CN106525427A - Direct current brushless motor bearing fault diagnosis method under variable rotational speed working condition - Google Patents

Abstract

The invention discloses a method for direct current brushless motor bearing fault diagnosis under a variable rotational speed working condition. The method includes the steps of (1) using a dual-channel data acquisition system to synchronously acquire a phase current signal and a bearing vibration signal of a direct current brushless motor; (2) adopting a zero-phase filter to filter a current signal, and adopting Hilbert conversion to calculate the phase of a filtering signal; (3) cutting off and aligning the current signal, the vibration signal and a current phase signal to reduce an endpoint error, and converting the current phase signal into a motor rotor mechanical angle signal; and (4) using the motor rotor mechanical angle signal to resample the bearing vibration signal, calculating an envelope order spectrum of the resampled vibration signal, and judging the bearing fault type according to the fault feature order. The method for direct current brushless motor bearing fault diagnosis under the variable rotational speed working condition has the advantages of simple principle, small calculation amount, high precision, and capability of realizing accurate diagnosis under the variable rotational speed working condition of a direct current brushless moor bearing.

Description

A kind of DC brushless motor Method for Bearing Fault Diagnosis under variable speed operating mode

Technical field

The present invention relates to motor bearings fault diagnosis technology field, and in particular to the brush DC under a kind of variable speed operating mode Motor bearings method for diagnosing faults.

Background technology

DC brushless motor instead of the mechanical commutation of traditional brushed DC motor using electronic commutation, thus reduce by The problems such as the spark, abrasion, noise that Mechanical Contact is produced.Therefore brshless DC motor has simple structure, and low noise is high to turn The advantages of speed, high reliability and long service live and be widely used in industry and civil area.DC brushless motor adopts six Step unidirectional driving, in order to realize correct six steps commutation sequential, needs to perceive the corner phase information of rotor.Corner information can be with Obtained by Hall element or from the counter electromotive force of the cold phase of motor estimating, the latter is also referred to as sensorless drive Technology.The commutation of six steps only needs to accurately perceive 60 degree of electric phase angles and its multiple, and need not perceive its within 0 to 360 degree Its phase angle.

Because instead of mechanical commutation using six step electronic commutations, thus brshless DC motor frame for movement be simple.Especially It is the motor for sensorless drive, only stator, rotor, motor housing, the several parts of bearing.During bearing is brushless electric machine When uniquely having the part of Mechanical Contact, i.e. motor to rotate, bearing roller is rolled between bearing inner race and outer ring.Therefore bearing event Barrier is one of major failure of brshless DC motor.For the bearing failure diagnosis of constant speed rotary, there is the skill of a large amount of maturations Art can be applied.But if motor is run under variable speed operating mode, bearing failure diagnosis will become relative difficulty.Order tracking technique Technology by carrying out angular domain resampling to time-domain signal, the bearing failure diagnosis that can be effectively realized under variable speed operating mode.

But order tracking technique technology needs the accurate instantaneous phase within 0 to 360 degree, and integrated brushless electric machine control is System either adopts Hall element or sensorless drive technology all provide required phase information.Thus brushless electric machine Diagnosis of the bearing in the case of variable working condition is with certain technological difficulties.Although the phase place of bearing can also by vibration or The analysis of acoustical signal is extracting, but these extracting method generally need the time-frequency analysis technology of complexity, while easily being made an uproar Sound interference, it is impossible to be advantageously applied in actual industrial scene, so prior art has limitation.How improved method, carries Brushless electric machine bearing failure diagnosis precision and efficiency under high variable speed operating mode, it is still necessary to further inquire into.

The content of the invention

In order to solve problems of the prior art, present invention aim at providing the direct current under a kind of variable speed operating mode Brushless electric machine Method for Bearing Fault Diagnosis.

The technical solution used in the present invention is：A kind of DC brushless motor bearing failure diagnosis side under variable speed operating mode Method, the method include four implementation steps：

Step (1), phase current signal and bearing that DC brushless motor is gathered using double channel data acquisition system synchronization Vibration signal, current signal representation are C [n], and n=0,1 ..., L-1, vibration signal are expressed as：V [n], n=0,1 ..., L-1, Wherein L=T × f_sFor data length, T and f_sRespectively sampling time length and sample frequency.

Step (2), current signal is filtered using zero-phase filters, initially with an infinite-duration impulse response filter Ripple device is filtered to C [n] and obtains an interim filtering signal FC₁[n], is shown below：

A in formula_j,b_g, N, M are the filter parameter depending on factors such as filter type, order, bandwidth.Subsequently to FC₁ [n] carries out reversion and obtains FC₂[n], is shown below：

FC₂[n]=FC₁[L-n+1], n=0,1 ..., L-1

Then using above-mentioned same filter to FC₂[n] is filtered and obtains FC₃[n], is shown below：

Again to FC₃[n] is inverted the filtering signal FC [n] that zero phase is obtained, and is shown below：

FC [n]=FC₃[L-n+1], n=0,1 ..., L-1

By with last forward filtering, an inverse filtering, the phase distortion of the zero-phase filtering signal for obtaining are use up Possibly reduce.The phase place of filtering signal FC [n] is calculated using Hilbert transform subsequently.Discrete Fu of FC [n] is calculated first In leaf transformation obtain X [k]：

Subsequently monolateral discrete time analytic signal conversion Z [m] of construction one, is shown below：

The inversefouriertransform for calculating Z [m] is obtained multiple discrete time analytic signal z [n] of FC [n], such as following formula institute Show：

In formula, Re (z [n]) and Im (z [n]) is respectively the real part and imaginary part of z [n].Subsequently, phase place arg [n] of z [n] can There is following formula to obtain：

Arg [n]=atan2 (Im (z [n]), Re (z [n])), n=1,2 ..., L-1

In formula, atan2 is the arctan function of two-parameter form.As the scope of arg [n] is that-π arrives π, arg [n] is done Solution winding computing obtains continuous current phase curve angle [n], is shown below：

In formula, K₁,K₂,…,K_pRespectively the 1st, the 2nd ..., p-th phase hit point.

Step (3), there is obvious end point error near n=0 and n=L-1 due to current phase curve angle [n], Further to current signal, vibration signal, current phase signal are blocked and are alignd to reduce end point error, while by electricity Stream phase signal is changed into rotor mechanical angle signal, is shown below：

C_T[n]=C [n+0.1L], n=0,1 ..., 0.8L-1

V_T[n]=V [n+0.1L], n=0,1 ..., 0.8L-1

N in formula_pFor the number of poles of DC brushless motor, C_T[n],angle_T[n],V_T[n] is respectively the electricity after blocking and aliging Stream signal, rotor mechanical angle signal, bearing vibration signal.

Step (4), using rotor mechanical angle signal angle_T[n] is to bearing vibration signal V_T[n] is adopted again Sample, carries out bandpass filtering according to resonant belt position to resampling signal, and subsequently filtering signal is calculated using Hilbert transform Envelope order spectrum is obtained, bearing fault type is judged according to fault signature order.

Advantages of the present invention and good effect are：

(1) the inventive method is filtered to brshless DC motor phase current signal using zero-phase filters, can be most Big degree ground reduces phase distortion, additionally, after phase place is calculated using Hilbert transform and winding is solved, blocking to signal And alignment, end point error is reduced, the estimated accuracy of rotor mechanical angle is finally improved.

(2) the inventive method obtains accurate rotor mechanical corner by the phase current of analysis brshless DC motor Information, and electric current can be measured by the non-contacting sensor based on Hall effect, thus original motor need not be changed System, without installing extra encoder additional.

(3) the rotor mechanical angle obtained using the inventive method is entered to the time domain bearing vibration signal of synchronous acquisition Time-domain signal is converted into angular domain signal, then the envelope order spectrum for calculating angular domain bearing vibration signal by row resampling, it is possible to obtain Bearing fault order information, realizes the brushless electric machine bearing failure diagnosis under variable speed operating mode.

Description of the drawings

Fig. 1 is the inventive method flowchart；

Fig. 2 is phase current signal of the brshless DC motor under variable speed operating mode；

Amplitude spectrums of the Fig. 3 for phase current；

Fig. 4 is using the filtered current signal obtained after the inventive method filtering；

Fig. 5 is using the calculated rotor mechanical angle signal of the inventive method；

Fig. 6 is the error and average angle error of the calculating angle and real angle before not blocking and aliging；

Fig. 7 is that the error of the calculating angle and real angle after being blocked and alignd using the inventive method and average angle are missed Difference；

Fig. 8 is bearing outer ring fault vibration signal and envelope spectrum under variable speed operating mode；

Fig. 9 is electric machine phase current signal corresponding with Fig. 8 and frequency spectrum；

Figure 10 is using the order letter obtained after the bearing outer ring fault-signal under the inventive method process variable speed operating mode Number and envelope order spectrum；

Figure 11 is bearing inner race fault vibration signal and envelope spectrum under variable speed operating mode；

Figure 12 is electric machine phase current signal corresponding with Figure 11 and frequency spectrum；

Figure 13 is using the order letter obtained after the bearing inner race fault-signal under the inventive method process variable speed operating mode Number and envelope order spectrum.

Specific embodiment

Below in conjunction with the accompanying drawings and specific embodiment further illustrates the present invention.

Embodiment one：

The inventive method is first verified that for extracting the effectiveness of motor rotatable phase information.Accompanying drawing 2 is brushless dc Phase current signal of the machine under variable speed operating mode, due to the change of motor speed, the current signal has obvious amplitude and frequency Rate modulation signature, while the pulsewidth modulation feature that visible electrical motor gearshift drives, and noise jamming is more obvious.Accompanying drawing 3 is the electricity The amplitude spectrum of stream signal, the frequency component energy from frequency spectrum between visible 20Hz to 50Hz are higher.Realization in 1 with reference to the accompanying drawings The step of flow chart 2, current signal is entered using the second order Butterworth zero phase band filter with a width of 18Hz to 52Hz Row filtering, as a result as shown in Figure 4.After after filtering, originally complicated current signal becomes near sinusoidal signal, but still Remain amplitude and frequency modulation(PFM) feature.The phase place of filtering signal is calculated using Hilbert transform subsequently, and according to motor Current phase is converted into rotor mechanical angle by number of poles, as a result as shown in Figure 5.In figure it is shown that curve and it is non-straight Line, illustrates that motor is run under variable speed operating mode.For the accuracy of the instantaneous corner of the motor for verifying extraction, in the shaft end of motor The each revolution of photoelectric encoder of 200 pulses is installed, using the instantaneous corner measured from encoder as reference signal, present invention side is calculated Instantaneous angular signal and the instantaneous angular error of reference signal, mean error and root-mean-square error that method is extracted, such as 6 institute of accompanying drawing Show.Within 0.1 second to 0.9 second, angular error is fluctuated up and down in mean error line, but in the interval of 0 to 0.1 second, has one Individual obvious angle saltus step, similarly, also had an angle saltus step within 0.9 second to 1 second.The two angle saltus steps are end points Error, so as to cause root-mean-square error to reach 18.56 °, end point error can affect follow-up bearing signal resampling and failure Diagnostic accuracy.In order to reduce end point error impact, signal is carried out blocking using the step 3 in flowchart in accompanying drawing 1 and Alignment, as a result as shown in Figure 7.It can be seen that end point error is eliminated well, root-mean-square error is also reduced to 1.89°.Resolution 1.8 ° of the error amount already close to photoelectric encoder, this result illustrate that the motor that the inventive method is extracted turns Angle has higher precision, so as to ensure that follow-up bearing signal resampling and fault diagnosis required precision.

Embodiment two：

Above example demonstrates the effectiveness that the inventive method estimates the instantaneous corner of motor, and the present embodiment is further verified Effectiveness of the inventive method in brshless DC motor bearing failure diagnosis application.Bearing to be detected is arranged on brshless DC motor Drive end, bearing fault adopts linear cutter, and its width and depth are respectively 0.18mm and 2mm, bearing designation and failure Feature order is as shown in the table：

1. bearing parameter to be detected of table

First outer ring faulty bearings are detected, variable speed bearing signal and its envelope range value are composed as shown in Figure 8, from Visible abundant frequency component in spectrogram, but due to noise jamming and rotation speed change, it is impossible to identify the fault signature of bearing Frequency, thus cannot also determine the fault type of bearing.The current signal and its amplitude spectrum of data collection system synchronizing collection is such as Shown in accompanying drawing 9.Step 2 in flowchart in 1 with reference to the accompanying drawings, using the second order Butterworth band of 20Hz to 40Hz bandwidth Bandpass filter is filtered to current signal, is carried out phase extraction to filtering signal and is converted to rotor mechanical angle, with After carry out signal cutout and alignment, and the bearing outer ring failure of time domain is believed with the rotor mechanical angle curve for finally giving Number carry out angular domain resampling and calculate envelope order spectrum, as a result as shown in Figure 10.Clearly can recognize from envelope order spectrum The outer ring fault signature order FCO of bearing_OAnd its frequency multiplication such that it is able to judge that motor bearings has outer ring failure.Contrast accompanying drawing 8 Understand that the inventive method extracts motor rotation angle information from current of electric, so as to pass through to adopt again with the frequency spectrum in accompanying drawing 10 Sample realizes that time domain bearing fault signal, to the transformation of angular domain signal, finally realizes the fault diagnosis of motor bearings well, Demonstrate the effectiveness of the inventive method.

Embodiment three：

Similarly, inner ring faulty bearings are analyzed, its result is as shown in accompanying drawing 11 to accompanying drawing 13.Wherein with reference to the accompanying drawings Current spectrum distribution in 10, second order Butterworth bandpass filtering of the zero-phase filters for adopting for bandwidth 30Hz to 55Hz Device.The failure-frequency of bearing cannot be accurately recognized from the envelope spectrum of 11 middle (center) bearing of accompanying drawing.It is processed through the inventive method Afterwards, bearing inner race fault signature order FCO can clearly be found from the envelope order spectrum of the resampling signal of accompanying drawing 13_IAnd its Frequency multiplication, thus confirm that bearing has inner ring failure.Three above embodiment demonstrates the inventive method for variable speed operating mode Under brshless DC motor corner extract and motor bearings fault diagnosis effectiveness.

The content not being described in detail in description of the invention belongs to prior art known to professional and technical personnel in the field.

Although disclosing embodiments of the invention and accompanying drawing for the purpose of illustration, those skilled in the art can manage Solution：Without departing from the spirit and scope of the invention and the appended claims, various replacements, to change and modifications all be possible 's.Therefore, the present invention should not be limited to the embodiment of the present invention and accompanying drawing disclosure of that.

Claims (1)

1. the DC brushless motor Method for Bearing Fault Diagnosis under a kind of variable speed operating mode, it is characterised in that the method include as Lower step：

Step (1), phase current signal and bear vibration that DC brushless motor is gathered using double channel data acquisition system synchronization Signal, current signal representation are C [n], and n=0,1 ..., L-1, vibration signal are expressed as：V [n], n=0, wherein 1 ..., L-1, L =T × f_sFor data length, T and f_sRespectively sampling time length and sample frequency；

Step (2), current signal is filtered using zero-phase filters, initially with an infinite impulse response filter An interim filtering signal FC is filtered and is obtained to C [n]₁[n], is shown below：

$a_0=1,{\mathrm{FC}}_1\[n\]=-\underset{j=1}{\overset{N}{\Σ}}{a}_j{\mathrm{FC}}_1\[n-j\]+\underset{g=0}{\overset{M}{\Σ}}{b}_{g}C\[n-g\],n=0,1,...,L-1$

A in formula_j,b_g, N, M are the filter parameter depending on factors such as filter type, order, bandwidth；Subsequently to FC₁[n] enters Row reversion obtains FC₂[n], is shown below：

FC₂[n]=FC₁[L-n+1], n=0,1 ..., L-1

Then using above-mentioned same filter to FC₂[n] is filtered and obtains FC₃[n], is shown below：

$a_0=1,{\mathrm{FC}}_3\[n\]=-\underset{j=1}{\overset{N}{\Σ}}{a}_j{\mathrm{FC}}_3\[n-j\]+\underset{g=0}{\overset{M}{\Σ}}{b}_g{\mathrm{FC}}_2\[n-g\],n=0,1,...,L-1$

Again to FC₃[n] is inverted the filtering signal FC [n] that zero phase is obtained, and is shown below：

FC [n]=FC₃[L-n+1], n=0,1 ..., L-1

By with last forward filtering, an inverse filtering, the phase distortion of the zero-phase filtering signal for obtaining is by as far as possible Ground is reduced；The phase place of filtering signal FC [n] is subsequently calculated using Hilbert transform, the discrete fourier of FC [n] is calculated first Conversion obtains X [k]：

$X\[k\]=\underset{n=0}{\overset{L-1}{\Σ}}FC\[n\]\exp (-j2\mathrm{\π}kn/L),k=0,1,...,L-1$

Subsequently monolateral discrete time analytic signal conversion Z [m] of construction one, is shown below：

$Z\[m\]=\left\{\begin{array}{cc}X\[0\],& m=0\\ 2X\[m\],& 1\≤m\≤L/2-1\\ X\[L/2\],& m=L/2\\ 0,& L/2+1\≤m\≤L-1\end{array}\right.$

The inversefouriertransform for calculating Z [m] is obtained multiple discrete time analytic signal z [n] of FC [n], is shown below：

$\begin{array}{c}z\[n\]=\frac{1}{L}\underset{k=0}{\overset{L-1}{\Σ}}Z\[k\]\exp (j2\mathrm{\π}kn/L)\\ =\mathrm{Re}(z\[n\])+\mathrm{Im}(zp\[n\]),n=0,1,\mathrm{...},L-1\end{array}$

In formula, Re (z [n]) and Im (z [n]) is respectively the real part and imaginary part of z [n].Subsequently, under phase place arg [n] of z [n] can have Formula is obtained：

Arg [n]=atan2 (Im (z [n]), Re (z [n])), n=1,2 ..., L-1

In formula, atan2 is the arctan function of two-parameter form, as the scope of arg [n] is that-π arrives π, does uncoiling to arg [n] Continuous current phase curve angle [n] is obtained around computing, is shown below：

$angle\&lsqb;n\&rsqb;=\left\{\begin{array}{cc}\arg \&lsqb;n\&rsqb;,& 0\&le;n<K_1\\ \arg \&lsqb;n\&rsqb;+2\mathrm{\&pi;},& K_1\&le;n<K_2\\ \mathrm{...}& \\ \arg \&lsqb;n\&rsqb;+2\mathrm{\&pi;}\&times;p,& K_p\&le;n\&le;L-1\end{array}\right.$

In formula, K₁,K₂,…,K_pRespectively the 1st, the 2nd ..., p-th phase hit point；

Step (3), because there is obvious end point error near n=0 and n=L-1 in current phase curve angle [n], enter one To current signal, vibration signal, current phase signal are blocked and are alignd to reduce end point error on step ground, while by electric current phase Position signal is changed into rotor mechanical angle signal, is shown below：

C_T[n]=C [n+0.1L], n=0,1 ..., 0.8L-1

V_T[n]=V [n+0.1L], n=0,1 ..., 0.8L-1

N in formula_pFor the number of poles of DC brushless motor, C_T[n],angle_T[n],V_T[n] is respectively the letter of the electric current after blocking and aliging Number, rotor mechanical angle signal, bearing vibration signal；

Step (4), using rotor mechanical angle signal angle_T[n] is to bearing vibration signal V_T[n] carries out resampling, right Resampling signal carries out bandpass filtering according to resonant belt position, subsequently filtering signal is calculated using Hilbert transform and is wrapped Network order is composed, and judges bearing fault type according to fault signature order.