CN101451920B - All-band characteristic prewarning method for mechanical vibration signal - Google Patents

Abstract

The invention discloses an early-warning method for mechanical vibration signal full frequency band characteristic, including acquisition multiple sets of the mechanical vibration signals, calculation of components of each weight of the vibration signals, and obtainment of a plurality of sets of vibration amplitudes and frequency value sequence sets, namely, for specific fi frequency, a plurality of sets of vibration amplitudes can be obtained, an amplitude early-warning value for the specific fi frequency can be obtained by calculating average value, deviation value of the plurality of sets of vibration amplitudes; for the plurality of sets of fi frequencies, an early-warning line for the mechanical vibration signal full frequency band characteristic is drawn on an original vibration signal spectrogram by manufacturing the spectrogram, in order to determine early warning for the mechanical vibration signal full frequency band characteristic. Hence, the method can realize early warning for the mechanical vibration signal full frequency band characteristic even when certain frequency component vibration amplitude exceeds standard to early warn.

Description

All-band characteristic prewarning method for mechanical vibration signal

Technical field

The present invention relates to the method for machine performance monitoring and diagnosis, relate in particular to a kind of method of mechanical oscillation signal all-band characteristic prewarning.

Background technology

At mechanical field, 60%～70% mechanical fault is arranged according to statistics all is by vibration and have the noise of vibration radiation to reflect, therefore, for other plant equipment Condition Monitoring Technology, whether in allowed limits and be in which rank vibration monitoring is simple and easy to do with diagnosis, normal adopt the actual measurement vibration level and differentiate this vibration.

Judge whether plant equipment exists the degree of fault and fault, must measured value and the specified standard value that characterize machine state be compared.A large amount of example explanations exists special association between the mechanical fault of vibration value and machine, and vibration values does not show that greatly fault must occur.Will inevitably in vibration, reflect to some extent when on the one hand, machine breaks down; On the other hand, vibration may not necessarily be broken down again to the machine of certain value greatly.In view of the situation, the former becomes the benchmark of vibration standard, and the latter becomes the difficult point of standard formulation again.

In ISO mechanical vibration standard, adopt vibration signal entire spectrum global vibration value OA to characterize the vibration of machine mostly,

$\mathrm{OA}=\frac{\sqrt{\underset{i=1}{\overset{n}{\mathrm{\Σ}}}{A}_i^2}}{\sqrt{N_{\mathrm{BF}}}}=\frac{\sqrt{A_1^2+A_2^2+\·\·\·+A_n^2}}{\sqrt{1.5}}$

OA=vibration signal full range global vibration value n=FFT resolution spectral line number wherein

A _i=every FFT spectral line amplitude N _BF=Hanning window function grass parameter

And stipulated the vibration early warning threshold value of all kinds of machines, its absolute vibration, axle of only having stipulated to estimate axle relatively and the vibration of structure member and the vibration of bearing seat and is only assessed the wide-band vibration value, does not have the evaluation criterion of frequency component and phase place; Its interpretational criteria is the amplitude of wide-band vibration, and another is the variation of considering amplitude, increases or reduces no matter be.

In engineering practice was used, the method for early warning of above-mentioned standard code had been obtained big effect, but has also been had certain deficiency in machine diagnostic, mainly contain:

Although certain frequency content vibration amplitude may exceed tolerance interval in 1 vibration signal, the global vibration value does not exceed standard, and early warning in time causes and fails to report, and is insensitive to initial failure.

Even 2 global vibrations exceed standard, can't know that also those concrete frequency content component amplitudes exceed standard, the disappearance diagnostic message causes accurately fault location position and reason.

3, former early warning belongs to the early warning of entire spectrum global vibration monodrome, can't realize vibration signal full range band arbitrfary point or the numerous points early warning is set.

Summary of the invention

Technical matters to be solved by this invention provides a kind of all-band characteristic prewarning method for mechanical vibration signal, and it can carry out the timely early warning that the global vibration value exceeds standard, and also can realize the timely early warning that certain frequency content vibration amplitude exceeds standard.

For solving the problems of the technologies described above, all-band characteristic prewarning method for mechanical vibration signal of the present invention comprises the steps,

Step 1, vibration signal pretreatment, the mechanical oscillation signal of collecting device, the mechanical vibration discrete signal of establishing collection is: x (n Δ t), n=1 wherein, 2....N, Δ t are sampling time interval, by the Fast Fourier Transform (FFT) formula $S\left(f\right)=\mathrm{\Δt}\underset{n=-\∞}{\overset{\∞}{\mathrm{\Σ}}}x\left(\mathrm{n\Δt}\right)e^{-j2\mathrm{\πf\Δt}}$

Calculate each frequency amplitude of vibration signal, and obtain

Group (A _i, f _i) vibration amplitude and frequency values;

Step 2, rumble spectrum data processing to an equipment, are gathered the n group vibration signal of its running status, form the n group length by said method and are

(A _i, f _i) vibration amplitude and frequency values sequence sets, promptly for specific f _iFrequency can obtain n group A _I1, A _I2, A _I3... .A _InVibration amplitude;

Step 3, employing statistical method draw f _iThe amplitude early warning value of frequency,

Calculate vibration amplitude mean value

${\stackrel{\‾}{M}}_i=\frac{A_{\mathrm{il}}+A_{i2}+A_{i3}+A_{i4}+\·\·\·\·\·\·+A_{\mathrm{in}}}{n}$

Calculate vibration amplitude and mean value Standard deviation δ _i:

${\mathrm{\δ}}_i=\sqrt{\frac{{({\mathrm{Ai}}_i-{\stackrel{\‾}{M}}_i)}^2+{(A_{i2}-{\stackrel{\‾}{M}}_i)}^2+\·\·\·+{(A_{\mathrm{in}}-{\stackrel{\‾}{M}}_i)}^2}{n-1}}$

Determine to judge f _iThe amplitude early warning value M of frequency _Ia:

$M_{\mathrm{ia}}={\stackrel{\‾}{M}}_i+(1~2){\mathrm{\δ}}_i$

Step 4, mechanical oscillation signal all-band characteristic prewarning,

Method according to step 3 is right

Group (f _i) all just calculate M separately _Ia, and make spectrogram, on the original vibration signal spectrogram, draw out mechanical oscillation signal all-band characteristic prewarning line;

To the machine vibration signal of new collection, calculate separately frequency values and respective magnitudes by step 1, promptly Group (A _i, f _i) vibration amplitude and frequency values, the early warning value (M that its value calculates with step 3 _Ia, f _i) relatively, exceed the all-band characteristic prewarning line, realize early warning.

Because all-band characteristic prewarning method for mechanical vibration signal of the present invention has adopted technique scheme, it is the mechanical oscillation signal of collecting device, and this mechanical oscillation signal carried out particular processing, calculate each frequency amplitude of vibration signal, form the frequency values sequence sets, calculate for specific f _iThe amplitude early warning value time, also can draw the mechanical oscillation signal all-band characteristic prewarning under the frequency by the making of spectrogram.

Description of drawings

The present invention is described in further detail below in conjunction with drawings and embodiments:

Fig. 1 is the spectrogram of each frequency amplitude of vibration signal of the present invention,

Fig. 2 is an all-band characteristic prewarning spectrogram of the present invention.

Embodiment

All-band characteristic prewarning method for mechanical vibration signal of the present invention comprises the steps,

Step 1, vibration signal pretreatment, the mechanical oscillation signal of collecting device, the mechanical vibration discrete signal of establishing collection is: x (n Δ t), n=1 wherein, 2....N, Δ t are sampling time interval, by the Fast Fourier Transform (FFT) formula $S\left(f\right)=\mathrm{\Δt}\underset{n=-\∞}{\overset{\∞}{\mathrm{\Σ}}}x\left(\mathrm{n\Δt}\right)e^{-j2\mathrm{\πf\Δt}}$

As shown in Figure 1, calculate each frequency amplitude of vibration signal, and obtain

Group (A _i, f _i) vibration amplitude and frequency values;

Step 2, rumble spectrum data processing to an equipment, are gathered the n group vibration signal of its running status, form the n group length by said method and are

(A _i, f _i) vibration amplitude and frequency values sequence sets, promptly for specific f _iFrequency can obtain n group A _I1, A _I2, A _I3... .A _InVibration amplitude;

Step 3, employing statistical method draw f _iThe amplitude early warning value of frequency,

Calculate vibration amplitude mean value

${\stackrel{\‾}{M}}_i=\frac{A_{i1}+A_{i2}+A_{i3}+A_{i4}+\·\·\·\·\·\·+A_{\mathrm{in}}}{n}$

Calculate vibration amplitude and mean value

Standard deviation δ _i:

${\mathrm{\δ}}_i=\sqrt{\frac{{({\mathrm{Ai}}_1-{\stackrel{\‾}{M}}_i)}^2+{(A_{i2}-{\stackrel{\‾}{M}}_i)}^2+\·\·\·+{(A_{\mathrm{in}}-{\stackrel{\‾}{M}}_i)}^2}{n-1}}$

Determine to judge f _iThe amplitude early warning value M of frequency _Ia:

$M_{\mathrm{ia}}={\stackrel{\‾}{M}}_i+(1~2){\mathrm{\δ}}_i$

Step 4, mechanical oscillation signal all-band characteristic prewarning,

Method according to step 3 is right

Group (f _i) all just calculate M separately _Ia, as shown in Figure 2, make spectrogram, on the original vibration signal spectrogram, draw out mechanical oscillation signal all-band characteristic prewarning line.

To the machine vibration signal of new collection, calculate separately frequency values and respective magnitudes by step 1, promptly

Group (A _i, f _i) vibration amplitude and frequency values, the early warning value (M that its value calculates with step 3 _Ia, f _i) relatively, exceed the all-band characteristic prewarning line, realize early warning.

In accompanying drawing 1 and the accompanying drawing 2, X-axis is frequency (Hz) or rotating speed (RPM), and its span is [0,1/ (2 Δ t)], and Δ t is a sampling interval; Y-axis is displacement (um) or speed (mm/s) or acceleration (mm/s ²), also can be its db decibel value.

Claims (1)

1. all-band characteristic prewarning method for mechanical vibration signal is characterized in that: this method comprises the steps,

Step 1, vibration signal pretreatment, the mechanical oscillation signal of collecting device, the mechanical vibration discrete signal of establishing collection is: x (n Δ t), n=1 wherein, 2....N, Δ t is a sampling time interval, selects suitable window function, by the Fast Fourier Transform (FFT) formula

Calculate each frequency amplitude of vibration signal, and obtain

Group vibration amplitude A _iWith frequency values f _i

Step 2, rumble spectrum data processing to an equipment, are gathered the n group vibration signal of its running status, form the n group length by first step and are

Vibration amplitude A _iWith frequency values f _iSequence sets is promptly for specific frequency values f _i, can obtain n group A _I1, A _I2, A _I3... .A _InVibration amplitude;

Step 3, employing statistical method draw frequency values f _iThe amplitude early warning value,

Calculate vibration amplitude mean value

$\stackrel{\‾}{M_i}=\frac{A_{i1}+A_{i2}+A_{i3}+A_{i4}+\·\·\·\·\·\·+A_{\mathrm{in}}}{n}$

Calculate vibration amplitude and vibration amplitude mean value

Standard deviation δ _i:

${\mathrm{\δ}}_i=\sqrt{\frac{{({\mathrm{Ai}}_1-\stackrel{\‾}{M_i})}^2+{(A_{i2}-\stackrel{\‾}{M_i})}^2+\·\·\·+{(A_{\mathrm{in}}-\stackrel{\‾}{M_i})}^2}{n-1}}$

Determine determination frequency value f _iAmplitude early warning value M _Ia:

$M_{\mathrm{ia}}=\stackrel{\‾}{M_i}+(1~2){\mathrm{\δ}}_i$

Step 4, mechanical oscillation signal all-band characteristic prewarning,

Method according to step 3 is right

Class frequency value f _iAll just calculate amplitude early warning value M separately _Ia, and make spectrogram, on the original vibration signal spectrogram, draw out mechanical oscillation signal all-band characteristic prewarning line;

To the machine vibration signal of new collection, calculate separately frequency values and corresponding vibration amplitude by step 1, promptly

Group vibration amplitude A _iWith frequency values f _i, the amplitude early warning value M that its value calculates with step 3 _IaWith frequency values f _iRelatively, exceed the all-band characteristic prewarning line, realize early warning.