CN116165486A - Method and system for recovering time domain waveform of partial discharge pulse electric field - Google Patents

Abstract

The invention discloses a method and a system for recovering a time domain waveform of a partial discharge pulse electric field, comprising the following steps: respectively decomposing the broadband electric field signal and the ultra-high frequency sensor output voltage signal into a plurality of first subband signals and a plurality of second subband signals with different frequency bands; respectively calculating the amplitude flatness and the group delay flatness of each sub-band signal based on the first sub-band signal and the second sub-band signal; judging whether the ultrahigh frequency sensor is a linear system or not based on the amplitude flatness and the group delay flatness, and acquiring a judging result; and recovering the time domain waveform according to a preset recovery strategy based on the judging result. Aiming at two situations of whether the transfer function of the UHF sensor is linear or not, the phase is recovered by adopting a minimum phase method under the linear undistorted condition, and the time domain electric field waveform is directly recovered by adopting a deep convolution network DCN under the nonlinear distorted condition, so that the problem of inaccurate subsequent signal processing caused by incomplete phase information in the traditional method is solved.

Description

Method and system for recovering time domain waveform of partial discharge pulse electric field

Technical Field

The invention relates to the technical field of electrical equipment tests, in particular to a method and a system for recovering a time domain waveform of a partial discharge pulse electric field.

Background

Partial discharge measurements for insulation of high voltage electrical equipment are a common means of assessing the insulation status of the equipment. The partial discharge measurement technology comprises a traditional pulse current method, and also comprises a series of novel electrified detection technologies such as an ultrahigh frequency method, a high frequency method and an ultrasonic method, wherein the ultrahigh frequency method partial discharge detection technology has proper quantification, positioning capability and portability and is widely and widely applied.

The time domain waveform of the pulse electric field excited by the ultrahigh frequency partial discharge contains abundant partial discharge characteristic information, the ultrahigh frequency sensor converts the pulse electromagnetic field signal into a pulse voltage signal and outputs the pulse voltage signal to the detector host, and the time domain waveform of the pulse electromagnetic field signal excited by the partial discharge is difficult to obtain in practice and needs to be calculated by adopting a time-frequency conversion technology. The conversion from the frequency domain spectrum to the time domain waveform can be achieved by inverse fourier transform (Inverse Fast Fourier Transform, IFFT), but this requires that the amplitude and phase information of the spectrum are known at the same time, whereas the equipment manufacturer or unit of measurement generally only provides the amplitude information of the sensor correction coefficient, and no phase information, so the phase information is lost during the data correction process, and the time domain waveform cannot be obtained directly by inverse fourier transform. This also presents difficulties in calculating the time domain waveform of the pulsed electric field to determine the sensitivity threshold. Therefore, there is a need to study methods for recovering a time domain waveform from amplitude spectrum data, wherein the lack of a phase frequency curve affects the recovery of a pulsed time domain waveform.

HAYESM H et al first proposed reconstructing phase information using a minimum phase method, and assuming that a signal transmission system satisfies a minimum phase condition, the magnitude and phase of its transfer function satisfy a hilbert transform relationship, and can be solved from one to the other. TESCHEF M assumes that the system is a minimum (or maximum) phase system, reconstructing its phase information from the Hilbert transform. The above conventional minimum phase estimation time domain response method generates additional numerical errors due to repeated use of fast fourier transform (Fast Fourier Transform, FFT) to convert signals between time and frequency domains during implementation, and the related parameters of the FFT must be carefully selected to reduce the introduced errors. Some improved processing methods are to use Prony method, best approximation element method, filter modeling and the like to carry out parameterized estimation on the frequency domain transfer function, obtain a discrete transfer function model of the system, and estimate impulse response of the system by the frequency domain parameter model. However, the parametric modeling methods such as the Prony method also have some limitations, such as poor fitting accuracy, unstable poles, and the convergence of the algorithm being particularly dependent on the choice of initial values. All the methods are difficult to adapt to the time domain waveform recovery of the ultrahigh frequency partial discharge pulse electric field.

Disclosure of Invention

The invention provides a method and a system for recovering a time domain waveform of a partial discharge pulse electric field, which are used for solving the problem of recovering the time domain waveform of the partial discharge pulse electric field.

In order to solve the above problems, according to an aspect of the present invention, there is provided a partial discharge pulse electric field time domain waveform recovering method, the method comprising:

respectively decomposing the broadband electric field signal and the ultra-high frequency sensor output voltage signal into a plurality of first subband signals and a plurality of second subband signals with different frequency bands;

respectively calculating the amplitude flatness and the group delay flatness of each sub-band signal based on the first sub-band signal and the second sub-band signal;

judging whether the ultrahigh frequency sensor is a linear system or not based on the amplitude flatness and the group delay flatness, and acquiring a judging result;

and recovering the time domain waveform according to a preset recovery strategy based on the judging result.

Preferably, wherein the method further comprises:

before the amplitude flatness and the group delay flatness of each sub-band signal are calculated based on the first sub-band signal and the second sub-band signal, noise judgment is performed on the first sub-band signal and the second sub-band signal respectively, and the first sub-band signal and the second sub-band signal which are determined to be noise are removed.

Preferably, the recovering of the time domain waveform according to a preset recovery strategy based on the judging result includes:

and if the judging result indicates that the pulse electric field belongs to a linear system, recovering the phase by adopting a minimum phase method, and then recovering the time domain waveform of the pulse electric field.

And if the judging result indicates that the pulse electric field waveform does not belong to the linear system, recovering the pulse electric field time domain waveform based on the depth convolution network.

Preferably, the recovering of the pulse electric field time domain waveform after the phase recovery by the minimum phase method includes:

performing Fast Fourier Transform (FFT) on the ultra-high frequency sensor output voltage signal to obtain a complex frequency spectrum sequence V (k);

interpolation calculation is carried out on the correction coefficient ln|H (k) | according to the frequency information of V (k), and the point number N of the correction coefficient is obtained; wherein H (k) is a discretized system transfer function, and is obtained by discretizing after dividing the Fourier transform of an output signal by the Fourier transform of an input signal;

performing scale transformation and continuation on ln|H (k) | according to N to obtain G (k);

performing inverse Fourier transform on the G (k), and performing windowing operation to obtain H (k) meeting the minimum phase condition;

and multiplying V (k) and H (k) in a complex spectrum form, and performing fast Fourier transform (IFFT) operation to obtain a corresponding pulse electric field time domain waveform e (t).

Preferably, the recovery of the pulse electric field time domain waveform based on the depth convolution network comprises:

collecting a pulse electric field waveform e (t) at the front end of the UHF sensor and a rear end output voltage waveform u (t);

converting a one-dimensional waveform signal into a two-dimensional time-frequency spectrogram by using a time-frequency conversion technology, and denoising;

taking a u (t) spectrogram set as input and a corresponding e (t) spectrogram set as output by using a DCN method, training a DCN network, and obtaining a trained depth network model;

and inputting the output voltage signal of the ultrahigh frequency sensor into a trained depth network model to obtain a corresponding pulse electric field time domain waveform.

According to another aspect of the present invention, there is provided a partial discharge pulse electric field time domain waveform recovering system, the system comprising:

the signal decomposition unit is used for respectively decomposing the broadband electric field signal and the ultra-high frequency sensor output voltage signal into a plurality of first subband signals and a plurality of second subband signals with frequency bands which are not mutually mixed;

a flatness calculation unit for calculating an amplitude flatness and a group delay flatness of each subband signal based on the first subband signal and the second subband signal, respectively;

the judging unit is used for judging whether the ultrahigh frequency sensor is a linear system or not based on the amplitude flatness and the group delay flatness, and obtaining a judging result;

and the recovery unit is used for recovering the time domain waveform according to a preset recovery strategy based on the judging result.

Preferably, wherein the system further comprises:

and the denoising unit is used for respectively carrying out noise judgment on the first sub-band signal and the second sub-band signal before respectively calculating the amplitude flatness and the group delay flatness of each sub-band signal based on the first sub-band signal and the second sub-band signal, and removing the first sub-band signal and the second sub-band signal which are determined to be noise.

Preferably, the recovery unit performs recovery of the time domain waveform according to a preset recovery policy based on the determination result, and includes:

and if the judging result indicates that the pulse electric field belongs to a linear system, recovering the phase by adopting a minimum phase method, and then recovering the time domain waveform of the pulse electric field.

And if the judging result indicates that the pulse electric field waveform does not belong to the linear system, recovering the pulse electric field time domain waveform based on the depth convolution network.

Preferably, the recovery unit performs recovery of the pulse electric field time domain waveform after recovering the phase by using a minimum phase method, and includes:

performing Fast Fourier Transform (FFT) on the ultra-high frequency sensor output voltage signal to obtain a complex frequency spectrum sequence V (k);

interpolation calculation is carried out on the correction coefficient ln|H (k) | according to the frequency information of V (k), and the point number N of the correction coefficient is obtained; wherein H (k) is a discretized system transfer function, and is obtained by discretizing after dividing the Fourier transform of an output signal by the Fourier transform of an input signal;

performing scale transformation and continuation on ln|H (k) | according to N to obtain G (k);

performing inverse Fourier transform on the G (k), and performing windowing operation to obtain H (k) meeting the minimum phase condition;

and multiplying V (k) and H (k) in a complex spectrum form, and performing fast Fourier transform (IFFT) operation to obtain a corresponding pulse electric field time domain waveform e (t).

Preferably, the recovery unit performs recovery of the pulse electric field time domain waveform based on a deep convolution network, and includes:

collecting a pulse electric field waveform e (t) at the front end of the UHF sensor and a rear end output voltage waveform u (t);

converting a one-dimensional waveform signal into a two-dimensional time-frequency spectrogram by using a time-frequency conversion technology, and denoising;

using a DCN system to train a DCN network by taking a u (t) spectrogram set as input and a corresponding e (t) spectrogram set as output, and obtaining a trained depth network model;

and inputting the output voltage signal of the ultrahigh frequency sensor into a trained depth network model to obtain a corresponding pulse electric field time domain waveform.

The invention provides a method and a system for recovering a time domain waveform of a partial discharge pulse electric field, comprising the following steps: respectively decomposing the broadband electric field signal and the ultra-high frequency sensor output voltage signal into a plurality of first subband signals and a plurality of second subband signals with different frequency bands; respectively calculating the amplitude flatness and the group delay flatness of each sub-band signal based on the first sub-band signal and the second sub-band signal; judging whether the ultrahigh frequency sensor is a linear system or not based on the amplitude flatness and the group delay flatness, and acquiring a judging result; and recovering the time domain waveform according to a preset recovery strategy based on the judging result. Aiming at two situations of whether the transfer function of the UHF sensor is linear or not, the phase is recovered by adopting a minimum phase method under the linear undistorted condition, and the time domain electric field waveform is directly recovered by adopting a Deep Convolutional Network (DCN) under the nonlinear distorted condition. The problem of inaccurate subsequent signal processing caused by incomplete phase information in the traditional method is solved; after the phase information is calculated, the time domain waveform can be recovered by the amplitude spectrum data, and the problem that the time domain waveform of the pulse electric field excited by the prior ultrahigh frequency partial discharge is difficult to accurately recover is solved.

Drawings

Exemplary embodiments of the present invention may be more completely understood in consideration of the following drawings:

FIG. 1 is a flow chart of a method 100 for recovering a partial discharge pulse electric field time domain waveform according to an embodiment of the present invention;

FIG. 2 is a flow chart of a solution minimum phase sequence according to an embodiment of the present invention;

FIG. 3 is a flow chart of a method for DCN-based broadband pulse recovery according to an embodiment of the present invention;

fig. 4 is a schematic structural diagram of a partial discharge pulse electric field time domain waveform recovery system 400 according to an embodiment of the present invention.

Detailed Description

The exemplary embodiments of the present invention will now be described with reference to the accompanying drawings, however, the present invention may be embodied in many different forms and is not limited to the examples described herein, which are provided to fully and completely disclose the present invention and fully convey the scope of the invention to those skilled in the art. The terminology used in the exemplary embodiments illustrated in the accompanying drawings is not intended to be limiting of the invention. In the drawings, like elements/components are referred to by like reference numerals.

Unless otherwise indicated, terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art. In addition, it will be understood that terms defined in commonly used dictionaries should be interpreted as having a meaning that is consistent with their meaning in the context of the relevant art and will not be interpreted in an idealized or overly formal sense.

Aiming at the problems that the voltage waveform of the output end of the sensor is easy to obtain and the pulse electric field waveform of the input end of the sensor is difficult to calculate accurately in the process of partial discharge measurement and calibration by the ultrahigh frequency method, the ultrahigh frequency partial discharge pulse electric field time domain waveform recovery method based on the signal processing technology is provided. Firstly, decomposing a time domain signal into a series of subband signals with mutually non-aliasing frequency bands, removing noise, and judging whether the system is linear or not. If the method belongs to a linear system (i.e. under the condition of no distortion), reconstructing a pulse electric field time domain waveform after the phase is recovered by adopting a minimum phase method; if the method does not belong to a linear system (namely under the distortion condition), a 'deep convolution network' (DCN) in a deep learning theory is adopted to realize the recovery of the broadband pulse electric field time domain signal.

Fig. 1 is a flowchart of a method 100 for recovering a time domain waveform of a partial discharge pulse electric field according to an embodiment of the present invention. As shown in fig. 1, in the method for recovering a time domain waveform of a partial discharge pulse electric field provided by the embodiment of the invention, for two cases of whether a transfer function of a UHF sensor is linear, a minimum phase method is adopted to recover a phase under the condition of no linear distortion, and a Deep Convolutional Network (DCN) is adopted to directly recover a time domain electric field waveform under the condition of nonlinear distortion. The problem of inaccurate subsequent signal processing caused by incomplete phase information in the traditional method is solved; after the phase information is calculated, the time domain waveform can be recovered by the amplitude spectrum data, and the problem that the time domain waveform of the pulse electric field excited by the prior ultrahigh frequency partial discharge is difficult to accurately recover is solved. In the method 100 for recovering a time domain waveform of a partial discharge pulse electric field provided in the embodiment of the present invention, starting from step 101, in step 101, a wideband electric field signal and an ultrahigh frequency sensor output voltage signal are respectively decomposed into a plurality of first subband signals and a plurality of second subband signals with frequency bands that are not aliased with each other.

In step 102, an amplitude flatness and a group delay flatness of each subband signal are calculated based on the first subband signal and the second subband signal, respectively.

Preferably, wherein the method further comprises:

before the amplitude flatness and the group delay flatness of each sub-band signal are calculated based on the first sub-band signal and the second sub-band signal, noise judgment is performed on the first sub-band signal and the second sub-band signal respectively, and the first sub-band signal and the second sub-band signal which are determined to be noise are removed.

In the invention, firstly, because the broadband pulse signal often contains a plurality of characteristic frequency bands, a time domain signal decomposition technology is adopted to decompose a broadband electric field signal e (t) and a UHF sensor output voltage signal u (t) into a series of sub-band signals with frequency bands which are not mutually mixed, and noise components in the sub-band signals are removed; then, judging whether the UHF sensor is a linear system or not by calculating the amplitude flatness and the group delay flatness of the corresponding subband signals in e (t) and u (t); finally, if the UHF sensor meets the linear condition, recovering the e (t) waveform by adopting a minimum phase method, otherwise, using a deep convolution network with excellent nonlinear approximation capability to realize the recovery of the e (t) waveform.

And step 103, judging whether the ultrahigh frequency sensor is a linear system or not based on the amplitude flatness and the group delay flatness, and obtaining a judging result.

In step 104, the recovery of the time domain waveform is performed according to a preset recovery strategy based on the determination result.

Preferably, the recovering of the time domain waveform according to a preset recovery strategy based on the judging result includes:

and if the judging result indicates that the pulse electric field belongs to a linear system, recovering the phase by adopting a minimum phase method, and then recovering the time domain waveform of the pulse electric field.

And if the judging result indicates that the pulse electric field waveform does not belong to the linear system, recovering the pulse electric field time domain waveform based on the depth convolution network.

Preferably, the recovering of the pulse electric field time domain waveform after the phase recovery by the minimum phase method includes:

performing Fast Fourier Transform (FFT) on the ultra-high frequency sensor output voltage signal to obtain a complex frequency spectrum sequence V (k);

interpolation calculation is carried out on the correction coefficient ln|H (k) | according to the frequency information of V (k), and the point number N of the correction coefficient is obtained; wherein H (k) is a discretized system transfer function, and is obtained by discretizing after dividing the Fourier transform of an output signal by the Fourier transform of an input signal;

performing scale transformation and continuation on ln|H) k| according to N to obtain G (k);

performing inverse Fourier transform on the G (k), and performing windowing operation to obtain H (k) meeting the minimum phase condition;

and multiplying V (k) and H (k) in a complex spectrum form, and performing fast Fourier transform (IFFT) operation to obtain a corresponding pulse electric field time domain waveform e (t).

Preferably, the recovery of the pulse electric field time domain waveform based on the depth convolution network comprises:

collecting a pulse electric field waveform e (t) at the front end of the UHF sensor and a rear end output voltage waveform u (t);

converting a one-dimensional waveform signal into a two-dimensional time-frequency spectrogram by using a time-frequency conversion technology, and denoising;

taking a u (t) spectrogram set as input and a corresponding e (t) spectrogram set as output by using a DCN method, training a DCN network, and obtaining a trained depth network model;

and inputting the output voltage signal of the ultrahigh frequency sensor into a trained depth network model to obtain a corresponding pulse electric field time domain waveform.

(1) Pulse time domain waveform recovery when determined to be linear

There are many methods for recovering the phase frequency curve from the amplitude-frequency characteristic curve, and the minimum phase method is one of the methods with better effect. Assuming that a signal transmission system satisfies a minimum phase condition, the magnitude and phase of its transfer function satisfy the hilbert transform relationship, one can solve for the other. The minimum phase system refers to a system in which all the zeros and poles of the discrete transfer function are within a unit circle. If a linear time-invariant system and its inverse are both causal and stable, then the system is a minimum phase system. Although the actual signal transmission system does not necessarily meet the minimum phase condition, any causal system may be represented as a cascade of an all-pass system and a minimum phase system. Since the all-pass system does not affect the amplitude characteristics of the signal, in most cases the original system can be approximated by the minimum phase system corresponding to it, i.e. assuming that the system meets the minimum phase condition.

Under the condition that the system meets the minimum phase, the phase frequency information can be recovered from the amplitude frequency information of the system through Hilbert transformation. The data can be conveniently converted between the time domain and the frequency domain by utilizing the FFT, and the minimum phase sequence corresponding to the amplitude spectrum can be obtained by using the method shown in figure 2. Wherein X is _c (n) is a real cepstrum,

complex cepstrum, x (n) is the minimum phase sequence. Firstly, taking natural logarithm of amplitude spectrum, then realizing Hilbert transformation by IFFT and FFT, obtaining Fourier transformation X (omega) of X (n) by taking index, obtaining X (n) after IFFT.

After the pulse measurement system is regarded as a signal transmission system and the minimum phase condition is assumed to be met, the phase frequency characteristic of the system can be obtained by utilizing the property of the minimum phase system and the amplitude-frequency characteristic (namely the correction coefficient) of the system, the missing phase information is complemented, and the transfer function of the system is further obtained. Because the transfer function and the correction coefficient are in reciprocal relation, the input and output of the system can be reversed for convenience, so that the amplitude spectrum of the transfer function is equivalent to the correction coefficient. Let the transfer function of the minimum phase system be:

H(ω)＝|H(ω)|·e ^jθ(ω) (45)

ω in the above formula is the signal frequency, and |h (ω) | and θ (ω) represent the amplitude-frequency response and the phase-frequency response of the system at the frequency ω, respectively.

Taking the logarithm from the two sides to obtain

ln(H(ω))＝ln(|H(ω)|)+jθ(ω)＝A(ω)+jP(ω) (46)

A (ω) and P (ω) reflect the frequency amplitude characteristic and the phase frequency characteristic of the system, respectively, which satisfy the hilbert transform therebetween,

wherein:

and->

Hilbert transforms for A (ω) and P (ω), respectively. Thus, P (omega) can be obtained by performing Hilbert transform on A (omega), the transfer function H (omega) of the system is obtained, then the complex spectrum of the measured voltage signal is multiplied by the transfer function H (omega), the complex spectrum of the pulse to be measured is obtained, and the result is transformed into the time domain by inverse Fourier transform, so that the recovered waveform is obtained.

Therefore, in the present invention, when linear transfer, i.e., no distortion, recovery of the pulse electric field time domain waveform is performed by the following means, including:

1) FFT is performed on the measured voltage signal to obtain a complex spectrum sequence V (k) thereof.

2) In order to facilitate the direct multiplication of V (k) and H (k) later, interpolation calculation is performed on the correction coefficient ln|H (k) | according to the frequency information of V (k), and the point number N of the correction coefficient is recorded, and the length of both is not required to be 2 ^N Because the number of points is automatically complemented during FFT operation; where H (k) is the discretized system transfer function divided by the Fourier transform of the output signalObtained by discretizing after fourier transformation of the input signal.

3) And (3) carrying out scale transformation and continuation on ln|H (k) | according to N to obtain G (k), wherein G (k) is in the form of discrete Fourier transformation of H (t). This step is important because the magnitude spectrum is not directly inverse fourier-computed and must be transformed into a satisfactory discrete sequence of times.

4) And (3) performing inverse Fourier transform on the G (k), and performing windowing operation to finally obtain the H (k) meeting the minimum phase condition.

5) And multiplying V (k) and H (k) in a complex spectrum form, and performing fast Fourier transform (IFFT) operation to obtain a waveform e (t) of the pulse to be detected.

(2) Pulse time domain waveform recovery when non-linearity is determined

When the transfer function of the UHF sensor is nonlinear, it is difficult to recover the pulse electric field time domain signal by the conventional minimum phase method.

The deep learning method has excellent feature learning capability, can automatically search the most suitable feature information from mass data, and avoids subjectivity of manually selecting feature quantity. The invention applies the deep learning theory to the recovery of UHF pulse electric field time domain signals, and provides a new idea for improving the accurate recovery of the pulse field under the nonlinear condition. The recovery of the wideband pulse electric field time domain signal is realized by adopting a 'deep convolutional network' (Deep convolutional network, DCN) in the deep learning theory, and the technical route is shown in figure 3.

The method specifically comprises the following steps:

1) Collecting a pulse electric field waveform e (t) at the front end of the UHF sensor and a rear end output voltage waveform u (t);

2) Converting a one-dimensional waveform signal into a two-dimensional time-frequency spectrogram by using a time-frequency conversion technology, and denoising; directly setting spectral components which are not in an ultrahigh frequency band (0.3 GHz-3 GHz) in the two-dimensional spectrum to zero to obtain a denoised time-frequency spectrogram;

3) Taking a u (t) spectrogram set as input and a corresponding e (t) spectrogram set as output by using a DCN method, and training a DCN network;

4) And the network parameters are adjusted, so that the accuracy of the cross-validation test is improved.

Finally, by using the trained depth network, the pulse electric field time domain waveform corresponding to the UHF sensor can be predicted when the UHF sensor outputs a certain voltage waveform.

The invention provides a technical route for a series of processes such as signal denoising, sensor transfer function linear judgment, phase recovery, waveform reconstruction and the like. Aiming at two cases of whether the transfer function of the UHF sensor is linear or not, a minimum phase method is adopted to recover the phase under the condition of no linear distortion, and a Depth Convolution Network (DCN) is adopted to directly recover the time domain electric field waveform under the condition of nonlinear distortion. The problem of inaccurate subsequent signal processing caused by incomplete phase information in the traditional method is solved; after the phase information is calculated, the time domain waveform can be recovered by the amplitude spectrum data, and the problem that the time domain waveform of the pulse electric field excited by the prior ultrahigh frequency partial discharge is difficult to accurately recover is solved.

Fig. 4 is a schematic structural diagram of a partial discharge pulse electric field time domain waveform recovery system 400 according to an embodiment of the present invention. As shown in fig. 4, a partial discharge pulse electric field time domain waveform recovery system 400 according to an embodiment of the present invention includes: a signal decomposition unit 401, a flatness calculation unit 402, a judgment unit 403, and a recovery unit 404.

Preferably, the signal decomposition unit 401 is configured to decompose the broadband electric field signal and the uhf sensor output voltage signal into a plurality of first subband signals and a plurality of second subband signals with frequency bands that are not aliased respectively.

Preferably, wherein the system further comprises:

and the denoising unit is used for respectively carrying out noise judgment on the first sub-band signal and the second sub-band signal before respectively calculating the amplitude flatness and the group delay flatness of each sub-band signal based on the first sub-band signal and the second sub-band signal, and removing the first sub-band signal and the second sub-band signal which are determined to be noise.

Preferably, the flatness calculating unit 402 is configured to calculate the amplitude flatness and the group delay flatness of each subband signal based on the first subband signal and the second subband signal, respectively.

Preferably, the determining unit 403 is configured to determine whether the uhf sensor is a linear system based on the amplitude flatness and the group delay flatness, and obtain a determination result.

Preferably, the recovery unit 404 is configured to perform recovery of the time domain waveform according to a preset recovery policy based on the determination result.

Preferably, the recovering unit 404 performs recovery of the time domain waveform according to a preset recovery policy based on the determination result, including:

and if the judging result indicates that the pulse electric field belongs to a linear system, recovering the phase by adopting a minimum phase method, and then recovering the time domain waveform of the pulse electric field.

And if the judging result indicates that the pulse electric field waveform does not belong to the linear system, recovering the pulse electric field time domain waveform based on the depth convolution network.

Preferably, the recovering unit 404 performs recovery of the pulse electric field time domain waveform after recovering the phase by using a minimum phase method, including:

performing Fast Fourier Transform (FFT) on the ultra-high frequency sensor output voltage signal to obtain a complex frequency spectrum sequence V (k);

interpolation calculation is carried out on the correction coefficient ln|H (k) | according to the frequency information of V (k), and the point number N of the correction coefficient is obtained; wherein H (k) is a discretized system transfer function, and is obtained by discretizing after dividing the Fourier transform of an output signal by the Fourier transform of an input signal;

performing scale transformation and continuation on ln|H (k) | according to N to obtain G (k);

performing inverse Fourier transform on the G (k), and performing windowing operation to obtain H (k) meeting the minimum phase condition;

and multiplying V (k) and H (k) in a complex spectrum form, and performing fast Fourier transform (IFFT) operation to obtain a corresponding pulse electric field time domain waveform e (t).

Preferably, the recovering unit 404 performs recovery of the pulse electric field time domain waveform based on a deep convolutional network, including:

collecting a pulse electric field waveform e (t) at the front end of the UHF sensor and a rear end output voltage waveform u (t);

converting a one-dimensional waveform signal into a two-dimensional time-frequency spectrogram by using a time-frequency conversion technology, and denoising;

using a DCN system to train a DCN network by taking a u (t) spectrogram set as input and a corresponding e (t) spectrogram set as output, and obtaining a trained depth network model;

and inputting the output voltage signal of the ultrahigh frequency sensor into a trained depth network model to obtain a corresponding pulse electric field time domain waveform.

The partial discharge pulse electric field time domain waveform recovering system 400 according to the embodiment of the present invention corresponds to the partial discharge pulse electric field time domain waveform recovering method 100 according to another embodiment of the present invention, and is not described herein.

The invention has been described with reference to a few embodiments. However, as is well known to those skilled in the art, other embodiments than the above disclosed invention are equally possible within the scope of the invention, as defined by the appended patent claims.

Generally, all terms used in the claims are to be interpreted according to their ordinary meaning in the technical field, unless explicitly defined otherwise therein. All references to "a/an/the [ means, component, etc. ]" are to be interpreted openly as referring to at least one instance of said means, component, etc., unless explicitly stated otherwise. The steps of any method disclosed herein do not have to be performed in the exact order disclosed, unless explicitly stated.

It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.

The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

Finally, it should be noted that: the above embodiments are only for illustrating the technical aspects of the present invention and not for limiting the same, and although the present invention has been described in detail with reference to the above embodiments, it should be understood by those of ordinary skill in the art that: modifications and equivalents may be made to the specific embodiments of the invention without departing from the spirit and scope of the invention, which is intended to be covered by the claims.

Claims (10)

1. A method for recovering a time domain waveform of a partial discharge pulsed electric field, the method comprising:

respectively decomposing the broadband electric field signal and the ultra-high frequency sensor output voltage signal into a plurality of first subband signals and a plurality of second subband signals with different frequency bands;

respectively calculating the amplitude flatness and the group delay flatness of each sub-band signal based on the first sub-band signal and the second sub-band signal;

judging whether the ultrahigh frequency sensor is a linear system or not based on the amplitude flatness and the group delay flatness, and acquiring a judging result;

and recovering the time domain waveform according to a preset recovery strategy based on the judging result.

2. The method according to claim 1, wherein the method further comprises:

before the amplitude flatness and the group delay flatness of each sub-band signal are calculated based on the first sub-band signal and the second sub-band signal, noise judgment is performed on the first sub-band signal and the second sub-band signal respectively, and the first sub-band signal and the second sub-band signal which are determined to be noise are removed.

3. The method according to claim 1, wherein the recovering the time domain waveform according to a preset recovery policy based on the determination result includes:

and if the judging result indicates that the pulse electric field belongs to a linear system, recovering the phase by adopting a minimum phase method, and then recovering the time domain waveform of the pulse electric field.

And if the judging result indicates that the pulse electric field waveform does not belong to the linear system, recovering the pulse electric field time domain waveform based on the depth convolution network.

4. A method according to claim 3, wherein the recovering of the pulse electric field time domain waveform after the recovering of the phase using the minimum phase method comprises:

performing Fast Fourier Transform (FFT) on the ultra-high frequency sensor output voltage signal to obtain a complex frequency spectrum sequence V (k);

interpolation calculation is carried out on the correction coefficient ln|H (k) | according to the frequency information of V (k), and the point number N of the correction coefficient is obtained; wherein H (k) is a discretized system transfer function, and is obtained by discretizing after dividing the Fourier transform of an output signal by the Fourier transform of an input signal;

performing scale transformation and continuation on ln|H (k) | according to N to obtain G (k);

performing inverse Fourier transform on the G (k), and performing windowing operation to obtain H (k) meeting the minimum phase condition;

and multiplying V (k) and H (k) in a complex spectrum form, and performing fast Fourier transform (IFFT) operation to obtain a corresponding pulse electric field time domain waveform e (t).

5. A method according to claim 3, wherein the recovery of the pulsed electric field time domain waveform based on the deep convolutional network comprises:

collecting a pulse electric field waveform e (t) at the front end of the UHF sensor and a rear end output voltage waveform u (t);

converting a one-dimensional waveform signal into a two-dimensional time-frequency spectrogram by using a time-frequency conversion technology, and denoising;

taking a u (t) spectrogram set as input and a corresponding e (t) spectrogram set as output by using a DCN method, training a DCN network, and obtaining a trained depth network model;

and inputting the output voltage signal of the ultrahigh frequency sensor into a trained depth network model to obtain a corresponding pulse electric field time domain waveform.

6. A partial discharge pulsed electric field time domain waveform recovery system, the system comprising:

the signal decomposition unit is used for respectively decomposing the broadband electric field signal and the ultra-high frequency sensor output voltage signal into a plurality of first subband signals and a plurality of second subband signals with frequency bands which are not mutually mixed;

a flatness calculation unit for calculating an amplitude flatness and a group delay flatness of each subband signal based on the first subband signal and the second subband signal, respectively;

the judging unit is used for judging whether the ultrahigh frequency sensor is a linear system or not based on the amplitude flatness and the group delay flatness, and obtaining a judging result;

and the recovery unit is used for recovering the time domain waveform according to a preset recovery strategy based on the judging result.

7. The system of claim 6, wherein the system further comprises:

and the denoising unit is used for respectively carrying out noise judgment on the first sub-band signal and the second sub-band signal before respectively calculating the amplitude flatness and the group delay flatness of each sub-band signal based on the first sub-band signal and the second sub-band signal, and removing the first sub-band signal and the second sub-band signal which are determined to be noise.

8. The system according to claim 6, wherein the recovery unit performs recovery of the time domain waveform according to a preset recovery policy based on the determination result, including:

and if the judging result indicates that the pulse electric field belongs to a linear system, recovering the phase by adopting a minimum phase method, and then recovering the time domain waveform of the pulse electric field.

And if the judging result indicates that the pulse electric field waveform does not belong to the linear system, recovering the pulse electric field time domain waveform based on the depth convolution network.

9. The system of claim 8, wherein the recovery unit performs recovery of the pulse electric field time domain waveform after recovering the phase using a minimum phase method, comprising:

performing Fast Fourier Transform (FFT) on the ultra-high frequency sensor output voltage signal to obtain a complex frequency spectrum sequence V (k);

interpolation calculation is carried out on the correction coefficient ln|H (k) | according to the frequency information of V (k), and the point number N of the correction coefficient is obtained; wherein H (k) is a discretized system transfer function, and is obtained by discretizing after dividing the Fourier transform of an output signal by the Fourier transform of an input signal;

performing scale transformation and continuation on ln|H (k) | according to N to obtain G (k);

performing inverse Fourier transform on the G (k), and performing windowing operation to obtain H (k) meeting the minimum phase condition;

and multiplying V (k) and H (k) in a complex spectrum form, and performing fast Fourier transform (IFFT) operation to obtain a corresponding pulse electric field time domain waveform e (t).

10. The system of claim 8, wherein the recovery unit performs recovery of the pulsed electric field time domain waveform based on a deep convolutional network, comprising:

collecting a pulse electric field waveform e (t) at the front end of the UHF sensor and a rear end output voltage waveform u (t);

converting a one-dimensional waveform signal into a two-dimensional time-frequency spectrogram by using a time-frequency conversion technology, and denoising;

using a DCN system to train a DCN network by taking a u (t) spectrogram set as input and a corresponding e (t) spectrogram set as output, and obtaining a trained depth network model;

and inputting the output voltage signal of the ultrahigh frequency sensor into a trained depth network model to obtain a corresponding pulse electric field time domain waveform.

Images