CN114136249A - Novel denoising method for transformer winding deformation ultrasonic detection signal - Google Patents
Novel denoising method for transformer winding deformation ultrasonic detection signal Download PDFInfo
- Publication number
- CN114136249A CN114136249A CN202111441256.2A CN202111441256A CN114136249A CN 114136249 A CN114136249 A CN 114136249A CN 202111441256 A CN202111441256 A CN 202111441256A CN 114136249 A CN114136249 A CN 114136249A
- Authority
- CN
- China
- Prior art keywords
- signal
- point
- signals
- value
- algorithm
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
- G01B17/00—Measuring arrangements characterised by the use of infrasonic, sonic or ultrasonic vibrations
- G01B17/04—Measuring arrangements characterised by the use of infrasonic, sonic or ultrasonic vibrations for measuring the deformation in a solid, e.g. by vibrating string
Abstract
The application relates to a novel denoising method for transformer winding deformation ultrasonic detection signals, which comprises the following steps: collecting the transmitted wave and the received echo signal transmitted by the probe by using a collecting card; dividing the signal into three sections, namely clutter signals, transmitted wave signals and echo signals according to the transmitted wave oscillation starting point and the echo response point; denoising the clutter signals; judging whether sampling points of the transmitted wave signal and the echo signal are both larger than a set value; if the sampling points of the emission wave signal and the echo signal are both larger than a set value, denoising the emission wave signal and the echo signal respectively by adopting an ITD-PE-PCA algorithm; if the sampling points of the emission wave signal and the echo signal are not both larger than a set value, combining the emission wave signal and the echo signal, and then denoising the combined signal by adopting an ITD-PE-PCA algorithm; and splicing the denoised signals. The denoising method combines a signal extreme value mutation segmentation algorithm with ITD-PE-PCA, so that the correlation coefficient of the ultrasonic effective signal after being filtered and the original signal is higher, and the denoising effect is better.
Description
Technical Field
The application relates to the technical field of signal processing, in particular to a novel denoising method for transformer winding deformation ultrasonic detection signals.
Background
Ultrasonic signals have a wide range of applications, including mechanical flaw detection in industry, disease diagnosis in medical treatment, and surface topography exploration in geography. However, because the ultrasonic detection environment is generally complex and the interference of the noise of the equipment is added, the finally obtained ultrasonic sampling signal is inevitably mixed with a certain amount of noise components, and the result of the ultrasonic detection is greatly interfered in a severe case.
In order to reduce the interference of noise to the ultrasonic detection signal, fourier transform, wavelet transform and other methods are used for denoising the ultrasonic signal, but they also have some unavoidable disadvantages. With the advent of new time-frequency analysis algorithms, such as Empirical Mode Decomposition (EMD) algorithms, Local Mean Decomposition (LMD) algorithms, and their improved algorithms, non-stationary signal analysis techniques have become more sophisticated. However, as the research on the EMD algorithm and its improved algorithm progresses, the problems of low processing efficiency, loss of partial time domain information, existence of false information in the vicinity of the end point of the decomposition component, etc. when the EMD processes signals also attract the attention of researchers. To address these issues with EMD, Frei, equal to 2006, proposed a new approach to nonlinear, non-stationary signals-inherent time scale decomposition (ITD). The method is more suitable for analyzing the nonlinear unstable signals with time-varying spectrums, and spline interpolation and screening processes are not needed, so that the method almost has no edge effect, has high calculation speed, and can process a large amount of data in real time. Konstantin et al combined variational knowledge in mathematics with modal decomposition in 2014 to form a Variational Modal Decomposition (VMD) algorithm. The method selects the decomposed frequency center and bandwidth by iteratively searching the optimal solution of the variation model, thereby avoiding the end effect and the spectrum aliasing.
Although the above-mentioned classical denoising methods are widely used in the engineering field, they all have various defects.
Disclosure of Invention
In order to solve or partially solve the problems in the related technology, the application provides a novel denoising method for transformer winding deformation ultrasonic detection signals, and a signal extreme value mutation segmentation algorithm is combined with ITD-PE-PCA, so that the correlation coefficient of ultrasonic effective signals after being filtered and the original signals is higher, and the denoising effect is better.
The application provides a transformer winding deformation ultrasonic detection signal novel denoising method in a first aspect, which comprises the following steps:
collecting the transmitted wave and the received echo signal transmitted by the probe by using a collecting card;
dividing the signal into three sections, namely clutter signals, transmitted wave signals and echo signals according to the transmitted wave oscillation starting point and the echo response point;
denoising the clutter signals;
judging whether sampling points of the transmitted wave signal and the echo signal are both larger than a set value;
if the sampling points of the emission wave signal and the echo signal are both larger than a set value, denoising the emission wave signal and the echo signal respectively by adopting an ITD-PE-PCA algorithm;
if the sampling points of the emission wave signal and the echo signal are not both larger than a set value, combining the emission wave signal and the echo signal, and then denoising the combined signal by adopting an ITD-PE-PCA algorithm;
and splicing the denoised signals.
Optionally, the sampling frequency of the acquisition card is greater than 2 times the center frequency of the ultrasonic wave.
Optionally, based on a signal extreme value detection algorithm, dividing the signal into three segments according to the determined launch wave oscillation starting point and the determined received wave response point.
Optionally, the signal extremum detecting algorithm is a piecewise method of abrupt change of signal extremum; the method specifically comprises the following steps:
s101, obtaining a maximum value point and a minimum value point of the ultrasonic signal through the characteristics of the extreme values, and recording corresponding moments of all the extreme values;
s102, finding the maximum absolute value of the maximum value point, calculating the absolute difference value of the adjacent extreme value points in the time reduction direction from the maximum value point of the maximum value point, and sequencing the absolute difference value sequence from large to small;
s103, finding out adjacent maximum point pairs corresponding to a part of large absolute difference value sequences (for example, the value in the sequences is larger than the absolute average value), calculating the absolute ratio of the large number divided by the small number in the adjacent maximum point pairs, and taking the minimum maximum value of the non-zero absolute value if the divisor is zero; setting the time corresponding to the minimum maximum value point in the adjacent maximum value points with the largest divisor as the starting oscillation point PH1 of the transmitted signal, and calculating the absolute difference value between the starting oscillation point PH1 and the next maximum value point as AH;
s104, similarly, determining a starting oscillation point PL1 of the transmitted signal by using the minimum value point, and solving an absolute difference AL between the starting oscillation point PL1 and the next minimum value point; comparing the sizes of AH and AL, if AH is larger than AL, the oscillation starting point is taken as PH1, otherwise, PL1 is taken;
s105, finding a first maximum point with an absolute value smaller than 0.1MH along with a corresponding time t1 from the time corresponding to the maximum MH according to the time increasing direction, finding all maximum points and minimum points from t1 to the sampling end, and recording the corresponding times of all the maximum points;
s106, the remaining steps of determining the echo signal response time are the same as steps S102 to S105.
Optionally, the algorithm for denoising the clutter signal is a wavelet threshold denoising algorithm, and the wavelet threshold denoising algorithm may extract white noise in the clutter signal.
Optionally, the set value is 1000; in order to deal with the problem that the sampling length of the ultrasonic effective signal may be insufficient, 1000 sampling points are selected as dividing points where both the transmitted wave and the echo can be divided.
Optionally, the ITD-PE-PCA algorithm includes:
decomposing the ultrasonic signal by using an inherent time scale decomposition (ITD) algorithm;
determining the number of ITD decompositions by using Permutation Entropy (PE);
and carrying out denoising processing on the signal component obtained by decomposing the ITD-PE by utilizing Principal Component Analysis (PCA).
Optionally, the inherent time-scale decomposition (ITD) algorithm comprises: and decomposing the ultrasonic signal sampling sequence to obtain a series of inherent rotation component sequences and a decomposition residue sequence.
Optionally, the Permutation Entropy (PE) can reflect the complexity of the one-dimensional time signal, is sensitive to the change of the complex signal, and can well amplify the micro-variability of the system.
Alternatively, the Principal Component Analysis (PCA) algorithm is a commonly used dimension reduction algorithm, since the noise signal has low energy and is generally not correlated to the ultrasound signal, and the noise signal can be mostly filtered in the PCA dimension reduction, so the PCA can also be used for denoising the signal.
The technical scheme provided by the application can comprise the following beneficial effects:
according to the method, the signal extreme value mutation segmentation algorithm is combined with the ITD-PE-PCA, so that the correlation coefficient of the ultrasonic effective signal after being filtered with the original signal is higher, and the denoising effect is better.
It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the application.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings used in the description of the embodiments will be briefly introduced below, and it is obvious that the drawings in the following description are only some embodiments of the present application, and it is obvious for those skilled in the art to obtain other drawings without creative efforts.
Fig. 1 is a flowchart of a segmentation denoising process for an ultrasonic signal in an embodiment of the present application.
Detailed Description
Embodiments of the present application will be described in more detail below with reference to the accompanying drawings. While embodiments of the present application are illustrated in the accompanying drawings, it should be understood that the present application may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.
The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. As used in this application and the appended claims, the singular forms "a", "an", and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It should also be understood that the term "and/or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.
It should be understood that although the terms "first," "second," "third," etc. may be used herein to describe various information, these information should not be limited to these terms. These terms are only used to distinguish one type of information from another. For example, first information may also be referred to as second information, and similarly, second information may also be referred to as first information, without departing from the scope of the present application. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of the present application, "a plurality" means two or more unless specifically limited otherwise.
The embodiment of the application provides a novel denoising method for transformer winding deformation ultrasonic detection signals, and a signal extreme value mutation segmentation algorithm is combined with ITD-PE-PCA, so that correlation coefficient with original signals is higher after ultrasonic effective signals are filtered, and denoising effect is better.
The technical solutions of the embodiments of the present application will be described in detail below with reference to the accompanying drawings.
Referring to fig. 1, the present embodiment provides a novel denoising method for ultrasonic detection signals of transformer winding deformation, including:
collecting the transmitted wave and the received echo signal transmitted by the probe by using a collecting card;
dividing the signal into three sections, namely clutter signals, transmitted wave signals and echo signals according to the transmitted wave oscillation starting point and the echo response point;
denoising the clutter signals;
judging whether sampling points of the transmitted wave signal and the echo signal are both larger than a set value;
if the sampling points of the emission wave signal and the echo signal are both larger than a set value, denoising the emission wave signal and the echo signal respectively by adopting an ITD-PE-PCA algorithm;
if the sampling points of the emission wave signal and the echo signal are not both larger than a set value, combining the emission wave signal and the echo signal, and then denoising the combined signal by adopting an ITD-PE-PCA algorithm;
and splicing the denoised signals.
In this embodiment, the transmitted wave and the received echo signal transmitted by the probe may be used to measure the deformation of the transformer winding. Specifically, a common transformer in an electric power system is an oil-immersed transformer, wherein a transformer winding is mainly manufactured by winding an insulated wire with good electrical conductivity on a silicon steel sheet. When the deformation of the winding is detected by adopting ultrasonic waves, the ultrasonic probe is selected as a transmitting-receiving integrated probe, and the ultrasonic waves are generated by the probe during actual measurement. The probe can continue to be used for receiving ultrasonic waves after the ultrasonic signals are transmitted, and the propagation distance of the ultrasonic waves can be calculated by recording the time (also called cross-over time) from the transmission of the signals to the reception of the signals and combining the propagation speed of the ultrasonic waves, so that the deformation of the winding can be obtained. It can be seen that the accuracy of the deformation of the transformer winding depends mainly on the calculation accuracy of the crossing time.
Ultrasonic wave is taken as a non-stationary time-varying signal, and when the ultrasonic wave propagates in different media, a reflection echo is generated at an interface between the media, wherein a bottom echo, a defect echo and a material scattering echo are main forms of the echo. The ultrasonic wave mainly passes through a transformer shell and insulating oil during propagation, the transformer shell is a non-uniform medium composed of cast iron metal with a large size in composition, the insulating oil is complex in composition, and oil molecules can also comprise other impurities and bubbles.
After the ultrasonic wave is emitted from the probe, the ultrasonic wave firstly needs to pass through the transformer shell, and only a small part of energy is lost. However, when the ultrasonic wave propagates in the insulating oil, because the insulating oil is not a uniform medium, the ultrasonic wave is easy to generate a scattering phenomenon, so that the amplitude of the ultrasonic echo is reduced, and a distorted signal of the ultrasonic wave is mixed. When ultrasonic waves are reflected by the winding, the surface of the winding can absorb part of signals, and echo signals pass through insulating oil and a shell and then have signal loss and signal distortion. In addition, vibration noise and environmental noise exist during normal operation of the transformer, thermal noise exists during operation of the acquisition equipment, and the noise is mixed in the sampling signal finally. Although the analog filter such as the high-pass filter can filter most of the noise in the low frequency band, the noise is introduced by the addition of the analog filter, and the analog filter is also difficult to filter the noise with smaller amplitude, so that a more complex noise signal needs to be further identified through an algorithm and denoised.
In the transformer winding deformation ultrasonic detection method, the ultrasonic signals collected by the acquisition card mainly comprise transmitted waves and echo signals. The ultrasonic signal can be divided into 3 segments, namely a transmitting wave signal, an echo signal and a clutter signal by utilizing the composition characteristics of the ultrasonic wave. Two division points are required to divide a time sequence into three small time sequences, and the division points are selected as a starting point of a transmitted wave signal and a response point of an echo signal. Although the objective of segmenting the ultrasonic signal can be achieved by directly observing the change of the extreme point of the ultrasonic signal, the observation method depends on the experience knowledge of the observer, so that the efficiency of observing the segmentation point is low, and the results obtained by different people are different. In order to quickly determine the segmentation point of the ultrasonic signal, a segmentation method of a signal extreme value mutation method is provided, which is carried out by following the observation process of an observation method and optimizing the observation result. The signal extreme value mutation method is realized as follows:
(1) and obtaining the maximum value point and the minimum value point of the ultrasonic signal through the characteristics of the extreme value points, and recording the corresponding moments of all the extreme value points.
(2) And finding the maximum absolute value of the maximum value point, calculating the absolute difference value of the adjacent maximum value points in the time reduction direction from the maximum value point of the maximum value point, and sequencing the absolute difference value sequence from large to small.
(3) And finding out adjacent maximum point pairs corresponding to partial large absolute difference value sequences (such as values in the sequences are larger than the absolute average value), calculating the absolute ratio of the large number divided by the small number in the adjacent maximum point pairs, and taking the minimum maximum value of the non-zero absolute value if the divisor is zero. And (3) determining the moment corresponding to the minimum maximum value in the adjacent maximum value points with the maximum divisor as the oscillation starting point PH1 of the transmitted signal, and calculating the absolute difference value between the oscillation starting point PH1 and the next maximum value point as AH.
(4) Similarly, the minimum value point may be used to determine the start point PL1 of the transmitted signal, and the absolute difference AL between the start point PL1 and the next minimum value point may be obtained. Comparing the sizes of AH and AL, if AH is larger than AL, the oscillation starting point is PH1, otherwise PL 1.
(5) Finding the maximum point with the absolute value less than 0.1MH along with the corresponding time t from the time corresponding to the maximum MH according to the time increasing direction1Find t1And all the maximum value points and minimum value points at the sampling end are reached, and the corresponding moments of all the maximum value points are recorded.
(6) The remaining steps of determining the echo signal response time are the same as steps (2) to (5).
The ITD algorithm has the characteristic of rapid decomposition, and is improved to form the ITD-PE-PCA algorithm in order to realize rapid denoising of the ultrasonic signals and ensure that the signal-to-noise ratio of denoising meets the requirements. The ITD-PE-PCA algorithm firstly decomposes an ultrasonic signal by using an inherent time scale decomposition (ITD) algorithm, the Permutation Entropy (PE) is used for determining the times of ITD decomposition, and the Principal Component Analysis (PCA) is used for denoising a signal component obtained by ITD-PE decomposition.
Decomposing the ultrasonic signal sample sequence by an inherent Time-scale Decomposition (ITD) can obtain a series of inherent rotation component sequences and a Decomposition margin sequence. In the ITD method, X is assumedt(t denotes time t) is the signal to be analyzed, let L be the baseline extraction operator. Applying L to the original signal XtThereafter, the signal remaining after the calculation is defined as an inherent rotation component. Let H be the eigen-rotation extraction operator, then H-1-L. Further obtaining XtThe first decomposition of (A) is as follows:
Xt＝LXt+(1-L)Xt＝Lt+Ht (1)
in the formula, LtRepresenting the baseline signal, HtRepresenting the inherent rotational component.
Let { taukK is 1, 2tAnd the extreme point is assumed to be at the end point. When X is presenttWhen the value in a certain time interval is constant, the extreme value tau is setkIs selected as the right end of the time interval. For simplification, X (τ) is separately addedk) And L (τ)k) Is represented by XtAnd Lt。
Let LtAnd HtDomain of [0, τ ]t]And X istIs determined bySense domain of [0, τ ]t+2]. At successive extreme points (τ)t，τt+1]Within the range, a defined baseline extraction operator L of
In the formula, Lk+1Is calculated as
In the formula, alpha is more than 0 and less than 1, and alpha is generally 0.5.
To initialize [0, τ ]1]The first point of the signal can be defined as the extreme point, and L is defined0＝X(τ0)+X(τ1)＝X0+X1. Then X is obtainedtIs expressed as
In the formula, HLkXtDenotes the intrinsic rotation component (PRC), L, obtained at the k +1 th decompositionpXtAnd (4) showing a monotonous trend term (margin) after the decomposition.
Similar to the Empirical Mode Decomposition (EMD) algorithm, the ITD algorithm also has an end-point effect. The end point effect means that the end point data is uncertain whether the end point is an extreme point or not, and influences the end point nearby value of the signal component obtained by decomposition. To suppress the end-point effect of ITD, continuation methods are generally used, including mirror continuation, parallel continuation, extremum continuation, etc. The extreme value continuation method is based on one characteristic wave of an end point, and two end points of the two end.
One disadvantage of the ITD algorithm is that the number of decompositions cannot be given by calculation. Although a better decomposition frequency of the ITD can be found out by a method of multiple tests, the method is low in efficiency, and the decomposition frequency tested is only suitable for a single application scene, so that the use of the ITD algorithm is limited. In order to avoid artificially specifying the number of times of decomposition of the ITD, it is considered to determine whether or not the decomposition is continued by the statistical characteristics of the decomposed signal. The method combines the permutation entropy and the ITD algorithm to form the ITD-PE algorithm, the ITD-PE algorithm judges whether the decomposition is continued or not by judging whether the permutation entropy of the surplus obtained by ITD decomposition is smaller than a certain threshold value, and the decomposition is stopped when the permutation entropy of the surplus is smaller than the certain threshold value.
The Permutation Entropy (PE) can reflect the complexity of a one-dimensional time signal, is sensitive to the change of the complex signal, and can well amplify the micro-denaturation of a system, and the algorithm principle is as follows.
Let the phase space reconstruction matrix corresponding to the temporal sequence { x (i), i ═ 1, 2.
Wherein d is the embedding dimension; τ is a time delay; m is the number of reconstruction components, and m is n- (d-1) τ. Each row in the matrix can be regarded as a reconstruction component, each row in the reconstruction matrix is arranged in ascending order, and jk (k ═ 1, 2, …, d) is defined as an index of a column where elements of the reconstruction component are located, that is, an index value of an element in an ith row of the matrix satisfies:
x(i+(j1-1)τ)≤x(i+(j2-1)τ)≤…≤x(i+(jd-1)τ)
when x (i + (j)p-1)τ)＝x(i+(jq-1) τ) (1 ≦ p, q ≦ d, and p, q ∈ Z), ordered by the size of the index value. I.e. if jp＜jq，x(i+(jp-1)τ)≤x(i+(jq-1)τ)。
Therefore, for any row in any time sequence and the corresponding reconstruction sequence Y, a set of symbol sequences can be obtained:
S(l)＝{j1，j2，...，jd}
s (1) has m! Calculating the frequency P of occurrence of each S (1)1，P2，…，PkIn this case, the permutation entropy Hp of the time series x (i) can be defined as:
and (4) normalizing the Hp to obtain:
the value of (c) may reflect the degree of randomness of the time series, with larger values being more random in the time series.
The permutation entropy algorithm needs to select the length n of the time sequence, the time delay tau and the number m of the reconstruction. The influence of the number of the reconstructed quantity on the permutation entropy algorithm is most obvious, the value of m is recommended to be between 3 and 7 by Bandt, if the value of m is too large, the time sequence can be homogenized by the reconstructed phase space matrix, and the random noise in the signal is inconvenient to evaluate; if m is too small, the state contained in the row vector in the reconstruction matrix is lost.
In the simulation echo signal denoising and real echo signal denoising tests, n is 2500, tau is 1, and m is 5. The PE threshold was taken to be 0.6 according to the research results [14] of the related papers.
Principal Component Analysis (PCA) is a commonly used dimensionality reduction algorithm. However, because the noise signal has low energy and is generally not correlated with the ultrasound signal, the noise signal can be mostly filtered in the dimensionality reduction of the PCA, and the PCA can also be denoised by the de-noising signal. The ITD-PE-PCA algorithm is formed by combining the ITD-PE and the PCA algorithm. The ITD-PE-PCA algorithm firstly carries out ITD-PE decomposition on an original sampling signal to obtain a plurality of signal components, the decomposed signal components are reconstructed and denoised by using PCA, and finally the denoised signal components are superposed to obtain a final denoised signal.
Time series
(j is not less than 1 and not more than m and Xj1X n row vectors), covariance matrix of X
Since C is a symmetric matrix, it can be decomposed using Singular Value Decomposition (SVD) as:
C＝UΛVT (6)
in expression (6), the matrix U is an n × n matrix, Λ is a diagonal matrix of n × n and including eigenvalues, and V is an n × n matrix including n unit-orthogonalized eigenvectors.
Taking larger k characteristic values in Lambda to form a k multiplied by k diagonal matrix Lambda'k×kThe matrix V takes k column vectors corresponding to the k eigenvalues to form a matrix V'n×kFrom this, a matrix X 'after dimension reduction can be obtained'm×k＝Xm×nV′n×k。
When the dimensionality of the dimensionality reduction is larger than 1, the method for analyzing the dimensionality reduction only retains the information of the first principal component, and the dimensionality reduction signals are superposed, so that effective signals are retained and noise is filtered.
The above description is only an example of the present application, and is not intended to limit the scope of the present application. Any modification, equivalent replacement, and improvement made within the spirit and scope of the present application are included in the protection scope of the present application.
Claims (10)
1. A novel denoising method for transformer winding deformation ultrasonic detection signals is characterized by comprising the following steps:
collecting the transmitted wave and the received echo signal transmitted by the probe;
dividing the signal into three sections, namely clutter signals, transmitted wave signals and echo signals according to the transmitted wave oscillation starting point and the echo response point;
denoising the clutter signals;
judging whether sampling points of the transmitted wave signal and the echo signal are both larger than a set value;
if the sampling points of the emission wave signal and the echo signal are both larger than a set value, denoising the emission wave signal and the echo signal respectively by adopting an ITD-PE-PCA algorithm;
if the sampling points of the emission wave signal and the echo signal are not both larger than a set value, combining the emission wave signal and the echo signal, and then denoising the combined signal by adopting an ITD-PE-PCA algorithm;
and splicing the denoised signals.
2. The novel denoising method for the ultrasonic detection signal of the transformer winding deformation as claimed in claim 1, wherein the sampling frequency of the acquisition card is greater than 2 times the ultrasonic center frequency.
3. The novel denoising method for ultrasonic detection signals of transformer winding deformation according to claim 1, wherein the signals are divided into three sections according to the determination of the excitation point of the transmitted wave and the response point of the received wave based on a signal extreme value detection algorithm.
4. The novel denoising method for ultrasonic detection signal of transformer winding deformation according to claim 1, wherein the signal extreme value detection algorithm is a segmentation method of signal extreme value mutation; the method specifically comprises the following steps:
s101, obtaining a maximum value point and a minimum value point of the ultrasonic signal through the characteristics of the extreme values, and recording corresponding moments of all the extreme values;
s102, finding the maximum absolute value of the maximum value point, calculating the absolute difference value of the adjacent extreme value points in the time reduction direction from the maximum value point of the maximum value point, and sequencing the absolute difference value sequence from large to small;
s103, finding out adjacent maximum point pairs corresponding to part of the larger absolute difference sequence, calculating the absolute ratio of the large number divided by the small number in the adjacent maximum point pairs, and taking the minimum maximum of the non-zero absolute value if the divisor is zero; setting the time corresponding to the minimum maximum value point in the adjacent maximum value points with the largest divisor as the starting oscillation point PH1 of the transmitted signal, and calculating the absolute difference value between the starting oscillation point PH1 and the next maximum value point as AH;
s104, similarly, determining a starting oscillation point PL1 of the transmitted signal by using the minimum value point, and solving an absolute difference AL between the starting oscillation point PL1 and the next minimum value point; comparing the sizes of AH and AL, if AH is larger than AL, the oscillation starting point is taken as PH1, otherwise, PL1 is taken;
s105, finding a first maximum point with an absolute value smaller than 0.1MH along with a corresponding time t1 from the time corresponding to the maximum MH according to the time increasing direction, finding all maximum points and minimum points from t1 to the sampling end, and recording the corresponding times of all the maximum points.
5. The method as claimed in claim 1, wherein the noise reduction algorithm for the clutter signals is a wavelet threshold noise reduction algorithm, and the wavelet threshold noise reduction algorithm can extract white noise from the clutter signals.
6. The novel denoising method for ultrasonic detection signals of transformer winding deformation according to claim 1, wherein the set value is 1000; in order to deal with the problem that the sampling length of the ultrasonic effective signal may be insufficient, 1000 sampling points are selected as dividing points where both the transmitted wave and the echo can be divided.
7. The novel denoising method for ultrasonic detection signals of transformer winding deformation according to claim 1, wherein the ITD-PE-PCA algorithm comprises:
decomposing the ultrasonic signals by using an inherent time scale decomposition algorithm;
determining the times of ITD decomposition by using a permutation entropy algorithm;
and performing denoising processing on the signal component obtained by decomposing the ITD-PE by using principal component analysis.
8. The novel denoising method for ultrasonic detection signals of transformer winding deformation according to claim 7, wherein the inherent time scale decomposition algorithm comprises: and decomposing the ultrasonic signal sampling sequence to obtain a series of inherent rotation component sequences and a decomposition residue sequence.
9. The novel denoising method for the ultrasonic detection signal of transformer winding deformation as claimed in claim 1, wherein the arrangement entropy can reflect the complexity of the one-dimensional time signal, is sensitive to the change of the complex signal, and can well amplify the micro-variability of the system.
10. The method as claimed in claim 1, wherein the principal component analysis algorithm is a general dimensionality reduction algorithm, and since the noise signal has low energy and is generally not correlated with the ultrasound signal, the noise signal can be mostly filtered in the dimensionality reduction of the PCA, and thus the PCA can also be used to denoise the signal.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111441256.2A CN114136249B (en) | 2021-11-30 | 2021-11-30 | Transformer winding deformation ultrasonic detection signal denoising method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111441256.2A CN114136249B (en) | 2021-11-30 | 2021-11-30 | Transformer winding deformation ultrasonic detection signal denoising method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114136249A true CN114136249A (en) | 2022-03-04 |
| CN114136249B CN114136249B (en) | 2023-08-22 |
Family
ID=80389697
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111441256.2A Active CN114136249B (en) | 2021-11-30 | 2021-11-30 | Transformer winding deformation ultrasonic detection signal denoising method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114136249B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117111048A (en) * | 2023-10-24 | 2023-11-24 | 中国电建集团华东勘测设计研究院有限公司 | Improved ITD-based multi-beam submarine topography measurement echo noise reduction method |
| CN117540220A (en) * | 2024-01-09 | 2024-02-09 | 巨洋神州科技集团有限公司 | Near-zero carbon park source network load matching method and system |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2004010327A2 (en) * | 2002-07-19 | 2004-01-29 | Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. | Device and method for characterising an information signal |
| CN101241872A (en) * | 2008-01-22 | 2008-08-13 | 西安交通大学 | A distributed online detection system for ultrasonic lead key connection quality |
| CN101637394A (en) * | 2009-08-26 | 2010-02-03 | 济南大学 | Method for positioning and segmenting heart sound signal |
| CN102937668A (en) * | 2012-11-08 | 2013-02-20 | 电子科技大学 | Electric system low-frequency oscillation detection method |
| CN110926386A (en) * | 2019-11-25 | 2020-03-27 | 国网吉林省电力有限公司电力科学研究院 | CEEMD-improved wavelet threshold denoising-based transformer winding ultrasonic detection three-dimensional imaging method |
| CN110956136A (en) * | 2019-12-02 | 2020-04-03 | 天津市计量监督检测科学研究院 | Hydrophone signal denoising method based on ESMD and roughness penalty smoothing technology |
| CN111123206A (en) * | 2019-12-18 | 2020-05-08 | 广东工业大学 | Millimeter wave radar anti-interference method in congested road conditions |
| CN112799078A (en) * | 2021-04-15 | 2021-05-14 | 深圳中科乐普医疗技术有限公司 | Detection method and system for shear wave propagation velocity and ultrasonic imaging equipment |
| CN113219054A (en) * | 2021-04-27 | 2021-08-06 | 清华大学深圳国际研究生院 | Magnetic shoe internal defect detection device and detection method |
| CN113288095A (en) * | 2021-05-24 | 2021-08-24 | 南京优博一创智能科技有限公司 | Sleep health management method and system for calculating sleep cycle based on physical activity and heart rate information |
-
2021
- 2021-11-30 CN CN202111441256.2A patent/CN114136249B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2004010327A2 (en) * | 2002-07-19 | 2004-01-29 | Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V. | Device and method for characterising an information signal |
| CN101241872A (en) * | 2008-01-22 | 2008-08-13 | 西安交通大学 | A distributed online detection system for ultrasonic lead key connection quality |
| CN101637394A (en) * | 2009-08-26 | 2010-02-03 | 济南大学 | Method for positioning and segmenting heart sound signal |
| CN102937668A (en) * | 2012-11-08 | 2013-02-20 | 电子科技大学 | Electric system low-frequency oscillation detection method |
| CN110926386A (en) * | 2019-11-25 | 2020-03-27 | 国网吉林省电力有限公司电力科学研究院 | CEEMD-improved wavelet threshold denoising-based transformer winding ultrasonic detection three-dimensional imaging method |
| CN110956136A (en) * | 2019-12-02 | 2020-04-03 | 天津市计量监督检测科学研究院 | Hydrophone signal denoising method based on ESMD and roughness penalty smoothing technology |
| CN111123206A (en) * | 2019-12-18 | 2020-05-08 | 广东工业大学 | Millimeter wave radar anti-interference method in congested road conditions |
| CN112799078A (en) * | 2021-04-15 | 2021-05-14 | 深圳中科乐普医疗技术有限公司 | Detection method and system for shear wave propagation velocity and ultrasonic imaging equipment |
| CN113219054A (en) * | 2021-04-27 | 2021-08-06 | 清华大学深圳国际研究生院 | Magnetic shoe internal defect detection device and detection method |
| CN113288095A (en) * | 2021-05-24 | 2021-08-24 | 南京优博一创智能科技有限公司 | Sleep health management method and system for calculating sleep cycle based on physical activity and heart rate information |
Non-Patent Citations (1)
| Title |
|---|
| 顾惠杰: "基于信号极值检测分段法和ITD-PE-PCA 算法的 变压器绕组变形超声检测信号去噪研究", 《变压器》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117111048A (en) * | 2023-10-24 | 2023-11-24 | 中国电建集团华东勘测设计研究院有限公司 | Improved ITD-based multi-beam submarine topography measurement echo noise reduction method |
| CN117111048B (en) * | 2023-10-24 | 2024-01-26 | 中国电建集团华东勘测设计研究院有限公司 | Improved ITD-based multi-beam submarine topography measurement echo noise reduction method |
| CN117540220A (en) * | 2024-01-09 | 2024-02-09 | 巨洋神州科技集团有限公司 | Near-zero carbon park source network load matching method and system |
| CN117540220B (en) * | 2024-01-09 | 2024-04-05 | 巨洋神州科技集团有限公司 | Near-zero carbon park source network load matching method and system |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114136249B (en) | 2023-08-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wang et al. | Fault diagnosis of diesel engine based on adaptive wavelet packets and EEMD-fractal dimension | |
| CN114136249A (en) | Novel denoising method for transformer winding deformation ultrasonic detection signal | |
| Kuo et al. | The application of wavelets correlator for ship wake detection in SAR images | |
| KR100646715B1 (en) | Method for improving a quality of a 2-d ultrasound image by post-processing | |
| CN109580787B (en) | Ultrasonic echo denoising method for ultrasonic detection of transformer high-voltage bushing lead | |
| CN1954235B (en) | Improved method and device for local spectral analysis of an ultrasonic signal | |
| Wang et al. | Analysis of low-order autoregressive models for ultrasonic grain signal characterization | |
| CN113358743B (en) | Lamb wave mode separation method based on time-frequency distribution similarity analysis | |
| Govindan et al. | Processing algorithms for three‐dimensional data compression of ultrasonic radio frequency signals | |
| Oruklu et al. | Ultrasonic flaw detection using discrete wavelet transform for NDE applications | |
| Yan et al. | Frequency division denoising algorithm based on VIF adaptive 2D-VMD ultrasound image | |
| Zhou et al. | A denoising scheme for DSPI fringes based on fast bi-dimensional ensemble empirical mode decomposition and BIMF energy estimation | |
| Sabzevari et al. | Locating the acoustic source in thin glass plate using low sampling rate data | |
| Zhang et al. | Unsupervised learning for 3D ultrasonic data compression | |
| CN108398183B (en) | Multi-line spectrum Duffing oscillator detection and frequency identification method with unknown parameters under strong noise background | |
| Ďuriš et al. | Wavelets and digital filters designed and synthesized in the time and frequency domains | |
| Perov et al. | Wavelet filtering of signals from ultrasonic flaw detector | |
| JP3108317B2 (en) | Signal processing device and signal processing method | |
| Weiss et al. | Wavelet-based denoising of underwater acoustic signals | |
| Auslander et al. | Comparison of the Gabor and short-time Fourier transforms for signal detection and feature extraction in noisy environments | |
| Fan et al. | Ultrasonic broadband time-reversal with multiple signal classification imaging using full matrix capture | |
| Mina et al. | Two dimensional multi-frequency eddy current data fusion | |
| Dewi et al. | Denoising cloud interference on landsat satellite image using discrete haar wavelet transformation | |
| Zi et al. | Self-supervised learning for seismic swell noise removal | |
| CN113409283B (en) | Defect quantification method, device and medium based on super-resolution ultrasonic image |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |