CN106224224A - A kind of based on Hilbert-Huang transform and quality the Hydraulic pump fault feature extracting method away from entropy - Google Patents

Abstract

The present invention relates to a kind of based on Hilbert-Huang transform and quality the Hydraulic pump fault feature extracting method away from entropy, hydraulic pump is in the oil circuit of closing, fluid structure interaction between the compressibility of hydraulic oil and hydraulic pump and servosystem, the fault signature making hydraulic pump is inconspicuous, and fault message extracts more difficulty.For this problem, the present invention propose a kind of based on Hilbert-Huang transform and quality the Hydraulic pump fault feature extracting method away from entropy: first hydraulic pump vibration signal is carried out empirical mode decomposition, obtains its natural mode of vibration component；Then, each natural mode of vibration component is carried out Hilbert transform, obtains hilbert spectrum；Finally, the quality of fault-signal Hilbert time-frequency distributions is calculated away from entropy.Experiment proves that, the Hydraulic pump fault feature acquired in the method for proposition has excellent sort feature, can well support that Fault Diagnosis of Hydraulic Pump works.

Description

A kind of Hydraulic pump fault feature extraction away from entropy based on Hilbert-Huang transform and quality Method

Technical field

The present invention relates to a kind of based on Hilbert-Huang transform and quality the Hydraulic pump fault feature extracting method away from entropy, belong to In Hydraulic system fault diagnosis technologies field.

Background technology

Hydraulic system is widely used in the systems such as Aero-Space, naval vessel, vehicle, and current forward high pressure, high power are close Degree and extensive, integrated direction development, the safety and reliability of hydraulic system is increasingly subject to the attention of people.Hydraulic pump quilt Being described as " heart " of hydraulic system, be the dynamical element of whole hydraulic system, the quality of its service behaviour will directly affect hydraulic pressure The overall work state of system.As high-speed rotating machine, it is the longest that hydraulic pump runs the time in hydraulic system, and live load is Greatly, so the abrasion degradation speed of hydraulic pump is very fast.Statistics shows, in the fault of all engineering machinery, and the fault of hydraulic pump Proportion accounts for 30%～40%, and therefore the fault diagnosis of hydraulic pump is the pith of Failure Diagnosis of Hydraulic System.Hydraulic pump one Denier breaks down, and is easily caused the inefficacy of whole hydraulic system, causes irremediable loss.As applied hydraulic system on aircraft Aircraft is carried out attitude manipulation, when fuel feeding fault occurs in hydraulic system, aircraft will be caused out of hand, the most then emergency landing, Heavy then cause the major accident of fatal crass.

Feature extraction is the core of fault diagnosis, and good fault signature is particularly significant to the precision improving fault diagnosis.So And, due to hydraulic pump be in the oil circuit of closing, fluid structurecoupling between the compressibility of hydraulic oil and hydraulic pump and servosystem Effect so that Hydraulic pump fault feature is inconspicuous, fault message extracts more difficulty, and ambiguity is strong, difficulty is big to cause its diagnosis.

Summary of the invention

The technology of the present invention solves problem: overcome the deficiencies in the prior art, it is provided that propose a kind of yellow based on Hilbert Conversion and the quality Hydraulic pump fault feature extracting method away from entropy, it can extract the event of sensitivity in the vibration signal of hydraulic pump Barrier feature, provides for fault diagnosis and supports.

The technology of the present invention solution: a kind of the Hydraulic pump fault feature away from entropy carries based on Hilbert-Huang transform and quality Access method, its step is as follows:

(1) propose to be applicable to the quality of troubleshooting issue away from entropy, fill when quantifying fault-signal time-frequency distributions complexity Dividing the positional information considering time-frequency block, three quality of fault-signal two dimension time-frequency distributions are away from entropy (s_t(q),s_f(q),s_o(q)) It is defined as follows:

Time-frequency plane is divided into the time-frequency block of N number of area equation, and every piece of interior energy is E_i, during this time-frequency block energy pair The quality of countershaft t, frequency axis f and initial point O is away from being respectively as follows:

$\left\{\begin{array}{c}{M}_{ti}=E_i.d_{ti},\\ M_{fi}=E_i.d_{fi},\\ M_{oi}=E_i.d_{oi},\end{array}\right.,i=1,...,N.$

Whole time-frequency plane to the quality of time shaft, frequency axis and initial point away from being respectively as follows:

$\left\{\begin{array}{c}{M}_t=\underset{i=1}{\overset{N}{\Σ}}{M}_{ti},\\ M_f=\underset{i=1}{\overset{N}{\Σ}}{M}_{fi},\\ M_o=\underset{i=1}{\overset{N}{\Σ}}{M}_{oi},\end{array}\right.,i=1,...,N.$

To the quality of each time-frequency block energy away from being normalized, obtain:

$\left\{\begin{array}{c}{q}_{ti}=M_{ti}/M_t,\\ q_{fi}=M_{fi}/M_i,\\ q_{oi}=M_{oi}/M_o,\end{array}\right.,i=1,...,N.$

Then have:

$\left\{\begin{array}{c}\underset{i=1}{\overset{N}{\Σ}}{q}_{ti}=1,\\ \underset{i=1}{\overset{N}{\Σ}}{q}_{fi}=1,\\ \underset{i=1}{\overset{N}{\Σ}}{q}_{oi}=1,\end{array}\right.,i=1,...,N.$

Fault-signal time-frequency distributions to time shaft quality away from entropy s_tQ (), to frequency axis quality away from entropy s_t(q) and to initial point O's Quality is away from entropy s_oQ () is defined respectively as:

$\left\{\begin{array}{c}{s}_t\left(q\right)=-\underset{i=1}{\overset{N}{\Σ}}{q}_{ti}\ln q_{ti},\\ s_f\left(q\right)=-\underset{i=1}{\overset{N}{\Σ}}{q}_{fi}\ln q_{fi},\\ s_o\left(q\right)=-\underset{i=1}{\overset{N}{\Σ}}{q}_{oi}\ln q_{oi}.\end{array}\right.$

In formula, q_ti, q_fiAnd q_oiIt is respectively the i-th time-frequency block energy quality to each coordinate axes or initial point in time accounting for whole Frequently distribution energy relative to respective coordinates axle or initial point quality away from ratio；

To the quality of time shaft away from entropy s_tQ () characterizes the time-frequency distributions complexity to frequency f, i.e. fault-signal energy is not The distribution situation tolerance of same frequency section；To the quality of frequency axis away from entropy s_fQ () characterizes the time-frequency distributions complexity to the time, i.e. event The time-varying characteristics tolerance of barrier signal energy distribution；To the quality of initial point O away from entropy s_oQ () characterizes the general complexity of time-frequency distributions；

(2) Hilbert-Huang transform and quality are combined away from entropy, propose a kind of fault being applicable to process non-stationary signal Feature extracting method, empirical mode decomposition is for being decomposed into a series of natural mode of vibration component adaptively by vibration signal, uncommon Your Bert conversion is used for calculating instantaneous amplitude and instantaneous frequency thus obtains hilbert spectrum, finally uses quality when entropy quantifies The complexity of frequency division cloth, as Hydraulic pump fault feature.

It is above-mentioned that to implement step as follows:

The first step, data prediction: gather hydraulic pump vibration signal, and vibration signal is carried out outlier rejecting and noise reduction Process；

Second step, empirical mode decomposition EMD: vibration signal is decomposed into adaptively a series of intrinsic mode function IMF Component and trend term；

3rd step, Hilbert transform: each intrinsic mode function IMF component is implemented Hilbert transform, it is thus achieved that wink Time amplitude and instantaneous frequency, thus obtain hilbert spectrum；

4th step, calculates the quality of fault-signal time-frequency distributions away from entropy: the Hilbert obtained according to Hilbert transform Spectrum, calculate fault-signal time-frequency distributions three quality away from entropy, i.e. time-frequency distributions to time shaft quality away from entropy s_tQ (), to frequency Axoplasm span entropy s_t(q) and to the quality of initial point O away from entropy s_o(q)；

5th step, draws the dendrogram of different faults state sample eigenvalue, analyzes method effectiveness.

Described second step, the process that EMD decomposes is as follows: by the composition of signal highest frequency, and decomposition obtains one by one The frequency range of IMF reduces successively, and decomposition method builds local maximum and the local pole of signal sequence by cubic spline interpolation The envelope of little value, removes after calculating the average of upper and lower envelope from primary signal；Same procedure is used to repeat remaining residual error Carry out till the average envelope at each point goes to zero, resulting in first IMF, from primary signal, deducting first IMF, uses identical screening technique to decomposite other IMF one by one, stops when residual signals amplitude is the least or becomes dullness dividing Solve.

In described 4th step, calculate three quality of fault-signal time-frequency distributions away from entropy (s_t(q),s_f(q),s_o(q)) process For: time-frequency plane is divided into the time-frequency block of N number of area equation, and every piece of interior energy is E_i, this time-frequency block self-energy is to the time Axle t, frequency axis f and to the quality of initial point O away from being respectively as follows:

$\left\{\begin{array}{c}{M}_{ti}=E_i.d_{ti},\\ M_{fi}=E_i.d_{fi},\\ M_{oi}=E_i.d_{oi},\end{array}\right.,i=1,...,N.$

Whole time-frequency plane to the quality of Two coordinate axle and initial point away from being respectively as follows:

$\left\{\begin{array}{c}{M}_t=\underset{i=1}{\overset{N}{\Σ}}{M}_{ti},\\ M_f=\underset{i=1}{\overset{N}{\Σ}}{M}_{fi},\\ M_o=\underset{i=1}{\overset{N}{\Σ}}{M}_{oi},\end{array}\right.,i=1,...,N.$

To the quality of each time-frequency block energy away from being normalized, obtain:

$\left\{\begin{array}{c}{q}_{ti}=M_{ti}/M_t,\\ q_{fi}=M_{fi}/M_i,\\ q_{oi}=M_{oi}/M_o,\end{array}\right.,i=1,...,N.$

Then have:

$\left\{\begin{array}{c}\underset{i=1}{\overset{N}{\Σ}}{q}_{ti}=1,\\ \underset{i=1}{\overset{N}{\Σ}}{q}_{fi}=1,\\ \underset{i=1}{\overset{N}{\Σ}}{q}_{oi}=1,\end{array}\right.,i=1,...,N.$

Fault-signal time-frequency distributions to time shaft quality away from entropy s_tQ (), to frequency axis quality away from entropy s_t(q) and to initial point O Quality away from entropy s_oQ () is defined respectively as:

$\left\{\begin{array}{c}{s}_t\left(q\right)=-\underset{i=1}{\overset{N}{\Σ}}{q}_{ti}\ln q_{ti},\\ s_f\left(q\right)=-\underset{i=1}{\overset{N}{\Σ}}{q}_{fi}\ln q_{fi},\\ s_o\left(q\right)=-\underset{i=1}{\overset{N}{\Σ}}{q}_{oi}\ln q_{oi}.\end{array}\right.$

In formula, q_ti, q_fiAnd q_oiIt is respectively each quality of i-th time-frequency block energy away from the matter accounting for whole time-frequency distributions energy The ratio of span；

To the quality of time shaft away from entropy s_tQ () characterizes the time-frequency distributions complexity to frequency f, i.e. fault-signal energy is not The distribution situation of same frequency section；To the quality of frequency axis away from entropy s_fQ () characterizes the time-frequency distributions complexity to the time, i.e. fault letter The time-varying characteristics of number Energy distribution；To the quality of initial point O away from entropy s_oQ () characterizes the general complexity of time-frequency distributions.

Present invention advantage compared with prior art is: hydraulic pump is in the oil circuit of closing, the compressibility of hydraulic oil And the fluid structure interaction between hydraulic pump and servosystem so that Hydraulic pump fault feature is inconspicuous, and fault message extracts relatively For difficulty.For this problem, it is special that the present invention proposes a kind of Hydraulic pump fault based on Hilbert-Huang transform and quality away from entropy Levy extracting method: first hydraulic pump vibration signal is carried out empirical mode decomposition, obtain natural mode of vibration component；Then, to each solid There is modal components to carry out Hilbert transform, obtain hilbert spectrum；Finally, the massic entropy of fault-signal time-frequency distributions is calculated. Experiment proves that, the present invention proposes the fault signature acquired in method and has excellent sort feature, can well support hydraulic pressure Failure of pump diagnostic work.

Accompanying drawing explanation

Fig. 1 is the Hydraulic pump fault feature extraction flow chart away from entropy based on Hilbert-Huang transform and quality of the present invention；

Fig. 2 is EMD algorithm flow chart；

Fig. 3 is time-frequency Entropy principle figure；

Fig. 4 is that quality is away from Entropy principle figure；

Fig. 5 is plunger hydraulic Test-bed for pump；

Fig. 6 is the vibration signal of each malfunction, i.e. under normal condition, valve plate wear-out failure and piston shoes wear-out failure Vibration signal figure；(a) normal (b) valve plate fault (c) piston shoes fault；

Fig. 7 is the hilbert spectrum of each malfunction: (a) normal (b) valve plate fault (c) piston shoes fault；

Fig. 8 is fault signature dendrogram.

Detailed description of the invention

As it is shown in figure 1, the present invention based on Hilbert-Huang transform and quality the Hydraulic pump fault feature extracting method away from entropy Mainly comprise the steps of

The first step, data prediction.Gather hydraulic pump vibration signal, and vibration signal is carried out outlier rejecting and noise reduction Process；

Second step, EMD decomposes.Vibration signal is decomposed into adaptively a series of intrinsic mode function IMF component and becomes Gesture item；

3rd step, Hilbert transform.Each intrinsic mode function IMF component is implemented Hilbert transform, it is thus achieved that its Instantaneous amplitude and instantaneous frequency, thus obtain hilbert spectrum；

4th step, calculates the quality of fault-signal time-frequency distributions away from entropy: the Hilbert obtained according to Hilbert transform Spectrum, calculate fault-signal time-frequency distributions three quality away from entropy, i.e. time-frequency distributions to time shaft quality away from entropy s_tQ (), to frequency Axoplasm span entropy s_t(q) and to the quality of initial point O away from entropy s_o(q)；

5th step, draws the dendrogram of different faults state sample eigenvalue, analyzes method effectiveness.

1. the Hydraulic pump fault feature extracting method away from entropy based on Hilbert-Huang transform and quality

1.1 Hilbert-Huang transform

Hilbert-Huang transform (HHT) is mainly made up of 2 parts: empirical mode decomposition (Empirical Mode Decomposition, EMD) and Hilbert transform.

(1) empirical mode decomposition

Time Series is become a series of and is referred to as intrinsic mode function (Intrinsic mode by empirical mode decomposition Function, IMF) simple component signal, a simple component signal represents one and is similar to harmonic function the most universal, most basic Oscillating function.Each IMF has different frequency components, comprises in signal by minimum frequency component, has instruction The potential of different faults information.They need to meet following two condition:

First, in signal, extreme point is identical with the number of zero crossing or at most differs one；Second, for appointing on signal A bit, local maximum and local minimum the average of the envelope defined respectively is zero to meaning.For oscillation data first Individual condition is the most necessary, limits with the stringent condition of satisfied calculating instantaneous frequency, i.e. provides shaking of signal at specific time point Swing frequency.Second condition requires that the upper and lower envelope of an IMF, relative to time shaft Local Symmetric, makes the signal can be along with decomposition IMF out is modulated.

EMD assumes based on following three points:

(a) signal at least minimum and a maximum (Non-monotonic function)；

B the time difference between () continuous threshold point defines characteristic time scale；

If (c) only flex point and there is no extreme point, data will be differentiated, then application EMD method and by obtain point Amount integration is to obtain final result.

The catabolic process of EMD method, by the composition of signal highest frequency, decomposes the frequency model of the IMFs obtained one by one Enclose and reduce successively.Decomposition method builds the minimizing envelope of local maximum and local of signal sequence by cubic spline interpolation, Remove from primary signal after calculating the average of upper and lower envelope.Remaining residual error use same procedure is repeated until each Till average envelope at Dian reasonably goes to zero, resulting in first IMF.First IMF is deducted from primary signal, Use identical screening technique to decomposite other IMF one by one, stop when residual signals amplitude is the least or becomes dullness decomposing.According to This, can sum up corresponding arthmetic statement as follows:

A () initializes: r₀(t)=x (t), i=1；

B () seeks i-th intrinsic mode function IMF_i=c_i(t):

A) initialize: h₀(t)=r_i-1(t), j=1；

B) h is found out_j-1Whole Local Extremum of (t)；

C) application cubic spline interpolation interpolation fitting h respectively_j-1T the very big and minimum point of (), tries to achieve lower envelope e₊ (t) and e-(t), and calculate its average

D) therefrom deduct the average of envelope, try to achieve h_j(t)=h_j-1(t)-m_j-1(t)；

E) judging whether the condition of convergence meets, if meeting, having c_i(t)=h_i(t)；If being unsatisfactory for, making j=j+1, returning step Suddenly (b)；

(c)r_i(t)=r_i-1(t)-c_i(t)；

If (d) r_iT the extreme point number more than one of (), orders i=i+1, return step (b), otherwise decomposed.

The idiographic flow of EMD algorithm is as shown in Figure 2.

By EMD method, original signal is broken down into:

$x\left(t\right)=\underset{i=1}{\overset{n}{\Σ}}{c}_i\left(t\right)+r_n\left(t\right)---\left(1\right)$

Wherein c_iT () is an IMF component, r_nT () is the average tendency of residual components, generally signal, for constant sequence Or monotonic sequence.

(2) Hilbert transform

Decomposed by EMD after obtaining IMF, it is possible to each component is done Hilbert transform, obtains its instantaneous frequency And instantaneous amplitude.If IMF component is c (t), then its complex analytic signal is:

$H\[c\left(t\right)\]=c\left(t\right)+j\tilde{c}\left(t\right)=a\left(t\right)e^{i\mathrm{\φ}\left(t\right)}---\left(2\right)$

WhereinFor:

$\tilde{c}\left(t\right)=\frac{1}{\mathrm{\π}}{\∫}_{-\mathrm{\∞}}^{+\mathrm{\∞}}\frac{c\left(\mathrm{\τ}\right)}{t-\mathrm{\τ}}d\mathrm{\τ}---\left(3\right)$

A (t) is magnitude function:

$a\left(t\right)=\sqrt{c{\left(t\right)}^2+\tilde{c}{\left(t\right)}^2}---\left(4\right)$

φ (t) is phase function:

$\mathrm{\φ}\left(t\right)={\tan }^{-1}\left(\tilde{c}\right(t)/c(t\left)\right)---\left(5\right)$

Instantaneous frequency is:

$f\left(t\right)=\frac{1}{2\mathrm{\π}}\frac{d\mathrm{\φ}\left(t\right)}{dt}---\left(6\right)$

Wherein magnitude function represents the instantaneous amplitude energy of each sampled point of signal；Phase function represents each sampling of signal The instantaneous phase of point, just obtains instantaneous frequency to its derivation.Each IMF component is Hilbert convert and ignore decomposition remainder, Data can be expressed as:

$x\left(t\right)=\underset{i=1}{\overset{n}{\Σ}}{a}_i\left(t\right)\exp (j\∫{\mathrm{\ω}}_i(t\left)dt\right)---\left(7\right)$

According to formula (7) can using amplitude and instantaneous phase as the function representation of time in three-dimensional planar, this of amplitude Plant time-frequency distributions and be referred to as Hilbert amplitude spectrum, referred to as hilbert spectrum.Square energy is represented traditionally by amplitude Density, if replacing the amplitude in Hilbert amplitude spectrum by amplitude square, will obtain Hilbert energy spectrum.

1.2 mass are away from entropy

(1) comentropy

The mathematical definition of comentropy is: set p (p₁,p₂,...,p_n) be the probability distribution of a random event, k be arbitrary often Number, is typically taken as 1, and the comentropy that this distribution is had is defined as:

$s\left(p\right)=-k\underset{i=1}{\overset{N}{\Σ}}{p}_i\ln p_i---\left(8\right)$

The size of comentropy can be used to portray the average degree of uncertainty of probability system.If in a certain probability system certain The probability that one event produces is 1, and the probability that other events produce is 0, formula (12) after calculating, and comentropy s of this system =0, thus be one and determine system, uncertainty is 0.If in a certain system, its probability distribution is uniform, then it represents that should The probability that in system, each event produces is equal, and the comentropy of this system has maximum, and i.e. this system is uncertain maximum. Theoretical according to this, the most uncertain probability distribution has the entropy of maximum, and information entropy reflects the inequality of its probability distribution Even degree.

(2) time-frequency entropy

The time-frequency distributions of signal describes signal Energy distribution situation within the sampling time at each frequency, different operating shape Under state, the time-frequency distributions of hydraulic pump is different, for quantitative this difference degree of description, information entropy theory is incorporated into fault-signal In time-frequency distributions.Different faults signal difference in time-frequency distributions shows as time-frequency fragment energy different on time-frequency plane and divides The difference of cloth, time-frequency entropy can quantify this species diversity, and then the running status of reflection machine.As it is shown on figure 3, by time-frequency plane etc. Being divided into the time-frequency block of N number of area equation, every piece of interior energy is E_i(i=1 ..., N), the energy of whole time-frequency plane is A, right Every piece carries out energy normalized, obtains q_i=E_i/ A (i=1 ..., N), then haveThe normalizing of according calculation comentropy Change condition, copies the computing formula of comentropy, and the computing formula of the time-frequency entropy of signal is defined as:

$s\left(q\right)=-\underset{i=1}{\overset{N}{\Σ}}{q}_i\ln q_i---\left(9\right)$

(3) quality is away from entropy

It is to carry out under the hypothesis of stochastic variable with the definition from comentropy, time-frequency entropy, namely not order between variable Difference.But, after comentropy is introduced fault diagnosis field, not only to distinguish the energy size of each energy block, also should close Noting the position at this energy block place, comprehensive coordinate and magnitude information weigh the distribution of fault-signal exactly.Conversely, Discounting for the position of each time-frequency block, by constant for the energy value of each time-frequency block of time-frequency plane, upset original order, The most calculated time-frequency entropy is constant, and order difference the most usually reflects different fault messages, and this explanation is concerned only with value Comentropy form of Definition can not portray fault signature exactly.

In order to comprehensively portray magnitude information and the positional information of fault-signal distribution, the present invention examines during definition entropy Consider the position at current time-frequency block place, propose a kind of quality of troubleshooting issue that is suitable for away from entropy.As shown in Figure 4, by time-frequency Plane is divided into the time-frequency block of N number of area equation, and every piece of interior energy is E_i(i=1 ..., N), during this time-frequency block self-energy pair Countershaft t, frequency axis f and to the quality of initial point O away from being respectively as follows:

$\left\{\begin{array}{c}{M}_{ti}=E_i.d_{ti},\\ M_{fi}=E_i.d_{fi},\\ M_{oi}=E_i.d_{oi},\end{array}\right.,i=1,...,N.---\left(10\right)$

Whole time-frequency plane to Two coordinate axle and to the quality of initial point away from being respectively as follows:

$\left\{\begin{array}{c}{M}_t=\underset{i=1}{\overset{N}{\Σ}}{M}_{ti},\\ M_f=\underset{i=1}{\overset{N}{\Σ}}{M}_{fi},\\ M_o=\underset{i=1}{\overset{N}{\Σ}}{M}_{oi},\end{array}\right.i=1,...,N.---\left(11\right)$

To the quality of each time-frequency block energy away from being normalized, obtain:

$\left\{\begin{array}{c}{q}_{ti}=M_{ti}/M_t,\\ q_{fi}=M_{fi}/M_i,\\ q_{oi}=M_{oi}/M_o,\end{array}\right.,i=1,...,N.---\left(12\right)$

Then have:

$\left\{\begin{array}{c}\underset{i=1}{\overset{N}{\Σ}}{q}_{ti}=1,\\ \underset{i=1}{\overset{N}{\Σ}}{q}_{fi}=1,\\ \underset{i=1}{\overset{N}{\Σ}}{q}_{oi}=1,\end{array}\right.,i=1,...,N.---\left(13\right)$

The quality of initial point O to time shaft, frequency axis and is defined respectively as by fault-signal time-frequency distributions away from entropy:

$\left\{\begin{array}{c}{s}_t\left(q\right)=-\underset{i=1}{\overset{N}{\Σ}}{q}_{ti}\ln q_{ti},\\ s_f\left(q\right)=-\underset{i=1}{\overset{N}{\Σ}}{q}_{fi}\ln q_{fi},\\ s_o\left(q\right)=-\underset{i=1}{\overset{N}{\Σ}}{q}_{oi}\ln q_{oi}.\end{array}\right.---\left(14\right)$

In formula, q_ti, q_fiAnd q_oiBe respectively i-th time-frequency block energy quality away from account for whole time-frequency distributions energy quality away from Ratio.

To the quality of time shaft away from entropy s_tQ () characterizes the time-frequency distributions complexity to frequency f, i.e. fault-signal energy is not The distribution situation of same frequency section；To the quality of frequency axis away from entropy s_fQ () characterizes the time-frequency distributions complexity to the time, i.e. fault letter The time-varying characteristics of number Energy distribution；To the quality of initial point O away from entropy s_oQ () characterizes the general complexity of time-frequency distributions.Quality is away from entropy (s_t(q),s_f(q),s_o(q)) complexity of fault-signal time-frequency distributions, and the relatively low applicable visualization of dimension can be measured all sidedly Analyzing, therefore the present invention is as fault feature vector during Fault Diagnosis of Hydraulic Pump.

2. case checking

The present invention uses vibration data during hydraulic plunger pump operation to verify effectiveness and the feasibility of proposition method, test Data acquisition is from plunger pump trouble injection testing platform.Testing stand is as it is shown in figure 5, stablize after 528r/min at motor speed, logical Cross the vibrating sensor being installed at plunger displacement pump end, obtain the vibration signal of testing stand with the sample frequency of 1000Hz.Divide successively Cai Ji not be analyzed by hydraulic pump system vibration data under normal condition, valve plate wear-out failure and piston shoes wear-out failure.

As shown in Figure 6, (a) is normal (b) for vibration signal under normal condition, valve plate wear-out failure and piston shoes wear-out failure Valve plate fault (c) piston shoes fault.

Vibration signal under each state is carried out Hilbert-Huang transform, obtains Hilbert Huang spectrum as it is shown in fig. 7, (a) Normally (b) valve plate fault (c) piston shoes fault.

The quality calculating each state Hilbert spectrum is such as schemed away from entropy, the fault signature dendrogram making each malfunction Shown in 8.

As it can be observed in the picture that the fault sample of health status of the same race flocks together, the fault sample spacing of different faults state From relatively big, quality is had excellent sort feature away from entropy as fault signature by this explanation, can be follow-up fault diagnosis work Good fault signature support is provided.

There is provided above case study on implementation to be only used to describe the purpose of the present invention, and be not intended to limit the scope of the present invention. The scope of the present invention is defined by the following claims.The various equivalents made without departing from spirit and principles of the present invention and Amendment, all should contain within the scope of the present invention.

Claims (1)

1. one kind based on Hilbert-Huang transform and quality the Hydraulic pump fault feature extracting method away from entropy, it is characterised in that:

(1) propose to be applicable to the quality of troubleshooting issue away from entropy, fully examine when quantifying fault-signal time-frequency distributions complexity Considering the positional information of time-frequency block, three quality of fault-signal two dimension time-frequency distributions are away from entropy (s_t(q),s_f(q),s_o(q)) concrete It is defined as follows:

Time-frequency plane is divided into the time-frequency block of N number of area equation, and every piece of interior energy is E_i, this time-frequency block energy is to time shaft The quality of t, frequency axis f and initial point O is away from being respectively as follows:

$\left\{\begin{array}{c}{M}_{ti}=E_i.d_{ti},\\ M_{fi}=E_i.d_{fi},\\ M_{oi}=E_i.d_{oi},\end{array}\right.,i=1,...,N.$

Whole time-frequency plane to the quality of time shaft, frequency axis and initial point away from being respectively as follows:

$\left\{\begin{array}{c}{M}_t=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{M}_{ti},\\ M_f=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{M}_{fi},\\ M_o=\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{M}_{oi},\end{array}\right.,i=1,...,N.$

To the quality of each time-frequency block energy away from being normalized, obtain:

$\left\{\begin{array}{c}{q}_{ti}=M_{ti}/M_t,\\ q_{fi}=M_{fi}/M_i,\\ q_{oi}=M_{oi}/M_o,\end{array}\right.,i=1,...,N.$

Then have:

$\left\{\begin{array}{c}\underset{i=1}{\overset{N}{\Σ}}{q}_{ti}=1,\\ \underset{i=1}{\overset{N}{\Σ}}{q}_{fi}=1,\\ \underset{i=1}{\overset{N}{\Σ}}{q}_{oi}=1,\end{array}\right.,i=1,...,N.$

Fault-signal time-frequency distributions to time shaft quality away from entropy s_tQ (), to frequency axis quality away from entropy s_t(q) and the quality to initial point O Away from entropy s_oQ () is defined respectively as:

$\left\{\begin{array}{c}{s}_t\left(q\right)=-\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{q}_{ti}\ln q_{ti},\\ s_f\left(q\right)=-\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{q}_{fi}\ln q_{fi},\\ s_o\left(q\right)=-\underset{i=1}{\overset{N}{\mathrm{\Σ}}}{q}_{oi}\ln q_{oi}.\end{array}\right.$

In formula, q_ti, q_fiAnd q_oiIt is respectively i-th time-frequency block energy quality frequency division in time accounting for whole to each coordinate axes or initial point Cloth energy relative to respective coordinates axle or initial point quality away from ratio；

To the quality of time shaft away from entropy s_tQ () characterizes the time-frequency distributions complexity to frequency f, i.e. fault-signal energy at difference frequency The distribution situation tolerance of rate section；To the quality of frequency axis away from entropy s_fQ () characterizes the time-frequency distributions complexity to the time, i.e. fault letter The time-varying characteristics tolerance of number Energy distribution；To the quality of initial point O away from entropy s_oQ () characterizes the general complexity of time-frequency distributions；

(2) Hilbert-Huang transform and quality are combined away from entropy, propose a kind of fault signature being applicable to process non-stationary signal Extracting method, empirical mode decomposition for being decomposed into a series of natural mode of vibration component, Martin Hilb adaptively by vibration signal Special conversion is used for calculating instantaneous amplitude and instantaneous frequency thus obtains hilbert spectrum, finally uses quality frequency division when entropy quantifies The complexity of cloth, as Hydraulic pump fault feature.