CN101782475B - Blade fault diagnosing method based on vibration of wind generating set - Google Patents

Abstract

The invention relates to a blade fault diagnosing method based on vibration of a wind generating set, which comprises the following steps of: (1) making a power spectrum of a vibratory response signal with the frequency range f of not more than 1,000Hz in the running state of the wind generating set; extracting a front M-order low-frequency power spectrum corresponding to a wind generating set blade from the power spectrum of the wind generating set by using a filtering method; (2) obtaining a wind generating set blade running state characteristic value of the front M-order low-frequency power spectrum of the wind generating set blade by using a correlation dimension or a length fractal dimension or a box dimension or a frequency translation distance method; (3) comparing the wind generating set blade running state characteristic value with a wind generating set blade normal state characteristic value to obtain a blade fault diagnosing result of the wind generating set. The diagnosing method is simple and easy, has higher sensitivity and precision and lower diagnosing cost and is the blade fault diagnosing method based on the vibration of the wind generating set, which can effectively improve the safety and the reliability of a blade of the wind generating set.

Description

Blade fault diagnosing method based on vibration of wind generating set

Technical field

The invention belongs to the wind power generating set field, especially a kind of blade fault diagnosing method based on vibration of wind generating set.

Background technology

Along with expanding economy, people are more and more to the demand of electric power energy, and for inexhaustible, nexhaustible wind-power electricity generation, the emphasis of clean electric power energy development especially, supply with in order better to realize electric power energy, wind power generating set is towards maximization, the ocean development, and the blade of the wind power generating set that adapts is with it also encouraging complicated development to maximization and stand under load, in order better to realize the safety supply of electric power energy, the monitoring of the wind power generating set link that is absolutely necessary, especially to the blade fault diagnosing of wind power generating set, analyze blade fault mainly by two big class reasons according to available data, the strong wind that one of them former because weather extremes causes and the rapid variation of environment temperature cause the fault of wind generator set blade mass eccentricity, and the another one factor is exactly the crack fault etc. that is subjected to the blade fatigue damage that vibration in various degree causes for a long time.Present blade injury fault can only be monitored the following frequency change of 100HZ, have only big wind-induced high vibration that monitor and alarm system is started, make whole wind power generating set brake and carry out maintenance maintenance, but for the rapid variation of environment temperature and be subjected to the blade injury fault that vibration in various degree causes for a long time and can not realize effective monitoring, the rapid variation of weather extremes of the U.S. in 2008 and environment temperature causes wind generator set blade to surpass the loss of 3,000 ten thousand U.S. dollars, reason is exactly that heavy showers and cooling cause the blade inner chamber to be full of water, and the cooling of the blade row mouth of a river is freezed, and the deformable blade on the whole wind power generating set is lost; If above-mentioned example can in time monitor and effectively safeguard, can effectively avoid the generation of huge loss so to a certain extent.

Summary of the invention

The objective of the invention is to overcome the deficiencies in the prior art, provide that a kind of diagnostic method is simple, sensitivity and precision is higher, the diagnosis cost is lower, the blade fault diagnosing method based on vibration of wind generating set that can effectively improve wind generator set blade security, reliability.

The present invention solves its technical matters and is achieved through the following technical solutions:

A kind of blade fault diagnosing method based on vibration of wind generating set, the step of its diagnostic method is

(1) gathers the vibration response signal of wind power generating set running status lower frequency scope f≤1000Hz by being arranged on vibration transducer in the wind generating set engine room and data collecting instrument, to its vibration response signal rate of doing work spectrum; From the power spectrum of wind power generating set, extract low frequency power spectrum in M rank before the corresponding wind generator set blade, 3≤M≤50 by filtering method;

(2) M rank low frequency power spectrum before the wind generator set blade is obtained wind generator set blade running status eigenwert with correlation dimension or length fractal dimension or box counting dimension or frequency translation distance method;

(3) wind generator set blade running status eigenwert and wind generator set blade normal condition eigenwert are compared obtain the wind generator set blade fault diagnosis result.

And wind generator set blade normal condition eigenwert acquisition methods is:

(1) by vibration transducer and data collecting instrument wind power generating set is carried out modal test analysis and dynamics FEM (finite element) calculation, obtain the mode and the natural frequency of wind power generating set, therefrom extract the preceding M rank mode and the natural frequency of wind generator set blade, 3≤M≤50;

(2) gather the vibration response signal rate of doing work spectrum of aerogenerator class frequency scope f≤1000Hz under the normal operating condition by being arranged on vibration transducer in the wind generating set engine room and data collecting instrument, by the preceding M rank mode and the natural frequency of the wind generator set blade of acquisition in the step (1), M rank low frequency power is composed the employing filtering method extracts corresponding wind generator set blade from the power spectrum of wind power generating set before;

(3) M rank low frequency power spectrum before the wind generator set blade is obtained wind generator set blade normal condition eigenwert with correlation dimension or length fractal dimension or box counting dimension or frequency translation distance method.

And, comprise also and obtain wind generator set blade malfunction eigenwert that the method that this malfunction eigenwert is obtained is:

(1) gathers the vibration response signal of aerogenerator class frequency scope f≤1000Hz under the malfunction and make power spectrum by being arranged on vibration transducer in the wind generating set engine room and data collecting instrument, from the power spectrum of wind power generating set, extract low frequency power spectrum in M rank before the corresponding wind generator set blade, 3≤M≤50 by filtering method;

(2) M rank low frequency power spectrum before the wind generator set blade is obtained wind generator set blade malfunction eigenwert with correlation dimension or length fractal dimension or box counting dimension or frequency translation distance method.

And the malfunction eigenwert is that blade end adds eccentric massblock malfunction eigenwert or root of blade bolt looseness malfunction eigenwert or blade crackle and end and adds eccentric massblock malfunction eigenwert.

Advantage of the present invention and beneficial effect are:

1, this blade fault diagnosing method can provide diagnosis to the misoperation of wind generator set blade, by the analyzing and processing to the vibration transducer acquired signal, diagnoses the fault of wind generator set blade fast and accurately, for brake in time provides foundation.

2, the sensor of this blade fault diagnosing method employing is arranged in the cabin of wind power generating set, can a plurality of sensors be set according to actual condition, and this sensor is the vibration transducer of market sale, and cost is lower.

3, this blade fault diagnosing method employing correlation dimension or length fractal dimension or box counting dimension or frequency translation distance method are obtained wind generator set blade running status eigenwert, can verify mutually between four kinds of methods, improve the reliability of whole diagnostic method.

4, this blade fault diagnosing method also comprises and obtains wind generator set blade malfunction eigenwert, can accurately find out the abort situation of wind generator set blade by blade fault status flag value, shorten maintenance maintenance and search the time, improve the efficient of maintenance maintenance.

5, simple, the sensitivity of diagnostic method of the present invention and precision is higher, the diagnosis cost is lower, be a kind of blade fault diagnosing method based on vibration of wind generating set that can effectively improve wind generator set blade security, reliability.

Description of drawings

Fig. 1 is the time-domain signal of wind generator set blade under normal condition;

Fig. 2 adds time-domain signal under the eccentric massblock malfunction for wind generator set blade at blade end;

Fig. 3 is the time-domain signal of wind generator set blade under root of blade bolt looseness malfunction;

Fig. 4 is wind generator set blade blade crackle and the terminal time-domain signal that adds under the eccentric massblock malfunction;

Fig. 5 is the power spectrum of wind generator set blade under normal condition;

Fig. 6 adds power spectrum under the eccentric massblock malfunction for wind generator set blade at blade end;

Fig. 7 is the power spectrum of wind generator set blade under root of blade bolt looseness malfunction;

Fig. 8 is wind generator set blade blade crackle and the terminal power spectrum that adds under the eccentric massblock malfunction;

Fig. 9, Figure 10, Figure 11, Figure 12 are respectively the power spectrum that extracts first three rank natural frequency of corresponding blade from Fig. 5, Fig. 6, Fig. 7, Fig. 8 after the filtering;

Figure 13 is four kinds of state relation Dimension Characteristics value result of calculations;

Figure 14 is four kinds of state length fractal dimension eigenvalue calculation results;

Figure 15 is the eigenvalue calculation result of four kinds of state box counting dimension methods;

Figure 16 is four kinds of state frequency translation distance method eigenvalue calculation results.

Embodiment

The invention will be further described below by specific embodiment, and following examples are descriptive, is not determinate, can not limit protection scope of the present invention with this.

A kind of blade fault diagnosing method based on vibration of wind generating set, the step of its diagnostic method is

(1) gathers the vibration response signal of wind power generating set running status lower frequency scope f≤1000Hz by being arranged on vibration transducer in the wind generating set engine room and data collecting instrument, this vibration transducer model is a PCB 352C68 type ICP piezoelectric acceleration sensor, to its vibration response signal rate of doing work spectrum; From the power spectrum of wind power generating set, extract low frequency power spectrum in M rank before the corresponding wind generator set blade, 3≤M≤50 by filtering method;

(2) M rank low frequency power spectrum before the wind generator set blade is obtained wind generator set blade running status eigenwert with correlation dimension or length fractal dimension or box counting dimension or frequency translation distance method;

(3) wind generator set blade running status eigenwert and wind generator set blade normal condition eigenwert are compared obtain this wind generator set blade normal condition eigenwert acquisition methods of wind generator set blade fault diagnosis result and be:

1. by vibration transducer and data collecting instrument wind power generating set is carried out modal test analysis and dynamics FEM (finite element) calculation, obtain the mode and the natural frequency of wind power generating set, therefrom extract the preceding M rank mode and the natural frequency of wind generator set blade, 3≤M≤50;

2. gather the vibration response signal rate of doing work spectrum of aerogenerator class frequency scope f≤1000Hz under the normal operating condition by being arranged on vibration transducer in the wind generating set engine room and data collecting instrument, by the preceding M rank mode and the natural frequency of wind generator set blade, M rank low frequency power is composed the employing filtering method extracts corresponding wind generator set blade from the power spectrum of wind power generating set before;

3. M rank low frequency power spectrum before the wind generator set blade is obtained wind generator set blade normal condition eigenwert with correlation dimension or length fractal dimension or box counting dimension or frequency translation distance method.

This comprises also obtains wind generator set blade malfunction eigenwert that based on the blade fault diagnosing method of vibration of wind generating set the method that this malfunction eigenwert is obtained is:

(1) gather the vibration response signal of aerogenerator class frequency scope f≤1000Hz under the malfunction and make power spectrum by being arranged on vibration transducer in the wind generating set engine room and data collecting instrument, from the power spectrum of wind power generating set, extracts corresponding wind generator set blade by filtering method before M rank low frequency power compose;

(2) M rank low frequency power spectrum before the wind generator set blade is obtained wind generator set blade malfunction eigenwert with correlation dimension or length fractal dimension or box counting dimension or frequency translation distance method, this malfunction eigenwert is that blade end adds eccentric massblock malfunction eigenwert or root of blade bolt looseness malfunction eigenwert or blade crackle and end and adds eccentric massblock malfunction eigenwert.

Above-mentioned method to wind power generating set vibration response signal rate of doing work spectrum under normal or fault or the running status is:

Vibration response signal X (t) with the wind power generating set obtained carries out Fourier transform by formula (1), obtains power spectrum X (j ω):

$X\left(\mathrm{j\ω}\right)=\underset{-\∞}{\overset{\∞}{\∫}}X\left(t\right)e^{-\mathrm{j\ωt}}\mathrm{dt}$ Formula (1)

According to the preceding M rank mode and the natural frequency of wind generator set blade, the power spectrum that is truncated to the preceding M rank of blade correspondence is G _x(f); Again to G _x(f) low-frequency component on M rank filters before the non-blade in, adopts Chebyshev's filtering method here:

The Chebyshev filter design:

Use Chebyshev I type low-pass filter to G _x(f) Filtering Processing filters the composition of non-blade frequencies correspondence, selects the fundamental function formula of Chebyshev filter to be:

${\left|H\left(\mathrm{j\ω}\right)\right|}^2=\frac{1}{1+{\mathrm{\ϵ}}^2{V}_N^2(\mathrm{\ω}/{\mathrm{\ω}}_c)}$ Formula (2)

In the formula (2), ω _cFor effective passband by frequency; N is a filter order; ε is a ripple coefficient, ripple fluctuating size in the decision passband, V _N(ω) be defined as:

$V_N\left(\mathrm{\&omega;}\right)=\left\{\begin{array}{cc}\cos \left({N\cos }^{-1}\mathrm{\&omega;}\right)& \left|\mathrm{\&omega;}\right|<1\\ \cosh \left(N{\cosh }^{-1}\mathrm{\&omega;}\right)& \left|\mathrm{\&omega;}\right|\&GreaterEqual;1\end{array}\right.$

F=ω/2 π again, so formula (2) can be written as:

${\left|H\left(f\right)\right|}^2=\frac{1}{1+{\mathrm{\&epsiv;}}^2{V}_N^2(f/f_c)}$ Formula (3)

The parameter of design Chebyshev bandpass filter: the passband upper cut off frequency is ω _uThe passband lower limiting frequency is ω _lThe stopband upper cut off frequency is ω _S2The stopband lower limiting frequency is ω _S1Maximum attenuation α=3db in the passband; Minimal attenuation β=30db in the stopband; The analog band-pass filter technical indicator is:

$\mathrm{\&Omega;}=2\tan \frac{1}{2}\mathrm{\&omega;}$ Formula (4)

It is as follows to convert the technical indicator of above-mentioned wave filter to the analog band-pass filter technical indicator by formula (4):

The passband upper cut off frequency ${\mathrm{\&Omega;}}_u=2\tan \frac{1}{2}{\mathrm{\&omega;}}_u$

The passband lower limiting frequency ${\mathrm{\&Omega;}}_l=2\tan \frac{1}{2}{\mathrm{\&Omega;}}_l$

The stopband upper cut off frequency ${\mathrm{\&Omega;}}_l=2\tan \frac{1}{2}{\mathrm{\&omega;}}_l$

The stopband lower limiting frequency ${\mathrm{\&Omega;}}_l=2\tan \frac{1}{2}{\mathrm{\&omega;}}_l$

Passband central frequency ${\mathrm{\&Omega;}}_0=\sqrt{{\mathrm{\&Omega;}}_u{\mathrm{\&Omega;}}_l}$

Bandwidth B=Ω _u-Ω _l

Above edge frequency to bandwidth B normalization, is obtained:

${\mathrm{\&eta;}}_u=\frac{{\mathrm{\&Omega;}}_u}{B},$ ${\mathrm{\&eta;}}_l=\frac{{\mathrm{\&Omega;}}_l}{B},$ ${\mathrm{\&eta;}}_{\mathrm{<figure-callout\; id="2"\; label="s"\; filenames="HSA00000035828900012.png,HSA00000035828900022.png"\; state="\{\{state\}\}">s</figure-callout>}2}=\frac{{\mathrm{\&Omega;}}_{s2}}{B},$ ${\mathrm{\&eta;}}_{\mathrm{<figure-callout\; id="1"\; label="s"\; filenames="HSA00000035828900011.png,HSA00000035828900012.png"\; state="\{\{state\}\}">s</figure-callout>}1}=\frac{{\mathrm{\&Omega;}}_{s1}}{B},$ ${\mathrm{\&eta;}}_0=\frac{{\mathrm{\&Omega;}}_0}{B}$

Normalization stopband cutoff frequency

Normalization cut-off frequecy of passband λ _p=1; f _c=λ _s/ λ _p, ε, N are calculated by formula (5), formula (6):

$\mathrm{\&epsiv;}={({10}^{\frac{\mathrm{\&alpha;}}{10}}-1)}^{1/2}$ Formula (5)

$N\&GreaterEqual;\frac{{\mathrm{ch}}^{-1}[{({10}^{\mathrm{\&beta;}/10}-1)}^{1/2}/\mathrm{\&epsiv;}}{{\mathrm{ch}}^{-1}(f/f_c)}$ Formula (6)

With f _c, ε, N substitution formula (3) gets digital band-pass filter, and Chebyshev filter is g _x(f) be filtered power spectrum suc as formula (7):

g _x(f)=G _x(f) | H (f) | ²Formula (7)

From the power spectrum of wind power generating set, extract low frequency power spectrum in M rank before the corresponding wind generator set blade, 3≤M≤50 by filtering method; Obtain the eigenwert of wind generator set blade running status, normal condition and malfunction respectively with correlation dimension or length fractal dimension or box counting dimension or frequency translation distance method, and as the fault diagnosis characteristic quantity to the blade running status.

Correlation dimension method, length fractal dimension method, box counting dimension method, frequency translation distance method are the prior art computing method, and regarding to four kinds of methods does simple introduction down:

1, correlation dimension method:

For the single argument sequence, the direct compute associations dimension from univariate sequence according to embedding theorems and phase space reconstruction thought.Be designated as D _c, suppose one group of data sequence: X ₁, X ₂, X ₃, L X _i, L, wherein X _iIt is the value that i records constantly.Now these group data are divided into different groups, for example, getting m=10 data is one group, X1 then, X ₂, X ₃, L X ₁₀As first group, be designated as Y ₁Move to right then a step, X ₂, X ₃, X ₄, L X ₁₁As second group, be designated as Y ₂So divide, can obtain large quantities of data set Y ₁, Y ₂, Y ₃, LY _kIn fact, these data sets itself are exactly a vector of m-dimensional space.Absolute value with their any differences between the two is designated as r now _{I, j}=Y _i-Y _j, Y _iAnd Y _jRepresent i and j group data respectively.r _{I, j}For m dimension hypersphere radius, so obtain a series of r _{I, j}Ball is that the sieve of r screens these r with pore radius _{I, j}Ball, the obviously little r of radius ratio r _{I, j}Ball just leaks down, and its number is designated as N _Down(r); The r that radius ratio r is big _{I, j}Above ball was just stayed, its number was designated as N _Up(r).r _{I, j}Total number N (the r)=N of ball _Down(r)+N _Up(r).Now the r of radius less than r _{I, j}Ball number and total r _{I, j}The ratio of ball number is designated as C (r), then has

$C\left(r\right)=\frac{N_{\mathrm{down}}\left(r\right)}{N\left(r\right)}$

Here C (r) is an important parameters.Suitable change r makes r in certain interval, makes C (r) ∝ r ^δ, index δ is correlation dimension D in the formula _cA kind of approaching.D _cMore strict definition is

$D_c=\underset{r\&RightArrow;0}{\lim }\frac{d\ln C_r}{d\ln r}$ Formula (8)

Wherein

$C(r,m)=\frac{1}{N_p(N_p-1)}\underset{i=1}{\overset{N_p}{\mathrm{\&Sigma;}}}\underset{j=1}{\overset{N_p}{\mathrm{\&Sigma;}}}H(r-r_{i,j})$

N in the formula _pBe the phase space vector number by time series reconstruct, H is the Heaviside function, promptly

$H(r-r_{i,j})=\left\{\begin{array}{cc}1& \left({r-r}_{i,j}\right)\&GreaterEqual;0\\ 0& \left({r-r}_{i,j}\right)<0|\end{array}\right\}$

C (r m) is a correlation integral, in the expression phase space on the attractor distance between two points less than the probability of r.

2, length fractal dimension method:

If nonlinear dynamic system vibrational waveform sampling A={a|a=is (x _i, y _i), i=1,2 ... N}, wherein x _iBe time-sampling point, y _iBe corresponding vibration amplitude, N is a sample points.Set A is carried out Topological Mapping by formula (9), (10), A → M then, M={b | b=(x _i ^*, y _i ^*), i=1,2 ... N} is the subclass on a unit plane.

${x_i}^{*}=\frac{x_i}{x_N}$ Formula (9)

$y_i^{*}=\frac{y_i-\min \left(V_I\right)}{\max \left(V_I\right)-\min \left(V_I\right)}$ Formula (10)

In the formula (5-2), V _I={ y ₁, y ₂, y _i, i=1,2 ... N

Linear transformation does not change the topological structure of set, so A, M dimension equate.

In set M, establishing vibrational waveform length is L, and the hypercube length of side is ε, and then the capping unit number is L/ ε, gets ε=1/N ^*, N ^*=N-1.When coating or measuring vibrational waveform, the hypercube of unit with one dimension long measure; For the nonlinear dynamic system vibrational waveform of limited sample point, the length fractal dimension is

$D_L=1+\underset{N^{*}\&RightArrow;\&infin;}{\lim }\ln (L/N^{*})=1+\frac{\ln L}{\ln N^{*}}$ Formula (11)

3, box counting dimension method:

An if discrete dynamical systems vibration signal sampling point set

If it can be the Ω dimension hypercube covering of ε by the individual length of side of N (ε), definition:

${\dim }_f\left(\mathrm{\&Omega;}\right)\stackrel{\mathrm{def}}{=}\underset{\mathrm{\&epsiv;}\&RightArrow;0}{\lim }\frac{\ln N\left(\mathrm{\&epsiv;}\right)}{\ln (1/\mathrm{\&epsiv;})}$ Formula (12)

And be called box counting dimension or the capacity dimension of sampling point set Ω.The fractal box of signal is in interval (1,2), and signal is irregular more, and fractal box is big more.Therefore, can judge the regularity of signal by the size of box counting dimension.

4, frequency translation distance method:

Power spectrum G (f) to arbitrary i status signal.Use coefficient R B _i(l) similarity of signature criteria spectrum BG (f) and power spectrum G (f),

${\mathrm{RB}}_i\left(l\right)=\frac{<G(f+l),\mathrm{BG}>\left(f\right)>}{{\left|\right|G\left(f\right)\left|\right|}_2{\left|\right|\mathrm{BG}\left(f\right)\left|\right|}_2}$ (i=1,2...50) formula (13)

In the formula (10), RB _i(l) be power spectrum G (f) and the standard related coefficient of standard spectrum BG (f), it has characterized G (f), the similarity of BG (f),＜G (f+l), BG (f)〉inner product of expression G (f) and BG (f), || G (f) || ₂With || BG (f) || ₂2 norms of then representing G (f) and BG (f).

This is based on the blade fault diagnosing method of vibration of wind generating set, can at first obtain wind generator set blade normal condition eigenwert and wind generator set blade malfunction eigenwert, again by obtaining wind generator set blade running status eigenwert and wind generator set blade normal condition eigenwert and wind generator set blade malfunction eigenwert compares, when obtaining diagnostic result and being malfunction, can compare the malfunction eigenwert and find out the blade fault position, accuracy of judgement, precision height.

Be research object with the 300W wind power generating set below, the rotating speed of rotor is 100r/min during the operating states of the units test, and sample frequency is 500Hz.Test is provided with four kinds of states, and they are respectively: a is that wind generator set blade runs well; B is that blade end adds eccentric massblock; C is the root of blade bolt looseness; D is a coupling fault: blade crackle and end add eccentric massblock, have 2cm long crack and blade end to add eccentric massblock apart from root of blade 2/3 place.Various states are respectively measured 50 sections time-domain signals; Fig. 1, Fig. 2, Fig. 3, Fig. 4 are the time-domain signal under four kinds of states of wind generator set blade; Fig. 5, Fig. 6, Fig. 7, Fig. 8 are the power spectrum under four kinds of states of corresponding blade.Adopt Chebyshev's filtering method, from Fig. 5, Fig. 6, Fig. 7, four kinds of states of Fig. 8 blade, extract the power spectrum of first three rank natural frequency of corresponding blade, as Fig. 9, Figure 10, Figure 11, shown in Figure 12.

The eigenwert of four kinds of states of blade is calculated, and can be verified mutually with correlation dimension method, length fractal dimension method, box counting dimension method, frequency translation distance method respectively:

(1) correlation dimension eigenvalue calculation:

To the power spectrum signal of Fig. 5, Fig. 6, Fig. 7, Fig. 8, get corresponding delay factor τ=5, embed dimension m=25; With correlation dimension principle and computing method, calculate the correlation dimension of four kinds of states, result of calculation is seen Figure 13, and the correlation dimension of power spectrum signal is stabilized in 0.02 error range under four kinds of states of wind generator set blade: a is that correlation dimension is 4.4523 ± 0.02 under the wind generator set blade normal condition; B is that blade end adds that correlation dimension is 3.7652 ± 0.02 under the eccentric massblock malfunction; C is that correlation dimension is 3.4572 ± 0.01 under the root of blade bolt looseness malfunction; D is that blade crackle and end add that correlation dimension is 1.0252 ± 0.005 under the eccentric massblock malfunction; Correlation dimension significant difference under each state.Can identify each running status of wind generator set blade.

(2) length FRACTAL DIMENSION eigenvalue calculation:

Power spectrum signal utilization length FRACTAL DIMENSION principle and computing method to Fig. 5, Fig. 6, Fig. 7, Fig. 8, four kinds of states to wind generator set blade calculate, result of calculation is seen Figure 14, a be under the blade of wind-driven generator normal condition when running length fractal dimension be 1.0710 ± 0.01; B is that blade end adds that the length fractal dimension is 1.1481 ± 0.03 under the eccentric massblock malfunction; C is that the length fractal dimension is 1.2215 ± 0.01 under the root of blade bolt looseness malfunction; D is that blade crackle and end add that the length fractal dimension is 1.0031 ± 0.003 under the eccentric massblock malfunction, and the length fractal dimension of each state has tangible difference under fractal meaning, can identify each running status of wind generator set blade with it.

(3) eigenvalue calculation of box counting dimension method:

The power spectrum signal of Fig. 5, Fig. 6, Fig. 7, Fig. 8 is used above-mentioned box dimension principle and computing method, four kinds of states of wind generator set blade are calculated, result of calculation is seen Figure 15, a be aerogenerator when running well box counting dimension be 1.4153 ± 0.005; B be blade end when adding eccentric massblock box counting dimension be 1.3142 ± 0.01; Box counting dimension was 1.3793 ± 0.01 when c was the root of blade bolt looseness; Box counting dimension was 1.4651 ± 0.005 when d was coupling fault, and d state box counting dimension＞a state box counting dimension＞c state box counting dimension＞b state box counting dimension.The box counting dimension of state has tangible difference under fractal meaning, can identify each running status of wind generator set blade with it.

(4) frequency translation distance method eigenvalue calculation:

To the power spectrum signal operating frequency translation distance method of Fig. 5, Fig. 6, Fig. 7, Fig. 8, four kinds of states of wind generator set blade to be calculated, result of calculation is seen Figure 16, the frequency translation distance of a wind generator set blade normal condition is ± 0.3; The frequency translation distance was-2.0 ± 0.3 when the b blade end added eccentric massblock; The frequency translation distance is-4.3 ± 0.4 during c root of blade bolt looseness; The frequency translation distance is 8.5 ± 0.9 during the d coupling fault.Frequency translation distance difference under each state significantly can identify each running status of wind generator set blade with it.

Claims (4)

1. blade fault diagnosing method based on vibration of wind generating set, it is characterized in that: the step of this diagnostic method is:

(1) gathers the vibration response signal of wind power generating set running status lower frequency scope f≤1000Hz by being arranged on vibration transducer in the wind generating set engine room and data collecting instrument, the vibration response signal rate of doing work is composed; From the power spectrum of wind power generating set, extract low frequency power spectrum in M rank before the corresponding wind generator set blade, 3≤M≤50 by filtering method;

(2) M rank low frequency power spectrum before the wind generator set blade is obtained wind generator set blade running status eigenwert with correlation dimension or length fractal dimension or box counting dimension or frequency translation distance method;

(3) wind generator set blade running status eigenwert and wind generator set blade normal condition eigenwert are compared obtain the wind generator set blade fault diagnosis result.

2. the blade fault diagnosing method based on vibration of wind generating set according to claim 1 is characterized in that: wind generator set blade normal condition eigenwert acquisition methods is:

1. by vibration transducer and data collecting instrument wind power generating set is carried out modal test analysis and dynamics FEM (finite element) calculation, obtain the mode and the natural frequency of wind power generating set, therefrom extract the preceding M rank mode and the natural frequency of wind generator set blade, 3≤M≤50;

2. gather the vibration response signal rate of doing work spectrum of aerogenerator class frequency scope f≤1000Hz under the normal operating condition by being arranged on vibration transducer in the wind generating set engine room and data collecting instrument, the preceding M rank mode and the natural frequency of the wind generator set blade that obtains in 1. by step, adopt filtering method from the power spectrum of wind power generating set, to extract corresponding wind generator set blade before M rank low frequency power compose;

3. M rank low frequency power spectrum before the wind generator set blade is obtained wind generator set blade normal condition eigenwert with correlation dimension or length fractal dimension or box counting dimension or frequency translation distance method.

3. the blade fault diagnosing method based on vibration of wind generating set according to claim 1, it is characterized in that: after obtaining the wind generator set blade fault diagnosis result, also comprise and obtain wind generator set blade malfunction eigenwert, the method that this malfunction eigenwert is obtained is:

1. gather the vibration response signal of aerogenerator class frequency scope f≤1000Hz under the malfunction and make power spectrum by being arranged on vibration transducer in the wind generating set engine room and data collecting instrument, from the power spectrum of wind power generating set, extract low frequency power spectrum in M rank before the corresponding wind generator set blade, 3≤M≤50 by filtering method;

2. M rank low frequency power spectrum before the wind generator set blade is obtained wind generator set blade malfunction eigenwert with correlation dimension or length fractal dimension or box counting dimension or frequency translation distance method.

4. the blade fault diagnosing method based on vibration of wind generating set according to claim 3 is characterized in that: wind generator set blade malfunction eigenwert is that blade end adds eccentric massblock malfunction eigenwert or root of blade bolt looseness malfunction eigenwert or blade crackle and end and adds eccentric massblock malfunction eigenwert.

Images