CN102607848A - Detection method for train bearing fault - Google Patents
Detection method for train bearing fault Download PDFInfo
- Publication number
- CN102607848A CN102607848A CN2012100894452A CN201210089445A CN102607848A CN 102607848 A CN102607848 A CN 102607848A CN 2012100894452 A CN2012100894452 A CN 2012100894452A CN 201210089445 A CN201210089445 A CN 201210089445A CN 102607848 A CN102607848 A CN 102607848A
- Authority
- CN
- China
- Prior art keywords
- bearing
- train
- signal
- fault
- spectrum
- 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.)
- Pending
Links
Images
Abstract
The invention provides a detection method for a train bearing fault. The method comprises the following steps of: comprehensively detecting an integral fault characteristic spectrum so as to ensure the safety of a train bearing; and when detection signals are implemented, performing further processing by utilizing related methods after a measured spectrum result is obtained by sampling, conversion and calculation so as to overcome the interference of random vibratory impulses in measurement. The method has the effects that a KW2002 bearing fault diagnostic apparatus designed and manufactured by utilizing the method is used for realizing the static no-disassembly detection of faults comprising the abrasion of a bearing component, and a KW0021 train bearing monitor designed and manufactured by utilizing the method can be used for monitoring the status of the bearing in real time in the running of a train to realize the early warning of the bearing fault, so that the safety of railway traffic is greatly improved; and minor fault characteristics can be detected in the running of the train, and the long-lasting thinking mode that the inference produced in the running of the train cannot be overcome in China is broken.
Description
Technical field
The present invention relates to a kind of detection method, particularly a kind of detection method of railroad train bearing fault.
Background technology
Railway is the important department of Chinese national economy, and train has become the main vehicles that are particularly suitable for China's national situation.Just because of the status of railway in national economy is so important, so the cardinal task that the safety of railway traffic has become the railway system and commonwealthn to be concerned about.In the today of constantly raising speed significantly, it is important that this problem especially seems.
And in railway traffic, a kind of serious accident is to cut axle (being off-axis), directly has influence on the safety of passenger and goods.In case cut an accident in transit, train derailment might take place, with producing enormous economic loss, casualties and extremely abominable social influence.According to statistics, the accident of cutting more than 90% is because the fault of bearing causes, and, is called " combustion axle " usually because at this moment bearing can send high heat by the professional person.
Therefore, research and development bearing fault, the scientific and effective method of diagnosis also actively drop into practical application, can produce huge economic benefit and social benefit.The present invention be exactly for an early diagnosis difficult problem that solves train bearing provide more comprehensively, more effectively, more practical method.
At present, the method for railroad train rolling bearing fault diagnosis mainly contains five kinds:
1, artificial diagnosis (comprising overall appearance inspection and abnormal sound diagnosis).Promptly there is the worker who enriches on-site experience to find different shape (for example getting rid of oil) or judge according to the sound that sends in the bearing rotation process whether the work of bearing element is normal through visual examination.
2, the visual examination of bearing element.Promptly after journey is decomposed bearing the impaired degree of all size of component and surface is checked and judged, except visual inspection, also adopt high-tech approaches such as magnetic powder inspection according to repairing of shed repair (each rolling stock section every at a distance from a year and a half once) to the maintenance of every joint compartment.The advantage of this method is directly perceived, but because bearing can destroy original break-in state after decomposing, so external expert generally believes and should not advocate.
3, electronic auscultation device diagnosis.Be the improvement of first method, the now more health check-up of not understanding fast that is used for freight car bearing is surveyed.In order to improve the sensitivity of human auditory system; Overcome the interference of neighbourhood noise, in the bearing rotating process, an audio sensor is adsorbed on the bearing (ball) cover diagnosis; Import people's ear into through multiplying arrangement and earphone then; Though this diagnostic method can be expanded the sensitivity of people's ear to a certain extent, to the change that diagnostic result can not produce essence, diagnostic result receives human factor to influence this disadvantage still to exist.
4, axle temperature detection method.It is found that the accident of cutting more than 90% is heated up rapidly by axletree and causes, so China all installs axle temperature annunciator since the eighties in last century in all passenger cars, when the temperature rise of bearing surpasses 40 ℃, sent alerting signal.This method is used so far always.
5, Spectral Analysis Method.Along with development of modern science and technology, people begin the method for spectrum analysis is introduced the bearing failure diagnosis field from the last century the eighties, and have obtained certain success.
Above-mentioned the 1st kind and the 3rd kind of method, the result of diagnosis is examiner's subjective assessment, and its correctness is relevant with examiner's experience, attitude, sense of responsibility even physical qualification, and randomness is stronger, has bigger risk.
Though the 2nd kind of method is directly perceived, owing to must bearing be decomposed, so can destroy the existing break-in condition of bearing.The bearing that re-assemblies even original state is good, also must carry out new break-in.And in running-in period, the probability that bearing produces fault is very high.As Japanese EastSea Passenger Railway Co., Ltd patent " bearing detecting device that is used for railway master the motor " (applying date: 1998.3.16; International publication day: disclosed all disadvantages in instructions 1999.9.23).
The 4th kind of dynamic monitoring that method is used for runing train.But for bearing fault, the amplitude of temperature rise neither adequate condition neither necessary condition.For example, recall Beijing-Shanghai express railway 54 row CRH380BL motor train unit trains in August, 2011, rectify and improve and return use again after qualified, wherein recur hot box and report to the police that to report by mistake promptly be one of the main reasons.Claim that its " wrong report " is inaccurate, because the temperature rise of these bearings exceeds standard really, but might not there be bearing fault in the bearing that temperature rise exceeds standard.Axle temperature annunciator " wrong ", not equal to method itself is wrong.But from another point of view, if temperature rise is to be caused by bearing fault really, will be after alarm gives the alarm so in extremely short time Nei Fashengqie axle accident.Related data shows, " Fig. 3-1 (this figure omits---the person of drawing annotates, down with) be depicted as the axle temperature change procedure of certain passenger train hot box accident.The axle box temperature rises to 197 ℃ by 89 ℃ (being actual temperatures, non-temperature rise) and has only 3 minutes when producing the combustion axle." (referring to the 33rd page in institution of higher education's teaching material " vehicle electric device ", Xi'an Communications University's chapter is because of compiling, and China Railway Press publishes, 1999) must be noted that, above-mentioned teaching material is to publish in 1999, the speed per hour of train is no more than 100 kilometers basically at that time.And physical basic theories is told everybody, square being directly proportional of fricative heat and speed.When the speed per hour of train is brought up to 300 kilometers, above-mentioned three minutes---Here it is leaves the All Time that people take urgent measure temporarily for---20 seconds of being about to reduce to limited! Therefore, no matter to see at which aspect axle temperature annunciator all is neither science, also unpractical makeshift from.
So-called " spectrum analysis " is called as " Fourier transform " in mathematics.We can obtain whole components of a sophisticated signal like a cork through spectrum analysis.Say that from principle Spectral Analysis Method and method 1,3 all are to come tracing trouble through the abnormal component in the discovery vibration signal.Different is, the latter is sound and a loudness thereof of finding different tones through people's ear, can only be qualitative, and can't be quantitative.And the former is frequency (being exactly tone) and an amplitude (being exactly loudness) of confirming abnormal signal through measuring.Therefore the latter more science, more objective, more accurately, sensitiveer.
For a long time, spectrum analysis is considered to detect the effective ways of bearing fault, and the fault characteristic frequency of deriving critical piece inner ring, outer ring and the roller of bearing is respectively:
f
1=Z (1+d/D) f
0/ 2 (for inner rings)
f
2=Z (1-d/D) f
0/ 2 (for outer rings)
f
3=D/d (1-(d/D)
2) f
0(for roller)
Wherein Z is a quantity of using roller in the bearing, and d is the diameter of roller, and D is the mean value (being the diameter of roller revolution track) of bearing inner race and race diameter, f
0It is the rotating speed of wheel shaft.
Above-mentioned formula is considered to the mathematical model of bearing fault characteristic frequency spectrum, and it is the result of theoretical derivation.But above-mentioned mathematical model has been ignored many actual failure modes that take place; For example the most common bearing element wearing and tearing (sur-face peeling) in the reality; And damaged for cutting the most dangerous retainer of an accident, also have the tired the most serious tread damage etc. that influences of bearing material.
In practical application, the wearing and tearing of bearing element are common phenomena the most, also are one of tendencys that causes the bearing catastrophic failure.The method that the Japanese EastSea Passenger Railway Co., Ltd patent " bearing detecting device that is used for railway master motor " that preceding text are quoted is provided is; Take out the grease in the bearing; Be placed on and detect its tenor in the spectroanalysis instrument; And testing result and previous testing result compared, see whether its tenor increases.The practical application property of this method is relatively poor, has shown that the patent inventor's is helpless.
Summary of the invention
To deficiency of the prior art, the purpose of this invention is to provide a kind of detection method of railroad train bearing fault, utilize the complete mathematical model of railroad train bearing fault in this method, the railroad train bearing fault is effectively detected.Method used in the present invention is based on traditional vibration signal spectrum analysis technique, and solved some still unsolved before this difficult point problems on this basis.
For realizing above-mentioned purpose, the technical scheme that the present invention adopts provides a kind of detection method of railroad train bearing fault, and this method includes following steps:
(1) the complete fault signature spectrum for ensureing that train bearing safety must detect comprises:
For train bearing inner ring fault signature spectrum f
1=Z (1+d/D) f
0/ 2 ... (1)
For train bearing outer ring fault signature spectrum f
2=Z (1-d/D) f
0/ 2 ... (2)
For train bearing roller fault signature spectrum f
3=D/d (1-(d/D)
2) f
0(3)
Contain the damaged characteristic spectrum f of retainer for above-mentioned bearing element wearing and tearing
4=0 ... (4)
For defect characteristic spectrum f inside and outside the train wheel set bearing tread
5=f
0(5)
In the formula: Z is the roller quantity in the bearing; D is the diameter of roller; D is the mean value of inner ring and race diameter; f
0Revolution for bearing.
When (two) stating detection signal on the implementation,, utilize the method for relevant treatment in order effectively to overcome the interference of random vibration pulse to measuring:
When time-domain signal was handled, algorithm was the correlativity with two functions of convolutional calculation:
z(t)＝∫x(t)y(t-T
0)dt
In the formula: x (t) is one of them function of participating in correlation computations; Y (t) is another function; Y (t-T
0) be that y (t) is along time shaft translation T
0After function; Z (t) is the related function of y (t) and y (t);
And when frequency-region signal was handled, the calculating of this correlativity only need be multiplied each other the spectrum component correspondence of two functions and got final product:
1, signal is carried out the sampling first time, obtain f
A=f
A1+ f
A2+ f
A3+ ... + f
AN, wherein N is a sampling number;
2, to the sampled result f first time of signal
ACarry out Fourier transform, obtain the power spectrum F that signal A samples for the first time
A=F
A1+ F
A2+ F
A3+ ... + F
AM(M=N/2);
3, signal is carried out the sampling second time, obtain f
B=f
B1+ f
B2+ f
B3+ ... + f
BN
4, to the secondary sampled result f of signal
BCarry out Fourier transform, obtain the power spectrum F of signal B
B=F
B1+ F
B2+ F
B3+ ... + F
BM(M=N/2);
5, with the power spectrum component of the double sampling back evolution that multiplies each other respectively, obtain F=(F
A1F
B1)
1/2+ (F
A2F
B2)
1/2+ (F
A3F
B3)
1/2+ ... + (F
AMF
BM)
1/2
F is the correlated results of twice measurement, i.e. the geometrical mean of twice measurement; Can access equally 3 times, 4 times ... N correlations result, the numerical value of N is big more, and anti-jamming capacity is strong more.
Effect of the present invention is that this method adopts at the operation train through KW2002 bearing failure diagnosis appearance, KW0021 train bearing monitor, with regard to the state of the real-time monitoring bearing of ability, realizes the early warning of bearing fault, the security that improves railway traffic greatly.This method has the ability in train operation, to detect less fault signature, has broken up the domestic outer for a long time thinking set about " interference that train produces in advancing is unconquerable ".
Description of drawings
Fig. 1 is that the KW0021 profile shaft that the present invention is used for kinetic measurement holds the failure diagnostic apparatus outside drawing;
Fig. 2 is that the KW2002 profile shaft that the present invention is used for static measurement holds the failure diagnostic apparatus outside drawing;
Fig. 3 is that the present invention uses the bearing outer ring of static measurement method measurement to peel off the frequency spectrum of fault;
Fig. 4 is the frequency spectrum that the present invention uses the bearing roller fault of static measurement method measurement;
Fig. 5 is that the present invention uses the detected freight car bearing of static measurement method outer ring to peel off fault;
Fig. 6 is the passenger vehicle tread fault spectrum that the present invention uses dynamic measurement method to monitor;
Fig. 7 is that the present invention peels off (grinding the skin fault) with the detected roller surface of static measurement method;
Fig. 8 is the frequency spectrum that the present invention uses the passenger vehicle bearing outer ring fault that dynamic measurement method monitors.
Embodiment
In conjunction with accompanying drawing the detection method of railroad train bearing fault of the present invention is explained.
Mechanism according to the bearing fault generation; Improve the mathematical model of bearing fault characteristic spectrum; This method can be applied to the detection and the maintenance of bearing in two ways: the first realizes not understanding health check-up and surveys in maintenance process, and it two is to be in operation the bearing of train is monitored in real time.A kind of product is to be the Static Detection product (Fig. 2) of representative with KW2002 bearing failure diagnosis appearance; A kind of is the KW0021 train bearing monitor (Fig. 1) that is used as detection of dynamic.
The detection method of railroad train bearing fault of the present invention, this method includes following steps:
(1) the complete fault signature spectrum for ensureing that train bearing safety must detect comprises:
For train bearing inner ring fault signature spectrum f
1=Z (1+d/D) f
0/ 2 ... (1)
For train bearing outer ring fault signature spectrum f
2=Z (1-d/D) f
0/ 2 ... (2)
For train bearing roller fault signature spectrum f
3=D/d (1-(d/D)
2) f
0(3)
Contain the damaged characteristic spectrum f of retainer for above-mentioned bearing element wearing and tearing
4=0 ... (4)
For defect characteristic spectrum f inside and outside the train wheel set bearing tread
5=f
0(5)
In the formula: Z is the roller quantity in the bearing; D is the diameter of roller; D is the mean value of inner ring and race diameter; f
0Revolution for bearing.
When (two) stating detection signal on the implementation,, utilize the method for relevant treatment in order effectively to overcome the interference of random vibration pulse to measuring:
When time-domain signal was handled, classical algorithm was the correlativity with two functions of convolutional calculation:
z(t)＝∫x(t)y(t-T
0)dt
In the formula: x (t) is one of them function of participating in correlation computations; Y (t) is another function; Y (t-T
0) be that y (t) is along time shaft translation T
0After function; Z (t) is the related function of y (t) and y (t).
And when frequency-region signal was handled, the calculating of this correlativity only need be multiplied each other the spectrum component correspondence of two functions and got final product, and specific practice is:
1, signal is carried out the sampling first time, obtain f
A=f
A1+ f
A2+ f
A3+ ... + f
AN, wherein N is a sampling number;
2, to the sampled result f first time of signal
ACarry out Fourier transform, obtain the power spectrum F that signal A samples for the first time
A=F
A1+ F
A2+ F
A3+ ... + F
AM(M=N/2);
3, signal is carried out the sampling second time, obtain f
B=f
B1+ f
B2+ f
B3+ ... + f
BN
4, to the secondary sampled result f of signal
BCarry out Fourier transform, obtain the power spectrum F of signal B
B=F
B1+ F
B2+ F
B3+ ... + F
BM(M=N/2);
5, with the power spectrum component of the double sampling back evolution that multiplies each other respectively, obtain F=(F
A1F
B1)
1/2+ (F
A2F
B2)
1/2+ (F
A3F
B3)
1/2+ ... + (F
AMF
BM)
1/2
F is the correlated results of twice measurement, i.e. the geometrical mean of twice measurement; Can access equally 3 times, 4 times ... N correlations result, the numerical value of N is big more, and anti-jamming capacity is strong more.
The KW2002 profile shaft that is used for static measurement holds failure diagnostic apparatus for f
4The detection effect remarkable.
The KW0021 profile shaft that is used for kinetic measurement holds failure diagnostic apparatus for f
5The detection effect remarkable.
The detection method of railroad train bearing fault of the present invention is achieved in that using the KW0021 profile shaft and holds in the kinetic measurement that failure diagnostic apparatus carries out, and this method adopts corresponding electronics disposal route " being correlated with " disposal route in the mathematics in addition.This method has overcome effectively that various random vibrations make and utilize the bearing fault vibration signal to detect the successfully realization under dynamical state of this classic method of bearing the strong interference of bearing fault signal during train is advanced.
So-called " being correlated with " is meant alike degree between two objects or the phenomenon.
For two time-domain signals in two functions (curve) or the physics in the mathematics, people measure its correlativity with their convolution, and formula is:
z(t)＝∫x(t)y(t-T
0)dt
Wherein to be two be the function of independent variable with t for x (t), y (t), T
0Be commonly called " initial phase ".
The physical significance of related function z (t) is, with curve y (t) translation T
0, cover then on the curve x (t), observe the alike degree of two curves.Its mathematical method is the back integration that multiplies each other of the corresponding point functional value with two functions.Last z (t) numerical value is big more, its correlativity high more (two functions are alike more).
Obviously, z (t) is not only the function of x (t), y (t), and is T
0Function because T
0Difference, the numerical value of z (t) are also different, must constantly change T
0Numerical value, maximum in the hope of the numerical value of z (t).
Two time-domain signals are carried out process of convolution, only if this signal is very simple, otherwise are very complicated by processed conventionally process.
According to mathematical principle, the convolution of time-domain signal is equivalent to the product of frequency-region signal.Signal is being carried out calculate its correlativity again after the spectrum analysis, just becoming very simple.Specific practice is:
1, signal is carried out the sampling first time, obtain f
A=f
A1+ f
A2+ f
A3+ ... + f
AN, N counts for sampling;
2, to the sampled result f first time of signal
ACarry out Fourier transform, obtain signal A, for the first time the power spectrum F of sampling
A=F
A1+ F
A2+ F
A3+ ... + F
AM(M=N/2);
3, signal is carried out the sampling second time, obtain f
B=f
B1+ f
B2+ f
B3+ ... + f
BN
4, to the secondary sampled result f of signal
BCarry out Fourier transform, obtain the power spectrum F of signal B
B=F
B1+ F
B2+ F
B3+ ... + F
BM(M=N/2);
5, with the power spectrum component of the double sampling back evolution that multiplies each other respectively, obtain F=(F
A1F
B1)
1/2+ (F
A2F
B2)
1/2+ (F
A3F
B3)
1/2+ ... + (F
AMF
BM)
1/2
Can know that by aforementioned calculation frequency-region signal relevant is actually result with twice measurement and carries out geometric mean and calculate.Geometrical mean is different with the arithmetic mean value, F
AMAnd F
BMAs long as in have one to be zero, its result just is zero.This just thoroughly eliminates the influence of unstable signal.The vibration signal of the normality constant amplitude that produces owing to object resonance, through the wave filter filtering in the said instrument, what do not eliminate is not the pulse signal of low frequency.
For example, when measuring for the first time, detected the pulse interference signal at random of 35Hz, so when measuring for the second time this pulse interference signal still to be rendered as the probability of 35Hz very little, for example 1%.That is to say that the relevant treatment of passing through exists the probability of the impulse disturbances of 35Hz to have only 1%.If increase the number of times of relevant treatment, for example twice measurement increased to 3 measurements, exist the probability of the impulse disturbances of 35Hz be reduced to ten thousand/.If measurement is increased to 4 measurements, then probability be reduced to 1,000,000/.KW0021 train bearing monitor adopts 4 measurements exactly.
Through such processing, the spectrum component that can keep only is the extremely stable pulse signal frequency spectrums of those frequencies.
The bearing fault signal just in time is to belong to the extremely stable pulse signal of frequency, except f
4=0 component quilt filtering then.As everyone knows, the quality of train itself is very big, only if in the finite time of starting (acceleration) and stop (deceleration), the speed per hour of train and the revolution of bearing all are sufficiently stable, are like this at least within measured cycle.
The present invention can be applied to the detection and the maintenance of bearing in two ways: the first realizes not understanding health check-up and surveys in maintenance process, and it two is to be in operation the bearing of train is monitored in real time.For this reason specialized designs two series products: one type is to be the Static Detection product of representative with KW2002 bearing failure diagnosis appearance; A kind of is the KW0021 train bearing monitor that is used as detection of dynamic.
1, the mathematical model of bearing element wearing and tearing
The outer ring of bearing element, inner ring and roller through each other long-term rolling friction, can produce local hardened layer and come off, and are commonly called as " stone roller skin " and cause the surface of element can produce some powder or disintegrating slag.In the bearing movable process, in a single day roller ground these powder or disintegrating slag, and powder or disintegrating slag are many more, and the vibration of generation is many more; The particle of powder or disintegrating slag is big more, and the Oscillation Amplitude of generation is big more.These vibrations are random fully, but theory and practice proves that through after the processing of spectrum analysis, the integration of these Oscillation Amplitudes is A=∫
0 TF (t) dt, promptly their total DC components are directly proportional with the product of grain size with the quantity of powder or disintegrating slag basically.What worry most for the maintenance personal of lorry is the breakage of retainer, and in the spectrum measurement of static state, its result is identical with " stone roller skin ".
Therefore, (so-called " static state " is not to be the bearing transfixion, and is meant that under the train non-operating state rotation of bearing is driven by measuring system in the static measurement of bear vibration.Relative with it, " kinetic measurement " is meant the real-time measurement in train is advanced.) in DC component in the measurement result (be f
4=0) can be referred to as the characteristic spectrum that bearing element weares and teares (containing the retainer breakage).
2, the mathematical model of the mechanism of bearing element fault and wheel tread fault
On the employed train of China; Though bearing is gyrating, the outer ring of bearing is static basically relatively, the total weight of vehicle body; All be pressed in the bearing outer ring top, transmit limited area and 1～2 roller that contacts with it that this pressure almost only leans on the top of bearing outer ring.Roller and inner ring rotate constantly, and its different at different position and different is born pressure in turn, but to bear the position of pressure almost be constant in the outer ring.Under normal situation, the limited area of bearing outer ring the top is easy fatigue.
Above-mentioned analysis also only is in normal duty.There is wound if take turns right tread; For example, train is after emergency brake, and the tread that contacts with track partly can be polished---be commonly called as " flat wheel "; Or for example the wheel tread depths exists trachoma or bubble and the internal injury that forms; Can produce instantaneous jolting when then this injury contacts with track, send the sound that vibration produces simultaneously, therefore this phenomenon is called " getting ready " by the professional person.When getting ready, instantaneous jolting can produce strong wallop, and this impulsive force also is added on the limited area of bearing outer ring the top through 1～2 roller with the pressure in compartment simultaneously.Because it is bigger that this impulsive force is compared with the normal pressure in compartment, the limited area that makes the bearing outer ring the top is easy fatigue more.
On the surface, it is irrelevant to take turns right tread and bearing, but the damage of tread to the fatigue of bearing, peel off or breakage is producing directly influence.Therefore in the detection that detects bearing fault, never can ignore check to tread damage.
In kinetic measurement, if having only place damage on a tread, to take turns so whenever turning around, tread will produce one-shot.Therefore impacting the frequency that produces should be identical with the right revolution of wheel.Therefore we can say the characteristic frequency f of tread fault
5=f
0In theory, even on a tread Liang Chu, three places even more many places damage are arranged, so long as not on circumference, being evenly distributed, its characteristic frequency still satisfies f
5=f
0
3, the effective ways of in the bearing fault detection of dynamic, getting rid of vibration interference
Can in train operation, gather and detect the bearing vibration signal effectively? For a long time, expert both domestic and external almost offers a negation with one voice.Bearing has produced its distinctive vibration because of fault, its amplitude is compared very little with other vibrations.Therefore, it is generally acknowledged under abominable like this work condition environment, want to gather the fault-signal of bearing, be tantamount to look for a needle in a haystack.
The maximum characteristics of this method are exactly through long-term and unremitting research, although find that the fault-signal of bearing is faint, it has own exclusive characteristic after all.Can find out its characteristic according to the modal analysis method of vibration, set up mathematical model and algorithm thereof,, can extract needed signal content through KW0021 train bearing monitor and KW2002 bearing failure diagnosis appearance.
The China railways system takes respectively to safeguard by the milimeter number of traveling or the time interval of operation for lorry and passenger vehicle.Up to now, except motor-car, bearing is reinstalled use after still adopting a series of processing such as decomposing back range estimation, cleaning.As aforementioned said, for the product of the precision-fit that needs break-in, bearing new potential faults possibly occur after decomposing.In fact, owing to reinstall the back mismatch takes place, thereby cause the incident of great near accident to exist.
As long as auxiliary with general wheel to drive unit, the KW2002 bearing failure diagnosis appearance of mentioning among the present invention just can be realized the puzzled health check-up survey of passenger vehicle or freight car bearing.The parameter that detects comprises the damaged and the most common stone roller skin of retainer that the maintenance personal worries the most.Although belong to Static Detection, this product still has good anti-interference, and its testing result can not receive the influence of rugged surroundings.
In order to guarantee train operating safety, be not enough only obviously by hard time maintenance.Just can measure cardiogram when going over regularly to hospital's health check-up, therefore can not eliminate the possibility of heart attack just as people.So for recessive patient, the doctor advises that he wears carry-on ECG tester.The KW0021 train bearing monitor that the present invention mentioned come to this a kind of " portable electrocardiogram machine ".This instrument is installed in the compartment, just can constantly goes the rounds to detect, also can stop to go the rounds, fixed point is monitored 8 the bearing vibration frequency spectrums in this compartment at any time to doubtful fault bearing.In case unusual frequency spectrum occurs, and its amplitude surpasses setting value, instrument will send alerting signal.Because this instrument has adopted effective interference protection measure, can guarantee basically can not report by mistake for bearing in good condition.Simultaneously, in order to satisfy crew member's custom for a long time, instrument has kept the measuring ability of axle temperature.Different with traditional axle temperature annunciator, what this instrument showed is the temperature rise of bearing, if demonstration is temperature, then also needs to calculate temperature rise according to environment temperature at that time.
Embodiment
This detection method application Tianjin opens the KW2002 bearing failure diagnosis appearance (Fig. 2) of pavilion Electronics Co., Ltd. production and two kinds of products of KW0021 train bearing monitor (Fig. 1) are realized.
All adopted the program of 4 measurements and relevant treatment in the built-in analysis software of said two products, its process is:
1, to carrying out sampling and A/D conversion at 512 through the sensor output signal after amplification, the Filtering Processing;
2, above-mentioned transformation results is carried out Fast Fourier Transform (FFT), obtain first group of 200 power spectrum component;
3, time-delay is 1 minute, repeats above-mentioned 1 and 2, obtains second group of 200 power spectrum component;
4, time-delay is 1 minute, repeats above-mentioned 1 and 2, obtains the 3rd group of 200 power spectrum components;
5, time-delay is 1 minute, repeats above-mentioned 1 and 2, obtains the 4th group of 200 power spectrum components;
6, the respective components with above-mentioned four groups of power spectrum components multiplies each other respectively, opens 4 powers then, and what obtain is 200 power spectrum components after 4 correlations are handled;
7, on LCD, draw 200 perpendicular line, it is the size of corresponding 200 power spectrum components highly respectively;
8, with the addition of 200 power spectrum components, the result of gained represents the bearing fault size of quantification.
Except the fault that detects bearing outer ring, inner ring, roller, KW2002 bearing failure diagnosis appearance can also detect the wearing and tearing of bearing element and fault (as shown in Figure 7) that retainer breaks effectively; KW0021 train bearing monitor can also detect the fault (as shown in Figure 6) of wheel tread effectively.
When calendar year 2001, rolling stock section carried out the Static Detection test in former Tianjin, the bearing of having selected an outer ring to peel off carried out repeatability as the model and detects, and the result of each detection can both reappear preferably, and is as shown in Figure 3.Its detected fundamental frequency is consistent with theoretic characteristic frequency, and each harmonic is also very clean, as the textbook typical case.
But the most serious responsible section chief that works is still half believing, half doubting to this method.In order to consider test result's correctness, he has taken out the roller of a collection, and an elongated crackle is arranged on the surface of this roller, and is parallel with the axle of roller.Carry out in the customary shed repair once at a distance from a year and a half every then, to such an extent as to, come to light through magnetic powder inspection at last because crackle has very carefully been sneaked out visual inspection personnel's inspection.The section chief is installed in it in the bearing, mixes with other bearings, and which is out of order judging actually to the unwitting situation of testing crew.Yet the result of test but is perfect, as shown in Figure 4 unexpectedly.Because the passenger vehicle bearing arrangement, the characteristic frequency of outer ring and roller fault much at one, the fault that wants to distinguish through the measurement of characteristic frequency bearing outer ring and roller is very difficult.But the outer ring has determined that with the function of roller the structure of their fault spectrum is different.The outer ring is fixed, so every roller during through abort situation, the Oscillation Amplitude of generation is identical, and the frequency spectrum graphics that therefore detects has stable first-harmonic harmonic.But roller rotates, and when the fault roller had only through the top, location of fault just can send strong collision.And after leaving the top, because the gap of bearing, even there is fault also can not produce tangible vibration.Therefore because the Oscillation Amplitude that the fault of roller causes can produce periodic variation, the periodic signal of this changes in amplitude is called as modulated wave.The perfect part of test result is that it has not only generated fault spectrum clearly, and its first-harmonic harmonic all demonstrates the characteristics (there is side frequency both sides) of typical modulated wave.
The above-mentioned fact fully proves: the testing result of the spectrum analysis of this method is superior to people's naked-eye observation.
The photo of Fig. 5 and Fig. 7 is that the Ministry of Railways gathers in the qualification process of former Qinhuangdao rolling stock section to product, and it is peeled off fault and grind the skin fault through the detected respectively freight car bearing of KW2002 bearing failure diagnosis appearance outer ring.Particularly the latter has confirmed that the mathematical model of the measurement stone roller skin that this method proposes is correct.Such a case is in other rolling stock sections, and for example Nanjing rolling stock section, Fuyang rolling stock section etc. also have a lot.
2002; Former Tianjin rolling stock section arranges; On a joint sleeping carriage of 253/254 express train in Tianjin-Guangzhou, a KW0021 train bearing trouble-monitoring instrument being installed makes an experiment; Handle owing to passed through four correlations, the fault spectrum that peel off 3 axle outer rings clearly shows, and is as shown in Figure 8.Its frequency spectrum shows as typical first-harmonic and 2 subharmonic, and in reaching the tracking time of some months, keeps stable, does not have to find to have the trend of deterioration.
When on No. 4417/4418 train in Tianjin-Qinhuangdao, testing; The steward reflects have a joint compartment to have " getting ready " phenomenon on this train to testing crew, but through inspection several times; All not having to find is that which tread has problem; Ask through the instrument of this method and assist calibrating? Testing crew is transferred to this joint compartment with instrument and attached sensor, and the position of just pinpointing the problems at once is as shown in Figure 6; Fault spectrum is clear, fully the mathematical model of the prior tread fault that proposes of Pass Test personnel.And find, belong to the testing result of other three bearings of a bogie together with this bearing, do not find a bit spectrum component, even with another bearing on same axle of this bearing also be like this.Each bearing signal is described and is measured passage accordingly and have good isolation performance, the phenomenon that can not disturb and judge by accident, misjudge.Even so, still do not find flat wheel phenomenon on this tread of taking turns, therefore can judge under tread to have bubble.
Claims (1)
1. the detection method of a railroad train bearing fault, this method includes following steps:
(1) the complete fault signature spectrum for ensureing that train bearing safety must detect comprises:
For train bearing inner ring fault signature spectrum f
1=Z (1+d/D) f
0/ 2 ... (1)
For train bearing outer ring fault signature spectrum f
2=Z (1-d/D) f
0/ 2 ... (2)
For train bearing roller fault signature spectrum f
3=D/d (1-(d/D)
2) f
0(3)
Contain the damaged characteristic spectrum f of retainer for above-mentioned bearing element wearing and tearing
4=0 ... (4)
For defect characteristic spectrum f inside and outside the train wheel set bearing tread
5=f
0(5)
In the formula: Z is the roller quantity in the bearing; D is the diameter of roller; D is the mean value of inner ring and race diameter; f
0Revolution for bearing;
When (two) stating detection signal on the implementation,, utilize the method for relevant treatment in order effectively to overcome the interference of random vibration pulse to measuring:
When time-domain signal was handled, algorithm was the correlativity with two functions of convolutional calculation:
z(t)＝∫x(t)y(t-T
0)dt
In the formula: x (t) is one of them function of participating in correlation computations; Y (t) is another function; Y (t-T
0) be that y (t) is along time shaft translation T
0After function; Z (t) is the related function of y (t) and y (t);
And when frequency-region signal was handled, the calculating of this correlativity only need be multiplied each other the spectrum component correspondence of two functions and got final product:
1, signal is carried out the sampling first time, obtain f
A=f
A1+ f
A2+ f
A3+ ... + f
AN, wherein N is a sampling number;
2, to the sampled result f first time of signal
ACarry out Fourier transform, obtain the power spectrum F that signal A samples for the first time
A=F
A1+ F
A2+ F
A3+ ... + F
AM(M=N/2);
3, signal is carried out the sampling second time, obtain f
B=f
B1+ f
B2+ f
B3+ ... + f
BN
4, to the secondary sampled result f of signal
BCarry out Fourier transform, obtain the power spectrum F of signal B
B=F
B1+ F
B2+ F
B3+ ... + F
BM(M=N/2):
5, with the power spectrum component of the double sampling back evolution that multiplies each other respectively, obtain F=(F
A1F
B1)
1/2+ (F
A2F
B2)
1/2+ (F
A3F
B3)
1/2+ ... + (F
AMF
BM)
1/2
F is the correlated results of twice measurement, i.e. the geometrical mean of twice measurement; Can access equally 3 times, 4 times ... N correlations result, the numerical value of N is big more, and anti-jamming capacity is strong more.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2012100894452A CN102607848A (en) | 2012-03-27 | 2012-03-27 | Detection method for train bearing fault |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2012100894452A CN102607848A (en) | 2012-03-27 | 2012-03-27 | Detection method for train bearing fault |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN102607848A true CN102607848A (en) | 2012-07-25 |
Family
ID=46525416
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2012100894452A Pending CN102607848A (en) | 2012-03-27 | 2012-03-27 | Detection method for train bearing fault |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102607848A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104568444A (en) * | 2015-01-28 | 2015-04-29 | 北京邮电大学 | Method for extracting fault characteristic frequencies of train rolling bearings with variable rotational speeds |
| CN105806604A (en) * | 2016-03-18 | 2016-07-27 | 唐智科技湖南发展有限公司 | Locomotive vehicle running gear bearing holder fault pre-alarm method |
| CN106441893A (en) * | 2016-09-22 | 2017-02-22 | 北京邮电大学 | Train rolling bearing fault and impurity vibration distinguishing method |
| CN107547108A (en) * | 2017-08-26 | 2018-01-05 | 南京天普机电产品制造有限公司 | Car axle report communicates comprehensive tester |
| CN108204897A (en) * | 2016-12-16 | 2018-06-26 | 唐智科技湖南发展有限公司 | A kind of bearing parameter correction judgement and multi-parameter diagnose matched method automatically |
| CN108225769A (en) * | 2016-12-15 | 2018-06-29 | 唐智科技湖南发展有限公司 | It is a kind of to prevent the Mechanism Diagnosis method that the mistaken diagnosis of retainer characteristic spectrum is fault spectrum |
| CN108363853A (en) * | 2018-01-31 | 2018-08-03 | 浙江浙大鸣泉科技有限公司 | A kind of engine speed measurement method based on multisensor correlation denoising |
| CN112013895A (en) * | 2019-05-28 | 2020-12-01 | 浙江德盛铁路器材股份有限公司 | Method for monitoring and evaluating quality condition of railway track basic equipment |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6967586B2 (en) * | 2000-10-20 | 2005-11-22 | Sankyo Seiki Mfg. Co., Ltd. | Bearing test method, bearing test device, bearing monitoring device and storage device |
| CN1825082A (en) * | 2006-03-31 | 2006-08-30 | 洛阳轴研科技股份有限公司 | Automatic diagnosing system for rolling bearing fault |
| CN102269655A (en) * | 2010-06-04 | 2011-12-07 | 北京化工大学 | Method for diagnosing bearing fault |
-
2012
- 2012-03-27 CN CN2012100894452A patent/CN102607848A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6967586B2 (en) * | 2000-10-20 | 2005-11-22 | Sankyo Seiki Mfg. Co., Ltd. | Bearing test method, bearing test device, bearing monitoring device and storage device |
| CN1825082A (en) * | 2006-03-31 | 2006-08-30 | 洛阳轴研科技股份有限公司 | Automatic diagnosing system for rolling bearing fault |
| CN102269655A (en) * | 2010-06-04 | 2011-12-07 | 北京化工大学 | Method for diagnosing bearing fault |
Non-Patent Citations (2)
| Title |
|---|
| 张辉 等: "基于小波包变换的滚动轴承故障诊断方法的研究", 《振动与冲击》 * |
| 李智慧: "列车轴承故障监测系统研究", 《中国优秀博硕士学位论文全文数据库（硕士）信息科技辑》 * |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104568444A (en) * | 2015-01-28 | 2015-04-29 | 北京邮电大学 | Method for extracting fault characteristic frequencies of train rolling bearings with variable rotational speeds |
| CN105806604B (en) * | 2016-03-18 | 2018-10-19 | 北京唐智科技发展有限公司 | A kind of rolling stock EEF bogie bearing retainer failure prediction alarm method |
| CN105806604A (en) * | 2016-03-18 | 2016-07-27 | 唐智科技湖南发展有限公司 | Locomotive vehicle running gear bearing holder fault pre-alarm method |
| CN106441893A (en) * | 2016-09-22 | 2017-02-22 | 北京邮电大学 | Train rolling bearing fault and impurity vibration distinguishing method |
| CN106441893B (en) * | 2016-09-22 | 2018-08-10 | 北京邮电大学 | Train rolling bearing fault vibrates differentiating method with impurity |
| CN108225769A (en) * | 2016-12-15 | 2018-06-29 | 唐智科技湖南发展有限公司 | It is a kind of to prevent the Mechanism Diagnosis method that the mistaken diagnosis of retainer characteristic spectrum is fault spectrum |
| CN108225769B (en) * | 2016-12-15 | 2020-02-14 | 唐智科技湖南发展有限公司 | Mechanism diagnosis method for preventing holder characteristic spectrum from being misdiagnosed as fault spectrum |
| CN108204897A (en) * | 2016-12-16 | 2018-06-26 | 唐智科技湖南发展有限公司 | A kind of bearing parameter correction judgement and multi-parameter diagnose matched method automatically |
| CN107547108A (en) * | 2017-08-26 | 2018-01-05 | 南京天普机电产品制造有限公司 | Car axle report communicates comprehensive tester |
| CN107547108B (en) * | 2017-08-26 | 2020-11-03 | 南京天普机电产品制造有限公司 | Comprehensive tester for passenger car shaft newspaper communication |
| CN108363853A (en) * | 2018-01-31 | 2018-08-03 | 浙江浙大鸣泉科技有限公司 | A kind of engine speed measurement method based on multisensor correlation denoising |
| CN108363853B (en) * | 2018-01-31 | 2021-10-08 | 浙江浙大鸣泉科技有限公司 | Engine rotating speed measuring method based on multi-sensor correlation denoising |
| CN112013895A (en) * | 2019-05-28 | 2020-12-01 | 浙江德盛铁路器材股份有限公司 | Method for monitoring and evaluating quality condition of railway track basic equipment |
| CN112013895B (en) * | 2019-05-28 | 2023-01-24 | 浙江德盛铁路器材股份有限公司 | Method for monitoring and evaluating quality condition of railway track basic equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102607848A (en) | Detection method for train bearing fault | |
| Amini et al. | Onboard detection of railway axle bearing defects using envelope analysis of high frequency acoustic emission signals | |
| Wang et al. | Compound faults detection in gearbox via meshing resonance and spectral kurtosis methods | |
| Zhao et al. | Detection and recovery of fault impulses via improved harmonic product spectrum and its application in defect size estimation of train bearings | |
| Fan et al. | A hybrid approach for fault diagnosis of planetary bearings using an internal vibration sensor | |
| Elasha et al. | Planetary bearing defect detection in a commercial helicopter main gearbox with vibration and acoustic emission | |
| EP2414807B1 (en) | System and method for gearbox health monitoring | |
| Loutridis | Damage detection in gear systems using empirical mode decomposition | |
| Liang et al. | Railway wheel-flat and rail surface defect modelling and analysis by time–frequency techniques | |
| Amarnath et al. | Detection and diagnosis of surface wear failure in a spur geared system using EEMD based vibration signalanalysis | |
| Entezami et al. | Perspectives on railway axle bearing condition monitoring | |
| Parey et al. | Variable cosine windowing of intrinsic mode functions: Application to gear fault diagnosis | |
| CN104833534A (en) | Train running fault diagnosis device based on multi-source information fusion, and method | |
| Bourdon et al. | Reconstruction of angular speed variations in the angular domain to diagnose and quantify taper roller bearing outer race fault | |
| CN102441579B (en) | The on-line monitoring method of hot tandem rolling mill running status | |
| Fan et al. | Machine fault feature extraction based on intrinsic mode functions | |
| Figlus et al. | A method for detecting damage to rolling bearings in toothed gears of processing lines | |
| Zhang et al. | Adaptive filtering enhanced windowed correlated kurtosis for multiple faults diagnosis of locomotive bearings | |
| Moumene et al. | Application of the wavelets multiresolution analysis and the high-frequency resonance technique for gears and bearings faults diagnosis | |
| Loutridis | Gear failure prediction using multiscale local statistics | |
| Guo et al. | Experimental investigation on double-impulse phenomenon of hybrid ceramic ball bearing with outer race spall | |
| Vaičiūnas et al. | Rail vehicle axle-box bearing damage detection considering the intensity of heating alteration | |
| Feng et al. | A novel vibration indicator to monitor gear natural fatigue pitting propagation | |
| Liang et al. | Railway wheel flat and rail surface defect detection by time-frequency analysis | |
| Wang et al. | A new DFT-based dynamic detection framework for polygonal wear state of railway wheel |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C12 | Rejection of a patent application after its publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20120725 |