Ultrasonic detection signal denoising method for transformer

Abstract

The invention discloses an ultrasonic detection signal denoising method for a transformer, and the method comprises the steps: obtaining an original noise-containing signal of the transformer, and carrying out the preprocessing of the original noise-containing signal, and obtaining a first signal; decomposing the first signal by using a variational mode decomposition algorithm to obtain a second signal, and optimizing a coefficient of the variational mode decomposition algorithm by using a Red-tail eagle optimization algorithm; dividing the second signal into a first noise and a first useful signal; further decomposing the first noise by adopting a singular value decomposition algorithm to obtain second noise and a second useful signal; and reconstructing the first useful signal and the second useful signal to obtain a third signal. According to the method, the Red-tail eagle optimization algorithm is used, and the optimized variational mode decomposition algorithm can inhibit the mode aliasing phenomenon; decomposing the first noise signal by adopting a singular value decomposition algorithm so as to extract a second useful signal from the first noise signal to the maximum extent; and the noise-reduced signal is obtained through reconstruction, so that the signal-to-noise ratio of the original signal is improved.

Description

Technical Field

[0001] The invention relates to the field of electrical equipment maintenance, and in particular to an ultrasonic detection signal denoising method for a transformer. Background Art

[0002] Transformers are an important part of power transmission. In transmission lines, there are a large number of transformers, which undertake the key tasks of power transmission and distribution. The operational reliability of transformers is closely related to the safety and stability of the power system.

[0003] Among them, the bushing is an important part of the power transformer. If there are insulation risks or lead deformation problems, it will directly affect the operational reliability of the power transformer. The deformation of the leads in the bushing will lead to insulation degradation and performance degradation, posing a threat to the safe operation of the power transformer. In severe cases, it may cause operational failures of the power transformer. In the actual operation and maintenance process, the main detection method for the leads in the bushing is regular preventive testing, but the existing method cannot detect internal defects of the equipment in time. When the test time interval is long, many potential or internal problems are difficult to be detected in time.

[0004] Currently, ultrasonic echo monitoring is used for casing. When the ultrasonic echo is processed, the noise in the echo will cause modal aliasing, which reduces the signal-to-noise ratio of the echo signal.

[0005] Therefore, an intuitive and quantitative live monitoring method is needed to effectively evaluate the operating status of the high-voltage bushing lead, which is of great significance for power transformers that need to operate stably for a long time. Summary of the invention

[0006] The object of the present invention is to provide a method for denoising ultrasonic detection signals for transformers, which has the advantage of high efficiency in effective signal extraction.

[0007] To achieve the above object, the present invention provides a method for denoising an ultrasonic detection signal for a transformer, comprising:

[0008] S10, obtaining an original transformer noisy signal, and preprocessing the original transformer noisy signal to obtain a first signal;

[0009] S20, using a variational mode decomposition algorithm to decompose the first signal to obtain a second signal, and using a red-tailed hawk optimization algorithm to optimize the coefficients of the variational mode decomposition algorithm;

[0010] S30, distinguishing the second signal into a first noise and a first useful signal;

[0011] S40, further decomposing the first noise using a singular value decomposition algorithm to obtain a second noise and a second useful signal;

[0012] S50: Reconstruct the first useful signal and the second useful signal to obtain a third signal.

[0013] Optionally, in step S10, the preprocessing includes removing DC components and filtering processing, wherein the filtering processing is used to remove high-frequency noise, low-frequency drift and other frequency components unrelated to the target signal; the filtering processing is used to further remove interference from non-target frequency components in the original transformer noisy signal.

[0014] Optionally, in step S20, the coefficients of the variational mode decomposition algorithm include the decomposition level k and the quadratic penalty factor α, and the red-tailed hawk optimization algorithm is used to optimize the values ​​of the decomposition level k and the quadratic penalty factor α.

[0015] Optionally, the step S20 specifically includes:

[0016] S21, using the Red-tailed Hawk optimization algorithm to optimize the decomposition level k and the quadratic penalty factor α of the variational mode decomposition algorithm;

[0017] S22, using the variational mode decomposition algorithm to find multiple intrinsic mode functions of the first signal, and decompose the first signal into multiple frequency components;

[0018] S23. Optimize the multiple frequency components by using the variational principle, and obtain the optimal frequency and amplitude of the multiple intrinsic mode functions by minimizing the difference between the first signal and the multiple frequency components.

[0019] Optionally, in step S22, the variational modal decomposition algorithm decomposes the first signal into k modal components u k (k), each of the modal components u k (t) is expressed as:

[0020]

[0021] In the formula, u k (t) is each modal component, A k (t) is the amplitude of each modal component, is the instantaneous frequency of the signal;

[0022] In order to obtain the limited bandwidth of each modal component, k (t) is subjected to Hilbert transform, and then u k (t) frequency shift to ω k The constraint expression established is:

[0023]

[0024] In the formula, u k (t) is each modal component, {ω k} is the set of center frequencies of k modal components. The constraint condition of the above formula is that the sum of the signals of each modal component is equal to the input signal;

[0025] In order to solve the above equation, the quadratic penalty factor α and the Lagrange multiplier λ are introduced, and the transformed expression is:

[0026]

[0027] In the formula, Represents the inner product of the two. The optimal solution can be obtained by taking the derivative of the above formula and setting the derivative to 0. The optimization iteration is performed using alternating direction multiplier operation. The expression for iterative update is:

[0028]

[0029] In the formula, and f(t), u k (t) and λ(t) in the frequency domain, ω is the frequency, n is the number of iterations, and τ is the fidelity parameter.

[0030] Optionally, the step S30 specifically includes: setting a permutation entropy threshold to distinguish useful first noise from a first useful signal; a component greater than the permutation entropy threshold is classified as the first noise, and a component less than the permutation entropy threshold is classified as the first useful signal.

[0031] Optionally, in step S30, the permutation entropy threshold (PE) is 0.55.

[0032] Optionally, after step S40, the following steps are further included:

[0033] S41. Use the second-order difference spectrum to select the singular value retention order to extract the second useful signal to a greater extent.

[0034] In summary, compared with the prior art, the ultrasonic detection signal denoising method for transformer provided by the present invention has the following beneficial effects:

[0035] The ultrasonic detection signal denoising method for transformers of the present invention is based on a joint denoising method of a red-tailed hawk optimization algorithm, a variational modal decomposition algorithm and a singular value decomposition algorithm. The red-tailed hawk optimization algorithm is used to optimize the decomposition layer k and the quadratic penalty factor α of the variational modal decomposition algorithm. The optimized variational modal decomposition algorithm can further suppress the modal aliasing phenomenon. Subsequently, a permutation entropy mechanism is introduced, a permutation entropy threshold is set to distinguish a first noise from a first useful signal, a singular value decomposition algorithm is used to decompose the first noise signal, and a second-order difference spectrum is used to determine the singular value retention order, so as to extract a second useful signal from the first noise signal to the maximum extent. Finally, the first useful signal and the second useful signal are reconstructed to obtain a denoised signal, thereby improving the signal-to-noise ratio of the original signal. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 The present invention is a flow chart of the method for denoising ultrasonic detection signals used for transformers.

[0037] Figure 2 This is a schematic diagram of the first viewing angle of the placement position of the ultrasonic detection probe for the transformer of the present invention.

[0038] Figure 3 This is a schematic diagram of the second viewing angle of the placement position of the ultrasonic detection probe for the transformer of the present invention.

[0039] Figure 4 The figure is a schematic diagram of a circuit module of an ultrasonic detection device for a transformer according to the present invention.

[0040] Figure 5 This is a schematic diagram of the ultrasonic emission module used for detection in the present invention.

[0041] Figure 6 This is a schematic diagram of the ultrasonic receiving module used for detection in the present invention. DETAILED DESCRIPTION

[0042] The following will be combined with the attached embodiment of the present invention Figure 1 ~Attached Figure 6 , the technical solutions, structural features, objectives achieved and effects in the embodiments of the present invention are described in detail.

[0043] It should be noted that the drawings are in a very simplified form and use non-precise proportions. They are only used to conveniently and clearly assist in explaining the embodiments of the present invention, and are not used to limit the conditions for the implementation of the present invention. Therefore, they have no substantive technical significance. Any structural modification, change in proportional relationship or adjustment of size, without affecting the effects and purposes that can be achieved by the present invention, should still fall within the scope of the technical content disclosed by the present invention.

[0044] It should be noted that, in the present invention, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only the elements explicitly listed, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0045] like Figure 1 As shown, the present invention provides a method for denoising an ultrasonic detection signal for a transformer, comprising:

[0046] S10, obtaining the original transformer noisy signal, and preprocessing the original transformer noisy signal to obtain the first signal. In step S10, the preprocessing includes removing the DC component and filtering, and the filtering is used to remove high-frequency noise, low-frequency drift and other frequency components irrelevant to the target signal; the filtering is used to further remove the interference of non-target frequency components in the original transformer noisy signal, and these non-target frequency components may include high-frequency noise, low-frequency drift or other frequency components irrelevant to the transformer signal analysis.

[0047] S20, using a variational mode decomposition algorithm (Variational Mode Decomposition, VMD) to decompose the first signal to obtain a second signal, and using a Red-Tailed Hawk Optimization Algorithm (Red-Tailed Hawk Optimization Algorithm, RTH) to optimize the coefficients of the variational mode decomposition algorithm. In step S20, the coefficients of the variational mode decomposition algorithm include the decomposition level k and the quadratic penalty factor α, and the Red-Tailed Hawk Optimization Algorithm is used to optimize the decomposition level k and the quadratic penalty factor α.

[0048] In this embodiment, step S20 specifically includes:

[0049] S21. The Red-tailed Hawk optimization algorithm is used to optimize the decomposition level k and the quadratic penalty factor α of the variational mode decomposition algorithm. The optimized variational mode decomposition algorithm can further suppress the phenomenon of mode aliasing.

[0050] S22. Using a variational mode decomposition algorithm, find multiple intrinsic mode functions of the first signal, and decompose the first signal into multiple frequency components.

[0051] Specifically, in step S22, the variational mode decomposition algorithm decomposes the first signal into k modal components u k(t), each modal component u k (t) is expressed as:

[0052]

[0053] In the formula, u k (t) is each modal component, A k (t) is the amplitude of each modal component, is the instantaneous frequency of the signal;

[0054] In order to obtain the limited bandwidth of each modal component, k (t) is subjected to Hilbert transform, and then u k (t) frequency shift to ω k The constraint expression established is:

[0055]

[0056] In the formula, u k (t) is each modal component, {ω k} is the set of center frequencies of k modal components. The constraint condition of the above formula is that the sum of the signals of each modal component is equal to the input signal;

[0057] In order to solve the above equation, the quadratic penalty factor α and the Lagrange multiplier λ are introduced, and the transformed expression is:

[0058]

[0059] In the formula, Represents the inner product of the two. The optimal solution can be obtained by taking the derivative of the above formula and setting the derivative to 0. The optimization iteration is performed using alternating direction multiplier operation. The expression for iterative update is:

[0060]

[0061] In the formula, and f(t), u k (t) and λ(t) in the frequency domain, ω is the frequency, n is the number of iterations, and τ is the fidelity parameter.

[0062] When using the VMD algorithm to decompose a signal, it is necessary to consider setting the values ​​of the parameters k and σ. If the value is too large, the signal will be decomposed. On the contrary, if the value is too small, part of the signal frequency band will be lost. In the red-tailed hawk optimization algorithm, the fitness function will affect the selection of VMD parameter values, and the envelope entropy value can better reflect the characteristics of the original signal coefficients. When the interference in the signal component is large and there is less effective information, the envelope entropy value becomes larger. Therefore, the envelope entropy value E of the signal x(i), i = 1, 2, ..., N p It can be expressed by the following formula:

[0063]

[0064] Where a(i) is the envelope signal of the IMF component after Hilbert mediation; ε(i) is the normalized probability distribution sequence; N is the number of sampling points; and the envelope entropy E is p That is the entropy value of ε(i).

[0065] The Red-tailed Hawk Optimization Algorithm is an intelligent optimization algorithm inspired by the hunting behavior of the Red-tailed Hawk in nature. The Red-tailed Hawk has a unique search strategy during the hunting process, and this algorithm simulates this strategy to find the optimal solution. Overall, the hunting process of the Red-tailed Hawk is divided into three stages. In the high-altitude soaring stage, the Red-tailed Hawk explores the search space and determines the area with the location of the prey; in the low-altitude soaring stage, the Red-tailed Hawk moves in the selected area around the prey to select the best position for hunting. Then, the Red-tailed Hawk swings and hits the target in the sharp turn and dive stages.

[0066] In the high-altitude soaring stage, the red-tailed hawk flies in a gentle dihedral, flapping its wings as little as possible to save energy. Unlike other hawks, the red-tailed hawk can fly long distances. During this stage, the red-tailed hawk will fly far into the sky to find the best location in terms of food supply. At this time, it will flap its wings as little as possible to save energy and explore the selected area. The following is the mathematical model of this stage:

[0067] X(t)=X best +(X mean -X(t-1))

[0068] Where X(t) represents the position of the red-tailed hawk at iteration t, X best To obtain the best position, X mean is the mean of the locations.

[0069] After the red-tailed hawk selects the target position in the previous stage, it will circle around the prey. This movement enables it to detect the best position and time to hit the target and give the prey a fatal blow. Its mathematical model is as follows:

[0070] X(t)=X best +(x(t)+y(t)·StepSize(t))

[0071] StepSize(t)=X(t)-X mean

[0072] Where X(t) represents the position of the red-tailed hawk at iteration t, X best To obtain the best position, x(t) and y(t) represent the direction coordinates.

[0073] In the dive hunting phase, the red-tailed hawk suddenly bends down and attacks the prey from the best position in the low-altitude flight phase after choosing the best position and time in the previous step. This phase is very fast, and its mathematical expression is as follows:

[0074] X(t)=α(t)X best +(x(t)·StepSize(t)+y(t)·StepSize2(t))

[0075] Each step length in this stage can be calculated as follows:

[0076] StepSize(t)=X(t)-TF(t)X mean

[0077] StepSize2(t)=G(t)·X(t)-TF(t)X best

[0078] Where α is the acceleration factor, which increases with the increase of t to improve the convergence speed. G is the gravity factor. When the red-tailed hawk is closer to the prey, the gravity effect decreases to reduce the diversity of prey.

[0079] S23. Optimize the multiple frequency components using the variational principle, and obtain the optimal frequencies and amplitudes of the multiple intrinsic mode functions by minimizing the difference between the first signal and the multiple frequency components.

[0080] In the variational mode decomposition (VMD) algorithm, each IMF (Intrinsic Mode Function) has its corresponding frequency characteristics. During the optimization process, the frequency parameters of each IMF are continuously adjusted to find the most suitable frequency value for each IMF under the goal of minimizing the difference. This optimal frequency enables the IMF to better reflect the time-frequency characteristics of the signal in the entire signal decomposition.

[0081] The amplitude indicates the vibration amplitude of the IMF at its corresponding frequency. It reflects the energy contribution of the mode in the signal. IMFs with larger amplitudes account for a relatively larger proportion of the original signal and have a more important impact on the overall shape and characteristics of the original signal. Finding the optimal amplitude of each IMF is related to the extraction of key features of the signal or the distinction between signal and noise.

[0082] S30, distinguishing the second signal into a first noise and a first useful signal. Step S30 specifically includes: setting a permutation entropy threshold to distinguish the useful first noise from the first useful signal; a component greater than the permutation entropy threshold is classified as the first noise, and a component less than the permutation entropy threshold is classified as the first useful signal. In this embodiment, the permutation entropy threshold (PE) is 0.55.

[0083] S40, further decomposing the first noise by using a singular value decomposition (SVD) algorithm to obtain a second noise and a second useful signal;

[0084] The core of SVD is matrix decomposition. SVD achieves signal noise reduction by decomposing a matrix into the product of three matrices. SVD is suitable for large data, can control the degree of dimensionality reduction, and can retain useful information to the greatest extent. Therefore, using SVD decomposition for secondary noise reduction has significant advantages. For the collected signal X(t) = {x1, x2, ..., x i}, i = 1, 2, ..., M. The following Hankel matrix can be constructed:

[0085]

[0086] Where M is the signal length and m is the sliding window width. By performing SVD decomposition on the matrix H, we can obtain:

[0087] H=A∑B T

[0088] Where A and B are both orthogonal matrices, and ∑ is a diagonal matrix.

[0089] By retaining the larger singular values ​​in the diagonal matrix and ignoring the smaller singular values, SVD is able to distinguish between signals and noise, because noise usually corresponds to smaller singular values. By removing these smaller singular values, the impact of noise on data analysis can be significantly reduced, and the signal-to-noise ratio of the signal can be improved.

[0090] Step S40 is followed by step S41, using the second-order difference spectrum to select the singular value retention order to extract the second useful signal to a greater extent. The second-order difference spectrum is an analysis method based on singular value differences, which calculates the second-order differences between the singular values ​​obtained after SVD decomposition (i.e., the differences between adjacent singular value differences). This method helps to highlight the mutation points in the singular value sequence, which are often associated with important structural changes in the signal. Assume that the singular value sequence after SVD decomposition is σ1, σ2, ..., σ n , then the second-order difference sequence d i It can be defined as:

[0091] d i =2σ i -σ i-1 -σ i+1 , i = 2, 3, ..., n-1

[0092] First, the original signal matrix is ​​decomposed by SVD to obtain the singular value sequence σ1, σ2, …, σ n , among which σ1≥σ2≥…≥σ n. According to the above formula, the second-order difference sequence d is calculated i . This sequence will reveal the rate of change in the singular value sequence, that is, which singular values ​​have the most significant changes. Observe the second-order difference spectrum and look for mutation points or significant peaks. These points usually correspond to the parts of the singular value sequence that fall or rise rapidly, and may mark important changes in the signal structure. Based on the analysis of the second-order difference spectrum, choose a suitable singular value retention order k. This order should be able to capture the main features of the signal while excluding smaller singular values ​​caused by noise.

[0093] Large singular values ​​represent useful information, while small singular values ​​represent noise. Therefore, large singular values ​​can be retained and small singular values ​​can be set to 0 to remove noise from the signal. The first-order difference spectrum and second-order difference spectrum of the singular values ​​are used to select the retained singular values, and the maximum value point and the previous singular values ​​of the second-order singular value difference spectrum are retained, and the singular values ​​after the maximum value are set to 0, so as to retain the useful signal to the greatest extent.

[0094] S50, reconstructing the first useful signal and the second useful signal to obtain a third signal. Specifically, the signal components of the first useful signal and the second useful signal identified in the two decomposition processes are merged together. The first useful signal and the second useful signal together constitute the effective part of the transformer sound signal, and most of the noise and interference are removed. The first useful signal and the second useful signal are reconstructed to obtain a complete third signal, which is the transformer sound signal after noise reduction. The final noise reduction signal should be able to accurately reflect the main characteristics of the transformer sound signal, while reducing or eliminating unnecessary noise and interference.

[0095] Figure 2 and Figure 3 The relative position relationship of the ultrasonic probe installed on the casing is shown. The ultrasonic transmitting probe and the ultrasonic receiving probe are respectively installed at different positions of the casing along the circumference. The ultrasonic transmitting probe transmits ultrasonic waves into the casing, and the ultrasonic receiving probe is used to receive echo signals.

[0096] Figure 4 The circuit module structure of the ultrasonic detection equipment for transformers is shown. The transmitting probe and the receiving probe are respectively connected to the fault detection main circuit. The fault detection main circuit controls the transmitting probe to send ultrasonic waves into the bushing. The receiving probe is used to receive the echo and transmit it back to the fault detection main circuit. The fault detection main circuit is also connected to a host computer and an LCD screen. The host computer is used to further process the echo signal received by the receiving probe, and the LCD screen is used to display the analysis results.

[0097] Figure 5 and Figure 6The schematic diagram of the ultrasonic transmitting module used for detection and the schematic diagram of the ultrasonic receiving module used for detection are shown.

[0098] Although the content of the present invention has been described in detail through the above preferred embodiments, it should be appreciated that the above description should not be considered as a limitation of the present invention. After reading the above content, it will be apparent to those skilled in the art that various modifications and substitutions of the present invention will occur. Therefore, the protection scope of the present invention should be limited by the appended claims.

Claims

1. A method for denoising ultrasonic detection signals for transformers, characterized in that: The method comprises: S10, obtaining an original transformer noisy signal, and preprocessing the original transformer noisy signal to obtain a first signal; S20, using a variational mode decomposition algorithm to decompose the first signal to obtain a second signal, and using a red-tailed hawk optimization algorithm to optimize the coefficients of the variational mode decomposition algorithm; S30, distinguishing the second signal into a first noise and a first useful signal; S40, further decomposing the first noise using a singular value decomposition algorithm to obtain a second noise and a second useful signal; S50: Reconstruct the first useful signal and the second useful signal to obtain a third signal.

2. The method for denoising an ultrasonic detection signal for a transformer according to claim 1, characterized in that: In step S10, the preprocessing includes removing DC components and filtering processing, wherein the filtering processing is used to remove high-frequency noise, low-frequency drift and other frequency components unrelated to the target signal; the filtering processing is used to further remove the interference of non-target frequency components in the original transformer noisy signal.

3. The method for denoising an ultrasonic detection signal for a transformer according to claim 1, characterized in that: In step S20, the coefficients of the variational mode decomposition algorithm include the decomposition level k and the quadratic penalty factor α, and the red-tailed hawk optimization algorithm is used to optimize the values ​​of the decomposition level k and the quadratic penalty factor α.

4. The method for denoising an ultrasonic detection signal for a transformer according to claim 3, characterized in that: The step S20 specifically includes: S21, using the Red-tailed Hawk optimization algorithm to optimize the decomposition level k and the quadratic penalty factor α of the variational mode decomposition algorithm; S22, using the variational mode decomposition algorithm to find multiple intrinsic mode functions of the first signal, and decompose the first signal into multiple frequency components; S23. Optimize the multiple frequency components by using the variational principle, and obtain the optimal frequency and amplitude of the multiple intrinsic mode functions by minimizing the difference between the first signal and the multiple frequency components.

5. The method for denoising an ultrasonic detection signal for a transformer according to claim 4, characterized in that: In step S22, the variational modal decomposition algorithm decomposes the first signal into k modal components u k (t), each of the modal components u k (t) is expressed as: In the formula, u k (t) is each modal component, A k (t) is the amplitude of each modal component, is the instantaneous frequency of the signal; In order to obtain the limited bandwidth of each modal component, k (t) is subjected to Hilbert transform, and then u k (t) frequency shift to ω k The constraint expression established is: In the formula, u k (t) is each modal component, {ω k } is the set of center frequencies of k modal components. The constraint condition of the above formula is that the sum of the signals of each modal component is equal to the input signal; In order to solve the above equation, the quadratic penalty factor α and the Lagrange multiplier λ are introduced, and the transformed expression is: In the formula, Represents the inner product of the two. The optimal solution can be obtained by taking the derivative of the above formula and setting the derivative to 0. The optimization iteration is performed using alternating direction multiplier operation. The expression for iterative update is: In the formula, and f(t), u k (t) and λ(t) in the frequency domain, ω is the frequency, n is the number of iterations, and τ is the fidelity parameter.

6. The method for denoising an ultrasonic detection signal for a transformer according to claim 1, characterized in that: The step S30 specifically includes: setting a permutation entropy threshold to distinguish between useful first noise and first useful signal; and classifying components greater than the permutation entropy threshold as first noise, and components less than the permutation entropy threshold as first useful signal.

7. The method for denoising an ultrasonic detection signal for a transformer according to claim 6, characterized in that: In step S30, the permutation entropy threshold (PE) is 0.

55.

8. The method for denoising an ultrasonic detection signal for a transformer according to claim 1, characterized in that: After step S40, the following steps are further included: S41. Use the second-order difference spectrum to select the singular value retention order to extract the second useful signal to a greater extent.

Images