CN103018669A - Method for detecting failure of rotor broken bar of cage type asynchronous motor at high reliability - Google Patents

Method for detecting failure of rotor broken bar of cage type asynchronous motor at high reliability Download PDF

Info

Publication number
CN103018669A
CN103018669A CN2012105031187A CN201210503118A CN103018669A CN 103018669 A CN103018669 A CN 103018669A CN 2012105031187 A CN2012105031187 A CN 2012105031187A CN 201210503118 A CN201210503118 A CN 201210503118A CN 103018669 A CN103018669 A CN 103018669A
Authority
CN
China
Prior art keywords
signal
component
rotor
esprit
frequency
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.)
Granted
Application number
CN2012105031187A
Other languages
Chinese (zh)
Other versions
CN103018669B (en
Inventor
孙丽玲
许伯强
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
North China Electric Power University
Original Assignee
North China Electric Power University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by North China Electric Power University filed Critical North China Electric Power University
Priority to CN201210503118.7A priority Critical patent/CN103018669B/en
Publication of CN103018669A publication Critical patent/CN103018669A/en
Application granted granted Critical
Publication of CN103018669B publication Critical patent/CN103018669B/en
Expired - Fee Related legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Landscapes

  • Tests Of Circuit Breakers, Generators, And Electric Motors (AREA)

Abstract

The invention discloses a method for detecting failure of a rotor broken bar of a cage type asynchronous motor at high reliability. The method comprises the steps of: firstly, filtering a direct current component from a three-phase current signal acquired according to a certain frequency by applying park transformation, simplifying the frequency content by applying Hilbert transformation; secondly, carrying out spectral analysis by applying ESPRIT to obtain an ESPRIT spectrogram; and finally, judging whether the failure of the rotor broken bar occurs according to the existence of a 2sf1 component spectrum peak and a 4sf1 component spectrum peak in the ESPRIT spectrogram. According to the invention, the failure of the rotor broken bar of the asynchronous motor can be detected at high reliability by only sampling a stator current signal in short time, and the adverse influence of load fluctuation on a detection result is effectively overcome; and the method is suitable for the operation condition of a low slip ratio of the asynchronous motor. Thus, the method is applicable for a large asynchronous motor widely applied to engineering practice, and has an important engineering practical value.

Description

A kind of high reliability ground detects the method for cage type asynchronous motor rotor strip-broken fault
Technical field
The present invention relates to a kind of method that can detect the cage type asynchronous motor rotor strip-broken fault, belong to the detection technique field.
Background technology
Cage type asynchronous motor is in operational process, rotor bar is subject to the effect of the alterante stresses such as radial electromagnetic force, electric rotating magnetic force, centrifugal force, thermal flexure amount of deflection power, the rotor manufacturing defect may cause broken bar fault in addition, and this kind fault rate is about 15%.
Rotor broken bar is typical gradual fault, common 1,2 bar failure of initial stage, then gradually development decline even shutdown so that motor is exerted oneself.Therefore, must implement rotor bar breaking fault detects.
After the cage type asynchronous motor generation rotor bar breaking fault, in its stator current, will occur (1 ± 2 Ks) f 1The extra current component of frequency ( sBe revolutional slip, f 1Be line frequency, k=1,2 ...), can be with it as the rotor bar breaking fault feature.And stator current signal is easy to gather, and therefore the stator current signal frequency spectrum analysis method based on Fast Fourier Transform (FFT) (FFT) is widely used in the rotor bar breaking fault detection.
Initial rotor strip-broken failure detecting method is that the stable state stator current signal is directly carried out the FFT spectrum analysis, according to whether having (1 ± 2 in the spectrogram Ks) f 1Frequency component judges that rotor has or not disconnected bar.When slightly breaking bar owing to rotor, (1 ± 2 Ks) f 1The amplitude of component with respect to f 1Component is very little, and asynchronous motor when operation revolutional slip
Figure 271012DEST_PATH_IMAGE001
Very little, (1 ± 2 Ks) f 1With f 1These two frequency numerical value approach, if directly do the FFT spectrum analysis, and then (1 ± 2 Ks) f 1Component may quilt f 1The leakage of component is flooded.This is the weak point of the method.
In order to remedy the deficiency of the method, development has formed the cage type asynchronous motor rotor strip-broken failure detecting method based on Park Transformation, and its core is: the stator current signal during with rotor bar breaking fault is considered as 2 Ksf 1Component (modulation signal) is right f 1The result of component (carrier signal) modulation, and modulation signal obviously comprises the rotor bar breaking fault feature; Therefore, carry out " demodulation " by proper method and obtain modulation signal, extract and analyze wherein 2 Ksf 1Component can realize that rotor bar breaking fault detects.
Above-mentioned Park Transformation detection method, it may further comprise the steps:
A. measure threephase stator electric current momentary signal i A, i B, i C
B. calculate threephase stator electric current momentary signal i A, i B, i CPark Transformation
Figure 348689DEST_PATH_IMAGE002
,
Figure 522182DEST_PATH_IMAGE003
, shown in (1);
Figure 126338DEST_PATH_IMAGE004
(1)
Figure 94294DEST_PATH_IMAGE002
,
Figure 659268DEST_PATH_IMAGE003
Frequency content very complicated, comprise 2 Sf 1, 4 Sf 1Frequency component and DC component are described as follows.
Behind the asynchronous motor generation rotor bar breaking fault, in its stator current, will occur (1 ± 2 Ks) f 1The extra current component of frequency.Therefore, the threephase stator current signal of asynchronous motor in the rotor bar breaking fault situation can adopt formula (2), formula (3) and formula (4) simulation.Herein, get k=1,2( k2 o'clock extra current component amplitude is too small, ignore).
Figure 370872DEST_PATH_IMAGE005
(2)
Figure 829535DEST_PATH_IMAGE006
(3)
Figure 968392DEST_PATH_IMAGE007
(4)
Wherein, I M1, I ML, I MR, I MLL, I MRRRespectively representative f 1, (1-2 s) f 1, (1+2 s) f 1, (1-4 s) f 1, (1+4 s) f 1The amplitude of component; f 1, f L, f R, f LL, f RRRepresent respectively the initial phase angle of each component.
With formula (2), formula (3), formula (4) substitution formula (1) and arrangement, can derive its Park Transformation
Figure 20662DEST_PATH_IMAGE002
,
Figure 270378DEST_PATH_IMAGE003
, be shown in formula (5), formula (6).
Figure 849127DEST_PATH_IMAGE008
(5)
(6)
According to formula (5), formula (6) as can be known,
Figure 698451DEST_PATH_IMAGE002
, Comprise 2 Sf 1, 4 Sf 1Frequency component and DC component are very complicated.
C. filtering
Figure 185113DEST_PATH_IMAGE002
In DC component, obtain a new signal M d, M d=
Figure 665773DEST_PATH_IMAGE002
-mean (
Figure 427055DEST_PATH_IMAGE002
), mean (
Figure 284153DEST_PATH_IMAGE002
) expression
Figure 571915DEST_PATH_IMAGE002
Mean value, signal M dBe shown in formula (7), its main frequency composition is 2 Sf 1Component, 4 Sf 1Component;
Figure 223476DEST_PATH_IMAGE010
(7)
D. filtering
Figure 737634DEST_PATH_IMAGE003
In DC component, obtain a new signal M q, M q=
Figure 726319DEST_PATH_IMAGE003
-mean ( ), mean (
Figure 831995DEST_PATH_IMAGE003
) expression
Figure 692504DEST_PATH_IMAGE003
Mean value, signal M qBe shown in formula (8), its main frequency composition is 2 Sf 1Component, 4 Sf 1Component;
Figure 891404DEST_PATH_IMAGE011
(8)
E. to signal M d, M qDo the FFT spectrum analysis, according to whether having 2 in the spectrogram Sf 1Component, 4 Sf 1Component is composed the peak and is carried out rotor bar breaking fault and detect.
So far as can be known, above-mentioned Park Transformation method is in essence with in the stator current signal f 1Component transformation is DC component, simultaneously with (1 ± 2 Ks) f 1Component transformation is 2 Ksf 1Component.But said method is take " DC component can the desirable filtering by deducting mean value " as prerequisite.This just requires---
Figure 763545DEST_PATH_IMAGE002
,
Figure 615963DEST_PATH_IMAGE003
In DC component held stationary during gathering.Particularly, in formula (5), the formula (6) I M1,
Figure 370292DEST_PATH_IMAGE012
All should be constant during gathering.So,
Figure 576146DEST_PATH_IMAGE002
, In DC component in DC component
Figure 122851DEST_PATH_IMAGE013
,
Figure 567739DEST_PATH_IMAGE014
During gathering, be only constant, can the desirable filtering by deducting mean value.Otherwise this DC component will fluctuate, can't the fully filtering by deducting mean value.Consider that from the angle of FFT spectrum analysis this will cause DC component " leakages " and make the FFT frequency spectrum be tending towards complexity even obscure the rotor bar breaking fault feature---2 Sf 1Component, 4 Sf 1Component affects the reliability that rotor bar breaking fault detects.But, in engineering reality, " I M1,
Figure 108441DEST_PATH_IMAGE012
All should be constant during gathering " this requirement can't satisfy, and reason is---and for real electrical machinery, load fluctuation to a certain degree is inevitably, that is: I M1,
Figure 345388DEST_PATH_IMAGE012
Certainly exist fluctuation to a certain degree.As everyone knows, a phase stator current effective value of real electrical machinery (
Figure 680554DEST_PATH_IMAGE015
) always with to a certain degree the fluctuation.Obviously, I M1Must fluctuate.And Be approximately equal to the power-factor angle of motor, follow motor stator current effective value fluctuation and fluctuate.Therefore, in engineering reality, above-mentioned Park Transformation method shortcoming reliability.
In addition, above-mentioned Park Transformation method is subject to frequency resolution (inverse of sampling duration) based on the FFT spectrum analysis, be explained as follows.
In engineering reality, extensively adopt large-sized asynchronous motor, and generally under 40% ~ 60% rate of load condensate, move revolutional slip sNumerical value very little (＜0.5%) causes frequency 2 Sf 1, 4 Sf 1Numerical value is very little.Situation is similar during medium and small asynchronous motor during fractional load operation.For above-mentioned situation, the Park Transformation method needs the asynchronous motor signal of the enough durations of continuous acquisition can guarantee that frequency resolution is enough high with practical resolution rotor bar breaking fault feature.But for real electrical machinery, load fluctuation to a certain degree is inevitably, and the signals collecting overlong time means with more high probability introducing load fluctuation, aggravation
Figure 957132DEST_PATH_IMAGE002
,
Figure 48584DEST_PATH_IMAGE003
The fluctuation of middle DC component, leakage affect the reliability that rotor bar breaking fault detects.This just shows that for the large-sized asynchronous motor that extensively adopts in the engineering reality, above-mentioned Park Transformation method faces a severe challenge, even loses efficacy.
At last, this Park Transformation method is directly to signal M d, M qDo the FFT spectrum analysis, but signal M in fact d, M qFrequency form still complicated, referring to formula (7), formula (8).With signal M dIn 2 Sf 1Component is example, and its expression formula is shifted onto as follows:
Figure 554652DEST_PATH_IMAGE016
(9)
Wherein,
Figure 239711DEST_PATH_IMAGE017
According to formula (9) as can be known, signal M dIn 2 Sf 1The amplitude of component is
Figure 122217DEST_PATH_IMAGE018
In actual conditions, I ML I MRLike this, exist
Figure 79895DEST_PATH_IMAGE019
The time, this amplitude is approximately equal to 0.This means: to signal M dDo the FFT spectrum analysis, will can not occur 2 in the spectrogram Sf 1Component spectrum peak (perhaps extremely not obvious), thus the reliability that rotor bar breaking fault detects affected.This possibility is really to exist.For signal M dIn 4 Sf 1There is same problem in component.
Similarly, to signal M qDo the FFT spectrum analysis, the same possibility that exists---can not occur 2 in the spectrogram Sf 1Component, 4 Sf 1Component spectrum peak (perhaps extremely not obvious), thus the reliability that rotor bar breaking fault detects affected.
More than the analysis showed that: there is significant deficiency in existing Park Transformation method based on the FFT spectrum analysis---and load fluctuation causes
Figure 22443DEST_PATH_IMAGE002
,
Figure 929219DEST_PATH_IMAGE003
The fluctuation of middle DC component, leak, cause the FFT frequency spectrum to be tending towards complicated and affect the reliability of rotor bar breaking fault detection.And the FFT spectrum analysis needs the motor signal of enough durations to guarantee frequency resolution, this means with more high probability introducing load fluctuation, causes this problem more outstanding.Particularly for the large-sized asynchronous motor that extensively adopts in the engineering reality, above-mentioned Park Transformation method faces a severe challenge, even loses efficacy.This is the current key issue that needs to be resolved hurrily, possesses deep engineering background, is the rotor strip-broken failure detecting method development, uses it " bottleneck ", and this solution of problem will have Important Project and be worth.At last, the Park Transformation method is carried out the object signal of FFT spectrum analysis, i.e. M in the preamble d, M qFrequency forms still complicated, may further worsen the reliability that rotor bar breaking fault detects.
Summary of the invention
The object of the present invention is to provide a kind of threephase stator current signal that can be short as far as possible according to duration, high reliability ground detects the method for cage type asynchronous motor rotor strip-broken fault.
The alleged problem of the present invention realizes with following technical proposals:
Detect to a kind of high reliability the method for cage type asynchronous motor rotor strip-broken fault, threephase stator current signal application Park Transformation, its DC component of filtering that it at first gathers pressing certain frequency, and use Hilbert transform and simplify its frequency composition; Then use ESPRIT(invariable rotary signal parameter estimation technique) carry out spectrum analysis, obtain the ESPRIT spectrogram; At last according to whether having 2 in the ESPRIT spectrogram Sf 1Component, 4 Sf 1Component spectrum peak judges that whether rotor bar breaking fault occurs.
Above-mentioned high reliability ground detects the method for cage type asynchronous motor rotor strip-broken fault, and it may further comprise the steps:
A. measure threephase stator electric current momentary signal i A, i B, i C
For high-voltage motor, adopt three current clamps to measure threephase stator electric current momentary signal in the Current Transformer Secondary side; For low voltage motor, adopt three current clamps directly to measure threephase stator electric current momentary signal at the connecting terminal of motor place;
B. adopt single power frequency period sliding window method to calculate the effective value of A phase stator current momentary signal
Figure 349836DEST_PATH_IMAGE020
, by analyzing the variation tendency of effective value, extract its one piece of data the most steady that is fluctuation is minimum, be designated as i AS,
C. extract in B, the C phase stator current momentary signal with i ASBe in the data of same time section, be designated as i BS, i CS
D. calculate threephase stator electric current momentary signal i AS, i BS, i CSPark Transformation
Figure 415881DEST_PATH_IMAGE002
,
Figure 263751DEST_PATH_IMAGE003
E. filtering
Figure 657824DEST_PATH_IMAGE002
In DC component, picked up signal M d, M d=
Figure 882132DEST_PATH_IMAGE002
-mean ( ), mean ( ) expression
Figure 968402DEST_PATH_IMAGE002
Mean value;
F. filtering
Figure 730822DEST_PATH_IMAGE003
In DC component, picked up signal M q, M q=
Figure 505880DEST_PATH_IMAGE003
-mean (
Figure 961132DEST_PATH_IMAGE003
), mean (
Figure 329796DEST_PATH_IMAGE003
) expression
Figure 489382DEST_PATH_IMAGE003
Mean value;
G. calculate M according to following formula qHilbert transform H (M q):
(10)
H. construct a new signal N according to following formula L:
Figure 620466DEST_PATH_IMAGE022
(11)
I. calculate M according to following formula dHilbert transform H (M d):
Figure 7585DEST_PATH_IMAGE023
(12)
J. construct a new signal N according to following formula R:
Figure 705283DEST_PATH_IMAGE024
(13)
In formula (10), the formula (12), tThe expression time,
Figure 330299DEST_PATH_IMAGE025
The expression time delay.
K. use ESPRIT respectively to signal N L, N RCarry out spectrum analysis, obtain 2 ESPRIT spectrograms;
L. according to whether having 2 in these 2 ESPRIT spectrograms Sf 1Component, 4 Sf 1Component spectrum peak judges that whether rotor bar breaking fault occurs: if having simultaneously 2 Sf 1Component spectrum peak, rotor broken bar then, otherwise rotor is normal.
Above-mentioned high reliability ground detects the method for cage type asynchronous motor rotor strip-broken fault, adopts single power frequency period sliding window method to calculate the effective value of A phase stator current momentary signal
Figure 127354DEST_PATH_IMAGE020
Method be:
Choose A phase stator current momentary signal i AIn continuous 20 points, calculate its effective value
Figure 329665DEST_PATH_IMAGE026
For selected i AIn continuous 20 points, keep rear 19 points, the order fill vacancies in the proper order i AIn back 1 point (the 21st point), thereby again obtain i AIn continuous 20 points, again calculate its effective value, by that analogy, determine i AThe effective value variation tendency.
Above-mentioned high reliability ground detects the method for cage type asynchronous motor rotor strip-broken fault, described threephase stator electric current momentary signal i A, i B, i CSample frequency be set as 1000Hz, the sampling duration is set as 10s, the duration of the most stably one piece of data that therefrom extracts is 4s.
Above-mentioned high reliability ground detects the method for cage type asynchronous motor rotor strip-broken fault, uses ESPRIT(invariable rotary signal parameter estimation technique) method that signal is carried out spectrum analysis is:
A. construct correlation matrix
Figure 503158DEST_PATH_IMAGE027
,
Figure 982680DEST_PATH_IMAGE028
, specific as follows;
Be without loss of generality signal N to be analyzed LWith N RAll can be expressed as the combination of a series of cosine harmonics components, be shown below.
Figure 950637DEST_PATH_IMAGE029
；
Wherein, The expression sampling period;
Figure 86269DEST_PATH_IMAGE031
The expression sampling number;
Figure 685877DEST_PATH_IMAGE032
Expression harmonic wave number;
Figure 824735DEST_PATH_IMAGE033
, ,
Figure 251354DEST_PATH_IMAGE035
Represent respectively The amplitude of individual harmonic wave, frequency, initial phase angle.
Herein, signal N LWith N RDuration be 4s, sample frequency is 1000Hz, therefore, sampling number is N=4000.In addition, preamble is pointed out: signal N LWith N RThe main frequency composition be 2 Sf 1Component, 4 Sf 1Component; In other words, signal N LWith N RMainly comprise 2 frequency components, this means p≈ 2.
Definition
Figure 15227DEST_PATH_IMAGE037
, below the introducing
Figure 679427DEST_PATH_IMAGE038
The rank matrix ( mShould much larger than p, get herein m=20):
Figure 732834DEST_PATH_IMAGE039
，
Figure 41455DEST_PATH_IMAGE040
，
Then
Figure 522115DEST_PATH_IMAGE041
Autocorrelation matrix be
Figure 408032DEST_PATH_IMAGE042
，
And
Figure 265129DEST_PATH_IMAGE041
With
Figure 428257DEST_PATH_IMAGE043
Cross-correlation matrix be
Figure 79818DEST_PATH_IMAGE044
，
Here,
Figure 718610DEST_PATH_IMAGE045
The expression mathematical expectation,
Figure 113819DEST_PATH_IMAGE046
The expression conjugate transpose.
B pair
Figure 131454DEST_PATH_IMAGE027
Carry out Eigenvalues Decomposition, determine its minimal eigenvalue
Figure 219496DEST_PATH_IMAGE047
C calculates
Figure 80004DEST_PATH_IMAGE048
,
Figure 482167DEST_PATH_IMAGE049
Expression
Figure 213362DEST_PATH_IMAGE050
The rank unit matrix.
D calculates
Figure 65781DEST_PATH_IMAGE051
,
Figure 554531DEST_PATH_IMAGE052
For
Figure 25964DEST_PATH_IMAGE050
The rank battle array,
Figure 549349DEST_PATH_IMAGE053
(herein,
Figure 307089DEST_PATH_IMAGE049
Expression
Figure 548715DEST_PATH_IMAGE054
The rank unit matrix).
E pair
Figure 558259DEST_PATH_IMAGE055
Carry out svd
Figure 998468DEST_PATH_IMAGE056
, herein
Figure 192689DEST_PATH_IMAGE057
,
Figure 656031DEST_PATH_IMAGE058
(
Figure 469266DEST_PATH_IMAGE059
By
Figure 560719DEST_PATH_IMAGE032
Individual main singular value forms),
Figure 535628DEST_PATH_IMAGE060
The f compute matrix
G pair
Figure 227827DEST_PATH_IMAGE062
Carry out generalized eigenvalue decomposition, determine Individual generalized eigenvalue
Figure 319596DEST_PATH_IMAGE063
(all the other Individual generalized eigenvalue is constantly equal to 0).
H determines the frequency of each component of sampled signal according to generalized eigenvalue
Figure 912569DEST_PATH_IMAGE065
,
Figure 713035DEST_PATH_IMAGE066
,
Figure 826484DEST_PATH_IMAGE067
Difference representation feature value
Figure 220556DEST_PATH_IMAGE068
Imaginary part, real part.
The i compute matrix
Figure 38340DEST_PATH_IMAGE069
The j compute matrix
Figure 834257DEST_PATH_IMAGE070
, herein
Figure 118608DEST_PATH_IMAGE071
It is a column vector
Figure 327873DEST_PATH_IMAGE072
, and
Figure 355871DEST_PATH_IMAGE073
Be column vector
Figure 130929DEST_PATH_IMAGE074
K determines the amplitude of each component of sampled signal
Figure 586182DEST_PATH_IMAGE075
,
Figure 954846DEST_PATH_IMAGE076
The present invention gathers the asynchronous motor stator current signal by data acquisition card, and data acquisition card is sent to portable computer with this signal, by portable computer current signal is processed, and judges whether to exist rotor bar breaking fault, and is simple to operation.The method is obtained frequency by Park Transformation, Hilbert transform and is formed simple, clear and definite signal to be analyzed, use ESPRIT and carry out spectrum analysis, can be according to the short as far as possible stator current signal of duration and high sensitivity, high reliability ground detect the asynchronous motor rotor strip-broken fault, effectively overcome the adverse effect of load fluctuation to testing result.
The most significant advantage of the present invention is exactly: only need very short stator current signal of sampling time to detect the asynchronous motor rotor strip-broken fault in high reliability ground, effectively overcome the adverse effect of load fluctuation to testing result, be applicable to asynchronous motor low revolutional slip ruuning situation.Therefore, the present invention is applicable to the extensive large-sized asynchronous motor (low revolutional slip operation) that adopts in the engineering reality, thereby possesses Important Project value and broad prospect of application.
Description of drawings
The invention will be further described below in conjunction with accompanying drawing.
Fig. 1 is the electrical schematic diagram of the used signal pickup assembly of the present invention;
Fig. 2 is experimental wiring figure;
Varied curve when Fig. 3 is the stator current effective value of motor when rotor is normal;
Varied curve when Fig. 4 is the stator current effective value of motor when rotor fault;
Fig. 5 is that motor is signal M when rotor is normal in the signals collecting duration 4s situation d(be threephase stator electric current momentary signal Park Transformation
Figure 520957DEST_PATH_IMAGE002
The signal that obtains after the filtering DC component) FFT frequency spectrum;
Fig. 6 is that motor is signal M when rotor is normal in the signals collecting duration 4s situation q(be threephase stator electric current momentary signal Park Transformation
Figure 150521DEST_PATH_IMAGE003
The signal that obtains after the filtering DC component) FFT frequency spectrum;
Fig. 7 is that motor is signal M when rotor fault in the signals collecting duration 4s situation d(be threephase stator electric current momentary signal Park Transformation
Figure 245516DEST_PATH_IMAGE002
The signal that obtains after the filtering DC component) FFT frequency spectrum;
Fig. 8 is that motor is signal M when rotor fault in the signals collecting duration 4s situation q(be threephase stator electric current momentary signal Park Transformation
Figure 367056DEST_PATH_IMAGE003
The signal that obtains after the filtering DC component) FFT frequency spectrum;
Fig. 9 is that motor is signal N when rotor is normal in the signals collecting duration 4s situation LThe ESPRIT frequency spectrum of (signal as ESPRIT spectrum analysis object of the new definition of the present invention);
Figure 10 is that motor is signal N when rotor is normal in the signals collecting duration 4s situation RThe ESPRIT frequency spectrum of (signal as ESPRIT spectrum analysis object of the new definition of the present invention);
Figure 11 is that motor is signal N when rotor fault in the signals collecting duration 4s situation LThe ESPRIT frequency spectrum of (signal as ESPRIT spectrum analysis object of the new definition of the present invention);
Figure 12 is that motor is signal N when rotor fault in the signals collecting duration 4s situation RThe ESPRIT frequency spectrum of (signal as ESPRIT spectrum analysis object of the new definition of the present invention);
Figure 13 is that motor is signal M when rotor is normal in the signals collecting duration 10s situation d(be threephase stator electric current momentary signal Park Transformation
Figure 330333DEST_PATH_IMAGE002
The signal that obtains after the filtering DC component) FFT frequency spectrum;
Figure 14 is that motor is signal M when rotor is normal in the signals collecting duration 10s situation q(be threephase stator electric current momentary signal Park Transformation
Figure 955349DEST_PATH_IMAGE003
The signal that obtains after the filtering DC component) FFT frequency spectrum;
Figure 15 is that motor is signal M when rotor fault in the signals collecting duration 10s situation d(be threephase stator electric current momentary signal Park Transformation The signal that obtains after the filtering DC component) FFT frequency spectrum;
Figure 16 is that motor is signal M when rotor fault in the signals collecting duration 10s situation q(be threephase stator electric current momentary signal Park Transformation
Figure 423556DEST_PATH_IMAGE003
The signal that obtains after the filtering DC component) FFT frequency spectrum.
Each label is among the figure: CT A, CT B, CT C, threephase current transformer, Motor, motor.
The meaning of used each symbol in the literary composition: , revolutional slip; , line frequency (fundamental frequency); i A, i B, i C, threephase stator electric current momentary signal; i AS, i BS, i CS, the most steady that is one piece of data that fluctuation is minimum in the threephase stator electric current momentary signal;
Figure 965899DEST_PATH_IMAGE002
,
Figure 62031DEST_PATH_IMAGE003
, threephase stator electric current momentary signal Park Transformation; , A phase stator current momentary signal effective value; M d, M q, threephase stator electric current momentary signal Park Transformation
Figure 966719DEST_PATH_IMAGE002
,
Figure 105576DEST_PATH_IMAGE003
The signal that obtains after the filtering DC component; Mean (
Figure 892267DEST_PATH_IMAGE002
), mean (
Figure 407562DEST_PATH_IMAGE003
), threephase stator electric current momentary signal Park Transformation
Figure 986311DEST_PATH_IMAGE002
,
Figure 296069DEST_PATH_IMAGE003
Mean value; H (M d), H (M q), signal M d, M qHilbert transform; N L, N R, the new definition of the present invention, as the new signal of ESPRIT spectrum analysis object; The meaning of other each symbol is all explained in preamble.
Embodiment
The present invention adopts circuit shown in Figure 1 to detect, this circuit is comprised of 3 current transformers, data acquisition card and portable computers, described current transformer is connected to respectively on three phase lines of asynchronous motor stator winding, its signal output part connects the simulating signal input channel ( input terminal 5 and 17, input terminal 6 and 18, input terminal 7 and 19) of data acquisition card, and the output port of described data acquisition card connects the USB mouth of portable computer.Data acquisition card adopts auspicious rich magnificent RBH8351 type data acquisition card, and the model of portable computer is Thinkpad X100e, data acquisition card is integrated circuit such as low-pass filter, signals collecting maintenance, mould/number conversion.The stator current momentary signal is delivered to data acquisition card, and data acquisition card is connected in portable computer by USB interface.Portable computing machine control signal capture card is with appropriate frequency sampling stator current momentary signal, and is stored in hard disk, by portable computer current signal processed again, judges whether to exist rotor bar breaking fault.This software kit is based on Windows XP operating system and adopt the establishment of Visual C++ application development platform.
This method may further comprise the steps:
A. measure threephase stator electric current momentary signal i A, i B, i C:
For high-voltage motor, adopt three current clamps to measure threephase stator electric current momentary signal in the Current Transformer Secondary side; For low voltage motor, adopt three current clamps directly to measure threephase stator electric current momentary signal at the connecting terminal of motor place; Sample frequency is set as 1000Hz, the sampling duration is set as 10s;
B. to A phase stator current momentary signal i ADo initial analysis, extract its most stably one piece of data, be designated as i AS:
This is by analyzing i AThe effective value variation tendency carry out effective value
Figure 570056DEST_PATH_IMAGE020
Adopt single power frequency period sliding window method to calculate according to following formula, monocycle sliding window method brief introduction is as follows.
Power frequency is generally 50 hertz, and then power frequency period is 0.02 second.Be that per second gathers 1000 points if frequency acquisition is 1000Hz(), then each power frequency period gathers stator current momentary signal 20 points.Choose i AIn continuous 20 points (single power frequency period) calculate its effective value; For selected i AIn continuous 20 points, keep rear 19 points, the order fill vacancies in the proper order i AIn back 1 point (the 21st point), thereby again obtain i AIn continuous 20 points, again calculate its effective value.By that analogy, can determine i AThe effective value variation tendency.
About i ASDuration, should lack as far as possible to avoid most possibly load fluctuation, but the signals collecting duration should be greater than lasting cycle of characteristic component to be extracted wherein.In engineering reality, extensively adopt large-sized asynchronous motor, and generally under 40% ~ 60% rate of load condensate, move revolutional slip sNumerical value very little (＜0.5%), but generally can be less than 0.25%, situation is similar during medium and small asynchronous motor during fractional load operation.And line frequency (fundamental frequency) f 1Be approximately equal to 50Hz.Therefore, rotor bar breaking fault characteristic component frequency to be extracted 2s f 10.25Hz, 4s f 10.50Hz its lasting cycle is respectively 4s, 2s.Obviously, the signals collecting duration should be chosen as 4s.
C. extract in B, the C phase stator current momentary signal with i ASBe in the data of same time section, be designated as i BS, i CS, the two can be considered as the most steady that is the fluctuation minimum;
D. calculate threephase stator electric current momentary signal i AS, i BS, i CSPark Transformation ,
Figure 549096DEST_PATH_IMAGE003
, specifically referring to formula (1);
E. filtering
Figure 575958DEST_PATH_IMAGE002
In DC component, picked up signal M d, M d=
Figure 167477DEST_PATH_IMAGE002
-mean (
Figure 720818DEST_PATH_IMAGE002
), mean (
Figure 372379DEST_PATH_IMAGE002
) expression
Figure 620958DEST_PATH_IMAGE002
Mean value;
F. filtering
Figure 16167DEST_PATH_IMAGE003
In DC component, picked up signal M q, M q=
Figure 424015DEST_PATH_IMAGE003
-mean (
Figure 715319DEST_PATH_IMAGE003
), mean (
Figure 982352DEST_PATH_IMAGE003
) expression
Figure 774727DEST_PATH_IMAGE003
Mean value;
G. calculate M qHilbert transform H (M q), specifically referring to formula (10);
H. construct a new signal N L, specifically referring to formula (11);
I. calculate M dHilbert transform H (M d), specifically referring to formula (12);
J. construct a new signal N R, specifically referring to formula (13);
K. use ESPRIT respectively to signal N L, N RCarry out spectrum analysis, obtain 2 ESPRIT spectrograms;
L. according to whether having 2 in these 2 ESPRIT spectrograms Sf 1Component, 4 Sf 1Component is composed the peak and is carried out rotor bar breaking fault and detect: if having simultaneously 2 Sf 1Component, 4 Sf 1Component spectrum peak, then rotor broken bar; Otherwise rotor is normal.
Therefore invariable rotary signal parameter estimation technique (ESPRIT, Estimation of Signal Parameters via Rotational Invariance Technique) is the key of the method, and brief introduction is as follows.
ESPRIT is by R. Roy, A. Paulraj, T. Kailath proposition and development, become at present the effective tool that cosine signal parameter (number and frequency) is estimated.
Sampled signal
Figure 443606DEST_PATH_IMAGE078
Can be expressed as the combination of a series of cosine harmonics components
Figure 905811DEST_PATH_IMAGE079
Figure 253616DEST_PATH_IMAGE080
(14)
In the formula, T SThe expression sampling period; NBe sampling number; pBe the harmonic wave number; A i , f i , f i Represent respectively The amplitude of individual harmonic wave, frequency, initial phase angle.
Definition
Figure 248434DEST_PATH_IMAGE037
, below the introducing
Figure 412699DEST_PATH_IMAGE038
The rank matrix (guarantees m p)
Figure 247800DEST_PATH_IMAGE081
(15)
Figure 788503DEST_PATH_IMAGE082
(16)
Then
Figure 635236DEST_PATH_IMAGE083
Autocorrelation matrix be
(17)
And With Cross-correlation matrix be
Figure 338433DEST_PATH_IMAGE086
(18)
In formula (17), (18),
Figure 703555DEST_PATH_IMAGE045
The expression mathematical expectation,
Figure 123035DEST_PATH_IMAGE046
The expression conjugate transpose.
The ESPRIT algorithm is as follows:
A constructs correlation matrix
Figure 5540DEST_PATH_IMAGE027
,
Figure 951500DEST_PATH_IMAGE028
B pair
Figure 362889DEST_PATH_IMAGE027
Carry out Eigenvalues Decomposition, determine its minimal eigenvalue
Figure 800824DEST_PATH_IMAGE047
C calculates
Figure 80496DEST_PATH_IMAGE048
,
Figure 756328DEST_PATH_IMAGE049
Expression
Figure 869777DEST_PATH_IMAGE050
The rank unit matrix.
D calculates
Figure 388483DEST_PATH_IMAGE051
,
Figure 81633DEST_PATH_IMAGE052
For
Figure 2184DEST_PATH_IMAGE050
The rank battle array, (herein,
Figure 902324DEST_PATH_IMAGE049
Expression
Figure 523798DEST_PATH_IMAGE054
The rank unit matrix).
E pair
Figure 705381DEST_PATH_IMAGE055
Carry out svd
Figure 629475DEST_PATH_IMAGE056
, herein
Figure 857194DEST_PATH_IMAGE057
,
Figure 423304DEST_PATH_IMAGE058
(
Figure 193814DEST_PATH_IMAGE059
By
Figure 819967DEST_PATH_IMAGE032
Individual main singular value forms),
Figure 534983DEST_PATH_IMAGE060
The f compute matrix
Figure 904784DEST_PATH_IMAGE061
G pair
Figure 529800DEST_PATH_IMAGE062
Carry out generalized eigenvalue decomposition, determine
Figure 326855DEST_PATH_IMAGE032
Individual generalized eigenvalue
Figure 529166DEST_PATH_IMAGE063
(all the other
Figure 437079DEST_PATH_IMAGE064
Individual generalized eigenvalue is constantly equal to 0).
H determines the frequency of each component of sampled signal according to generalized eigenvalue , ,
Figure 574166DEST_PATH_IMAGE067
Difference representation feature value
Figure 754611DEST_PATH_IMAGE068
Imaginary part, real part.
The i compute matrix
Figure 213274DEST_PATH_IMAGE069
The j compute matrix , herein
Figure 404401DEST_PATH_IMAGE071
It is a column vector , and
Figure 232866DEST_PATH_IMAGE073
Be column vector
Figure 542625DEST_PATH_IMAGE074
K determines the amplitude of each component of sampled signal
Figure 82190DEST_PATH_IMAGE075
,
Key concept, step according to above-mentioned ESPRIT, can infer---to carry out spectrum analysis different by frequency from FFT, ESPRIT is conceived to full frequency band, the Eigenvalues Decomposition by signal correlation matrix carries out spectrum analysis, and this can suppress spectrum leakage to a certain extent.And with FFT contrast, ESPRIT possesses the original signal extrapolability, thereby its frequency resolution can break away from the restriction of sampling duration, even for short signal, also can reach the high-frequency resolving power.Therefore, ESPRIT is applied to the rotor bar breaking fault detection and possesses feasibility.At first, even introduce load fluctuation, ESPRIT also can suppress to a certain extent DC component and leak, and guarantees the reliability that rotor bar breaking fault detects.Secondly, owing to only need gather short signal, ESPRIT possesses larger possibility and avoids load fluctuation than FFT, and this also is favourable.
Use the method an experiment motor is carried out the rotor bar breaking fault detection, effect is satisfactory.
Wiring is shown in Fig. 2.Motor adopts a Y100L-2 type threephase asynchronous (3kW, 380V, 50Hz), except normal rotor, be equipped with separately a failure rotor with the simulation broken bar fault, there is a fracture sliver (apart from the boring of end ring 10mm place, diameter 6mm, degree of depth 10mm) in this failure rotor.
In experiment, motor load approaches unloaded and does not do and regulate in the hope of keeping constant, revolutional slip sBe about 0.33%.In order to avoid as far as possible load fluctuation, shorten the signals collecting duration as far as possible, be chosen as 4s.Like this, the signals collecting duration is greater than rotor bar breaking fault feature to be extracted 2s f 1A lasting cycle of component.At this moment, 2s f 10.33Hz ( f 150Hz), a corresponding lasting cycle is about 3s.
Fig. 3, Fig. 4 represent motor varied curve when the normal phase stator current effective value during with fault of rotor.
Fig. 5, Fig. 6 represent motor signal M when rotor is normal dWith M q(be threephase stator electric current momentary signal Park Transformation
Figure 568852DEST_PATH_IMAGE002
,
Figure 49512DEST_PATH_IMAGE003
The signal that obtains after the filtering DC component) FFT frequency spectrum, concrete data are shown in table 1, and this is to adopt existing Park Transformation method to obtain.
Fig. 7, Fig. 8 represent motor signal M when rotor fault dWith M q(be threephase stator electric current momentary signal Park Transformation
Figure 76374DEST_PATH_IMAGE002
,
Figure 667892DEST_PATH_IMAGE003
The signal that obtains after the filtering DC component) FFT frequency spectrum, concrete data are shown in table 2, and this is to adopt existing Park Transformation method to obtain.
Fig. 9, Figure 10 represent motor signal N when rotor is normal LWith N RThe ESPRIT frequency spectrum of (signal as ESPRIT spectrum analysis object of the new definition of the present invention), concrete data are shown in table 3, and this is to adopt the present invention to obtain.
Figure 11, Figure 12 represent motor signal N when rotor fault LWith N RThe ESPRIT frequency spectrum of (signal as ESPRIT spectrum analysis object of the new definition of the present invention), concrete data are shown in table 4, and this is to adopt the present invention to obtain.
Figure 13, Figure 14 are that motor is signal M when rotor is normal in the signals collecting duration 10s situation dWith M q(be threephase stator electric current momentary signal Park Transformation
Figure 955654DEST_PATH_IMAGE002
,
Figure 341636DEST_PATH_IMAGE003
The signal that obtains after the filtering DC component) FFT frequency spectrum, concrete data are shown in table 5, and this is to adopt existing Park Transformation method to obtain.
Figure 15, Figure 16 are that motor is signal M when rotor fault in the signals collecting duration 10s situation dWith M q(be threephase stator electric current momentary signal Park Transformation
Figure 121373DEST_PATH_IMAGE002
,
Figure 110058DEST_PATH_IMAGE003
The signal that obtains after the filtering DC component) FFT frequency spectrum, concrete data are shown in table 6, and this is to adopt existing Park Transformation method to obtain.
According to Fig. 3, Fig. 4 as can be known: at rotor under the normal and failure condition, although in experimentation, motor load approaches unloaded and does not do and regulate to try hard to keeping constant, but varied curve all comprises fluctuation to a certain degree during a phase stator current effective value, this explanation---for real electrical machinery, load fluctuation to a certain degree is inevitably, that is: in formula (5), the formula (6) I M1Fluctuate.Like this,
Figure 924430DEST_PATH_IMAGE002
,
Figure 215734DEST_PATH_IMAGE003
In DC component in also fluctuate, can't be by deducting mean value fully filtering.Consider that from the angle of FFT spectrum analysis this will cause DC component " leakage " and obscure the rotor bar breaking fault feature---2 Sf 1Component, 4 Sf 1Component affects the reliability that rotor bar breaking fault detects.The defective of current asynchronous motor rotor strip-broken fault Park Transformation detection method namely derives from this.
According to Fig. 5, Fig. 6 and table 1 as can be known: under normal circumstances, non-fault characteristic spectrum peak in the FFT frequency spectrum---2 Sf 1, 4 Sf 1Component, therefore correct judgement " rotor is normal "; But Fig. 7, Fig. 8 and table 2 show: under failure condition, and also non-fault characteristic spectrum peak in the FFT frequency spectrum---2 Sf 1, 4 Sf 1Component (this moment 2 Sf 10.33Hz, 4 Sf 10.66Hz), so false judgment " rotor is normal ".And comparison diagram 9, Figure 10 and Figure 11, Figure 12, and associative list 3, table 4 are as can be known: the normal or fault of rotor no matter, ESPRIT frequency spectrum all succinct, understand, and provide correct judgement.
And, existing Park Transformation method the FFT frequency spectrum (Fig. 5, Fig. 6, Fig. 7, Fig. 8) that obtains and the ESPRIT frequency spectrum (Fig. 9, Figure 10, Figure 11, Figure 12) that the present invention obtains are contrasted, can find---the FFT frequency spectrum is too complicated, the ESPRIT frequency spectrum is then succinct, clear, thereby can guarantee the reliability that rotor bar breaking fault detects.
In addition, according to Fig. 7, Fig. 8 and associative list 2, as can be known: for the motor signal that only continues 4s, existing Park Transformation method is the detection rotor broken bar fault correctly.Prolong the signals collecting time to 10s, the experimental result under the normal and failure condition of rotor is shown in Figure 13, Figure 14, table 5 and Figure 15, Figure 16, table 6.According to Figure 15, Figure 16, table 6 as can be known, in the rotor fault situation, existing Park Transformation method can provide correct testing result.But rotor has the error-detecting result that the Park Transformation method then provides " disconnected bar " now under normal circumstances, consults Figure 13, Figure 14 and table 5.And frequency spectrum shown in the FFT frequency spectrum (Figure 15, Figure 16) when the FFT frequency spectrum when rotor is normal (Figure 13, Figure 14) and rotor fault is too complicated, and difference is very little, can't judge accordingly that in fact normally whether rotor.Different therewith, frequency spectrum shown in the ESPRIT frequency spectrum (Figure 11, Figure 12) when the ESPRIT frequency spectrum when rotor is normal (Fig. 9, Figure 10) and rotor fault is then succinct, clear, and difference significantly, can judge reliably accordingly rotor fully normally or fault.This has proved absolutely the place of the present invention with respect to the advantage of existing Park Transformation method.
In summary, the Park Transformation method of current asynchronous motor rotor strip-broken fault detect is because there is defective in spectrum analysis based on FFT---and load fluctuation causes in the restituted signal DC component fluctuation, leaks, and causes the FFT frequency spectrum to be tending towards complicated and affect the reliability of rotor bar breaking fault detection.Owing to the FFT spectrum analysis is subject to the motor signal that frequency resolution needs enough durations, mean with more high probability introducing load fluctuation, cause the problems referred to above more outstanding.Particularly, for the large-sized asynchronous motor that extensively adopts in the engineering reality, above-mentioned Park Transformation method faces a severe challenge, even loses efficacy.The present invention introduces ESPRIT overcoming above-mentioned limitation, and only needs the short signal can high reliability ground detection rotor broken bar fault, is applicable to the large-sized asynchronous motor that extensively adopts in the engineering reality, and this is significant advantage of the present invention.
With respect to current Park Transformation method, the present invention has made significant improvement, is described as follows.
At first, current Park Transformation method is take the FFT spectrum analysis as the basis, and the present invention has then overturned this basis and adopted the ESPRIT spectrum analysis.
Secondly, current Park Transformation method is subject to load fluctuation (this is inevitable) in engineering reality based on the FFT spectrum analysis, DC component is leaked and frequency resolution, loses efficacy because of the motor signal that needs the enough durations of continuous collecting under asynchronous motor low revolutional slip ruuning situation; The present invention has then broken away from above-mentioned restriction based on the ESPRIT spectrum analysis, but also high reliability ground detection rotor broken bar fault under asynchronous motor low revolutional slip ruuning situation, thereby be applicable to the extensive large-sized asynchronous motor (low revolutional slip operation) that adopts in the engineering reality, possess Important Project value, broad prospect of application.
At last, current Park Transformation method object signal selected, that implement spectrum analysis has complicated frequency and forms, the present invention then utilizes Hilbert transform to define frequency to form simple, clear and definite object signal to implement spectrum analysis, and this has further guaranteed the reliability of asynchronous motor rotor strip-broken fault detect.
Signal M when table 1 rotor is normal d, M qThe FFT frequency spectrum data
Figure 76243DEST_PATH_IMAGE087
Signal M during table 2 rotor fault d, M qThe FFT frequency spectrum data
Figure 275143DEST_PATH_IMAGE088
Signal N when table 3 rotor is normal L, N RThe ESPRIT frequency spectrum data
Figure 6339DEST_PATH_IMAGE089
Signal N during table 4 rotor fault L, N RThe ESPRIT frequency spectrum data
Figure 999703DEST_PATH_IMAGE090
In the table 5 signals collecting duration 10s situation, signal M when rotor is normal d, M qThe FFT frequency spectrum data
Figure 222874DEST_PATH_IMAGE091
In the table 6 signals collecting duration 10s situation, signal M during rotor fault d, M qThe FFT frequency spectrum data
Figure 553361DEST_PATH_IMAGE092
。

Claims (5)

1. detect to a high reliability method of cage type asynchronous motor rotor strip-broken fault, it is characterized in that, threephase stator current signal application Park Transformation, its DC component of filtering that it at first gathers pressing certain frequency, and use Hilbert transform and simplify its frequency composition; Then use ESPRIT invariable rotary signal parameter estimation technique and carry out spectrum analysis, obtain the ESPRIT spectrogram; At last according to whether having 2 in the ESPRIT spectrogram Sf 1Component, 4 Sf 1Component spectrum peak judges that whether rotor bar breaking fault occurs.
2. detect to described high reliability according to claim 1 the method for cage type asynchronous motor rotor strip-broken fault, it is characterized in that it may further comprise the steps:
A. measure threephase stator electric current momentary signal i A, i B, i C
For high-voltage motor, adopt three current clamps to measure threephase stator electric current momentary signal in the Current Transformer Secondary side; For low voltage motor, adopt three current clamps directly to measure threephase stator electric current momentary signal at the connecting terminal of motor place;
B. adopt single power frequency period sliding window method to calculate the effective value of A phase stator current momentary signal
Figure 2012105031187100001DEST_PATH_IMAGE001
, by analyzing the variation tendency of effective value, extract its one piece of data the most steady that is fluctuation is minimum, be designated as i AS,
C. extract in B, the C phase stator current momentary signal with i ASBe in the data of same time section, be designated as i BS, i CS
D. calculate threephase stator electric current momentary signal i AS, i BS, i CSPark Transformation
Figure 86300DEST_PATH_IMAGE002
,
E. filtering
Figure 65757DEST_PATH_IMAGE002
In DC component, picked up signal M d, M d=
Figure 845495DEST_PATH_IMAGE002
-mean (
Figure 834179DEST_PATH_IMAGE002
), mean (
Figure 648551DEST_PATH_IMAGE002
) expression
Figure 939855DEST_PATH_IMAGE002
Mean value;
F. filtering
Figure 206889DEST_PATH_IMAGE003
In DC component, picked up signal M q, M q=
Figure 999264DEST_PATH_IMAGE003
-mean ( ), mean (
Figure 130348DEST_PATH_IMAGE003
) expression
Figure 884678DEST_PATH_IMAGE003
Mean value;
G. calculate M according to following formula qHilbert transform H (M q):
Figure 215165DEST_PATH_IMAGE004
，
H. construct a new signal N according to following formula L:
Figure 2012105031187100001DEST_PATH_IMAGE005
；
I. calculate M according to following formula dHilbert transform H (M d):
Figure 207392DEST_PATH_IMAGE006
；
J. construct a new signal N according to following formula R:
Figure 2012105031187100001DEST_PATH_IMAGE007
；
K. use ESPRIT respectively to signal N L, N RCarry out spectrum analysis, obtain 2 ESPRIT spectrograms;
L. according to whether having 2 in these 2 ESPRIT spectrograms Sf 1Component, 4 Sf 1Component spectrum peak judges that whether rotor bar breaking fault occurs: if having simultaneously 2 Sf 1Component, 4 Sf 1Component spectrum peak, rotor broken bar then, otherwise rotor is normal.
3. detect to described high reliability according to claim 2 the method for cage type asynchronous motor rotor strip-broken fault, it is characterized in that, adopt single power frequency period sliding window method to calculate the effective value of A phase stator current momentary signal
Figure 230711DEST_PATH_IMAGE001
Method be:
Choose A phase stator current momentary signal i AIn continuous 20 points, calculate its effective value
Figure 941178DEST_PATH_IMAGE008
For selected i AIn continuous 20 points, keep rear 19 points, the order fill vacancies in the proper order i AIn the back a bit, i.e. the 21st point, thus again obtain i AIn continuous 20 points, again calculate its effective value, by that analogy, determine i AThe effective value variation tendency.
4. detect to described high reliability according to claim 3 the method for cage type asynchronous motor rotor strip-broken fault, it is characterized in that described threephase stator electric current momentary signal i A, i B, i CSample frequency be set as 1000Hz, the sampling duration is set as 10s, the duration of the most stably one piece of data that therefrom extracts is 4s.
5. detect to described high reliability according to claim 4 the method for cage type asynchronous motor rotor strip-broken fault, it is characterized in that, application ESPRIT to the method that signal carries out spectrum analysis is:
A. construct correlation matrix
Figure 2012105031187100001DEST_PATH_IMAGE009
,
Figure 809777DEST_PATH_IMAGE010
, specific as follows;
Be without loss of generality, signal to be analyzed can be expressed as the combination of a series of cosine harmonics components, is shown below:
Figure DEST_PATH_IMAGE011
，
Wherein,
Figure 922090DEST_PATH_IMAGE012
The expression sampling period;
Figure DEST_PATH_IMAGE013
The expression sampling number; Expression harmonic wave number;
Figure DEST_PATH_IMAGE015
, ,
Figure DEST_PATH_IMAGE017
Represent respectively
Figure DEST_PATH_IMAGE019
The amplitude of individual harmonic wave, frequency, initial phase angle;
Herein, signal to be analyzed mainly comprises 2 frequency components, namely p≈ 2;
Definition
Figure 189626DEST_PATH_IMAGE020
, below the introducing
Figure DEST_PATH_IMAGE021
The rank matrix ( mShould much larger than p, get herein m=20):
Figure 281079DEST_PATH_IMAGE022
，
Figure DEST_PATH_IMAGE023
，
Then Autocorrelation matrix be;
Figure DEST_PATH_IMAGE025
，
And With Cross-correlation matrix be
Figure DEST_PATH_IMAGE027
，
In the formula,
Figure 362987DEST_PATH_IMAGE028
The expression mathematical expectation,
Figure DEST_PATH_IMAGE029
The expression conjugate transpose;
B pair
Figure 774377DEST_PATH_IMAGE009
Carry out Eigenvalues Decomposition, determine its minimal eigenvalue
Figure 212312DEST_PATH_IMAGE030
C calculates
Figure DEST_PATH_IMAGE031
,
Figure 226404DEST_PATH_IMAGE032
Expression
Figure DEST_PATH_IMAGE033
The rank unit matrix;
D calculates , For
Figure 874740DEST_PATH_IMAGE033
The rank battle array,
Figure 268812DEST_PATH_IMAGE036
, herein, Expression
Figure 2012105031187100001DEST_PATH_IMAGE037
The rank unit matrix;
E pair Carry out svd
Figure 2012105031187100001DEST_PATH_IMAGE039
, herein
Figure 901285DEST_PATH_IMAGE040
,
Figure 2012105031187100001DEST_PATH_IMAGE041
, in the formula, By
Figure 404127DEST_PATH_IMAGE014
Individual main singular value forms,
The f compute matrix
G pair
Figure 2012105031187100001DEST_PATH_IMAGE045
Carry out generalized eigenvalue decomposition, determine
Figure 837700DEST_PATH_IMAGE014
Individual generalized eigenvalue
Figure 737523DEST_PATH_IMAGE046
, all the other
Figure 2012105031187100001DEST_PATH_IMAGE047
Individual generalized eigenvalue is constantly equal to 0;
H determines the frequency of each component of sampled signal according to generalized eigenvalue
Figure 897109DEST_PATH_IMAGE048
, ,
Figure 667618DEST_PATH_IMAGE050
Difference representation feature value Imaginary part, real part;
The i compute matrix
Figure 621668DEST_PATH_IMAGE052
The j compute matrix
Figure DEST_PATH_IMAGE053
, herein
Figure 477629DEST_PATH_IMAGE054
It is a column vector , and Be column vector
Figure DEST_PATH_IMAGE057
K determines the amplitude of each component of sampled signal
Figure 393818DEST_PATH_IMAGE058
,
CN201210503118.7A 2012-11-30 2012-11-30 Method for detecting failure of rotor broken bar of cage type asynchronous motor at high reliability Expired - Fee Related CN103018669B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201210503118.7A CN103018669B (en) 2012-11-30 2012-11-30 Method for detecting failure of rotor broken bar of cage type asynchronous motor at high reliability

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201210503118.7A CN103018669B (en) 2012-11-30 2012-11-30 Method for detecting failure of rotor broken bar of cage type asynchronous motor at high reliability

Publications (2)

Publication Number Publication Date
CN103018669A true CN103018669A (en) 2013-04-03
CN103018669B CN103018669B (en) 2015-03-25

Family

ID=47967496

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201210503118.7A Expired - Fee Related CN103018669B (en) 2012-11-30 2012-11-30 Method for detecting failure of rotor broken bar of cage type asynchronous motor at high reliability

Country Status (1)

Country Link
CN (1) CN103018669B (en)

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103235260A (en) * 2013-04-15 2013-08-07 哈尔滨工业大学 Submersible motor rotor broken bar fault recognition method based on HHT (Hilbert-Huang transform)
CN104777396A (en) * 2015-03-12 2015-07-15 浙江大学 Rotor fault diagnosis method giving comprehensive consideration to current and voltage of motor
CN106646222A (en) * 2016-09-22 2017-05-10 华北电力大学 Method for diagnosing number of broken conducting bars of rotor of three-phase squirrel-cage asynchronous motor
CN106646224A (en) * 2016-09-22 2017-05-10 华北电力大学 Three-phase asynchronous motor load fluctuation detection method
CN108802523A (en) * 2018-05-31 2018-11-13 山西大学 Wind power generator unit frequency converter is monitored on-line and trouble-shooter
CN109031118A (en) * 2018-06-29 2018-12-18 珠海格力电器股份有限公司 A kind of Method of Motor Fault Diagnosis, device and motor
RU2724988C1 (en) * 2019-07-09 2020-06-29 федеральное государственное бюджетное образовательное учреждение высшего образования "Ивановский государственный энергетический университет имени В.И. Ленина" (ИГЭУ) Method of detecting broken rods in short-circuited winding of asynchronous motor rotor
CN111537881A (en) * 2020-05-26 2020-08-14 华润智慧能源有限公司 Fault diagnosis method, device and equipment for asynchronous motor and readable storage medium
CN111579990A (en) * 2020-05-28 2020-08-25 中国人民解放军国防科技大学 Induction motor rotor broken bar fault detection method based on stator current absolute phase
CN113678007A (en) * 2019-04-12 2021-11-19 三菱电机株式会社 Short circuit detection device and short circuit detection method
CN114784778A (en) * 2022-04-01 2022-07-22 福州大学 Flexible composite arc extinction method for power distribution network ground fault adapting to structure and load change
CN115144750A (en) * 2022-09-02 2022-10-04 北京科锐特科技有限公司 Asynchronous motor rotor broken bar fault detection method, device, equipment and medium

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101025430A (en) * 2007-03-28 2007-08-29 华北电力大学 Cage type asynchronous motor rotor strip-broken failure detecting method
CN102279341A (en) * 2011-07-23 2011-12-14 华北电力大学（保定） Cage asynchronous motor rotor broken-bar fault detection method based on electronic stability program rotation invariant technology (ESPRIT) and pattern search algorithm (PSA)
CN102279364A (en) * 2011-07-23 2011-12-14 华北电力大学（保定） Method for detecting broken bar fault of cage type asynchronous motor rotor based on multiple signal classification (MUSIC) and pattern search algorithm (PSA)
CN102279343A (en) * 2011-07-23 2011-12-14 华北电力大学（保定） Online detection method of broken rotor bar fault of cage-type asynchronous motor based on ESPRIT-SAA

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101025430A (en) * 2007-03-28 2007-08-29 华北电力大学 Cage type asynchronous motor rotor strip-broken failure detecting method
CN102279341A (en) * 2011-07-23 2011-12-14 华北电力大学（保定） Cage asynchronous motor rotor broken-bar fault detection method based on electronic stability program rotation invariant technology (ESPRIT) and pattern search algorithm (PSA)
CN102279364A (en) * 2011-07-23 2011-12-14 华北电力大学（保定） Method for detecting broken bar fault of cage type asynchronous motor rotor based on multiple signal classification (MUSIC) and pattern search algorithm (PSA)
CN102279343A (en) * 2011-07-23 2011-12-14 华北电力大学（保定） Online detection method of broken rotor bar fault of cage-type asynchronous motor based on ESPRIT-SAA

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
BEN SALEM, S: "Induction motor fault diagnosis based on a Hilbert current space vector pattern analysis", 《ELECTROTECHNICAL CONFERENCE (MELECON), 2012 16TH IEEE MEDITERRANEAN》 *
宋佳伟: "变频调速笼型异步电动机转子断条故障诊断", 《中国优秀硕士论文全文数据库（工程科技Ⅱ辑）》 *

Cited By (16)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN103235260A (en) * 2013-04-15 2013-08-07 哈尔滨工业大学 Submersible motor rotor broken bar fault recognition method based on HHT (Hilbert-Huang transform)
CN104777396A (en) * 2015-03-12 2015-07-15 浙江大学 Rotor fault diagnosis method giving comprehensive consideration to current and voltage of motor
CN104777396B (en) * 2015-03-12 2017-05-17 浙江大学 Rotor fault diagnosis method giving comprehensive consideration to current and voltage of motor
CN106646222A (en) * 2016-09-22 2017-05-10 华北电力大学 Method for diagnosing number of broken conducting bars of rotor of three-phase squirrel-cage asynchronous motor
CN106646224A (en) * 2016-09-22 2017-05-10 华北电力大学 Three-phase asynchronous motor load fluctuation detection method
CN106646222B (en) * 2016-09-22 2019-06-21 华北电力大学 A kind of three-phase cage asynchronous motor rotor broken splits the diagnostic method of conducting bar number
CN108802523A (en) * 2018-05-31 2018-11-13 山西大学 Wind power generator unit frequency converter is monitored on-line and trouble-shooter
CN109031118A (en) * 2018-06-29 2018-12-18 珠海格力电器股份有限公司 A kind of Method of Motor Fault Diagnosis, device and motor
CN113678007A (en) * 2019-04-12 2021-11-19 三菱电机株式会社 Short circuit detection device and short circuit detection method
CN113678007B (en) * 2019-04-12 2024-06-04 三菱电机株式会社 Short circuit detection device and short circuit detection method
RU2724988C1 (en) * 2019-07-09 2020-06-29 федеральное государственное бюджетное образовательное учреждение высшего образования "Ивановский государственный энергетический университет имени В.И. Ленина" (ИГЭУ) Method of detecting broken rods in short-circuited winding of asynchronous motor rotor
CN111537881A (en) * 2020-05-26 2020-08-14 华润智慧能源有限公司 Fault diagnosis method, device and equipment for asynchronous motor and readable storage medium
CN111579990A (en) * 2020-05-28 2020-08-25 中国人民解放军国防科技大学 Induction motor rotor broken bar fault detection method based on stator current absolute phase
CN114784778A (en) * 2022-04-01 2022-07-22 福州大学 Flexible composite arc extinction method for power distribution network ground fault adapting to structure and load change
CN115144750A (en) * 2022-09-02 2022-10-04 北京科锐特科技有限公司 Asynchronous motor rotor broken bar fault detection method, device, equipment and medium
CN115144750B (en) * 2022-09-02 2022-11-22 北京科锐特科技有限公司 Asynchronous motor rotor broken bar fault detection method, device, equipment and medium

Also Published As

Publication number Publication date
CN103018669B (en) 2015-03-25

Similar Documents

Publication Publication Date Title
CN103018669A (en) Method for detecting failure of rotor broken bar of cage type asynchronous motor at high reliability
CN102944842A (en) Detecting method for rotor broken bar fault of cage-type asynchronous motor
Xu et al. Improvement of the Hilbert method via ESPRIT for detecting rotor fault in induction motors at low slip
Gu et al. A new method of accurate broken rotor bar diagnosis based on modulation signal bispectrum analysis of motor current signals
Yang et al. An online technique for condition monitoring the induction generators used in wind and marine turbines
CN102279341B (en) Cage asynchronous motor rotor broken-bar fault detection method based on electronic stability program rotation invariant technology (ESPRIT) and pattern search algorithm (PSA)
CN102279364B (en) Method for detecting broken bar fault of cage type asynchronous motor rotor based on multiple signal classification (MUSIC) and pattern search algorithm (PSA)
Affijulla et al. Development of phasor estimation algorithm for P-class PMU suitable in protection applications
CN102279343B (en) Online detection method of broken rotor bar fault of cage-type asynchronous motor based on ESPRIT-SAA
CN101363901B (en) Method for detecting early failure of generator by enhancing transformations by electrical current characteristic
Hu et al. A novel transform demodulation algorithm for motor incipient fault detection
CN101915884A (en) Identification method of ground fault phases in three-phase ungrounded system and identification device thereof
CN104749435A (en) DFT harmonic detection method without accumulated error sliding window
CN109856538A (en) A kind of induction machine strip-broken failure detecting method based on correction FFT in short-term
Wang et al. Faulty feeder detection based on the integrated inner product under high impedance fault for small resistance to ground systems
Xie et al. A novel adaptive auto-reclosing scheme for transmission lines with shunt reactors
CN107677904A (en) A kind of voltage dip origin cause of formation discrimination method and system
Wang et al. Induction motor broken rotor bar fault diagnosis based on third-order energy operator demodulated current signal
CN104407293A (en) Hybrid particle swarm optimization based method for detecting broken bar faults of induction motors
CN107017808B (en) The continuous estimation method of synchronous motor rotor position based on pulsation exciting current response
Xu et al. Investigation of signal injection methods for fault detection of PMSM drives
CN201740840U (en) Earth fault phase identification device in three-phase unearthed system
CN112255495B (en) Micro-grid high-resistance fault detection method
Mariun et al. An experimental study of induction motor current signature analysis techniques for incipient broken rotor bar detection
CN107271830B (en) Method for rapidly calculating transformation ratio of special transformer in unbalanced state

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
C14 Grant of patent or utility model
GR01 Patent grant
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20150325

Termination date: 20151130

EXPY Termination of patent right or utility model