speed with from inner each element of bearing carries out the comparison till 4 times, thereby judges consistent, inconsistent (S418: the bearing defect determination processing).
Wheel diagnostic process (S420) is to detect the abnormity processing of wheel 501 from the phenomenon that the rotational synchronization with wheel 501 takes place to impact.The main occurrence cause of the impact that produces synchronously with the rotation of wheel 501 is to exist in the flat flat that is known as that produces on the plane of rotation of wheel 501.In wheel diagnostic process (S420), at first carry out the low-pass filtering (LPF) of from the vibration data that is taken into, extracting the component below the assigned frequency (1kHz) and handle (S421), the data that extract are being carried out selected processing (S422) afterwards with the extraction yield of regulation, as shown in Figure 43, carry out the initial crossover (overlap) that last 1/3 data with the previous sampling interval of current sampling interval are connected on the data of current sampling interval in order to ensure the data in the interval (period T 2) longer and handle (S423) than 1 sampling interval (period T 1).Then, will through this crossover handle the data that surpass threshold value in the data of (S423) by peak value keep (peak hold) handle (S424) carry out absolute value and only certain hour (τ) remain value above threshold value.This retention time (τ) is picked as than short value of one week of wheel by the rotating speed decision of wheel 501.This carries out absolute value and keeps the peak value of certain hour to keep processing can carry out stable peak value metering.And paired pulses surpasses the number of times of threshold value and counts (S425: surpass the threshold number counting and handle), judges whether consistent (S426: the wheel fault determination processing) with the rotation number of wheel 501 of count number.
The vibration data of 4 channels that are updated at step S401 is carried out bearing diagnosis repeatedly handle (S410) and wheel diagnostic process (S420).That is, when Data Update each time, implement No. 4 bearing diagnostic process (S410) and wheel diagnostic process (S420) respectively.Then, the result of determination of the determination processing of each time (S418, S426) is stored and remains in diagnostic process unit (MPU) 550B (S404).Diagnostic process unit (MPU) 550B storage keeps reviewing over N time the result of determination processing (S418, S426) from up-to-date result of determination, carries out abnormality juding according to this result of determination of N time on adding up, and exports this result (S405).
That is, in the present embodiment, axle bearing 530, wheel 501 all have the unanimity with consistent, the wheel number of once defect frequency, are not judged to be unusual.Because the unanimity of frequency is based on the probability process, so need statistically to judge according to aggregate-value repeatedly.
As the statistics determination methods, the accumulative total that generally can enumerate frequency spectrum is average, but in the present embodiment in the determination methods of Shi Yonging, if bearing, then add the data of repeatedly for example representing the consistent degree of frequency spectrum for 16 times with round values, if reach reference value then be judged as unusually, otherwise be not judged as unusually, but can fully be applied to the unusual judgement of the axle bearing of rail truck.Even little peeling off takes place in bearing, yet can not lubricate or seal and just carry out always as long as fully carried out, travelling of rail truck brought the dangerous little of influence, bring the unusual generation of effect by other parts perception such as typical temperature fuses travelling of rail truck.
As mentioned above, the apparatus for diagnosis of abnormality 550 of present embodiment is by the vibration of vibration transducer 511 inspection vehicle axle bearings 530 or wheel 501,550A samples to the output signal of vibration transducer 511 by the sensor signal processing unit, and diagnostic process unit (MPU) 550B carries out the abnormity diagnosis of axle bearing 530 and wheel 501 based on this vibration data.At this moment, diagnostic process unit (MPU) 550B is taken into the vibration data from sensor signal processing unit 550A continuously, be divided into the interval of each fixed cycle simultaneously, the vibration data in 1 interval is handled as the vibration data that is used for bearing diagnosis, and the data of data that will connect the last stipulated time in its previous interval simultaneously in the beginning of the vibration data in 1 interval are handled as the vibration data that is used for the wheel diagnosis.Like this, handle, thereby can determine that abnormal vibrations still is to be caused by axle bearing 530 can implement to diagnose accurately by flat the causing of wheel 501 by being divided into vibration data that is used for bearing diagnosis and the vibration data that is used for the wheel diagnosis.
In addition, in the apparatus for diagnosis of abnormality of present embodiment, (roughly simultaneously) is taken into the sensor signal from 4 channels of 4 vibration transducers 511 installing respectively on 4 rotary supporting devices 510 on each chassis simultaneously, implement simultaneously in data are taken into the time, to finish the real-time processing that diagnostic data is handled for all channels, so do not have the accident failure of data, can carry out the very high abnormity diagnosis of reliability.
[the 15th embodiment]
Figure 46 is the block scheme of the 15th embodiment (variation of above-mentioned the 14th embodiment).This apparatus for diagnosis of abnormality 550 uses and comprises that the MPU of Port Multiplier (MUX) and AD transducer (ADC) is as diagnostic process unit 550B.That is, MPU has the function of the part of sensor signal processing unit 550A concurrently.According to this structure, can simplify the circuit in the apparatus for diagnosis of abnormality 550, and MPU built-in circuit collaborative that can realize waiting with dma controller (DMAC) 557 other by software simply, so can carry out the software control higher than the structure efficiency of the 14th embodiment.
[the 16th embodiment]
Figure 47 is the block scheme of the 16th embodiment (variation of above-mentioned the 15th embodiment).This apparatus for diagnosis of abnormality 550 comprises that also static RAM (SRAM) 562 with reserve battery (Batt) 561 is as memory element except the structure of Figure 46.In addition, make the built-in calendar clock circuit (RTC) of MPU 563 effective hardware configurations, can preserve the data when unusual by employing.
Figure 48 represents the wheel of the diagnostic process unit 550B among the 16th embodiment, the content of bearing diagnosis data processing.Diagnostic process unit 550B implements bearing diagnosis and handles (S510) and wheel diagnostic process (S520) when the renewal (S401) of at every turn carrying out the vibration data of 4 channels.Then, whether the spectrum intensity that judgement is handled the envelope waveform of the bear vibration that (S510) obtain by bearing diagnosis is reference value above (S511), (false among the S511) stores the result (S404) who keeps bearing diagnostic process (S510) for N time accumulative total under the situation less than reference value.In addition, the count number of judging the incident that surpasses the vibration class threshold value that obtains by wheel diagnostic process (S520) whether with the rotation number of wheel 501 consistent (S521), (false among the S521) stores maintenance (S404) for N time accumulative total under inconsistent situation.
On the other hand, at spectrum intensity be under the situation of (true among the S511) more than the reference value, the envelope intensity of the envelope waveform of bear vibration together is kept at (S530) among the SRAM562 with the date temporal information that reads from calendar clock circuit (RTC) 563.In addition, under the count number of the incident that surpasses the vibration class threshold value situation consistent (true among the S521), the data of the time waveform in the wheel diagnosis and the date temporal information that reads together are kept at (S530) among the SRAM562 from calendar clock circuit (RTC) 563 with the rotation number of wheel 501.After preserving the allowance that data volume reaches SRAM562, delete past data (S531).
According to this embodiment, the abnormality juding result is sent to warning horn and carry out the alarm processing, read simultaneously and be kept at the data among the SRAM562 and send to the computing machine that is used to safeguard, thereby the repair message that can be used as vehicle utilizes for the content of vector etc.
[the 17th embodiment]
Figure 49 is the block scheme of the 17th embodiment (variation of above-mentioned the 16th embodiment).This apparatus for diagnosis of abnormality 550 respectively comprises two groups of Port Multipliers (MUX) 552 and AD transducer (ADC) 553 in MPU, thereby can carry out real-time diagnosis by the sensor signal of 8 channels by a module.If the computing power of MPU allows, the demultiplexing of then such sensor signal input also can increase the number of AD transducer, thereby or by using the fast AD transducer of conversion rate and several channels of Port Multiplier all may.In addition, in the example of Figure 49, calendar clock circuit (RTC) 563 is not built in MPU and will be to the external structure that has reserve battery (Batt) of MPU.
[the 18th embodiment]
Figure 50 is the block scheme of the 18th embodiment (variation of above-mentioned the 15th embodiment).Added 1 turn signal generation frequency dividing circuit 565 in the structure of the apparatus for diagnosis of abnormality that this apparatus for diagnosis of abnormality 550 illustrates in Figure 46.The output of waveform shaping circuit 555 is transfused to diagnostic process unit (MPU) 550B and 1 turn signal generation frequency dividing circuit 565.1 turn signal generation frequency dividing circuit 565 will be counted the ratio sine wave by the rotation after waveform shaping circuit 555 shapings and carry out frequency division, the moving rotational synchronization signal that a week diagnostic process unit (MPU) 550B is provided 1 pulse of revolution.Diagnostic process unit (MPU) 550B serves as to trigger to carry out the sampling of data in the interval of certain speed with this rotational synchronization signal, and these data is carried out summation averaging handle and carry out abnormity diagnosis.To the rotational synchronization signal that takes place with moving 1 week of wheel 501 revolutions serves as to trigger the data of sampling to carry out the summation averaging processing, thereby be eliminated with the component beyond the signal of the rotational synchronization of wheel 501, and the component of the rotational synchronization of only residual and wheel 501, so the flat detection of wheel 501 is carried out in the judgement of threshold value that can be by impacting grade accurately.
[the 19th embodiment]
Figure 51 is the block scheme of the 19th embodiment.This apparatus for diagnosis of abnormality 550 has sensor signal processing unit 550A, diagnostic process unit (MPU) 550B.Sensor signal processing unit 550A has 571 and wave filters of an amplifier (Amp) (LPF) 572.And the output signal of 4 vibration transducers 511 (simulating signal) is transfused to a wave filter (LPF) 572 after being transfused to an amplifier (Amp) 571 and amplification.That is, in this embodiment, for the output signal from 4 channels of 4 vibration transducers 511 is amplified, filtering, use amplifier (Amp) 571 and wave filter (LPF) 572.Then, amplify and filtered simulating signal is taken into diagnostic process unit (MPU) 550B, be transformed to digital signal by the AD transducer (ADC) 553 in diagnostic process unit (MPU) 550B by amplifier (Amp) 571 and wave filter (LPF) 572.On the other hand, be taken into diagnostic process unit (MPU) 550B from passing on the rotational speed pulse signal of sensor after by waveform shaping circuit 555 shapings, by the umber of pulse of the 573 digit's times of time counter (TCNT) in diagnostic process unit (MPU) 550B, this value is used as tach signal and handles.Diagnostic process unit (MPU) 550B is based on coming the execute exception diagnosis by vibration transducer 511 detected vibrational waveforms with by the detected tach signal of speed probe.The diagnostic result of diagnostic process unit (MPU) 550B is output to communication line 520 (with reference to Figure 40) via line driver (LD) 556.Communication line 520 is connected to warning horn, the alarm action of carrying out when carrying out taking place unusually in the flat grade at wheel 501.
Can be from the output signal of vibration transducer 511 detected be peel off flat (wearing and tearing) with wheel 501 of axle bearing 530 unusually.Can survey as near the vibration signal of the frequency band 1kHz.Therefore, in the 19th embodiment, for the output signal of vibration transducer 511 is amplified, filtering, use amplifier (Amp) 571 and wave filter (LPF) 572.And, will be by software processes by wave filter (LPF) 572 filtering and the data separating that is transformed to digital signal by AD transducer (ADC) 533 for being used for the axle bearing diagnosis and being used for the wheel diagnosis, thus carry out both abnormity diagnosis.
Take place in the axle bearing 530 unusual in, the peeling off of the easiest outer wheel track that causes stationary wheel.Therefore, about axle bearing 530, can be detected object with the peeling off of outer wheel track of stationary wheel.
Axle bearing 530 peel off with wheel 501 flat in, the frequency band of defective differs from 10 times of degree.Rotating speed (the sec of wheel 501 -1) equate with the flat fundamental frequency of wheel.The scope of the rotating speed that should diagnose is 4~10sec -1(fundamental frequency: 4~10Hz).With respect to this, exist under the situation of defective in the outer wheel track of the stationary wheel of axle bearing 530, even the scope (4~40sec of identical rotating speed -1), the fundamental frequency of defective also is 33~83Hz.All checking under the situation of the higher hamonic wave till 4 times, for wheel 501,4~40Hz is the frequency analysis scope of the DFT (discrete Fourier transform (DFT)) of necessity, is the frequency analysis scope of the DFT (discrete Fourier transform (DFT)) of necessity for axle bearing 530,33~330Hz.Frequency resolution during the diagnosis of axle bearing 530 is that 1.0Hz is just enough.But, in the diagnosis of wheel 501, lack of resolution under 1.0Hz, and receive the influence that is offset the DC component in the FFT low frequency that causes easily.
Therefore, in the 19th embodiment, will peel off parsings (being used for axle bearing diagnoses) and handle for being used for axletree outside wheel track for the data conversion of digital signal by AD transducer (ADC) 553 conversion (sampling) with the two kinds of different data of sample frequency that are used for the flat parsing of wheel (being used for wheel diagnoses).
Figure 52 represents the motion flow of diagnostic process unit (MPU) 550B among the 19th embodiment.Diagnostic process unit (MPU) 550B will be transformed to digital signal (S601) from 511 outputs of 4 vibration transducers and the sensor signal that sends via amplifier (Amp) 571 and wave filter (LPF) 572 by AD transducer (ADC) 553.Then, the output signal of AD transducer (ADC) 553 is implemented to select processing (S602) by the FIR low-pass filtering that is realized by software.In this example, the sampling in the AD transducer (ADC) 553 under the frequency of 8kHz being 3 seconds that unit implements.In addition, in selecting processing (S602), reduce to 2kHz, will select rate M and be made as 4 and the data number is reduced to 1/4 in order to make sample frequency fs.
Diagnostic process unit (MPU) 550B will be via the data conversion of selecting processing (S602) for being used for the different two kinds of data (with reference to Figure 53 (a)) of sample frequency that wheel track axletree outside is peeled off parsings (following note is made ' bearing with ') and is used for the flat parsing of wheel (following note work ' wheel is used ').
By will be via selecting processing (S602) thus data carry out 4 and cut apart and be divided into per 0.75 second data interval and obtain the data (S611) that bearing is used.The data that obtain are implemented absolute value successively handle (S612) and ACization processing (S613).Then, further 0 be made as about 1 second data interval long (S614), and frequency resolution is made as about 1.0Hz carries out FFT (S615) by what append 1 interval about 0.25 second (sec).The input data number of FFT is 2048.Carrying out peaceful (Hanning) window of the Chinese before FFT handles.Behind FFT, usually ask foreign steamer defect frequency Zfc according to the speed of a motor vehicle and each unit of bearing, carry out detecting (S616) from the peak value of first-harmonic till 4 times.Then, carry out the comparison of foreign steamer defect frequency Zfc and frequency peak, and calculate both consistent degree (S617).The consistent degree that obtains based on this processing is repeated certain number of times adds up to counts, and carries out the abnormality juding of axle bearing 530.
By will being that the data of 2kHz are handled (S621) in absolute value and will be selected rate M by wave filter (LPF) afterwards and be made as 8 and select processing (S622) through the sample frequency fs that selects processing (S602), fs reduces to 250Hz with sample frequency, thereby obtains the data that wheel is used.Data number in this moment is 750, is made as about 4 seconds data by carrying out 0 cover interpolation (S624), implements FFT (S625) thereby frequency resolution is made as about 0.25Hz.Carrying out peaceful (Hanning) window of the Chinese before FFT handles.Behind FFT, carry out peak value and detect (S626).Then, carry out the higher hamonic wave till the flat fundamental frequency to 4 of wheel time and the comparison of frequency peak, and calculate both consistent degree (S627).The consistent degree total that obtains based on repeating this processing of certain number of times is counted and is carried out the abnormality juding of wheel 501.By by time counter (TCNT) thus the umber of pulse of the unit interval of 573 pairs of rotational speed pulse signals is counted the fundamental frequency of asking wheel flat.
As mentioned above, for each of 4 vibration transducers 511, respectively have one group of amplifier (Amp) 571 and wave filter (LPF) 572, to be two kinds of different data of sample frequency of using with wheel of bearing by AD transducer (ADC) 553 conversion (sampling) for the data conversion of digital signal via Port Multiplier (MUX) 553, and be divided into the processing that two systems comprise FFT, thereby, can high precision and carry out the abnormity diagnosis of bearing and wheel expeditiously.With respect to this, under the situation of the frequency range of having investigated most of different bearing of frequency domain and wheel by FFT, the precision (resolution) (with reference to Figure 53 (b)) that can not realize and assess the cost and match.
In addition, in above-mentioned example, the major part of carrying out digital processing by software, but also can realize its part or all by FPGA hardware such as (Field Programmable Cate Array).
[the 20th embodiment]
Figure 54 represents the motion flow of diagnostic process unit (MPU) 550B among the 20th embodiment.In this example, the sampling in the AD transducer (ADC) 553 under the frequency of 16kHz being 3 seconds that unit implements (S701).In addition, in selecting processing (S702), reduce to 4kHz, will select rate M and be made as 4 and the data number is reduced to 1/4 in order to make sample frequency fs.
Diagnostic process unit (MPU) 550B will be the bearing different two kinds of data of using with wheel (with reference to Figure 53 (a)) of sample frequency via the data conversion of selecting processing (S702).
By will be via selecting processing (S702) thus data carry out 3 and cut apart and be divided into per 1.0 seconds data interval and obtain the data (S711) that bearing is used.The data that obtain are implemented absolute value successively handle (S712) and ACization processing (S713).Then, omit 0 interpolation processing fully or carry out only interpolation mantissa slightly, for example 96 00 such interpolations are handled to 4000 data interpolations, and frequency resolution is made as about 1.0Hz and carries out FFT (S714).Carrying out peaceful (Hanning) window of the Chinese before FFT handles.Behind FFT, usually ask foreign steamer defect frequency Zfc according to the speed of a motor vehicle and each unit of bearing, carry out detecting (S715) from the peak value of first-harmonic till 4 times.Then, carry out the comparison of foreign steamer defect frequency Zfc and frequency peak, and calculate both consistent degree (S716).The consistent degree that obtains based on this processing is repeated certain number of times adds up to counts, and carries out the abnormality juding of axle bearing 530.
By being that the data of 4kHz are handled (S721) in absolute value and selected processing (S722) by wave filter (LPF) afterwards through the sample frequency fs that selects processing (S702), fs reduces to 500Hz with sample frequency, thereby obtains the data that wheel is used.The data that obtain are implemented ACization processing (S723) successively.Then, carry out 0 cover interpolation (S724) and be made as about 4 seconds data, implement FFT (S725) thereby frequency resolution is made as about 0.25Hz.Carrying out peaceful (Hanning) window of the Chinese before FFT handles.Behind FFT, carry out peak value and detect (S726).Then, carry out the higher hamonic wave till the flat fundamental frequency to 4 of wheel time and the comparison of frequency peak, and calculate both consistent degree (S727).The consistent degree total that obtains based on repeating this processing of certain number of times is counted and is carried out the abnormality juding of wheel 501.
As the 20th embodiment, in the data processing that bearing is used,, can cut down the number of times that FFT handles by omitting 0 interpolation processing fully or carrying out 0 such interpolation of only interpolation mantissa slightly and handle.That is, in the comparison of pre-the 19th embodiment, the number of times that the FFT that carries out with the time handles is reduced to 3 times from 4 times.But, as the 19th embodiment, carry out that 0 interpolation is handled and the FFT interval division can be increased the FFT interval that can avoid track noise etc. for the short time.
[the 21st embodiment]
Figure 55 (a), Figure 55 (b) are the part block schemes of diagnostic process unit (MPU) 550B among the 21st embodiment.In Figure 55 (a), in the hardware configuration of Figure 51, the prime (input end) of the AD transducer (ADC) 553 in diagnostic process unit (MPU) 550B is provided with absolute value circuit (ABS) 581, and then is provided with low-pass filter (LPF) 582 in level thereafter.In Figure 55 (b), in the hardware configuration of Figure 51, the prime (input end) of the AD transducer (ADC) 553 in diagnostic process unit (MPU) 550B is provided with envelope circuit (ENV) 591, and then is provided with Hi-pass filter (HPF) 592 in its prime.
Figure 56 represents the motion flow of diagnostic process unit (MPU) 550B among the 21st embodiment.Diagnostic process unit (MPU) 550B will be sampled and will be transformed to digital signal (S801) by AD transducer (ADC) 553 from 4 vibration transducers 511 output and via the sensor signal that amplifier (Amp) 571 and wave filter (LPF) 572 send.In this example, the sampling in the AD transducer (ADC) 553 under the frequency of 2kHz being 3 seconds that unit implements.Thereby cut apart and be divided into per 0.75 second data interval and obtain the data (S811) that bearing is used by carrying out 4 through the data that low-pass filter 582 has been carried out the AD conversion.The data that obtain are implemented ACization processing (S812).Then, further 0 be made as the long (S813: the cover interpolation), and frequency resolution is made as about 1.0Hz carries out FFT (S814) of about 1 second data interval by what append 1 interval about 0.25 second (sec).Carrying out peaceful (Hanning) window of the Chinese before FFT handles.Behind FFT, usually ask foreign steamer defect frequency Zfc according to the speed of a motor vehicle and each unit of bearing, carry out detecting (S815) from the peak value of first-harmonic till 4 times.Then, carry out the comparison of foreign steamer defect frequency Zfc and frequency peak, and calculate both consistent degree (S816).The consistent degree that obtains based on this processing is repeated certain number of times adds up to counts, and carries out the abnormality juding of axle bearing 530.
Handle by implementing absolute value by absolute value circuit 581 from the sensor signal of vibration transducer 111 outputs, and with after the 2kHz sampling, to select rate M and be made as 8 and select processing (S821), fs reduces to 250Hz with sample frequency, thereby obtains the data that wheel is used.The data that obtain are implemented ACization processing (S723) successively.Then, carry out 0 cover interpolation (S823) and be made as about 4 seconds data, implement FFT (S824) thereby frequency resolution is made as about 0.25Hz.Carrying out peaceful (Hanning) window of the Chinese before FFT handles.Carrying out detecting (S825) behind the FFT from the higher hamonic wave peak value of first-harmonic till 4 times.Then, carry out the higher hamonic wave till the flat fundamental frequency to 4 of wheel time and the comparison of frequency peak, and calculate both consistent degree (S826).The consistent degree total that obtains based on repeating this processing of certain number of times is counted and is carried out the abnormality juding of wheel 501.
In the 21st embodiment, by be implemented in by the software that can carry out high speed processing among Figure 52 by software implementation select processing (S602) and (S612) handled in absolute value, simplify signal Processing by software.Even the sample frequency fs in the AD transducer (ADC) 553 is reduced to its 2kHz of 1/4 from the 8kHz of the situation of Figure 52, also can carries out high precision and high efficiency abnormality juding.
[the 22nd embodiment]
The apparatus for diagnosis of abnormality of the 22nd embodiment at first, is described with reference to Figure 40,41,45,51,57,58.
As shown in figure 40, a rail truck 500 has been installed 4 wheels 501 by former and later two chassis supportings on each chassis.The vibration transducer 511 that the on-stream vibration that takes place from rotary supporting device 510 is detected has been installed in the rotary supporting device of each wheel 501 (bearing housing) 510.
Being equipped with two whiles (roughly simultaneously) on the console panel 515 of rail truck 500 is taken into the sensor signal of 4 channels and the apparatus for diagnosis of abnormality 550 of implementing diagnostic process.That is, the output signal of 4 vibration transducers 511 that are provided with on each chassis is transfused to the different apparatus for diagnosis of abnormality in each chassis 550 via signal wire 516 respectively.In addition, also be transfused to the rotational speed pulse signal of the speed probe (omitting diagram) that detects from rotating speed in the apparatus for diagnosis of abnormality 550 to wheel 501.
As shown in figure 41, in the rotary supporting device 510 as an example, be provided with the axle bearing 530 as rotatable parts, axle bearing 530 comprises: as the interior wheel 531 that is embedded in the rotor wheel on the rotation axis (not shown) outward, as the foreign steamer 532 that is embedded in the fast pulley in the shell (not shown), the retainer (not shown) that rollably keeps as the rolling body 533 of a plurality of rotors between wheel 531 and the foreign steamer 532 in being configured in, with rolling body 533 freedom.Vibration transducer 511 is retained as the attitude of the vibration acceleration that can detect gravity direction, and is fixed near the foreign steamer 532 of shell.Vibration transducer 511 uses various sensors such as acceleration transducer, AE (acousticEmission) sensor, ultrasonic sensor, shock pulse sensor.
Shown in Figure 51, apparatus for diagnosis of abnormality 550 has sensor signal processing unit 550A, diagnostic process unit (MPU:Micro Processing Unit) 550B.Sensor signal processing unit 550A is that a vibration transducer 511 has 571 and wave filters of an amplifier (Amp) (LPF) 572{ promptly, comprises 571 and four wave filters of four amplifiers (Amp) (LPF) 572}.And, after the output signal of 4 vibration transducers 511 (simulating signal) is transfused to corresponding amplifier (Amp) 571 respectively and is exaggerated, be transfused to corresponding wave filter (LPF) 572 respectively.Amplify and filtered simulating signal is taken into diagnostic process unit (MPU) 550B by amplifier (Amp) 571 and wave filter (LPF) 572, be transformed to digital signal by the AD transducer (ADC) 553 in diagnostic process unit (MPU) 550B via Port Multiplier (MUX) 552.On the other hand, from the rotational speed pulse signal of speed probe by after waveform shaping circuit 511 shapings, be taken into diagnostic process unit (MPU) 550B, by the umber of pulse of the 573 digit's times of time counter (TCNT) in diagnostic process unit (MPU) 550B, this value is used as tach signal and handles.Diagnostic process unit (MPU) 550B is based on coming the execute exception diagnosis by vibration transducer 511 detected vibrational waveforms with by the detected tach signal of speed probe.The diagnostic result of diagnostic process unit (MPU) 550B is transfused to communication line 520 (with reference to Figure 40) via line driver (LD) 556.Communication line 520 is connected to warning horn, the alarm action of carrying out when carrying out taking place unusually in the flat grade at wheel 501.
Can be from the output signal of vibration transducer 511 detected be peel off flat (wearing and tearing) with wheel 501 of axle bearing 530 unusually.The diagnosis of axle bearing 530 is described here.Take place in the axle bearing 530 unusual in, the peeling off of the easiest outer wheel track that causes stationary wheel is so be detected object with the peeling off of outer wheel track of the stationary wheel of axle bearing 530.
In this embodiment, for the output signal of vibration transducer 511 is amplified, filtering, use amplifier (Amp) 571 and wave filter (LPF) 572.Then, be transformed to the data of digital signal by handling by wave filter (LPF) 572 filtering and by AD transducer (ADC) 553, and carry out abnormity diagnosis based on the output signal of each vibration transducer 511 by the calculation function of software realization.
The output signal of vibration transducer 511 is imported into the interior AD transducer (ADC) 553 of diagnostic process unit (MPU) 550B by amplifier (Amp) 571 and wave filter (LPF) 572.The resolution of the AD transducer (ADC) 533 among this embodiment is 8.Diagnostic process unit (MPU) 550B reads in vibration data as 8 value.In addition, certain and suppress the load of CPU558 for the sample frequency that makes AD transducer (ADC) 553, use comparison match timer (CMT) 554 and direct memory access (DMA) controller (DMAC) 557.Sample frequency position 8kHz.Wave filter (LPF) 572 works as antialiasing filter, reduces the above band component of 1kHz.
The input range of AD transducer (ADC) 553 is 0~3.3V.Vibration transducer 511, amplifier (Amp) 571 and wave filter (LPF) 572 are designed to the input range that vibrational waveform is fit to AD transducer (ADC) 553, and the center voltage of vibrational waveform is 1.65V.
Figure 57 represents the motion flow of diagnostic process unit (MPU) 550B.Diagnostic process unit (MPU) 550B will sample by Port Multiplier (MUX) 552 switching channels respectively simultaneously from 4 vibration transducers 511 output and via the sensor signal that amplifier (Amp) 571 and wave filter (LPF) 572 send, thereby roughly simultaneously multichannel be sampled and be transformed to digital signal (signless 8 bit data) (step S901) by AD transducer (ADC) 553.
Then, at first will be transformed to 16 bit data (step S902) from signless 8 bit data of AD transducer (ADC) 553 outputs.Specifically, shown in Figure 58 (a), make 8 bit data recompiles so that become after the 0V, thereby add 8 values that are transformed to 16 at its low level as the 1.65V of the center voltage of vibrational waveform.
Then, implementing fixed-point mathematics Filtering Processing (step S903), and implementing envelope (absolute value) and handled (step S904) afterwards, implementing 16 fixed point FFT and handle (step S905).Then, ask the peak value (step S906) of frequency according to the result of FFT processing (step S905).In addition, according to axletree rotating speed and each element of bearing (with reference to Figure 45) calculation bearing defect frequency (step S907).Then, with consistent the counting of degree (step S908) of the peak value and the bearing defect function of frequency, judge (NG) (step S909) unusually according to the aggregate-value (cumulative points) of certain number of times.
Handling the fractional fixed point computing of (step S905) from 16 fixed-point mathematics Filtering Processing (step S903) to 16 fixed point FFT, 15 of low levels are used to show below the radix point in 16.When the coefficient of digital filter shows with real number more than or equal to-1.0 less than 1.0, but during with the performance of this fractional fixed point, be in computing machine more than or equal to-2 15And less than 2 15-1.If remain 8, under signed situation, for more than or equal to-2 7And less than 2 7-1.Because Filtering Processing reduces the amplitude of waveform,, the precision that frequency peak detects is brought obstacle so in the data of 8 bit widths still, become the littler data of amplitude.Therefore, the amplitude range of AD conversion is made as more than or equal to-1.0 less than 1.0 with real number, consistent with the data width of CPU558.Following 7 of the position of the most significant digit of signed 8 bit data and radix point still are high-order 8, and making 8 of low levels all is 0.Key is that the integer with-128~127 scope amplifies 256 times, and the integer that is transformed to-32768~32767 scope carries out computing.With respect to this, shown in Figure 58 (b), also be full of sign extended even expand to 16, not amplifying does not then have effect.
FFT handles (step S905) and is undertaken by the fixed-point arithmetic of 16 bit data.Its reason is, because the CPU558 that uses is 32 bit CPUs, so the multiplication of 16bit * 16bit do not overflow, and owing to do not comprise floating point arithmetic device (FPU), so also do not use floating-point, this is ideal aspect computing velocity.
In addition, handle in (step S905), carry out scaling and handle at FFT.In other words, be made as 2 n power and carry out under the situation of FFT in that computing is counted, carry out the butterfly operation of N section, but dwindle data in order to prevent overflowing of this moment.
Like this, in fixed-point arithmetic, owing to have the restriction of bit wide, so dynamic range reduces easily.And then, if the input data are half 8, then imbed abnormal signal in the error of calculation, the probability that the detection of the peak value of vibration can not be carried out well becomes very high.Therefore, in the present embodiment, 8 data in advance are enlarged into 16 carry out computing, thereby prevent and to disappear by detected peak value.
In this abnormity diagnosis was handled, frequency analysis and peak value thereof detected very important, did not require verily original waveform to be sampled and restore, when so AD transform data originally is 8, at least when computing, as mentioned above, can fully catch the feature of frequency by amplification.
As an one check example, table 6 is together represented the peeling off of taper roll bearing that is used for rail truck detected the result who has carried out test with comparative example.
[table 6]
Test 16 ADC 8 ADC are 8 integers (only is-symbol expansion) still 8 ADC are enlarged into 16 (256 times of amplitudes)
Abnormal vibrations 1 Detect Do not detect Detect
Abnormal vibrations 2 Detect Do not detect Detect
Abnormal vibrations 3 Detect Do not detect Detect
Abnormal vibrations 1 is bearing that the foreign steamer orbital plane of bearing the is peeled off vibration signal when rotating with 240rpm.Abnormal vibrations 2 is the bearing that forms the artificial defect by electrodischarge machining on the foreign steamer orbital plane of bearing vibration signals when rotating with 360rpm.Abnormal vibrations 3 is the bearing that forms the artificial defect by electrodischarge machining on the foreign steamer orbital plane of bearing vibration signals when rotating with 990rpm.
The situation of all abnormal vibrations all is still to be used for situation success on abnormality detection of computing from 16 round valuess that 16 AD transducers obtain.On the other hand, 8 round values former states that obtain from 8 AD transducers only being carried out sign extended carries out the situation of computing and can not detect unusually.Relative therewith, by will behind coding, expanding to 16, thereby in fact scope is enlarged into 256 times of situations successes in unusual detection of carrying out computing from 8 round valuess that the AD transducer obtains.
As mentioned above, in the future the simulating signal of self-excited oscillation sensor 511 resolution (being 8 in this example) the growth data width (expanding to 16 in this example) of output signal ratio AD transducer (ADC) 553 that is transformed to the AD transducer (ADC) 553 of digital signal carries out Fourier transform processing, and carry out abnormity diagnosis based on its result, thereby use the AD transducer of low resolution to realize the cost degradation and the save spaceization of circuit, and can carry out abnormity diagnosis and can not cause that precision reduces.
[the 23rd embodiment]
Figure 59 is the major part block scheme of the 23rd embodiment of apparatus for diagnosis of abnormality of the present invention.This embodiment represents not use the example of the microcomputer system (microcomputer/system) of AD transducer, amplify by amplifier (Amp) 571 from the simulating signal (waveform signal) of vibration transducer 511, by after the wave filter (LPF) 572 immediately via the ports (Port) of 673 times of inputs of comparer diagnostic process unit (MPU) 650B.That is, in the present embodiment, diagnostic process unit (MPU) 650B does not have 553 in AD transducer and be provided with comparer 673 in sensor signal processing unit 550A.Other structure is identical with the 22nd embodiment.
Comparer 673 uses hysteresis loop comparator in order to get rid of The noise.The voltage of the simulating signal of comparer 673 self-excited oscillation in the future sensors 511 (with reference to the waveform on the top of Figure 60 (a)) and certain reference voltage ref compare, and the voltage ratio reference voltage ref height of this simulating signal of output expression still is low 1 signal (with reference to the waveform of the bottom of Figure 60 (a)).Reference voltage ref for example is the center voltage (1.65V) of vibrational waveform.The sample frequency of comparer 673 is 32kHz.In addition, the signal from above-mentioned 1 (2 value) of comparer 673 that is transfused to the port (Port) of diagnostic process unit (MPU) 650B is handled by digital filtering in diagnostic process unit (MPU) 650B, becomes the waveform shown in Figure 60 (b).
Figure 61 represents the motion flow of diagnostic process unit (MPU) 650B among the 23rd embodiment.Diagnostic process unit (MPU) 650B is from comparer 673 received signals (step S910).The value of the port of diagnostic process unit (MPU) 650B only is 0 and 1, but because this is equivalent to sign bit in the AD conversion, so consider positive and negatively merely, i.e. 0 expression-1,1 expression 1 is transformed to tape symbol 16 bit data (step S920).Begin computings with 16 integers of tape symbol from-32768 and 32767.
Then, implementing FIR digital filtering processing (step S930), and implementing envelope (absolute value) and handled (step S940) afterwards, implementing 16 fixed point FFT and handle (step S950).Then, ask the peak value (step S960) of frequency according to the result of FFT processing (step S950).In addition, according to axletree rotating speed and each element of bearing (with reference to Figure 45) calculation bearing defect frequency (step S970).Then, with the consistent number of degrees value (step S980) of the peak value and the bearing defect function of frequency, judge (NG) (step S990) unusually according to the aggregate-value of certain number of times.
To be object below the 1kHz, from the vibration of generations such as parts of bearings or sensor outer housing, comprise the vibration of the frequency that much is higher than 1kHz as the defect frequency of axle bearing 530.Propagation by vibration transducer 511 detected vibrations is undertaken by the vibration of these parts, and the vibration frequency of the low frequency that defective causes can be thought these high-frequency vibrations (carrier wave) are modulated.Therefore, in this embodiment, the sample frequency of comparer 673 is set at 32kHz than the highland.By improving sample frequency, even the data of 2 values also can be recovered defect frequency.Its principle and PWM{Pulse Width Modulation (width modulation) } principle identical.FIR low-pass filtering treatment (step S930) narrows down to the scope of defect frequency for the component of removing above-mentioned carrier wave and with waveform signal and implements.
Like this, even use not using the AD transducer under the situation of comparer 673 more cheaply, carry out calculation process by the data that will expand to 16 bit wides, thereby can carry out by frequency analysis to the enough FFT processing of the peak value that detects abnormal signal from 2 Value Datas of comparer 673 outputs.
[the 24th embodiment]
Figure 62 is the major part block scheme of the 24th embodiment of apparatus for diagnosis of abnormality of the present invention.Same with the 23rd embodiment, diagnostic process unit (MPU) 650B does not have AD transducer 553 and be provided with comparer 673 in sensor signal processing unit 550A.In the 23rd embodiment, reference voltage ref is certain, but in the present embodiment, and the sine wave of the frequency that the simulating signal of self-excited oscillation sensor 511 recently is high is as reference voltage ref.Comparer 673 with the frequency that is higher than reference voltage ref to from the analog signal sampling of vibration transducer 511 and digitizing (binaryzation).
Diagnostic process unit (MPU) 650B handles by the binary signal from comparer 673 being carried out digital low-pass filtering, thereby realizes the function of the AD transducer of multidigit by software.Among above-mentioned the 23rd embodiment, the level of the characteristic frequency that bearing is peeled off is up to 1kHz, the high fdrequency component of the proper vibration of the rail wheel of bearing 530, rotor and vibration transducer 511 is added in the vibrational waveform, because software implementation low-pass filtering treatment by diagnostic process unit (MPU) 650B, so on the whole, carry out equal processing with the 24th embodiment.But, we can say that the 23rd embodiment is not needing on the sine wave generating circuit this point, favourable on aspect the cost.
In addition, in the above-described embodiments, the abnormality diagnostic situation of carrying out axle bearing 530 has been described, but apparatus for diagnosis of abnormality of the present invention also can be effectively applied to the abnormity diagnosis of wheel and other mechanical hook-up.
Utilizability on the industry
According to the present invention, though the S/N of abnormal signal or unusual omen signal and noise signal than little condition under, also can implement abnormity diagnosis accurately and can be not unusual omen signal with noise signal error detection error detection.
According to abnormity diagnostic system of the present invention, can high accuracy and implement expeditiously bearing in the plant equipment or the abnormity diagnosis of the related member of bearing.
According to the present invention, thereby can implement accurately abnormity diagnosis to carrying out FFT from the detected signal of diagnosis object with frequency resolution arbitrarily.
According to apparatus for diagnosis of abnormality of the present invention, only by every vehicle being arranged a vibrating sensing device, just can detect based on the waveform signal of this vibrating sensing device the flat etc. unusual of the peeling off of bearing in this vehicle, wheel, so can construct abnormity diagnostic system with low cost.
According to apparatus for diagnosis of abnormality of the present invention, cost degradation and the save space of realizing circuit with AD converter or the simple comparator of low resolution, and can carry out abnormity diagnosis and can not cause that precision reduces.

Claims (6)

1. the abnormity diagnostic system of a plant equipment by detecting sound or the vibration that takes place from plant equipment, and is analyzed its detection signal, thereby diagnoses the unusual of bearing in this plant equipment or the related member of bearing, it is characterized in that this system comprises:
Filter processing unit is taken out the signal of diagnosing required frequency band from described detected signal;
The envelope processing unit is asked the envelope signal by the signal of this filter processing unit taking-up;
Select processing unit, the envelope signal that is obtained by this envelope processing unit is extracted processing;
The FFT arithmetic element is carried out frequency resolution to selected extraction that processing unit the carries out envelope signal after handling by this; And
Diagnosis unit comes diagnosing abnormal based on the analysis result of this FFT arithmetic element.
2. the abnormity diagnostic system of plant equipment as claimed in claim 1 is characterized in that, this system also comprises the digital filtering processing unit of the frequency band low-frequency bandization that makes described envelope signal.
3. the abnormity diagnostic system of plant equipment as claimed in claim 1 or 2 is characterized in that, realizes described FFT arithmetic element by DSP, and the data number that will import described FFT arithmetic element simultaneously is as the data number in the storer that can be contained in this DSP.
4. the abnormity diagnostic system of a plant equipment by detecting sound or the vibration that takes place from plant equipment, and is analyzed its detection signal, thereby diagnoses the unusual of bearing in this plant equipment or the related member of bearing, it is characterized in that this system comprises:
Sample processing unit is sampled to described detected signal with the sample frequency higher than the sample frequency of necessity in advance;
Filter processing unit is taken out the signal of diagnosing required frequency band from the signal of being sampled out by this sample processing unit;
Select processing unit, the signal that is taken out by this filter processing unit is extracted processing;
The envelope processing unit is asked by this and is selected the envelope signal that processing unit has carried out extracting the signal of handling;
The FFT arithmetic element is carried out frequency resolution to the envelope signal that is obtained by this envelope processing unit; And
Diagnosis unit comes diagnosing abnormal based on the analysis result of this FFT arithmetic element.
5. the abnormity diagnostic system of plant equipment as claimed in claim 4 is characterized in that, this system also comprises the digital filtering processing unit of the frequency band low-frequency bandization that makes described envelope signal.
6. as the abnormity diagnostic system of claim 4 or 5 described plant equipment, it is characterized in that, realize described FFT arithmetic element by DSP, the data number that will import described FFT arithmetic element simultaneously is as the data number in the storer that can be contained in this DSP.
CNB2005800062882A 2004-10-18 2005-10-17 Abnormality diagnosis system for machinery Active CN100516789C (en)

Applications Claiming Priority (8)

Application Number Priority Date Filing Date Title
JP2004302804A JP2006113002A (en) 2004-10-18 2004-10-18 Anomaly diagnosis system for mechanical equipment
JP302871/2004 2004-10-18
JP302804/2004 2004-10-18
JP313875/2004 2004-10-28
JP004268/2005 2005-01-11
JP251910/2005 2005-08-31
JP289152/2005 2005-09-30
JP293796/2005 2005-10-06

Publications (2)

Publication Number Publication Date
CN1926413A CN1926413A (en) 2007-03-07
CN100516789C true CN100516789C (en) 2009-07-22

Family

ID=36381618

Family Applications (1)

Application Number Title Priority Date Filing Date
CNB2005800062882A Active CN100516789C (en) 2004-10-18 2005-10-17 Abnormality diagnosis system for machinery

Country Status (2)

Country Link
JP (1) JP2006113002A (en)
CN (1) CN100516789C (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108885157A (en) * 2016-03-23 2018-11-23 Ntn株式会社 Railway abnormity diagnostic system and transacter
CN109313071A (en) * 2016-06-21 2019-02-05 Ntn株式会社 Vibration measurement device and abnormity diagnostic system

Families Citing this family (93)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP4810389B2 (en) * 2006-10-03 2011-11-09 株式会社リコー Remote management system for image forming apparatus
KR100942287B1 (en) * 2008-02-14 2010-02-16 엠케이전자 주식회사 System and method for diagnosing a rotating machine using vibration signals
JP4560110B2 (en) * 2008-09-17 2010-10-13 ジヤトコ株式会社 Abnormality diagnosis apparatus and abnormality diagnosis method
CN101685042B (en) * 2008-09-25 2012-04-04 上海宝钢工业检测公司 On-line monitoring method of cold rolled heating furnace roller running state
US7827008B2 (en) * 2008-10-21 2010-11-02 General Electric Company System including phase signal saving during anomaly and related method
EP4033212A1 (en) 2008-12-22 2022-07-27 S.P.M. Instrument AB An analysis system
EP2370801B1 (en) 2008-12-22 2017-11-22 S.P.M. Instrument AB An analysis system
EP4024013A1 (en) 2008-12-22 2022-07-06 S.P.M. Instrument AB An analysis system
US8810396B2 (en) 2008-12-22 2014-08-19 S.P.M. Instrument Ab Analysis system
CN101858939B (en) * 2009-04-10 2012-11-07 华为技术有限公司 Method and device for detecting harmonic signal
CN102449445B (en) * 2009-05-05 2014-12-17 S.P.M.仪器公司 An apparatus and a method for analysing the vibration of a machine having a rotating part
FR2953289B1 (en) 2009-11-30 2012-04-27 Snecma METHOD AND DEVICE FOR MONITORING TORSION VIBRATIONS OF A ROTARY SHAFT OF A TURBOMACHINE.
SE535559C2 (en) 2010-01-18 2012-09-25 Spm Instr Ab Method and apparatus for analyzing the condition of rotating part machine
CN102893136A (en) * 2010-03-19 2013-01-23 Abb有限公司 Health monitoring method and system for drives
JP5428994B2 (en) * 2010-03-26 2014-02-26 新日鐵住金株式会社 Anomaly diagnosis method, apparatus and program for machine equipment
DE102010015208A1 (en) * 2010-04-16 2011-10-20 Schaeffler Technologies Gmbh & Co. Kg Method for monitoring a linear guide
CN101915727B (en) * 2010-07-02 2012-01-25 宁波大学 FPGA-based particle detection statistical system and method
CN101915726B (en) * 2010-07-02 2012-01-25 宁波大学 Particle detection statistical method based on embedded system
JP5553002B2 (en) * 2010-11-05 2014-07-16 日本精工株式会社 Bearing abnormality diagnosis device, railway vehicle equipped with the same, and bearing abnormality diagnosis method
CN102022143B (en) * 2010-11-24 2012-10-10 华北电力大学 Method for analyzing low-frequency vibration stability of steam turbine generator unit in real time
CN102243140B (en) * 2011-04-18 2013-01-23 杨彦利 Mechanical equipment state monitoring method based on sub-band signal analysis
CN102393214B (en) * 2011-06-09 2014-11-05 郝玉山 Method and device for acquiring continuous physical quantity data
EP2726835A4 (en) * 2011-06-28 2015-09-09 Ibm Vibration monitoring system
CN103688144B (en) 2011-07-14 2017-04-12 S.P.M.仪器公司 A method and a system for analysing the condition of a rotating machine part
EP2581724B1 (en) 2011-10-13 2020-03-25 Moventas Gears Oy A method and a system for the purpose of condition monitoring of gearboxes
CN102426095A (en) * 2011-11-08 2012-04-25 昆明理工大学 Method for extracting fluctuation of rotating speed through impact analysis
CN103134676B (en) * 2011-11-30 2016-10-05 上海宝钢工业检测公司 The on-line monitoring method for early warning of running state of gear box
WO2013115716A1 (en) 2012-01-30 2013-08-08 S.P.M. Instrument Ab Apparatus and method for analysing the condition of a machine having a rotating part
JP5930789B2 (en) * 2012-03-23 2016-06-08 三菱電機株式会社 Abnormal sound diagnosis device
CN102788671B (en) * 2012-07-26 2015-09-30 北京卫星环境工程研究所 Based on the structure failure modality diagnostic method of spacecraft vibration test sound spectrum
JP5977138B2 (en) * 2012-10-10 2016-08-24 日本信号株式会社 On-vehicle device and train control device using the same
CN103728134B (en) * 2012-10-16 2016-05-18 华锐风电科技(集团)股份有限公司 The detection method of unit bearing and device and wind-powered electricity generation unit
CN103116032A (en) * 2013-01-18 2013-05-22 北京金风科创风电设备有限公司 Method and device for obtaining revolving speed of wind generating set
JP6221652B2 (en) * 2013-11-08 2017-11-01 セイコーエプソン株式会社 Life prediction method, life prediction device, life prediction system, life calculation device, and rotating machine
CN103674538A (en) * 2013-12-18 2014-03-26 北京航天测控技术有限公司 Multi-fault mode identification method and device of swashplate of helicopter
CN103969046B (en) * 2014-05-20 2016-05-04 北京康拓红外技术股份有限公司 A kind of bearing acoustics diagnose system and method for and the coupling of wheel set bearing running-in machine
JP6283591B2 (en) * 2014-09-30 2018-02-21 Jfeプラントエンジ株式会社 Automatic vibration diagnostic equipment for rotating machinery
CN105651376B (en) * 2014-11-10 2019-08-06 上海宝钢工业技术服务有限公司 The analysis of mechanical equipment off-line checking system vibration signals spectrograph and alarm method
KR101663820B1 (en) * 2014-12-26 2016-10-10 주식회사 아이티매직 Method and apparatus of machine diagnosis using sound signal
CN104535176A (en) * 2015-01-22 2015-04-22 珠海格力电器股份有限公司 Vibration and noise detection device and method
CN105984474B (en) * 2015-02-10 2018-06-22 昆明耐维科技有限公司 A kind of broken rail detection method based on wideband encoding electric signal
GB2543521A (en) * 2015-10-20 2017-04-26 Skf Ab Method and data processing device for severity assessment of bearing defects using vibration energy
CN105548739B (en) * 2015-12-04 2019-01-15 江苏省电力公司宿迁供电公司 A kind of arrester operating state signal processing method
JP5954604B1 (en) * 2015-12-14 2016-07-20 富士ゼロックス株式会社 Diagnostic device, diagnostic system and program
US10110757B2 (en) * 2016-02-16 2018-10-23 Canon Kabushiki Kaisha Printing apparatus and notification method of printing apparatus abnormality in a printing apparatus
JP2017173076A (en) * 2016-03-23 2017-09-28 Ntn株式会社 State monitoring system and wind generator system with the same
GB2551112B (en) * 2016-05-25 2020-04-15 Ge Aviat Systems Ltd Aircraft component monitoring system
WO2017221588A1 (en) * 2016-06-21 2017-12-28 三菱電機株式会社 Load abnormality detection device
JP6718107B2 (en) * 2016-09-28 2020-07-08 エヌティーエンジニアリング株式会社 Vibration monitoring method and system for work machine
CN106597093B (en) * 2016-11-28 2018-11-13 深圳供电局有限公司 It is a kind of identification each span of transmission line of electricity between transmit vibration spectrum information method and system
WO2018100655A1 (en) 2016-11-30 2018-06-07 株式会社日立製作所 Data collection system, abnormality detection system, and gateway device
JP7017851B2 (en) * 2016-12-01 2022-02-09 住友重機械工業株式会社 Fault diagnosis system and processing unit
US10788395B2 (en) * 2017-02-10 2020-09-29 Aktiebolaget Skf Method and device of processing of vibration sensor signals
JP6388103B1 (en) * 2017-03-24 2018-09-12 日本精工株式会社 Vibration analyzer and machine parts diagnostic system
WO2018229836A1 (en) * 2017-06-12 2018-12-20 株式会社日立製作所 Bearing diagnosis device
CN107831013B (en) * 2017-10-11 2019-11-22 温州大学 Utilize the Method for Bearing Fault Diagnosis of probability principal component analysis enhancing cyclic bispectrum
JP6905445B2 (en) * 2017-10-16 2021-07-21 日本化薬株式会社 Crash detectors, methods for detecting aircraft crashes, parachute or paraglider deployment devices, and airbag devices
CN107894279B (en) * 2017-10-25 2018-12-07 北京金风科创风电设备有限公司 The synchronization detecting method and device of vibration of wind generating set data and floor data
JP7049138B2 (en) * 2018-03-02 2022-04-06 三菱重工機械システム株式会社 Deck crane condition judgment method, deck crane condition judgment device and deck crane system
US20210048792A1 (en) * 2018-03-05 2021-02-18 Nec Corporation Diagnostic apparatus, system, diagnostic method, and program
JP7134652B2 (en) * 2018-03-12 2022-09-12 曙ブレーキ工業株式会社 Anomaly detection method and anomaly detection device
CN108469329A (en) * 2018-03-20 2018-08-31 江苏新安电器有限公司 A kind of detecting system of washing machine non-equilibrium state
DE102018204648A1 (en) * 2018-03-27 2019-10-02 Robert Bosch Gmbh Guide, sensor assembly and method
CN108597055A (en) * 2018-04-20 2018-09-28 铨宝工业股份有限公司 Intelligent machine monitoring device with custom tolerance value
DE102018206434A1 (en) * 2018-04-25 2019-10-31 Aktiebolaget Skf Signal processing method and apparatus
CN110487546B (en) * 2018-05-10 2021-12-14 上汽通用汽车有限公司 Gearbox knocking noise testing method, testing device and evaluation method
CN109059839B (en) * 2018-05-23 2021-05-04 唐智科技湖南发展有限公司 Method, device and system for diagnosing out-of-roundness fault of wheel tread
CN110646201B (en) * 2018-06-08 2021-10-01 西门子股份公司 Bearing defect detection system and method
CN110658006B (en) * 2018-06-29 2021-03-23 杭州萤石软件有限公司 Sweeping robot fault diagnosis method and sweeping robot
KR102152352B1 (en) * 2018-09-28 2020-09-04 부경대학교 산학협력단 Apparatus and method for diagnosing machine fault
KR102126129B1 (en) * 2018-10-18 2020-06-24 현대위아 주식회사 Method for detecting tool status of machine tools
GB2578457B (en) * 2018-10-29 2021-01-13 Siemens Healthcare Ltd Anomaly detection and correlation system.
US11054339B2 (en) 2018-11-13 2021-07-06 GM Global Technology Operations LLC Method and apparatus for monitoring a machine bearing on-vehicle
JP7263271B2 (en) * 2019-01-25 2023-04-24 株式会社Cdsi Arithmetic unit and program
CN109738058A (en) * 2019-02-15 2019-05-10 江苏弘冉智能科技有限公司 A kind of self study vibrating failure diagnosis method
CN109799098B (en) * 2019-03-22 2021-08-10 唐智科技湖南发展有限公司 Method, device and medium for detecting transverse instability of bogie
CN109813544B (en) * 2019-03-26 2019-11-05 武汉众犇慧通科技有限公司 A kind of rotating machinery Incipient Fault Diagnosis method and system based on on-line monitoring
JP7352371B2 (en) * 2019-04-04 2023-09-28 株式会社日立製作所 diagnostic equipment
CN110108467B (en) * 2019-05-20 2020-10-20 长沙理工大学 Active sounding speed measurement method based on portable mobile equipment
FR3097960B1 (en) * 2019-06-27 2021-06-18 Safran Aircraft Engines Data acquisition method for detecting damage to a bearing
KR20210017577A (en) * 2019-08-09 2021-02-17 주식회사 엘지화학 Quantitatively diagnosis method for manufacturing facility
JP7339811B2 (en) * 2019-08-27 2023-09-06 株式会社荏原製作所 Abnormality detection method and polishing device for roller that transmits local load to retainer ring
CN110807901B (en) * 2019-11-08 2021-08-03 西安联丰迅声信息科技有限责任公司 Non-contact industrial abnormal sound detection method
CN111260835B (en) * 2020-01-19 2022-08-09 上海瑞皇管业科技股份有限公司 Monitoring and alarming system for urban comprehensive pipe gallery and control method thereof
CN112113655B (en) * 2020-09-21 2021-06-01 西南交通大学 Vibration signal detection device and health assessment method for subway track fastener
CN114433656A (en) * 2020-10-30 2022-05-06 深圳富桂精密工业有限公司 Punching abnormity detection system
CN112958840B (en) * 2021-02-10 2022-06-14 西南电子技术研究所(中国电子科技集团公司第十研究所) Automatic segmentation method for cutting force signal in precision part machining
CN115479788A (en) * 2021-05-31 2022-12-16 深圳富桂精密工业有限公司 Punching equipment abnormality detection method and electronic device
CN113392813B (en) * 2021-07-30 2022-04-26 湖北工业大学 Method and system for accurately identifying main frequency of vibration signal
CN113899446B (en) * 2021-12-09 2022-03-22 北京京仪自动化装备技术股份有限公司 Wafer transmission system detection method and wafer transmission system
JP2023127857A (en) * 2022-03-02 2023-09-14 日立建機株式会社 Performance diagnostic device of machine
CN116184140A (en) * 2023-04-21 2023-05-30 北京西能电子科技发展有限公司 Multifunctional monomer sensor suitable for GIS equipment defect detection
CN116976684B (en) * 2023-09-25 2024-01-02 尚古智造(山东)智能装备有限公司 Risk model predictive control method and system for logistics conveyor

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPS5432373A (en) * 1977-08-16 1979-03-09 Mitsubishi Electric Corp Abnormality judging and recognizing apparatus
JP2992954B1 (en) * 1998-12-22 1999-12-20 東日本旅客鉄道株式会社 Bridge pier abnormality detection method, system and recording medium
JP2004093255A (en) * 2002-08-30 2004-03-25 Nsk Ltd Abnormalities diagnostic equipment and method of machine
JP2004211813A (en) * 2002-12-27 2004-07-29 Nsk Ltd Bearing device for axle of rolling stock

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108885157A (en) * 2016-03-23 2018-11-23 Ntn株式会社 Railway abnormity diagnostic system and transacter
CN109313071A (en) * 2016-06-21 2019-02-05 Ntn株式会社 Vibration measurement device and abnormity diagnostic system

Also Published As

Publication number Publication date
JP2006113002A (en) 2006-04-27
CN1926413A (en) 2007-03-07

Similar Documents

Publication Publication Date Title
CN100516789C (en) Abnormality diagnosis system for machinery
Han et al. Fault feature extraction of low speed roller bearing based on Teager energy operator and CEEMD
WO2006043511A1 (en) Abnormality diagnosis system for machinery
Yang et al. ARX model-based gearbox fault detection and localization under varying load conditions
Bendjama et al. Application of wavelet transform for fault diagnosis in rotating machinery
EP2743669B1 (en) Fault detection system and associated method
William et al. Identification of bearing faults using time domain zero-crossings
US9453760B2 (en) Device and method for evaluation of vibrations
CN103502827B (en) For the situation method and apparatus of monitoring machine electric system
WO2015015987A1 (en) Bearing-device vibration analysis method, bearing-device vibration analysis device, and rolling-bearing status-monitoring device
Qu et al. A new acoustic emission sensor based gear fault detection approach
Kia et al. A comparative study of acoustic, vibration and stator current signatures for gear tooth fault diagnosis
WO2018222341A1 (en) Wind turbine fault detection using acoustic, vibration, and electrical signals
WO2013180723A1 (en) Methods apparatuses for defect diagnosis in a mechanical system
CN104081173A (en) Apparatus and method for analysing the condition of a machine having a rotating part
Feng et al. A novel order spectrum-based Vold-Kalman filter bandwidth selection scheme for fault diagnosis of gearbox in offshore wind turbines
CN103411774A (en) On-line early warning method of wind turbine generating unit on fluctuation working condition
Trajin et al. Comparison between vibration and stator current analysis for the detection of bearing faults in asynchronous drives
JP2005233789A (en) Abnormality diagnosis method of rotary machine, abnormality diagnosis apparatus, and abnormality diagnosis system
CN113049953B (en) Method, device, medium and equipment for fault location through noise
JP2009115537A (en) Vibration measuring method
Leite et al. Bearing fault detection in induction machine using squared envelope analysis of stator current
GB2122749A (en) Electrical condition monitoring of electric motors
CN111610038B (en) Fault diagnosis system, method, device, equipment and storage medium thereof
JPS63238522A (en) Device for monitoring axial torsional vibration

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant