CN115267446A - Power equipment partial discharge detection method based on multi-signal classification positioning algorithm - Google Patents

Abstract

The invention discloses a partial discharge detection method for power equipment based on a multi-signal classification and positioning algorithm, comprising the following steps: S1. Receive the partial discharge signal sent by the power equipment through an eight-element circular microphone array sensor; S2, receive the partial discharge signal sent by the power equipment. Fourier transform is performed on the time-domain digital signal with sound source information, and it is converted from the time domain to the frequency domain; S3. Extract the characteristic frequency domain of the frequency-domain discharge signal obtained in step S2, and find the best representative of the received signal. The frequency value of the discharge signal; S4, using a sound source localization algorithm suitable for multi-signal classification under the condition of small signal-to-noise ratio and small snapshot number, combined with the signal characteristic frequency value obtained in step S3, to the received signal The discharge signal is analyzed and located; the detection method proposed by the present invention adopts multi-parameter joint estimation, which is beneficial to more accurately locate the partial discharge position of the power equipment and achieve the purpose of effectively detecting partial discharge.

Description

A partial discharge detection method for power equipment based on multi-signal classification and location algorithm

technical field

The invention relates to the field of partial discharge detection of electric power equipment, in particular to a partial discharge detection method of electric power equipment based on a multi-signal classification and positioning algorithm.

Background technique

Abnormal partial discharge is an important factor leading to the reduction of the operating life of power equipment, and it is also the most common abnormal operating state of various power equipment. The location detection of partial discharge is a key engineering technology and an important basis for judging whether the operating status of many power equipment is normal. Timely detection of abnormal discharge and its accurate location are of great significance to prevent serious failures of power equipment.

At present, the methods for detecting partial discharge mainly include: electrical, optical, photo-acoustic and UHF methods. Among them, the ultrasonic wave itself has the characteristics of high frequency and short wavelength, and has a strong sense of direction for signal transmission, so the detection process is relatively simple. At the same time, the ultrasonic measurement method has been widely studied because it is easy to realize on-line detection, convenient for spatial positioning, and less affected by electrical interference.

The invention proposes a partial discharge detection method for electric equipment based on a multi-signal classification and positioning algorithm. The detection method adopts multi-parameter joint estimation, which is beneficial to more accurately locate the partial discharge position of the electric equipment, and achieves the purpose of effectively detecting partial discharge.

Contents of the invention

The purpose of the present invention is to provide a partial discharge detection method of electric power equipment based on a multi-signal classification and positioning algorithm, so as to obtain better detection effect in the process of detecting and locating the partial discharge of electric power equipment.

To achieve the above object, the present invention provides the following technical solutions:

A partial discharge detection method for electrical equipment based on a multi-signal classification and positioning algorithm, comprising the following steps:

S1. Receive the partial discharge signal sent by the power equipment through the eight-element circular microphone array sensor;

S2. Perform Fourier transform on the received time-domain digital signal with sound source information, and convert it from time domain to frequency domain;

S3. Extracting the characteristic frequency of the frequency-domain discharge signal obtained in step S2, and finding the frequency value that best represents the received discharge signal;

S4. Using a sound source localization algorithm suitable for multi-signal classification in the case of a small signal-to-noise ratio and a small number of snapshots, combined with the signal characteristic frequency value obtained in step S3, analyzing and locating the received discharge signal;

As a further solution of the present invention: in step S1, the eight-element circular array used is a planar microphone sensor array.

As a further solution of the present invention: in step S1, the eight-element circular array can receive two position parameters of the signal, the pitch angle and the azimuth angle, and the joint estimation of these two parameters can realize three-dimensional effective positioning in space.

As a further solution of the present invention: in step S2, the object of Fourier transform is the time-domain state of the received discharge digital signal, wherein the amount of signal data received by a single microphone sensor is associated with the number of snapshots, and is Signal length parameter in the Fourier transform.

As a further solution of the present invention: in step S2, the signal strength converted into the frequency domain is represented by the size of the ordinate, and the frequency corresponding to the part with high signal strength should be more considered in step S3.

As a further scheme of the present invention: in step S4, the application condition of the sound source localization algorithm applicable to the multi-signal classification under the situation of small signal-to-noise ratio and small number of snapshots is that when the signal can be considered as a narrowband signal, for the narrowband signal Sk(t) can be expressed as:

S _k (tt ₁ )≈S _k (t)

Wherein, t1 is the required delay time between array microphone units.

As a further solution of the present invention: in step S4, the received discharge signal will first pass through to obtain a covariance matrix, and then perform subsequent signal classification.

Wherein, X(i) is the received i-th signal data, X ^H (i) is the Hermitian matrix of X(i), and N is the total number of received data.

As a further solution of the present invention: since it is a planar array, in step S4, the time delay τ _{m, k} involved in the algorithm should be:

为信号源k的俯仰角和方位角，r为圆形阵列的半径，M为阵元数目。Among them, τ _m,k is the relative delay between signal source k and array element m relative to the center of the array, θ _k and

is the pitch angle and azimuth angle of the signal source k, r is the radius of the circular array, and M is the number of array elements.

As a further solution of the present invention: since it is a planar array, in step S4, the direction matrix A involved in the algorithm should be:

为阵列接收第i个信号源信息而产生的方向向量。in,

The direction vector generated for the array receiving the i-th signal source information.

As a further solution of the present invention: in step S4, the spectral function that determines the final search result of the multi-signal classification algorithm is:

为传统谱函数P对方位角

求二阶偏导的结果，

为传统谱函数P对俯仰角θ_k求二阶偏导的结果。in,

For the traditional spectral function P versus azimuth

Find the result of the second order partial derivative,

It is the result of calculating the second-order partial derivative of the traditional spectral function P with respect to the pitch angle θ _k .

Compared with prior art, the beneficial effect of the present invention is:

1. The eight-element cross microphone array is used to receive the discharge signal. Its planar array structure enables the estimation of the position of the sound source to be extended to the two parameters of pitch angle and azimuth angle. The joint estimation method of multiple parameters is beneficial to more accurately determine the discharge sound. The location information of the source can effectively complete the spatial three-dimensional positioning of partial discharge.

2. Through the Fourier transform of the received signal, the spectrum information of the discharge signal is obtained, and the main characteristic frequency is extracted from it as the value of the frequency parameter in the positioning algorithm, which refines the value of the frequency parameter in the multi-signal classification algorithm. It is beneficial to strengthen the positioning reliability of the signal.

3. In the step S4, a new spatial spectrum estimation function is used as the basis for determining the peak value after the global search. Compared with the traditional spatial spectrum function, this method has a small signal-to-noise ratio and a small number of snapshots. , it has a better positioning and detection effect for the partial discharge of power equipment.

Description of drawings

Fig. 1 is the algorithm flowchart of the present invention.

FIG. 2 is a simulation diagram of the eight-element circular sensor array used in the present invention when receiving far-field discharge signals.

Detailed ways

Below in conjunction with accompanying drawing, the technical scheme of this patent is described in further detail.

Referring to accompanying drawing 1, a kind of electric equipment partial discharge detection method based on multi-signal classification positioning algorithm comprises the following steps:

S1. Receive the partial discharge signal sent by the power equipment through the eight-element circular microphone array sensor;

When detecting partial discharge of electrical equipment, firstly, the general orientation of the eight-element circular microphone array is directed towards the electrical equipment to be detected, and lasts for a period of time to receive the discharge signal of the electrical equipment.

S2. Perform Fourier transform on the received time-domain digital signal with sound source information, and convert it from time domain to frequency domain;

The signal received by each microphone is Fourier transformed, and it is converted from the time domain to the frequency domain for analysis. Among them, the object of Fourier transform is the time domain state of the received discharge digital signal, the signal data volume received by a single microphone sensor is associated with the number of snapshots, and is the length parameter of the signal in Fourier transform.

S3. Extracting the characteristic frequency of the frequency-domain discharge signal obtained in step S2, and finding the frequency value that best represents the received discharge signal;

Due to the certain volatility of the discharge signal and the interference of surrounding noise, even after filtering, the obtained frequency domain information will not be a specific frequency, but the superposition of many frequency signals with different intensities. Among them, the larger signal strength is the part of the discharge signal. In the region where the signal amplitude is larger, select an appropriate frequency as the main frequency of the received discharge signal, and in the algorithm of the next step S4, use this value Bring it into the signal frequency parameters involved in the algorithm.

S4. Utilize a sound source localization algorithm suitable for multi-signal classification in the case of small signal-to-noise ratio and small number of snapshots, and combine the signal characteristic frequency value obtained in step S3 to analyze and locate the received discharge signal

The concrete improved multi-signal classification algorithm principle in step S4 of the present invention is as follows:

The multi-signal classification algorithm has the characteristics of high resolution, high precision and high stability, so it is suitable for the abnormal discharge location scene in power inspection.

In this scenario, when it is used to determine the incoming wave direction of a signal, it has the following advantages:

1. Can locate multiple sound sources;

2. The detection effect has high precision;

3. The antenna beam signal has high resolution;

4. Applicable to short data situations.

For the convenience of theoretical analysis, we make the following provisions:

1. Each test signal source has the same but unrelated polarization.

2. The signal sources are narrowband, while each source has the same center frequency ω0.

3. Assume that the number of test signal sources is D;

The acquisition array is a circular array composed of M (M>D) array elements; it should be noted that the value of M in the present invention is specifically 8, that is, the present invention uses an eight-element circular array to receive the discharge signal, but When describing the principle, let the number of array elements in the array be M.

Each element has the same properties and is isotropic in every direction.

4. The distance between the microphone elements is d, and the distance between the elements is not greater than half of the wavelength of the highest frequency signal.

5. The microphone array belongs to the far-field scene, that is, the received sound source signal can be simulated as a plane wave;

6. Both array elements and test signals are uncorrelated;

The variance σ ² is zero mean Gaussian noise n _m(t) ;

7. The characteristics of the receiving signal branches are the same.

Assuming that the number of signal sources propagating to the microphone array is k (k=1,2,...,D), and the wavefront signal is S _k (t), we assume it is a narrowband signal. Due to the simplified condition of the narrowband signal, S _k ( t) can be expressed as:

S _k (t) = s _k (t) exp[jω _k (t)]

where s _k (t) is the complex envelope of S _k (t).

ω _k (t) is the angular frequency of S _k (t).

The center frequency ω ₀ is the same for all signals.

therefore

Among them, c is the signal wave speed.

λ is the wavelength.

at the same time,

λ=c/f

Bring in the frequency value obtained in step S3 of the present invention, and c is 340m/s, and λ can be obtained.

The time required for the delay between the array microphone units is assumed to be t ₁ .

Under the condition that the signal is narrowband, there are:

S _k (tt ₁ )≈S _k (t)

Therefore, the delayed wavefront signal is

The center of the circular array in the present invention is used as the reference point of the microphone sensor.

At time t, the induction signal of the kth signal source by the array element m (m=1,2,...,M) is

Among them, a _k is the influence coefficient of array element m on signal source k.

Because each array element does not have directionality, so set a _k =1;

τ _m,k is the relative delay between signal source k and array element m relative to the center of the array, which can be expressed as

Among them, r is the array radius.

和θ_k分别为信号源k的方位角和俯仰角。

and θ _k are the azimuth and elevation angles of the signal source k, respectively.

为附图2中信号源与正x轴间的负角，范围在0°-360°。

It is the negative angle between the signal source and the positive x-axis in Fig. 2, and the range is 0°-360°.

θ _k is the angle between the signal source and the positive direction of the array center normal in Figure 2, and the range is 0°-90°.

Considering the noise and incoming waves from all signal sources, the output signal of the mth array element is:

where n _m(t) is the measurement noise.

Write the above formula in vector form, and let a _k = l, then we have

X(t)＝α(θ _k )S _k (t)+N(t)

in,

X(t)=[x ₁ (t), x ₂ (t), ... x _M (t)] ^T

N(t)=[n ₁ (t), n ₂ (t), ..., n _M (t)] ^T

but,

X(t)=AS(t)+N(t)

in

S(t)=[S ₁ (t), S ₂ (t), ..., S _D (t)] ^T

So far, the problem becomes to sample x _m (t), and then estimate the direction of arrival of signal source k from {x _m (i), i=1, 2, . . . , M}.

For the array output x(t), its covariance matrix R is

R=E[X(t)X ^H (t)]

The signal and noise are uncorrelated, and the noise is zero-mean white noise.

Therefore the following results can be obtained:

R＝E[(AS+N)(AS+N) ^H ]

＝AE[SS ^H ]A ^H +E[NN ^H ]

＝ ^APAH +RN

Among them, P is the correlation matrix of the spatial signal, and R _N is the correlation matrix of the noise, which can be expressed as follows

P=E[S(t)S ^H (t)]

R _N =σ ² I

Among them, σ ² is the noise power, and I is the identity matrix of M×M.

i≠j时，矩阵A是由上式定义的范德蒙德矩阵。When θ _i ≠ θ _j ,

When i≠j, matrix A is a Vandermonde matrix defined by the above formula.

Since A is a Vandermonde matrix, each column is independent of each other.

In this way, if P is a non-singular matrix, there are:

rank(APA ^H )=D

Since P is positive definite, the eigenvalues of the matrix APA ^H are positive, that is, there are D positive eigenvalues in total. Therefore, R is a full-rank matrix with M eigenvalues.

We then arrange the eigenvalues in descending order, we have

λ ₁ ≥λ ₂ ≥…≥λ _M >0

Among the M eigenvalues, the larger D eigenvalues correspond to the signal, and the remaining smaller M-D eigenvalues correspond to the noise.

Therefore, the eigenvalues (eigenvectors) of R can be decomposed into signal eigenvalues (eigenvectors) and noise eigenvalues (eigenvectors).

Let λi be the ith eigenvalue of matrix R, and vi be the eigenvector corresponding to λi, then:

Rv _i =λ _i v _i

Let λ _i = σ ² be the minimum eigenvalue of R, then we have

Rv _i = σ ² v _i , i = D+1, D+2, ..., M

σ ² v _i =(APA ^H +σ ² I)v _i , i=D+1, D+2, . . . , M

Organized to get:

APA ^H v _z =0

Because A ^H A is a D×D dimensional full-rank matrix, (A ^H A) ^-1 exists, and so does P ^-1 . Seen from the above, multiplying P ^-1 (A ^H A) ^-1 A ^H on both sides of the above formula at the same time, we can get:

P ^-1 (A ^H A) ^-1 A ^H AP ^H v _i =0

therefore,

A ^H v _i =0, i=D+1, D+2,..., M

That is, the noise eigenvectors are perpendicular to the column vectors of the signal matrix.

That is to say, we find the vector orthogonal to the noise subspace (or the closest to it), and its direction represents the direction of arrival.

Construct a noise matrix using noise eigenvalues for each column:

E _n =[V _D+1 , V _D+2 , . . . , V _M ]

Then define the two-dimensional spatial spectrum:

where the denominator is the inner product of the signal matrix and the noise matrix.

Ideally, the value of the denominator is zero.

有一个峰值。However, due to the presence of noise in the environment when detecting electrical equipment, the denominator only has a minimum value, so

There is a peak.

发生变化，寻找二维空间谱的峰值，进而来确定声源位置。Search the two-dimensional space such that θ and

Change, find the peak of the two-dimensional spatial spectrum, and then determine the location of the sound source.

However, when the signal-to-noise ratio of the received signal is low and the number of signal data is small, the above-mentioned spatial spectrum estimation of the sound source position cannot achieve good results. The present invention proposes a multi-signal classification algorithm that improves the spatial spectrum function .

The invention calculates the second-order derivatives for the pitch angle and the azimuth angle respectively at the peak of the spectrum, constructs a new space spectrum function, and further improves the resolution of the algorithm.

的范围为

搜索间隔为

俯仰角θ的范围为Rθ，搜索间隔为△θ。Assume that the range of azimuth angles is angle

in the range of

The search interval is

The range of the pitch angle θ is Rθ, and the search interval is Δθ.

order

Then the spatial spectral function can be expressed as:

处离散函数P对自变量

的一阶偏导数为：It can be obtained by calculating the partial derivative of the variable by the binary discrete function, in

The discrete function P at the independent variable

The first partial derivative of is:

为P′_φk。remember

is P′ _φk .

处对

的二阶导数为：exist

right

The second derivative of is:

为P″_φk。remember

is P″ _φk .

处离散函数P对自变量θ的一阶导数和二阶导数分别为：in the same way

The first and second derivatives of the discrete function P with respect to the independent variable θ are:

与

分别为

remember

and

respectively

It can be seen from the analysis that the second derivative of the maximum point of the original spectral function forms a sharp negative spectral peak at the arrival angle of the azimuth and elevation angles.

From this, the new spectral function can be sorted out: that is, the values with partial derivatives greater than 0 are classified as 0, so the new spectral function can be obtained as:

和θ求二阶偏导是相互独立的，所以可以将原谱函数的二阶导数P”表示为：Since P is right

and θ are independent of each other, so the second derivative P" of the original spectral function can be expressed as:

A new spatial spectrum function can be formed by the above formula;

In step S4, after traversing the spatial angles through the algorithm, a new spectral peak of the spatial spectral function is found, and its corresponding azimuth and elevation angles are the estimated spatial position of the electric equipment discharge sound source.

The beneficial effects of the present invention are:

1. An eight-element cross microphone array is used to receive signals. Its planar array structure enables the estimation of the position of the sound source to be extended to the two parameters of pitch angle and azimuth angle. The joint estimation method of multiple parameters is beneficial to more accurately determine the discharge sound source The location information of the device effectively completes the spatial three-dimensional positioning of partial discharge.

2. Through the Fourier transform of the received signal, the spectrum information of the discharge signal is obtained, and the main characteristic frequency is extracted from it as the value of the frequency parameter in the positioning algorithm, which refines the value of the frequency parameter in the multi-signal classification algorithm. It is beneficial to strengthen the positioning reliability of the signal.

3. In the step S4, a new spatial spectrum estimation function is used as the basis for determining the peak value after the global search. Compared with the traditional spatial spectrum function, this method has a small signal-to-noise ratio and a small number of snapshots. , it has a better positioning and detection effect for the partial discharge of power equipment.

The above is only a specific embodiment of the present invention, but the design concept of the present invention is not limited thereto, and any non-substantial changes made to the present invention by using this concept should be an act of violating the protection scope of the present invention.

Claims (10)

1. A power equipment partial discharge detection method based on a multi-signal classification positioning algorithm is characterized by comprising the following steps: s1, receiving a partial discharge signal sent by power equipment through an eight-element circular microphone array sensor; s2, carrying out Fourier transformation on the received time domain digital signal with the sound source information, and converting the time domain digital signal into a frequency domain; s3, extracting characteristic frequency of the frequency domain discharge signal obtained in the step S2, and finding out a frequency value which can represent the received discharge signal most; and S4, analyzing and positioning the received discharge signal by using a multi-signal classification sound source positioning algorithm suitable for the conditions of small signal-to-noise ratio and small snapshot number and combining the signal characteristic frequency value obtained in the step S3. Wherein the more detailed operation of S4 is as follows: and (4) extracting the signal characteristic frequency value obtained in the step (S3), taking the frequency value as an integral frequency value representing the received discharge signal, bringing the frequency value into a signal frequency parameter designed by a multi-signal classification sound source positioning algorithm suitable for the conditions of small signal-to-noise ratio and small snapshot number, and further analyzing and positioning the received discharge signal.

2. The partial discharge detection method for the power equipment based on the multi-signal classification positioning algorithm according to claim 1, characterized in that: in step S1, the eight-element circular array employed is a planar microphone sensor array.

3. The partial discharge detection method for the power equipment based on the multi-signal classification positioning algorithm as claimed in claim 1, wherein: the eight-element circular array in the step S1 can receive two position parameters of a pitch angle and an azimuth angle of a signal, and the two position parameters can be jointly estimated, so that space three-dimensional effective positioning can be realized.

4. The partial discharge detection method for the power equipment based on the multi-signal classification positioning algorithm according to claim 1, characterized in that: in step S2, the object for which the fourier transform is directed is the time domain state of the received discharge digital signal, wherein the amount of signal data received by a single microphone sensor is associated with the number of snapshots and is a signal length parameter in the fourier transform.

5. The partial discharge detection method for the power equipment based on the multi-signal classification positioning algorithm as claimed in claim 1, wherein: in step S2, the signal strength after conversion into the frequency domain is represented by the size of the ordinate, and the frequency corresponding to the portion having a large signal strength should be considered more in step S3.

6. The partial discharge detection method for the power equipment based on the multi-signal classification positioning algorithm according to claim 1, characterized in that: in step S4, the sound source localization algorithm applicable to multi-signal classification in the case of small signal-to-noise ratio and small snapshot count is applied on the condition that when the signal can be regarded as a narrowband signal, for the narrowband signal S_k(t) has:

S_k(t-t₁)≈S_k(t)

wherein, t₁The time required for the delay between the array microphone units.

7. The partial discharge detection method for the power equipment based on the multi-signal classification positioning algorithm according to claim 1, characterized in that: in step S4, the received discharge signal is first processed by obtaining a covariance matrix, and then is subjected to subsequent signal classification.

Wherein X (i) is the ith received signal data, X^H(i) The Hermite matrix of X (i), N being the total number of data received.

8. The partial discharge detection method for the power equipment based on the multi-signal classification positioning algorithm according to claim 1, characterized in that: due to the planar array, in step S4, the time delay tau involved in the algorithm_m,kThe method comprises the following steps:

wherein, tau_m,kIs the relative delay, theta, of the signal source k and the array element m relative to the center of the array_kAnd

the pitch angle and azimuth angle of a signal source k, r is the radius of the circular array, and M is the number of array elements.

9. The partial discharge detection method for the power equipment based on the multi-signal classification positioning algorithm according to claim 1, characterized in that: since the array is a planar array, in step S4, the direction matrix a involved in the algorithm should be:

wherein,

a direction vector generated for the array to receive the D-th signal source information.

10. The partial discharge detection method for the power equipment based on the multi-signal classification positioning algorithm as claimed in claim 1, wherein: the spatial spectrum function of the improved multi-signal classification algorithm in step S4 is

Wherein,

and is disclosed in

Respectively, is a conventional spatial spectrum function P

Is aligned with

And theta_nThe second derivative of (a).

Images