CN103454537A - Wind power generation low-voltage ride-through detection equipment and method based on wavelet analysis - Google Patents

Abstract

The invention relates to wind power generation low-voltage ride-through detection equipment and method based on wavelet analysis. The work process of the equipment includes that a detected signal is converted into an analog signal through a case wiring terminal, firstly after voltage sensor and spectrum-aliasing-resistant filter processing procedures, data collection is carried out through a data collecting panel, the collected data are sent to an industrial personal computer, and then the industrial personal computer sends the collected data for network communication, instrument control and data output through a LabVIEW software program. The method of the equipment includes the steps of (1) designing a filter to carry out denoising processing on the signal, matching or partially revising the signal through a defined structural element, and decomposing or extracting or deforming components of various forms, and (2) positioning and detecting transient state power quality disturbance through multi-layer decomposition by means of a wavelet transform singularity detection principle. The equipment can accurately position and analyze amplitude declination, phase hopping and three-phase voltage imbalance of power grid voltage.

Description

Wind-power electricity generation low voltage crossing checkout equipment and method based on wavelet analysis

Technical field

The invention belongs to wind-power electricity generation voltage and fall the monitoring field, especially a kind of wind-power electricity generation low voltage crossing checkout equipment and method based on wavelet analysis.

Background technology

Along with the continuous growth of wind-power electricity generation proportion, the operation of wind energy turbine set becomes further important to the impact of system stability.And one of the low voltage ride-through capability key that interconnection technology requires just.In the countries concerned's standard, this transient state standard is not really also had to complete related content now.Due to the defect of traditional fourier transform analysis method, can't complete the determination and analysis work to transient power quality.The difficult point that voltage falls detection is the signal of sudden change and non-stationary.And along with the development of wavelet transformation theory, with constantly improving, made wavelet transformation further extensive in the application of transient power quality detection field in recent years, developing rapidly of wind-powered electricity generation makes the shared power supply proportion of wind-powered electricity generation increase rapidly in recent years, the operation of wind energy turbine set also becomes and can not be ignored the impact of system stability, except the technical merit of wind-powered electricity generation unit self being improved, also the wind-powered electricity generation unit has been had to stricter interconnection technology requirement, and one of formal this key wherein of low voltage crossing (LVRT) ability.Low voltage ride-through capability refers to that, when the grid-connected point voltage of blower fan falls, it is grid-connected that blower fan can keep, and even can provide the part reactive power to electrical network, supports line voltage to recover, until power system restoration is normal, thus " passing through " this low-voltage process.Therefore the measurement of rated voltage being fallen to characteristic quantity is very important, is also progressive one to take to administer and foundation and the prerequisite of control measure.

Due to reasons such as equipment installation, external electromagnetic, make in transient signal often to be mingled with noise, random noise is presented as singular point in signal.And comprise a large amount of high-frequency informations in transient signal simultaneously, to retain accordingly and delete with useless information useful, therefore in transmission and storage, packed data remains an indispensable part, and its denoising and to be compressed on its way of thinking be basically identical.And for the difficult point of these detections to sudden change, input transient state, non-stationary and classification, traditional Fourier transform can not be extracted the transient state feature of signal completely, is presenting the defect that time-frequency can not localize aspect the transient power quality signal analysis.And wavelet transformation has been widely used in the detection and Identification electrical energy power quality disturbance because of its good Time-Frequency Localization feature and the ability of processing jump signal.But the problems such as the inhibition that wavelet transformation exists again block overlap of frequency bands, leakage effect and paired pulses to disturb is not ideal enough, and the mathematical morphology of founding on the integral geometry basis provides a kind of filtering white noise and the very effective nonlinear filtering technique of impulsive noise, and mathematical morphology is fast to the processing speed of signal, time delay is little, being easy to hardware realizes, signal after processing, its phase place and amplitude Characteristics can not change.

Summary of the invention

The object of the invention is to overcome the deficiencies in the prior art, a kind of wind-power electricity generation low voltage crossing checkout equipment and method based on wavelet analysis is provided.

The present invention solves its technical matters and takes following technical scheme to realize:

A kind of wind-power electricity generation low voltage crossing checkout equipment based on wavelet analysis, this equipment comprises cabinet connection terminal and the voltage sensor, anti-spectral aliasing wave filter, data acquisition board and the industrial computer that connect successively, measured signal changes simulating signal into through the cabinet connection terminal, at first after superpotential passes sensor and anti-spectral aliasing filter process link, carry out data acquisition by data acquisition board, the data that collect are delivered to industrial computer, by industrial computer, by the LabVIEW software program of installing within it, send to network service, instrument control and data output.

And the model that described voltage sensor adopts is WBV411D07, the model that described data acquisition board adopts is NIPCI-6133.

A kind of method of the wind-power electricity generation low voltage crossing checkout equipment based on wavelet analysis comprises that step is as follows:

(1) design a kind of wave filter based on mathematical morphology signal is carried out to denoising, utilize predefined structural elements to be mated or local correction signal, the various form components that concentrate in the object form are decomposed, extracted or are out of shape, thereby reached the purpose of extracting signal, keeping details and inhibition noise;

(2) utilize wavelet transformation Singularity Detection principle to decompose transient power quality disturbance is positioned and detects through multilayer.

And the concrete grammar step of described step (1) is:

1. pre-define structural unit

Structural element based on the morphology processing method is filled and is surveyed, the key factor that affects filtering performance is choosing of structural element shape, the selection of structural element mainly comprises the shape of structural element, the namely straight line of structural element, rectangle, ellipse, cosine, width and height, in order to improve computing velocity, choose the linear structure element;

2. signal is mated, and the data after coupling are revised

Different signals is determined to choosing of structural element length, the range of choice of structural element length:

L=f _s/2/f ₁

In formula, f _sfor sample frequency, the structural element of this length can be eliminated frequency and be greater than f ₁signal and noise, and, for filtering impulsive noise, the width of linear structure element is greater than the width of maximum impulse in sampled signal,

3. signal is carried out to denoising

The computing of mathematical morphology take the corrosion and expand these two kinds of fundamental operations as the basis, and draw the cascading of open and close and switching, morphological transformation is divided into binary morphological transform and many-valued morphological transformation, and power sampling signal is one-dimensional signal, for the many-valued morphological transformation in the one-dimensional discrete situation, is:

If f (n) is the one dimension multi-valued signal that sampling obtains, g (x) is the one-dimentional structure element sequence, and A, B are respectively the field of definition of f (n) and g (x), A={0,1,2, ..., N-1}, B={0,1,2 ..., M-1}, and M<N is arranged, and discrete signal f (n) (n ∈ A) is defined as respectively about morphological erosion and the dilation operation of g (x):

(fΘg)(n)=min[f(n+m)-g(m)]

((n+m)∈A,m∈B)

(f⊕g)(n)=max[f(n-m)+g(m)]

((n-m)∈A,m∈B)

In formula, Θ is erosion operation; ⊕ is dilation operation; N=1,2 ..., N.

First after the corrosion, expansion has just formed opening operation, and the post-etching that first expands has just formed closed operation, and opening operation and closed operation are defined as respectively:

(fоg)(n)=(fΘg⊕g)(n)

(f·g)(n)=(f⊕gΘg)(n)

Here we construct a class broad sense open-close and close-Kai wave filter, are defined as:

G _OC(f(n))=(fоg ₁·g ₂)(n)

G _CO(f(n))=(f·g ₁оg ₂)(n)

For the open-close wave filter, first utilize opening operation filtering positive pulse noise, strengthened the negative pulse noise simultaneously, the structural element that recycling length is greater than prime carries out closed operation, eliminates the negative pulse noise strengthened, and is also same for close-Kai wave filter;

When electric power signal detects, first should determine the best weights coefficient, but, in order to improve computing velocity, according to the statistics shift phenomenon existed in Generalized Morphological, the output amplitude of broad sense open-close wave filter is little and output amplitude Generalized Closed-Kai wave filter is large, can directly get:

$y\left(n\right)=\frac{G_{\mathrm{oc}}\left(f\left(n\right)\right)}{2}+\frac{G_{\mathrm{co}}\left(f\left(n\right)\right)}{2}$

Choose top two formula mean value filtering white noise and impulsive noises.

And the concrete grammar step of described step (2) is:

1. determine wavelet transformation base small echo

If θ (t) is a low pass smooth function that plays smoothing effect, and meets the following conditions

${\&Integral;}_{-\&infin;}^{\&infin;}\mathrm{\&theta;}\left(t\right)\mathrm{dt}=1---(3-28)$

With $\underset{\left|t\right|\&RightArrow;\&infin;}{\lim }\mathrm{\&theta;}\left(t\right)=0---(3-29)$

From the principle of Fourier conversion, its derivative ψ ⁽¹⁾(t) must be the band pass function, and meet the tolerable condition of small echo;

2. transient power quality disturbance is positioned, now to signal f (t) after denoising, corresponding continuous wavelet transform is:

$W_{\mathrm{\&psi;}}f(s,u)=f\left(t\right)*{\mathrm{\&psi;}}_s^{\left(1\right)}\left(t\right)=f\left(t\right)*\left(s^{1/2}\frac{{\mathrm{d\&theta;}}_s\left(t\right)}{\mathrm{dt}}\right)=s^{1/2}\frac{d}{\mathrm{dt}}[f\left(t\right)*{\mathrm{\&theta;}}_s\left(t\right)]$

It is signal differentiate again after level and smooth, it is equivalent to directly with the derivative of smooth function, signal be processed, for various transient power quality signals, occurring constantly and a sudden change all can occur in voltage waveform the finish time, by wavelet transformation, this sudden change can be amplified out, thereby can detect this sudden change, can detect the moment and the amplitude of LVRT Capability of Wind Turbine Generator start-stop.

Advantage of the present invention and good effect are:

The present invention utilizes a kind of method for detecting voltage drop based on the morphological wavelet algorithm, improved the defect of conventional Fourier transform to not localizing in the transient power quality analysis, a kind of nonlinear filtering technique that recycling provides on mathematical morphology, solved the not ideal enough and transient power quality signal characteristic of the interference inhibiting effect of block overlap of frequency bands that wavelet transformation exists, leakage effect, paired pulses with the yardstick increase problem such as weakened gradually.Accurately electrical network is broken down and produce that the line voltage amplitude is fallen, phase hit, imbalance of three-phase voltage positions and analyze, and the low voltage ride-through capability of wind-powered electricity generation unit has been done to systematic assessment.

The accompanying drawing explanation

Fig. 1 is apparatus of the present invention hardware structure diagrams.

Embodiment

Below in conjunction with accompanying drawing, the embodiment of the present invention is further described, it is emphasized that following embodiment is illustrative, rather than determinate, can not using this embodiment as limitation of the invention.

A kind of wind-power electricity generation low voltage crossing checkout equipment based on wavelet analysis, as shown in Figure 1, this equipment comprises cabinet connection terminal and the voltage sensor connected successively, anti-spectral aliasing wave filter, data acquisition board and industrial computer, measured signal changes simulating signal into through the cabinet connection terminal, at first after superpotential passes sensor and anti-spectral aliasing filter process link, carry out data acquisition by data acquisition board, the data that collect are delivered to industrial computer, send to network service by industrial computer by the LabVIEW software program of installing within it, instrument is controlled and data output.

In specific embodiment of the invention, the model that described voltage sensor adopts is WBV411D07, and the model that described data acquisition board adopts is NI PCI-6133.

A kind of method of the wind-power electricity generation low voltage crossing checkout equipment based on wavelet analysis comprises that step is as follows:

(1) design a kind of wave filter based on mathematical morphology signal is carried out to denoising.Utilize predefined structural elements (being equivalent to spectral window) to be mated or local correction signal, the various form components that concentrate in the object form are decomposed, extracted or are out of shape, thereby reached the purpose of extracting signal, keeping details and inhibition noise.Its concrete steps are:

1. pre-define structural unit

The thought of morphology processing method is based on structural element and fills detection, and the key factor that affects filtering performance is choosing of structural element shape.The selection of structural element mainly comprises shape (straight line, rectangle, ellipse, cosine etc.), width (width of structural element field of definition) and height (amplitude of structural element) these key elements of structural element, if the shape of structural element is more complicated, width is longer, its filter capacity is just stronger, calculated amount is also larger, in order to improve computing velocity, choose the linear structure element;

2. signal is mated, and the data after coupling are revised;

Mainly that different signals is determined to choosing of structural element length, the general range of choice of structural element length:

L=f _s/2/f ₁

In formula, f _sfor sample frequency, the structural element of this length can be eliminated frequency and be greater than f ₁signal and noise,

And, for filtering impulsive noise, the width of linear structure element is greater than the width of maximum impulse in sampled signal.3. signal is carried out to denoising

The computing of mathematical morphology be take corrosion and these two kinds of fundamental operations of expanding as basis, and draws the cascading of open and close and switching.Morphological transformation is divided into binary morphological transform and many-valued morphological transformation.Power sampling signal is generally one-dimensional signal, first introduces the many-valued morphological transformation in the one-dimensional discrete situation;

If f (n) is the one dimension multi-valued signal that sampling obtains, g (x) is the one-dimentional structure element sequence, and A, B are respectively the field of definition of f (n) and g (x), A={0, and 1,2 ..., N-1}, B={0,1,2 ..., M-1}, and M<N is arranged.Discrete signal f (n) (n ∈ A) is defined as respectively about morphological erosion and the dilation operation of g (x):

(fΘg)(n)=min[f(n+m)-g(m)]

((n+m)∈A,m∈B)

(f⊕g)(n)=max[f(n-m)+g(m)]

((n-m)∈A,m∈B)

In formula, Θ is erosion operation; ⊕ is dilation operation; N=1,2 ..., N;

First after the corrosion, expansion has just formed opening operation, and the post-etching that first expands has just formed closed operation.Opening operation can be obtained and closed operation is defined as respectively

(fоg)(n)=(fΘg⊕g)(n)

(f·g)(n)=(f⊕gΘg)(n)

Here we construct a class broad sense open-close and close-Kai wave filter, are defined as

G _OC(f(n))=(fоg ₁·g ₂)(n)

G _CO(f(n))=(f·g ₁оg ₂)(n)

For the open-close wave filter, first utilize opening operation filtering positive pulse noise, strengthened the negative pulse noise simultaneously.The structural element that recycling length is greater than prime carries out closed operation, eliminates the negative pulse noise strengthened, and is also same reason for close-Kai wave filter;

When electric power signal detects, first should determine the best weights coefficient, but, in order to improve computing velocity, according to the statistics shift phenomenon existed in Generalized Morphological, the output amplitude of broad sense open-close wave filter is little and output amplitude Generalized Closed-Kai wave filter is large, can directly get

$y\left(n\right)=\frac{G_{\mathrm{oc}}\left(f\left(n\right)\right)}{2}+\frac{G_{\mathrm{co}}\left(f\left(n\right)\right)}{2}$

Choose top two formula mean value filtering white noise and impulsive noises.

(2) utilize wavelet transformation Singularity Detection principle to decompose transient power quality disturbance is positioned and detects through multilayer, concrete steps are:

1. determine wavelet transformation base small echo

If θ (t) is a low pass smooth function that plays smoothing effect, and meets the following conditions

${\&Integral;}_{-\&infin;}^{\&infin;}\mathrm{\&theta;}\left(t\right)\mathrm{dt}=1---(3-28)$

With $\underset{\left|t\right|\&RightArrow;\&infin;}{\lim }\mathrm{\&theta;}\left(t\right)=0---(3-29)$

From the principle of Fourier conversion, its derivative ψ ⁽¹⁾(t) must be the band pass function, and meet the tolerable condition of small echo, so can be used as the wavelet of wavelet transformation.

2. transient power quality disturbance is positioned.Now to signal f (t) after denoising, corresponding continuous wavelet transform is:

$W_{\mathrm{\&psi;}}f(s,u)=f\left(t\right)*{\mathrm{\&psi;}}_s^{\left(1\right)}\left(t\right)=f\left(t\right)*\left(s^{1/2}\frac{{\mathrm{d\&theta;}}_s\left(t\right)}{\mathrm{dt}}\right)=s^{1/2}\frac{d}{\mathrm{dt}}[f\left(t\right)*{\mathrm{\&theta;}}_s\left(t\right)]$

Be signal differentiate again after level and smooth, it is equivalent to directly with the derivative of smooth function, signal be processed, like this, and wavelet transformation W _ψf (s, u) is exactly that signal f (t) is under yardstick s, by smooth function θ _s(t) first order derivative after level and smooth, the flex point that corresponding point is exactly function during for extreme value due to the absolute value (mould) of function first order derivative, and the catastrophe point that first order derivative absolute value (mould) corresponding point when the maximum value is exactly function, like this, can obtain when the base small echo is taken as the first order derivative of smooth function its wavelet transformation W _ψf (s, u) under each yardstick the modulus maximum point of coefficient corresponding to the position of signal f (t) singular point, for various transient power quality signals, occurring constantly and a sudden change all can occur in voltage waveform the finish time, by wavelet transformation, this sudden change can be amplified out, thereby can detect this sudden change, and under small scale, the wavelet coefficient modulus maximum is corresponding more accurate with the position of singular point, but it is larger to be subject to noise effect, can produce many pseudo-extreme points, contrary, under large scale, noise has been carried out to certain level and smooth extreme point relatively stable, but can make its position to singular point produce deviation, but preposition Generalized Morphological Filters has carried out filtering to signal, just can obtain effect preferably under less yardstick, can detect the moment and the amplitude of LVRT Capability of Wind Turbine Generator start-stop.

Claims (5)

1. the wind-power electricity generation low voltage crossing checkout equipment based on wavelet analysis, it is characterized in that: this equipment comprises cabinet connection terminal and the voltage sensor connected successively, anti-spectral aliasing wave filter, data acquisition board and industrial computer, measured signal changes simulating signal into through the cabinet connection terminal, at first after superpotential passes sensor and anti-spectral aliasing filter process link, carry out data acquisition by data acquisition board, the data that collect are delivered to industrial computer, send to network service by industrial computer by the LabVIEW software program of installing within it, instrument is controlled and data output.

2. the wind-power electricity generation low voltage crossing checkout equipment based on wavelet analysis according to claim 1 is characterized in that: the model that described voltage sensor adopts is WBV411D07, and the model that described data acquisition board adopts is NI PCI-6133.

3. the method for the wind-power electricity generation low voltage crossing checkout equipment based on wavelet analysis is characterized in that comprising that step is as follows:

(1) design a kind of wave filter based on mathematical morphology signal is carried out to denoising, utilize predefined structural elements to be mated or local correction signal, the various form components that concentrate in the object form are decomposed, extracted or are out of shape, thereby reached the purpose of extracting signal, keeping details and inhibition noise;

(2) utilize wavelet transformation Singularity Detection principle to decompose transient power quality disturbance is positioned and detects through multilayer.

4. the method for the wind-power electricity generation low voltage crossing checkout equipment based on wavelet analysis according to claim 3, it is characterized in that: the concrete grammar step of described step (1) is:

1. pre-define structural unit

Structural element based on the morphology processing method is filled and is surveyed, the key factor that affects filtering performance is choosing of structural element shape, the selection of structural element mainly comprises the shape of structural element, the namely straight line of structural element, rectangle, ellipse, cosine, width and height, in order to improve computing velocity, choose the linear structure element;

2. signal is mated, and the data after coupling are revised

Different signals is determined to choosing of structural element length, the range of choice of structural element length:

L=f _s/2/f ₁

In formula, f _sfor sample frequency, the structural element of this length can be eliminated frequency and be greater than f ₁signal and noise, and, for filtering impulsive noise, the width of linear structure element is greater than the width of maximum impulse in sampled signal,

3. signal is carried out to denoising

The computing of mathematical morphology take the corrosion and expand these two kinds of fundamental operations as the basis, and draw the cascading of open and close and switching, morphological transformation is divided into binary morphological transform and many-valued morphological transformation, and power sampling signal is one-dimensional signal, for the many-valued morphological transformation in the one-dimensional discrete situation, is:

If f (n) is the one dimension multi-valued signal that sampling obtains, g (x) is the one-dimentional structure element sequence, and A, B are respectively the field of definition of f (n) and g (x), A={0,1,2, ..., N-1}, B={0,1,2 ..., M-1}, and M<N is arranged, and discrete signal f (n) (n ∈ A) is defined as respectively about morphological erosion and the dilation operation of g (x):

(fΘg)(n)=min[f(n+m)-g(m)]

((n+m)∈A,m∈B)

(f⊕g)(n)=max[f(n-m)+g(m)]

((n-m)∈A,m∈B)

In formula, Θ is erosion operation; ⊕ is dilation operation; N=1,2 ..., N;

First after the corrosion, expansion has just formed opening operation, and the post-etching that first expands has just formed closed operation, and opening operation and closed operation are defined as respectively:

(fоg)(n)=(fΘg⊕g)(n)

(f·g)(n)=(f⊕gΘg)(n)

Here we construct a class broad sense open-close and close-Kai wave filter, are defined as:

G _OC(f(n))=(fоg ₁·g ₂)(n)

G _CO(f(n))=(f·g ₁оg ₂)(n)

For the open-close wave filter, first utilize opening operation filtering positive pulse noise, strengthened the negative pulse noise simultaneously, the structural element that recycling length is greater than prime carries out closed operation, eliminates the negative pulse noise strengthened, and is also same for close-Kai wave filter;

When electric power signal detects, first should determine the best weights coefficient, but, in order to improve computing velocity, according to the statistics shift phenomenon existed in Generalized Morphological, the output amplitude of broad sense open-close wave filter is little and output amplitude Generalized Closed-Kai wave filter is large, can directly get:

$y\left(n\right)=\frac{G_{\mathrm{oc}}\left(f\left(n\right)\right)}{2}+\frac{G_{\mathrm{co}}\left(f\left(n\right)\right)}{2}$

Choose top two formula mean value filtering white noise and impulsive noises.

5. the method for the wind-power electricity generation low voltage crossing checkout equipment based on wavelet analysis according to claim 3, it is characterized in that: the concrete grammar step of described step (1) is:

1. determine wavelet transformation base small echo

If θ (t) is a low pass smooth function that plays smoothing effect, and meets the following conditions

${\&Integral;}_{-\&infin;}^{\&infin;}\mathrm{\&theta;}\left(t\right)\mathrm{dt}=1---(3-28)$

With $\underset{\left|t\right|\&RightArrow;\&infin;}{\lim }\mathrm{\&theta;}\left(t\right)=0---(3-29)$

From the principle of Fourier conversion, its derivative ψ ⁽¹⁾(t) must be the band pass function, and meet the tolerable condition of small echo;

2. transient power quality disturbance is positioned, now to signal f (t) after denoising, corresponding continuous wavelet transform is:

$W_{\mathrm{\&psi;}}f(s,u)=f\left(t\right)*{\mathrm{\&psi;}}_s^{\left(1\right)}\left(t\right)=f\left(t\right)*\left(s^{1/2}\frac{{\mathrm{d\&theta;}}_s\left(t\right)}{\mathrm{dt}}\right)=s^{1/2}\frac{d}{\mathrm{dt}}[f\left(t\right)*{\mathrm{\&theta;}}_s\left(t\right)]$

It is signal differentiate again after level and smooth, it is equivalent to directly with the derivative of smooth function, signal be processed, for various transient power quality signals, occurring constantly and a sudden change all can occur in voltage waveform the finish time, by wavelet transformation, this sudden change can be amplified out, thereby can detect this sudden change, can detect the moment and the amplitude of LVRT Capability of Wind Turbine Generator start-stop.

Images