CN109238698A - A kind of motor bearings method for diagnosing faults based on current signal - Google Patents
A kind of motor bearings method for diagnosing faults based on current signal Download PDFInfo
- Publication number
- CN109238698A CN109238698A CN201811197326.2A CN201811197326A CN109238698A CN 109238698 A CN109238698 A CN 109238698A CN 201811197326 A CN201811197326 A CN 201811197326A CN 109238698 A CN109238698 A CN 109238698A
- Authority
- CN
- China
- Prior art keywords
- bearing
- signal
- frequency
- fault
- current
- 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
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01M—TESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
- G01M13/00—Testing of machine parts
- G01M13/04—Bearings
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R23/00—Arrangements for measuring frequencies; Arrangements for analysing frequency spectra
- G01R23/16—Spectrum analysis; Fourier analysis
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The invention discloses a kind of motor bearings method for diagnosing faults based on current signal obtains the current of electric historical signal under different bearing fault states in model training stage；HARMONIC SIGNAL EXTRACTION FROM is carried out to current history signal, fundamental frequency and harmonic wave are eliminated from primary current historical signal, obtains residual signals；To residual signals carry out when, frequency-domain analysis, extract bearing fault characteristics index；It is trained based on bearing fault characteristics index and in conjunction with bearing fault type, obtains bearing failure diagnosis model；In the fault diagnosis stage, the current of electric live signal for treating diagnosis carries out processing identical with model training stage and obtains bearing fault characteristics index, bearing fault characteristics index is inputted into trained bearing failure diagnosis model and carries out pattern-recognition, diagnoses bearing fault state.The present invention is able to solve existing motor bearings method for diagnosing faults and is difficult to effectively extract fault characteristic signals from the current signal of low signal-to-noise ratio, so as to cause accidentally, fail to report the technical issues of.
Description
Technical field
The present invention relates to fault detection technique fields, carry out traction motor bearings using current signal more particularly, to a kind of
The method of fault diagnosis.
Background technique
In electric power rail vehicles, traction electric machine is the most crucial component for realizing electric energy and mechanical energy conversion, is undertaken
The important task of power output in rolling stock operation, operational efficiency and safety for rail vehicles have important shadow
It rings.Operative practice shows that motor bearings failure is the most common and most dangerous failure of traction electric machine, the generation of these failures and hair
Exhibition, does not only result in motor damage, and may cause other equipment damages, to cause very big loss.How to traction electricity
Machine bearing failure carries out timely, effective status monitoring and fault diagnosis, causes to avoid serious accident and unnecessary shutdown
Economic loss be solve traction electric machine State Maintenance key technical problem.
Currently, being traction motor bearings fault diagnosis field based on analysis of vibration signal method and stator current signal analytic approach
The two methods being most widely used.When using analysis of vibration signal method, need additionally to install the equipment such as vibrating sensor additional, in this way
Cost is not only increased, and brings new security risk.And stator current signal analysis method is then a kind of non-intrusion type
Method for diagnosing faults does not need additionally to increase sensor, have many advantages, such as low at cheap, easy to implement.However, by pulse
Width modulated (PWM) power supply and operating condition influence complicated and changeable contain motor abundant in traction motor current signal
Running state information, and bearing fault characteristics signal relative weak, are often submerged in the interference signals such as current harmonics.Traditional
Current of electric signature analysis (MCSA) algorithm, such as: Fourier transformation (FFT), PARK Vector Mode Analysis scheduling algorithm are difficult to effectively
Ground extracts fault characteristic signals from the current signal of low signal-to-noise ratio, the occurrence of so as to cause reporting by mistake or failing to report.
In the prior art, mainly have following technical scheme related to the present patent application:
The prior art 1 is applied for University of Anhui on December 23rd, 2016, and, publication number open on May 31st, 2017
For a kind of Chinese invention application " the permanent magnet synchronous electric machine bearing on-line fault diagnosis dress under variable speed operating condition of CN106769041A
Set and method ".The patent application discloses the permanent magnet synchronous electric machine bearing on-line fault diagnosis device under a kind of variable speed operating condition
And method.Analog-digital converter 1 connects current probe, and constant duration acquires current signal, and microcontroller 1 carries out current signal
Low-pass filtering and dipole inversion.Calculate conversion after single polarity current signal angle and be rounded, angle it is every variation 1 degree when, micro-control
Device 1 processed generates a trigger signal.Microcontroller 2 receives the trigger signal that microcontroller 1 generates, and control analog-digital converter 2 is right
Microphone carries out triggering sampling, obtains bearing voice signal.Envelope demodulation is carried out to the bearing signal of angular domain, calculates envelope signal
Order spectrum, bearing fault type is judged according to fault signature order, and show on a display screen.The invention passes through to bearing sound
Sound signal is analyzed and processed, and judges bearing fault type according to fault signature order, needs additionally to increase sensor, modulus turns
The components such as parallel operation are unfavorable for the reduction of cost.Meanwhile using bearing voice signal judge bearing fault type be difficult to effectively from
Fault characteristic signals are extracted in the current signal of low signal-to-noise ratio.
The prior art 2 was Nanjing Aero-Space University in application on 03 28th, 2017, and on 07 28th, 2017 public affairs
It opens, Chinese invention application " the permanent magnetic motor bearing spot corrosion based on stator current wavelet packet analysis of Publication No. CN106989923A
Fault detection method ".The patent application discloses a kind of permanent magnetic motor bearing pitting fault based on stator current wavelet packet analysis
Detection method.Fault diagnosis is carried out using the stator current of permanent magnetic motor, letter is carried out to stator current by analysis method of wavelet packet
Number analysis, wavelet packet analysis is a kind of multiresolution algorithm, can be removed by frequency range to signal.When determining electrical fault
After frequency, corresponding small echo packet node is calculated, root mean square is asked to small echo packet node coefficient, failure feelings are judged according to root mean square
Condition.The invention carries out signal analysis to stator current using analysis method of wavelet packet, according to the root mean square of small echo packet node coefficient
Judge fault condition, there is also the technologies for being difficult to effectively extract fault characteristic signals from the current signal of low signal-to-noise ratio to ask
Topic.
The micro- Zhejiang University of the prior art 3 applied on 01 20th, 2017, and, publication number open on May 17th, 2017
For a kind of Chinese invention application " side of the magneto bearing failure diagnosis based on position-sensor-free of CN106680716A
Method ".The patent application discloses a kind of method of magneto bearing failure diagnosis based on position-sensor-free.Firstly, from
Voltage and electric current are obtained in motor control chip, and are obtained the rotor position angle of motor using position-sensor-free algorithm and turned
Speed.Then, the alternating component in revolving speed is extracted by moving average filter, and angular domain weight is carried out to it according to position angle
Sampling.Finally, counterweight sample rate signal carries out frequency-domain analysis, judge whether bearing fault occurs according to frequency domain information.The hair
The bright rotor position angle and revolving speed for needing to obtain motor by position-sensor-free algorithm according to voltage and electric current, according to position angle
Degree carries out angular domain resampling to it, carries out angular domain resampling to it further according to position angle.Not only algorithm is complicated, and there is also
It is difficult to the technical issues of effectively extracting fault characteristic signals from the current signal of low signal-to-noise ratio.
Summary of the invention
In view of this, the purpose of the present invention is to provide a kind of motor bearings method for diagnosing faults based on current signal,
It is difficult to effectively extract fault signature from the current signal of low signal-to-noise ratio to solve existing motor bearings method for diagnosing faults
Signal, the technical issues of so as to cause reporting by mistake or failing to report.
In order to achieve the above-mentioned object of the invention, the present invention specifically provides a kind of motor bearings failure based on current signal and examines
The technic relization scheme of disconnected method, the motor bearings method for diagnosing faults based on current signal, comprising the following steps:
A) model training stage
S101 the multiple groups current of electric historical signal under different bearing fault states, the training sample as model) are obtained
Notebook data collection；
S102 HARMONIC SIGNAL EXTRACTION FROM) is carried out to training sample data collection, is eliminated from original current of electric historical signal
Fundamental frequency and harmonic signal obtain residual signals；
S103 time and frequency domain analysis) is carried out to the residual signals that step S102) is obtained, the fault signature for extracting bearing refers to
Mark；
S104 fault signature index and combination bearing fault type) based on the step S103) bearing extracted are instructed
Practice, obtains bearing failure diagnosis model.
B) the fault diagnosis stage
S201 the current of electric live signal for) treating diagnosis carries out HARMONIC SIGNAL EXTRACTION FROM, real-time from original current of electric
Fundamental frequency and harmonic signal are eliminated in signal, obtain residual signals；
S202 time and frequency domain analysis) is carried out to the residual signals that step S201) is obtained, the fault signature for extracting bearing refers to
Mark；
S203 the fault signature index of the step S202) bearing extracted) is input to step S104) trained bearing
Fault diagnosis model carries out pattern-recognition, is diagnosed to be the malfunction of the bearing.
Further, the step S102) further comprise following procedure:
The fundamental frequency of current of electric historical signal and the amplitude of harmonic wave, initial are obtained from original current of electric historical signal
Phase and frequency, and then obtain fundamental frequency and harmonic signal；Fundamental frequency and harmonic wave letter are subtracted from original current of electric historical signal
Number, to obtain residual signals.
Further, the step S103) further comprise following procedure:
S1031 bearing fault characteristics frequency) is calculated first, then calculates bearing fault characteristics frequency sequence；
S1032 the frequency spectrum that FFT transform obtains residual signals) is carried out to residual signals, is obtained further according to the frequency spectrum of residual signals
Obtain corresponding bearing fault characteristics frequency sequence amplitude；
S1033 the minimum value I_min and virtual value I_rms of residual signals) are calculated, and according to bearing fault characteristics frequency sequence
Column amplitude I_f (n) calculates maximum value I_f_max, peak-to-peak value I_f_peark and the peak of bearing fault characteristics frequency sequence amplitude
Value coefficient I_f_crest.
Further, the bearing fault characteristics frequency sequence f (k) is | fe ± kfv |, k=1,2,3, wherein fe is residual
The electric current fundamental frequency of difference signal, fv are bearing fault characteristics frequency.
Further, the step S1033) in,
The minimum value I_min of the residual signals is calculated according to the following formula:
I_min=min (I (t))
In formula, I (t) is the time series of residual signals, whereinN is the sampling number of signal.
The virtual value I_rms of the residual signals is calculated according to the following formula:
The maximum value of the bearing fault characteristics frequency sequence amplitude calculates according to the following formula:
I_f_max=max (I_f (n))
In formula, I_f (n) is bearing fault characteristics frequency sequence amplitude.
The peak-to-peak value of the bearing fault characteristics frequency sequence amplitude calculates according to the following formula:
I_f_peark=I_f_max-I_min
The peak factor of the bearing fault characteristics frequency sequence amplitude calculates according to the following formula:
I_f_crest=I_f_max/I_rms.
Further, the malfunction type of the bearing includes bearing retainer failure, bearing outer ring failure, in bearing
Enclose failure and bearing roller failure.In the step S1031) in,
The bearing retainer fault characteristic frequency calculates according to the following formula:
The bearing outer ring fault characteristic frequency calculates according to the following formula:
The bearing inner race fault characteristic frequency calculates according to the following formula:
The bearing roller fault characteristic frequency calculates according to the following formula:
Wherein, fr is motor rotation frequency, DBFor the diameter of rolling element in bearing, DPFor the circular diameter of bearing section, NBFor axis
The number of middle rolling element is held, θ is contact angle.
Further, the step S201) further comprise following procedure:
The fundamental frequency of current of electric live signal and the amplitude of harmonic wave, just are obtained from current of electric live signal to be diagnosed
Beginning phase and frequency, and then obtain fundamental frequency and harmonic signal.Fundamental frequency and harmonic wave are subtracted from original current of electric live signal
Signal, to obtain residual signals.
Further, the step S202) further comprise following procedure:
S2021 bearing fault characteristics frequency) is calculated first, then calculates bearing fault characteristics frequency sequence；
S2022 the frequency spectrum that FFT transform obtains residual signals) is carried out to residual signals, is obtained further according to the frequency spectrum of residual signals
Obtain corresponding bearing fault characteristics frequency sequence amplitude；
S2023 the minimum value I_min and virtual value I_rms of residual signals) are calculated, and according to bearing fault characteristics frequency sequence
Column amplitude I_f (n) calculates maximum value I_f_max, peak-to-peak value I_f_peark and the peak of bearing fault characteristics frequency sequence amplitude
Value coefficient I_f_crest.
Further, in the step S102) in, current of electric history is obtained using full phase time shift phase difference correction method
The fundamental frequency of signal and amplitude, initial phase and the frequency of harmonic wave, and then obtain fundamental frequency and harmonic signal.In the step S201)
In, the fundamental frequency of current of electric live signal and amplitude, the initial phase of harmonic wave are obtained using full phase time shift phase difference correction method
And frequency, and then obtain fundamental frequency and harmonic signal.
Further, in the step 104), by the fault signature index and bearing fault type of the bearing extracted
It is input in gradient promotion Tree Classifier and is trained, obtain bearing failure diagnosis model.In the step 203), it will extract
To the fault signature index of bearing be input in the trained bearing failure diagnosis model based on gradient boosted tree and carry out mould
Formula identification, is diagnosed to be the malfunction of bearing.
By implementing the technical solution for the motor bearings method for diagnosing faults based on current signal that aforementioned present invention provides,
It has the following beneficial effects:
(1) present invention carries out HARMONIC SIGNAL EXTRACTION FROM using to motor current signal, from original current of electric historical signal
Middle elimination fundamental frequency and harmonic signal obtain residual signals, effectively reduce by PWM (PulseWidthModulation, pulsewidth tune
The abbreviation of system) interference of the harmonic signal introduced to bearing fault characteristics signal of powering, the signal-to-noise ratio of current signal is improved, into
And the fault signature index extracted is more significant；
(2) present invention reflects that traction motor bearings are strong using the variation of the bearing fault characteristics index of residual signals extraction
The variation of health state, which can accurately characterize bearing fault state, using full phase time shift phase difference school
It executes extraction and removes the harmonic signal in current signal, further improve the conspicuousness of fault signature index；
(3) gradient boosted tree fault diagnosis model established by the present invention can automatic identification bearing fault state, without
It wants business expert to observe spectrogram to diagnose, can automatically identify bearing fault according to the self-characteristic of motor current signal
Whether, the interference of artificial subjective factor is effectively reduced, and improve the objectivity and accuracy of fault diagnosis.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described.It should be evident that the accompanying drawings in the following description is only this
Some embodiments of invention for those of ordinary skill in the art without creative efforts, can be with
Other embodiments are obtained according to these attached drawings.
Fig. 1 is a kind of process principle of specific embodiment of motor bearings method for diagnosing faults the present invention is based on current signal
Schematic diagram；
Fig. 2 is the model instruction in a kind of specific embodiment of motor bearings method for diagnosing faults the present invention is based on current signal
Practice the program flow diagram in stage；
Fig. 3 is that the failure in a kind of specific embodiment of motor bearings method for diagnosing faults the present invention is based on current signal is examined
The program flow diagram of faulted-stage section；
Fig. 4 is working principle of the motor bearings trouble-shooter that is based on of the method for the present invention under fault diagnosis model
Block diagram；
Fig. 5 is working principle of the motor bearings trouble-shooter that is based on of the method for the present invention under model training mode
Block diagram；
In figure: 1- current signal acquiring unit, 2- residual signals computing unit, 3- fault signature extraction unit, 4- failure
Diagnostic model unit.
Specific embodiment
For the sake of quoting and understanding, will hereafter used in technical term, write a Chinese character in simplified form or abridge and be described below:
Full phase time shift phase difference correction method: being a kind of sinusoidal signal algorithm of interference eliminated in signal, for calculating base
Amplitude, frequency and the initial phase of frequency and harmonic signal；
Gradient boosted tree: being the member of a kind of machine learning algorithm and integrated study Boosting family, before being utilization
One takes turns the error rate of the weak learner of iteration to update the weight of training set.
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
In attached drawing, the technical scheme in the embodiment of the invention is clearly and completely described.Obviously, described embodiment is only
It is only a part of the embodiments of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, the common skill in this field
Art personnel all other embodiment obtained without creative efforts belongs to the model that the present invention protects
It encloses.
As shown in attached drawing 1 to attached drawing 5, the motor bearings method for diagnosing faults the present invention is based on current signal is given, and
The specific embodiment of device based on this method, the present invention will be further explained below with reference to the attached drawings and specific examples.
Embodiment 1
As shown in attached drawing 1 to attached drawing 3, a kind of embodiment of the Method for Bearing Fault Diagnosis based on current signal is specific to wrap
Include following steps:
A) model training stage
S101 the multiple groups current of electric under different bearing fault states (corresponding to different bearing fault types)) is obtained
Historical signal, the training sample data collection as model；
S102 HARMONIC SIGNAL EXTRACTION FROM) is carried out to training sample data collection, is eliminated from original current of electric historical signal
Fundamental frequency and harmonic signal obtain residual signals；
S103 time and frequency domain analysis) is carried out to the residual signals that step S102) is obtained, the fault signature for extracting bearing refers to
Mark；
S104 fault signature index (vector) and combination bearing fault type) based on the step S103) bearing extracted
It is trained, obtains bearing failure diagnosis model.As a kind of preferable specific embodiment of the present invention, by the bearing extracted
Fault signature index (vector) and bearing fault type be input to gradient promoted Tree Classifier (algorithm be it is in the prior art
Have algorithm, details are not described herein) in be trained, obtain bearing failure diagnosis model.
B) the fault diagnosis stage
S201 the current of electric live signal for) treating diagnosis carries out HARMONIC SIGNAL EXTRACTION FROM, real-time from original current of electric
Fundamental frequency and harmonic signal are eliminated in signal, obtain residual signals；
S202 time and frequency domain analysis) is carried out to the residual signals that step S201) is obtained, the fault signature for extracting bearing refers to
Mark；
S203 the fault signature index (vector) of the step S202) bearing extracted) is input to step S104) it trains
Bearing failure diagnosis model carry out pattern-recognition, be diagnosed to be the malfunction of the bearing.As a kind of preferably tool of the invention
The fault signature index (vector) of the bearing extracted is input to trained based on the (calculation of gradient boosted tree by body embodiment
Method be existing algorithm in the prior art, details are not described herein) bearing failure diagnosis model in carry out pattern-recognition, be diagnosed to be
The malfunction (i.e. bearing fault type) of bearing.
Step S102) further comprise following procedure:
Use full phase time shift phase difference correction method (for the algorithm for existing algorithm in the prior art, details are not described herein)
Obtained from original current of electric historical signal current of electric historical signal fundamental frequency fe and harmonic wave n*fe, n=2,3,
4 ..., 50 amplitude, initial phase and frequency, and then obtain fundamental frequency and harmonic signal.From original current of electric historical signal
In subtract fundamental frequency and harmonic signal, to obtain residual signals.
Step S103) further comprise following procedure:
S1031 bearing fault characteristics frequency) is calculated first, then calculates bearing fault characteristics frequency sequence；
Bearing fault characteristics frequency sequence f (k) is | fe ± kfv |, k=1,2,3 ..., n, wherein and and n is 3~5 times desirable,
Fe is the electric current fundamental frequency of residual signals, and fv is bearing fault characteristics frequency；
S1032 the frequency spectrum that FFT transform obtains residual signals) is carried out to residual signals, is obtained further according to the frequency spectrum of residual signals
It obtains corresponding bearing fault characteristics frequency sequence amplitude and (finds amplitude of the residual signals at each Frequency point f (k), frequency
Point f (k) is identical as a certain frequency in residual signals frequency spectrum)；
S1033 the minimum value I_min and virtual value I_rms of residual signals) are calculated, and according to bearing fault characteristics frequency sequence
Column amplitude I_f (n) calculates maximum value I_f_max, peak-to-peak value I_f_peark and the peak of bearing fault characteristics frequency sequence amplitude
Value coefficient I_f_crest.
In step S104), under the different bearing fault states extracted for step S103) includes that bearing fault is special
Levy frequency sequence amplitude I_f (n), the maximum value I_f_max of bearing fault characteristics frequency sequence amplitude, peak-to-peak value I_f_peark
It is special with the failure of the bearing including the minimum value I_min and virtual value I_rms of peak factor I_f_crest and residual signals
It levies index (vector), and is trained in conjunction with bearing fault type, obtain bearing failure diagnosis model.
In step S1033),
The minimum value I_min of residual signals is further calculated according to the following formula:
I_min=min (I (t))
In formula, I (t) is the time series of residual signals, whereinN is the sampling number of signal.
The virtual value I_rms of residual signals is further calculated according to the following formula:
The maximum value of bearing fault characteristics frequency sequence amplitude further calculates according to the following formula:
I_f_max=max (I_f (n))
In formula, I_f (n) is bearing fault characteristics frequency sequence amplitude；
The peak-to-peak value of bearing fault characteristics frequency sequence amplitude further calculates according to the following formula:
I_f_peark=I_f_max-I_min
The peak factor of bearing fault characteristics frequency sequence amplitude further calculates according to the following formula:
I_f_crest=I_f_max/I_rms
The malfunction type of bearing further comprises bearing retainer failure, bearing outer ring failure, bearing inner race failure
With bearing roller failure.In step S1031),
Bearing retainer fault characteristic frequency further calculates according to the following formula:
Bearing outer ring fault characteristic frequency further calculates according to the following formula:
Bearing inner race fault characteristic frequency further calculates according to the following formula:
Bearing roller fault characteristic frequency further calculates according to the following formula:
Wherein, fr is motor rotation frequency, DBFor the diameter of rolling element in bearing, DPFor the circular diameter of bearing section, NBFor axis
The number of middle rolling element is held, θ is contact angle.
Step S201) further comprise following procedure:
Use full phase time shift phase difference correction method (for the algorithm for existing algorithm in the prior art, details are not described herein)
Obtained from current of electric live signal to be diagnosed current of electric live signal fundamental frequency fe and harmonic wave n*fe, n=2,3,
4 ..., 50 amplitude, initial phase and frequency, and then obtain fundamental frequency and harmonic signal.From original current of electric live signal
In subtract fundamental frequency and harmonic signal, to obtain residual signals.
Step S202) further comprise following procedure:
S2021 bearing fault characteristics frequency) is calculated first, then calculates bearing fault characteristics frequency sequence；
S2022 FFT transform (FastFourierTransformation, Fast Fourier Transform (FFT)) are carried out to residual signals
Abbreviation) frequency spectrum that obtains residual signals, obtain corresponding bearing fault characteristics frequency sequence further according to the frequency spectrum of residual signals
Amplitude；
S2023 the minimum value I_min and virtual value I_rms of residual signals) are calculated, and according to bearing fault characteristics frequency sequence
Column amplitude I_f (n) calculates maximum value I_f_max, peak-to-peak value I_f_peark and the peak of bearing fault characteristics frequency sequence amplitude
Value coefficient I_f_crest.
In step S203), by step S202) extract include bearing fault characteristics frequency sequence amplitude I_f (n),
Maximum value I_f_max, peak-to-peak value I_f_peark and the peak factor I_f_crest of bearing fault characteristics frequency sequence amplitude, with
And the minimum value I_min of residual signals and the fault signature index (vector) of the bearing including virtual value I_rms are input to training
Good bearing failure diagnosis model carries out pattern-recognition, is diagnosed to be the malfunction of the bearing.
In the technical solution for the motor bearings method for diagnosing faults based on current signal that the embodiment of the present invention 1 describes, adopt
Carry out bearing failure diagnosis model training and pattern-recognition with gradient boosted tree intelligent algorithm, can also using neural network, with
The intelligent algorithms alternate process such as machine forest and support vector machines.The motor bearings based on current signal that embodiment 1 describes
Method for diagnosing faults can be advantageously applied to offline or online health monitoring and the fault pre-alarming of locomotive traction motor bearing fault
In.
Embodiment 2
As shown in Fig. 4, a kind of embodiment of the bearing fault diagnosing apparatus based on 1 the method for embodiment is specific to wrap
It includes: current signal acquiring unit 1, residual signals acquiring unit 2, fault signature extraction unit 3 and fault diagnosis model unit 4.
When device is in fault diagnosis state:
Current signal acquiring unit 1 obtains current of electric live signal；
Residual signals acquiring unit 2 carries out harmonic wave letter to the current of electric live signal that current signal acquiring unit 1 obtains
It number extracts, fundamental frequency and harmonic signal is eliminated from original current of electric live signal, obtains residual signals；
Fault signature extraction unit 3 carries out time domain and frequency domain point to the residual signals that residual signals acquiring unit 2 obtains
Analysis, extracts the fault signature index of bearing；
Fault diagnosis model unit 4 extracts fault signature extraction unit 3 using trained bearing failure diagnosis model
The fault signature index (vector) of the bearing arrived carries out pattern-recognition, is diagnosed to be the malfunction of the bearing.
As shown in Fig. 5, when device is in fault model physical training condition:
Current signal acquiring unit 1 obtains the multiple groups current of electric historical signal under different bearing fault states, as
The training sample data collection of model；
Residual signals acquiring unit 2 carries out harmonic signal to the training sample data collection that current signal acquiring unit 1 obtains
It extracts, fundamental frequency and harmonic signal is eliminated from original current of electric historical signal, obtain residual signals；
Fault signature extraction unit 3 carries out time domain and frequency domain point to the residual signals that residual signals acquiring unit 2 obtains
Analysis, extracts the fault signature index of bearing；
Fault diagnosis model unit 4, the bearing extracted based on fault signature extraction unit 3 fault signature index (to
Amount) and be trained in conjunction with bearing fault type, obtain bearing failure diagnosis model.
It can be referring in particular to the associated description of embodiment 1, herein no longer about the more detailed technical solution of rest part
It repeats.
By the skill for implementing the motor bearings method for diagnosing faults based on current signal of specific embodiment of the invention description
Art scheme can have the following technical effects:
(1) the motor bearings method for diagnosing faults based on current signal of specific embodiment of the invention description, using to electricity
Machine current signal carries out HARMONIC SIGNAL EXTRACTION FROM, and fundamental frequency and harmonic signal are eliminated from original current of electric historical signal, obtains
Residual signals effectively reduce interference of the harmonic signal introduced by PWM power supply to bearing fault characteristics signal, improve electric current
The signal-to-noise ratio of signal, and then the fault signature index extracted is more significant；
(2) the motor bearings method for diagnosing faults based on current signal of specific embodiment of the invention description, utilizes residual error
The variation of the bearing fault characteristics index of signal extraction changed to reflect traction motor bearings health status, the fault signature refer to
Mark can accurately characterize bearing fault state, extracted and removed in current signal using full phase time shift phase difference correction method
Harmonic signal further improves the conspicuousness of fault signature index；
(3) the motor bearings method for diagnosing faults based on current signal of specific embodiment of the invention description, is established
Gradient boosted tree fault diagnosis model can automatic identification bearing fault state, without business expert observe spectrogram examine
It is disconnected, whether capable of automatically identifying bearing fault according to the self-characteristic of motor current signal, effectively reduce it is artificial it is subjective because
The interference of element, and improve the objectivity and accuracy of fault diagnosis.
Each embodiment in this specification is described in a progressive manner, the highlights of each of the examples are with other
The difference of embodiment, the same or similar parts in each embodiment may refer to each other.
The above described is only a preferred embodiment of the present invention, being not intended to limit the present invention in any form.Though
So the present invention is disclosed as above with preferred embodiment, and however, it is not intended to limit the invention.It is any to be familiar with those skilled in the art
Member, in the case where not departing from Spirit Essence of the invention and technical solution, all using in the methods and techniques of the disclosure above
Appearance makes many possible changes and modifications or equivalent example modified to equivalent change to technical solution of the present invention.Therefore,
Anything that does not depart from the technical scheme of the invention are made to the above embodiment any simple according to the technical essence of the invention
Modification, equivalent replacement, equivalence changes and modification still fall within the range of technical solution of the present invention protection.
Claims (10)
1. a kind of Method for Bearing Fault Diagnosis based on current signal, which comprises the following steps:
A) model training stage
S101 the multiple groups current of electric historical signal under different bearing fault states, the number of training as model) are obtained
According to collection；
S102 HARMONIC SIGNAL EXTRACTION FROM) is carried out to training sample data collection, eliminates fundamental frequency from original current of electric historical signal
And harmonic signal, obtain residual signals；
S103 time and frequency domain analysis) is carried out to the residual signals that step S102) is obtained, extracts the fault signature index of bearing；
S104 fault signature index and combination bearing fault type) based on the step S103) bearing extracted are trained, and are obtained
To bearing failure diagnosis model；
B) the fault diagnosis stage
S201 the current of electric live signal for) treating diagnosis carries out HARMONIC SIGNAL EXTRACTION FROM, from original current of electric live signal
Middle elimination fundamental frequency and harmonic signal obtain residual signals；
S202 time and frequency domain analysis) is carried out to the residual signals that step S201) is obtained, extracts the fault signature index of bearing；
S203 the fault signature index of the step S202) bearing extracted) is input to step S104) trained bearing fault
Diagnostic model carries out pattern-recognition, is diagnosed to be the malfunction of the bearing.
2. the Method for Bearing Fault Diagnosis according to claim 1 based on current signal, which is characterized in that the step
S102) further comprise following procedure:
The fundamental frequency of current of electric historical signal and the amplitude of harmonic wave, initial phase are obtained from original current of electric historical signal
And frequency, and then obtain fundamental frequency and harmonic signal；Fundamental frequency and harmonic signal are subtracted from original current of electric historical signal, from
And obtain residual signals.
3. the Method for Bearing Fault Diagnosis according to claim 1 or 2 based on current signal, which is characterized in that the step
Rapid S103) further comprise following procedure:
S1031 bearing fault characteristics frequency) is calculated first, then calculates bearing fault characteristics frequency sequence；
S1032 the frequency spectrum that FFT transform obtains residual signals) is carried out to residual signals, further according to the frequency spectrum acquisition pair of residual signals
The bearing fault characteristics frequency sequence amplitude answered；
S1033 the minimum value (I_min) and virtual value (I_rms) of residual signals) are calculated, and according to bearing fault characteristics frequency sequence
Column amplitude (I_f (n)) calculates maximum value (I_f_max), the peak-to-peak value (I_f_ of bearing fault characteristics frequency sequence amplitude
) and peak factor (I_f_crest) peark.
4. the Method for Bearing Fault Diagnosis according to claim 3 based on current signal, it is characterised in that: the bearing event
Hindering characteristic frequency sequence f (k) is | fe ± kfv |, k=1,2,3, wherein fe is the electric current fundamental frequency of residual signals, and fv is axis
Hold fault characteristic frequency.
5. the Method for Bearing Fault Diagnosis according to claim 4 based on current signal, which is characterized in that the step
S1033 in),
The minimum value (I_min) of the residual signals calculates according to the following formula:
I_min=min (I (t))
In formula, I (t) is the time series of residual signals, whereinN is the sampling number of signal；
The virtual value (I_rms) of the residual signals calculates according to the following formula:
The maximum value of the bearing fault characteristics frequency sequence amplitude calculates according to the following formula:
I_f_max=max (I_f (n))
In formula, I_f (n) is bearing fault characteristics frequency sequence amplitude；
The peak-to-peak value of the bearing fault characteristics frequency sequence amplitude calculates according to the following formula:
I_f_peark=I_f_max-I_min
The peak factor of the bearing fault characteristics frequency sequence amplitude calculates according to the following formula:
I_f_crest=I_f_max/I_rms.
6. the Method for Bearing Fault Diagnosis according to claim 4 or 5 based on current signal, which is characterized in that the axis
The malfunction type held includes bearing retainer failure, bearing outer ring failure, bearing inner race failure and bearing roller failure；
In the step S1031) in,
The bearing retainer fault characteristic frequency calculates according to the following formula:
The bearing outer ring fault characteristic frequency calculates according to the following formula:
The bearing inner race fault characteristic frequency calculates according to the following formula:
The bearing roller fault characteristic frequency calculates according to the following formula:
Wherein, fr is motor rotation frequency, DBFor the diameter of rolling element in bearing, DPFor the circular diameter of bearing section, NBFor in bearing
The number of rolling element, θ are contact angle.
7. according to claim 1, the described in any item Method for Bearing Fault Diagnosis based on current signal in 2,4 or 5, feature exist
In the step S201) further comprise following procedure:
The fundamental frequency of current of electric live signal and the amplitude of harmonic wave, initial phase are obtained from current of electric live signal to be diagnosed
Position and frequency, and then obtain fundamental frequency and harmonic signal；Fundamental frequency and harmonic signal are subtracted from original current of electric live signal,
To obtain residual signals.
8. the Method for Bearing Fault Diagnosis according to claim 7 based on current signal, which is characterized in that the step
S202) further comprise following procedure:
S2021 bearing fault characteristics frequency) is calculated first, then calculates bearing fault characteristics frequency sequence；
S2022 the frequency spectrum that FFT transform obtains residual signals) is carried out to residual signals, further according to the frequency spectrum acquisition pair of residual signals
The bearing fault characteristics frequency sequence amplitude answered；
S2023 the minimum value (I_min) and virtual value (I_rms) of residual signals) are calculated, and according to bearing fault characteristics frequency sequence
Column amplitude (I_f (n)) calculates maximum value (I_f_max), the peak-to-peak value (I_f_ of bearing fault characteristics frequency sequence amplitude
) and peak factor (I_f_crest) peark.
9. according to claim 1, the described in any item Method for Bearing Fault Diagnosis based on current signal in 2,4,5 or 8, feature
It is:
In the step S102) in, using full phase time shift phase difference correction method obtain current of electric historical signal fundamental frequency and
Amplitude, initial phase and the frequency of harmonic wave, and then obtain fundamental frequency and harmonic signal；
In the step S201) in, using full phase time shift phase difference correction method obtain current of electric live signal fundamental frequency and
Amplitude, initial phase and the frequency of harmonic wave, and then obtain fundamental frequency and harmonic signal.
10. according to claim 1, the described in any item Method for Bearing Fault Diagnosis based on current signal in 2,4,5 or 8, special
Sign is:
In the step 104), the fault signature index of the bearing extracted and bearing fault type are input to gradient and promoted
It is trained in Tree Classifier, obtains bearing failure diagnosis model；
In the step 203), the fault signature index of the bearing extracted is input to trained based on gradient boosted tree
Bearing failure diagnosis model in carry out pattern-recognition, be diagnosed to be the malfunction of bearing.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811197326.2A CN109238698A (en) | 2018-10-15 | 2018-10-15 | A kind of motor bearings method for diagnosing faults based on current signal |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811197326.2A CN109238698A (en) | 2018-10-15 | 2018-10-15 | A kind of motor bearings method for diagnosing faults based on current signal |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109238698A true CN109238698A (en) | 2019-01-18 |
Family
ID=65052786
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811197326.2A Pending CN109238698A (en) | 2018-10-15 | 2018-10-15 | A kind of motor bearings method for diagnosing faults based on current signal |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109238698A (en) |
Cited By (19)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110197194A (en) * | 2019-04-12 | 2019-09-03 | 佛山科学技术学院 | A kind of Method for Bearing Fault Diagnosis and device based on improvement random forest |
| CN110907826A (en) * | 2019-11-14 | 2020-03-24 | 中车株洲电力机车研究所有限公司 | Motor fault diagnosis method and system based on convolutional neural network filtering |
| WO2020078132A1 (en) * | 2018-10-15 | 2020-04-23 | 株洲中车时代电气股份有限公司 | Motor bearing failure diagnosis device |
| CN111157092A (en) * | 2020-01-02 | 2020-05-15 | 深圳市汉德网络科技有限公司 | Vehicle-mounted weighing automatic calibration method and computer readable storage medium |
| CN111208427A (en) * | 2020-04-23 | 2020-05-29 | 湖南中车时代通信信号有限公司 | Traction motor fault diagnosis method and device |
| CN111879514A (en) * | 2020-07-31 | 2020-11-03 | 南京机电职业技术学院 | Brushless direct current motor bearing fault diagnosis method based on ELM model |
| CN112034340A (en) * | 2019-06-03 | 2020-12-04 | 中国人民解放军63756部队 | Method for screening fault characteristics of measurement and control antenna motor |
| CN112052796A (en) * | 2020-09-07 | 2020-12-08 | 电子科技大学 | Permanent magnet synchronous motor fault diagnosis method based on deep learning |
| CN112287775A (en) * | 2020-10-14 | 2021-01-29 | 中车株洲电力机车研究所有限公司 | Motor fault diagnosis method and device and related components |
| CN112504678A (en) * | 2020-11-12 | 2021-03-16 | 重庆科技学院 | Motor bearing fault diagnosis method based on knowledge distillation |
| CN113189483A (en) * | 2021-04-25 | 2021-07-30 | 西安交通大学 | Frequency conversion spectrum peak and current range joint inference axial flow fan fault diagnosis method |
| CN113283292A (en) * | 2021-04-13 | 2021-08-20 | 江苏大学镇江流体工程装备技术研究院 | Method and device for fault diagnosis of underwater micro propeller |
| CN113295413A (en) * | 2021-06-24 | 2021-08-24 | 北京交通大学 | Traction motor bearing fault diagnosis method based on indirect signals |
| CN113557420A (en) * | 2019-03-14 | 2021-10-26 | 西门子股份公司 | Expert knowledge based automatic health status classifier for rotating machines |
| CN113639999A (en) * | 2021-08-16 | 2021-11-12 | 北京信息科技大学 | Fault diagnosis method, system, medium and equipment for rolling bearing under variable rotating speed |
| CN114112396A (en) * | 2021-10-29 | 2022-03-01 | 江苏大学 | Bearing fault diagnosis method under quasi-steady working condition of rotating speed fluctuation by adopting current signal analysis |
| CN114152437A (en) * | 2020-08-17 | 2022-03-08 | Abb瑞士股份有限公司 | Method for detecting bearing failure |
| CN114295981A (en) * | 2022-01-05 | 2022-04-08 | 湖南师范大学 | Motor fault diagnosis method based on RN network current multi-feature form recognition |
| CN116106742A (en) * | 2023-04-07 | 2023-05-12 | 国家石油天然气管网集团有限公司 | Motor bearing fault diagnosis method and system based on DQ transformation |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101034038A (en) * | 2007-03-28 | 2007-09-12 | 华北电力大学 | Failure testing method of asynchronous motor bearing |
| CN103033745A (en) * | 2011-08-29 | 2013-04-10 | 通用电气公司 | Fault detection based on current signature analysis for a generator |
| CN106769041A (en) * | 2016-12-23 | 2017-05-31 | 安徽大学 | Permanent magnet synchronous motor bearing online fault diagnosis device and method under variable rotating speed working condition |
| CN106989923A (en) * | 2017-03-28 | 2017-07-28 | 南京航空航天大学 | Permanent magnetic motor bearing spot corrosion fault detection method based on stator current wavelet packet analysis |
-
2018
- 2018-10-15 CN CN201811197326.2A patent/CN109238698A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101034038A (en) * | 2007-03-28 | 2007-09-12 | 华北电力大学 | Failure testing method of asynchronous motor bearing |
| CN103033745A (en) * | 2011-08-29 | 2013-04-10 | 通用电气公司 | Fault detection based on current signature analysis for a generator |
| CN106769041A (en) * | 2016-12-23 | 2017-05-31 | 安徽大学 | Permanent magnet synchronous motor bearing online fault diagnosis device and method under variable rotating speed working condition |
| CN106989923A (en) * | 2017-03-28 | 2017-07-28 | 南京航空航天大学 | Permanent magnetic motor bearing spot corrosion fault detection method based on stator current wavelet packet analysis |
Non-Patent Citations (1)
| Title |
|---|
| 张海刚: ""基于定子电流监控的轴承故障在线监测"", 《控制理论与应用》 * |
Cited By (25)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020078132A1 (en) * | 2018-10-15 | 2020-04-23 | 株洲中车时代电气股份有限公司 | Motor bearing failure diagnosis device |
| US11898932B2 (en) | 2018-10-15 | 2024-02-13 | ZhuZhou CRRC Times Electric Co., Ltd. | Motor bearing failure diagnosis device |
| CN113557420A (en) * | 2019-03-14 | 2021-10-26 | 西门子股份公司 | Expert knowledge based automatic health status classifier for rotating machines |
| CN110197194A (en) * | 2019-04-12 | 2019-09-03 | 佛山科学技术学院 | A kind of Method for Bearing Fault Diagnosis and device based on improvement random forest |
| CN112034340A (en) * | 2019-06-03 | 2020-12-04 | 中国人民解放军63756部队 | Method for screening fault characteristics of measurement and control antenna motor |
| CN110907826A (en) * | 2019-11-14 | 2020-03-24 | 中车株洲电力机车研究所有限公司 | Motor fault diagnosis method and system based on convolutional neural network filtering |
| CN111157092A (en) * | 2020-01-02 | 2020-05-15 | 深圳市汉德网络科技有限公司 | Vehicle-mounted weighing automatic calibration method and computer readable storage medium |
| CN111208427A (en) * | 2020-04-23 | 2020-05-29 | 湖南中车时代通信信号有限公司 | Traction motor fault diagnosis method and device |
| CN111208427B (en) * | 2020-04-23 | 2020-07-24 | 湖南中车时代通信信号有限公司 | Traction motor fault diagnosis method and device |
| WO2021213142A1 (en) * | 2020-04-23 | 2021-10-28 | 中车株洲电力机车研究所有限公司 | Traction motor fault diagnosis method and apparatus |
| CN111879514A (en) * | 2020-07-31 | 2020-11-03 | 南京机电职业技术学院 | Brushless direct current motor bearing fault diagnosis method based on ELM model |
| CN114152437A (en) * | 2020-08-17 | 2022-03-08 | Abb瑞士股份有限公司 | Method for detecting bearing failure |
| CN112052796A (en) * | 2020-09-07 | 2020-12-08 | 电子科技大学 | Permanent magnet synchronous motor fault diagnosis method based on deep learning |
| CN112287775A (en) * | 2020-10-14 | 2021-01-29 | 中车株洲电力机车研究所有限公司 | Motor fault diagnosis method and device and related components |
| CN112504678A (en) * | 2020-11-12 | 2021-03-16 | 重庆科技学院 | Motor bearing fault diagnosis method based on knowledge distillation |
| CN113283292A (en) * | 2021-04-13 | 2021-08-20 | 江苏大学镇江流体工程装备技术研究院 | Method and device for fault diagnosis of underwater micro propeller |
| CN113283292B (en) * | 2021-04-13 | 2023-08-08 | 江苏大学镇江流体工程装备技术研究院 | Method and device for diagnosing faults of underwater micro-propeller |
| CN113189483A (en) * | 2021-04-25 | 2021-07-30 | 西安交通大学 | Frequency conversion spectrum peak and current range joint inference axial flow fan fault diagnosis method |
| CN113189483B (en) * | 2021-04-25 | 2022-10-28 | 西安交通大学 | Fault diagnosis method for joint inference of axial flow fan by frequency conversion spectrum peak and current range |
| CN113295413A (en) * | 2021-06-24 | 2021-08-24 | 北京交通大学 | Traction motor bearing fault diagnosis method based on indirect signals |
| CN113639999A (en) * | 2021-08-16 | 2021-11-12 | 北京信息科技大学 | Fault diagnosis method, system, medium and equipment for rolling bearing under variable rotating speed |
| CN114112396A (en) * | 2021-10-29 | 2022-03-01 | 江苏大学 | Bearing fault diagnosis method under quasi-steady working condition of rotating speed fluctuation by adopting current signal analysis |
| CN114112396B (en) * | 2021-10-29 | 2023-09-26 | 江苏大学 | Bearing fault diagnosis method under rotating speed fluctuation quasi-stationary working condition by adopting current signal analysis |
| CN114295981A (en) * | 2022-01-05 | 2022-04-08 | 湖南师范大学 | Motor fault diagnosis method based on RN network current multi-feature form recognition |
| CN116106742A (en) * | 2023-04-07 | 2023-05-12 | 国家石油天然气管网集团有限公司 | Motor bearing fault diagnosis method and system based on DQ transformation |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109238698A (en) | A kind of motor bearings method for diagnosing faults based on current signal | |
| CN109061474A (en) | A kind of motor bearings trouble-shooter | |
| Yang et al. | Cost-effective condition monitoring for wind turbines | |
| Kia et al. | A comparative study of acoustic, vibration and stator current signatures for gear tooth fault diagnosis | |
| Wang et al. | Multiscale filtering reconstruction for wind turbine gearbox fault diagnosis under varying-speed and noisy conditions | |
| CN104677623B (en) | A kind of blade of wind-driven generator fault acoustics in place diagnostic method and monitoring system | |
| CN103698699A (en) | Asynchronous motor fault monitoring and diagnosing method based on model | |
| Yang et al. | Vibration condition monitoring system for wind turbine bearings based on noise suppression with multi-point data fusion | |
| Wang et al. | Bearing fault diagnosis of direct-drive wind turbines using multiscale filtering spectrum | |
| Ibrahim et al. | Neural networks for wind turbine fault detection via current signature analysis | |
| Pietrzak et al. | PMSM stator winding fault detection and classification based on bispectrum analysis and convolutional neural network | |
| CN103234702B (en) | Method for diagnosing imbalance faults of blades | |
| Li et al. | Quantitative evaluation on the performance and feature enhancement of stochastic resonance for bearing fault diagnosis | |
| Amirat et al. | Condition monitoring of wind turbines based on amplitude demodulation | |
| Huang et al. | Frequency phase space empirical wavelet transform for rolling bearings fault diagnosis | |
| Liu et al. | Kernel regression residual decomposition-based synchroextracting transform to detect faults in mechanical systems | |
| Duan et al. | Adaptive morphological analysis method and its application for bearing fault diagnosis | |
| CN109765054A (en) | A kind of Fault Diagnosis of Roller Bearings | |
| CN110160767A (en) | Impulse period automatic identification and extracting method and system based on Envelope Analysis | |
| Wang et al. | Weak fault diagnosis of rolling bearing under variable speed condition using IEWT-based enhanced envelope order spectrum | |
| CN104111391A (en) | Three-phase instantaneous power based motorized spindle fault monitoring and diagnosing method | |
| Qiao et al. | Prognostic condition monitoring for wind turbine drivetrains via generator current analysis | |
| Abolghasemi et al. | Recursive singular spectrum analysis for induction machines unbalanced rotor fault diagnosis | |
| Wang et al. | Improved cyclostationary analysis method based on TKEO and its application on the faults diagnosis of induction motors | |
| Pei et al. | Bearing running state recognition method based on feature-to-noise energy ratio and improved deep residual shrinkage network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190118 |
|
| RJ01 | Rejection of invention patent application after publication |