CN113608073A - Cable partial discharge pulse separation method under variable frequency resonance - Google Patents
Cable partial discharge pulse separation method under variable frequency resonance Download PDFInfo
- Publication number
- CN113608073A CN113608073A CN202110604569.9A CN202110604569A CN113608073A CN 113608073 A CN113608073 A CN 113608073A CN 202110604569 A CN202110604569 A CN 202110604569A CN 113608073 A CN113608073 A CN 113608073A
- Authority
- CN
- China
- Prior art keywords
- pulse
- variable frequency
- partial discharge
- cluster
- sequence
- 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.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/12—Testing dielectric strength or breakdown voltage ; Testing or monitoring effectiveness or level of insulation, e.g. of a cable or of an apparatus, for example using partial discharge measurements; Electrostatic testing
- G01R31/1227—Testing dielectric strength or breakdown voltage ; Testing or monitoring effectiveness or level of insulation, e.g. of a cable or of an apparatus, for example using partial discharge measurements; Electrostatic testing of components, parts or materials
Abstract
The invention discloses a cable partial discharge pulse separation method under variable frequency resonance, which comprises the steps of obtaining an original pulse type voltage signal in a variable frequency type series resonance circuit; searching pulse edges of the pulse type voltage signals, and analyzing the pulse type according to the search result; extracting characteristic quantities of different types of pulses, and performing CFSFDP clustering analysis according to different ranges of the characteristic quantities; according to the clustering analysis result, separating the local discharge signal from the interference pulse of the variable frequency power supply; the method can effectively separate the interference source from the partial discharge signal source, thereby realizing synchronous partial discharge test on the basis of the existing variable frequency series resonance system.
Description
Technical Field
The invention relates to the technical field of cable tests, in particular to a method for separating partial discharge pulses of a cable under variable frequency resonance.
Background
With the increase of national electric power demand and the transformation of urban power grids, cross-linked polyethylene (XLPE) cables are widely applied to power transmission and distribution networks of electric power systems due to the excellent electrical properties, heat resistance, mechanical properties and other factors. In order to ensure the safe reliability of the operation of the power cable, the CIGRE technical report, the IEC standard, the national fund subsidy project standard and the enterprise standard all indicate that: the insulation performance of the newly completed power cable project must be tested by a withstand voltage test. The variable frequency series resonance voltage-withstanding method is considered to be the most effective voltage-withstanding test method at present due to the advantages of small equipment power, good equivalence with power frequency, high feasibility of field implementation and the like. However, latent defects such as insulating air gaps, semiconductor protrusions, insulating interface conductor/semiconductor suspension particles, etc. may be introduced during XLPE cable production, cabling and installation of cable accessories. Experimental studies have shown that the latent defects mentioned above are very likely to pass the withstand voltage test, but they develop further under the effect of a long-term voltage, resulting in breakdown accidents within a few years or even months of operation after the handover test has passed. Therefore, although the variable frequency series resonance withstand voltage is one of the most commonly used test methods in the power cable handover test, it cannot detect latent defects, and the Partial Discharge (PD) test, which is an effective means for detecting latent defects, faces the problem of strong interference under the variable frequency series resonance condition. Therefore, a method for separating partial discharge pulses of the cable under variable-frequency series resonance is provided.
Disclosure of Invention
The invention aims to provide a method for separating partial discharge pulses of a cable under variable frequency resonance, which comprises the steps of firstly extracting pulses based on a second-order envelope curve and removing partial interference by adopting a threshold strategy; then extracting the pulse frequency domain and time domain waveform characteristic quantity, and carrying out self-adaptive clustering and discrete noise elimination by utilizing a CFSFDP (circulating fast search and fine of noise peaks) based on an improved density peak value, thereby realizing the separation of partial discharge pulse and interference signal; the method can realize separation and detection of partial discharge signals under strong interference of a variable frequency power supply, and provides a feasible solution for realizing synchronous test of resonance withstand voltage and partial discharge of the power cable.
In order to achieve the purpose, the invention provides the following technical scheme: a method for separating partial discharge pulses of a cable under variable frequency resonance comprises the following steps:
acquiring an original pulse type voltage signal in the variable-frequency series resonant circuit;
searching pulse edges of the pulse type voltage signals, and analyzing the pulse type according to the search result;
extracting characteristic quantities of different types of pulses, and performing CFSFDP clustering analysis according to different ranges of the characteristic quantities;
and separating the local discharge signal from the interference pulse of the variable frequency power supply according to the clustering analysis result.
Preferably, the method also comprises the steps of presetting a voltage threshold, eliminating interference voltage, and setting a voltage threshold U in the process of extracting the single-wave pulsethFiltering the original pulse-type voltage signal above the voltage threshold UthThe pulse waveform of (a) is considered to be an interference signal generated by the variable frequency power supply.
In any one of the above embodiments, preferably, when performing the pulse edge search on the pulse-type voltage signal, a pulse edge search method based on a second-order envelope is adopted, and includes the following steps:
carrying out maximum value search on the processed sequence according to a preset search window to obtain a sequence V 'consisting of all first-order maximum values'peak；
In order to eliminate partial oscillation of the first-order maximum value sequence, maximum value search is carried out on the first-order maximum value sequence to obtain a second-order maximum value sequence Vpeak；
For sequence VpeakLinear interpolation is carried out to obtain a second-order envelope line sequence V with the length of NN×1；
Searching a maximum value sequence with the second-order envelope curve amplitude value higher than the white noise level of the processed pulse type voltage signal
m is the number of maxima whose second-order envelope is higher than white noise level, and the sequence is used
Middle arbitrary point Vi′(i∈[1,2,…,m]) Respectively searching zero point of second-order envelope curve to left as starting point, recording its index as single-pulse starting point PsiSearching a zero point of a second-order envelope line rightwards, recording the index of the zero point as a single-pulse end point Pei, and obtaining a single-pulse start-stop index matrix: pIndex＝[Ps1,Pe1；Ps2,Pe2；L；Psm,Pem]。
In any of the above embodiments, preferably, when analyzing the pulse type according to the search result, the following method is adopted: deleting repeated rows according to the obtained monopulse start-stop index matrix;
for the extracted pulse interval distance dpluseComparing the magnitude with a preset threshold value when the pulse interval d is two timespluse>When a threshold value is preset, two pulses are judged, and when the two pulses are separated by dpluseJudging the same pulse when the pulse is less than or equal to a preset threshold value;
and judging each pulse type according to the calculated pulse number.
Preferably, in any of the above embodiments, the pulse type includes a variable frequency power interference signal and a PD pulse signal according to the order of magnitude of the signal.
In any of the above embodiments, preferably, when extracting the feature quantities of the pulses of different types, a frequency domain equivalent bandwidth is selected as the frequency domain feature quantity, where the frequency domain equivalent bandwidth is calculated by using the following formula:
wherein s (f) is a fourier transform of the time domain signal.
Preferably, in any one of the above embodiments, the performing CFSFDP cluster analysis includes the following steps:
forming the extracted characteristic quantities into a cluster data set D ═ { x ═ xi}，i＝1，…，N；
Computing any two data x in a clustered data set DiAnd xjA distance d betweenij＝dist(xi,xj) For any data point x in DiDefining the local density piIs composed of
Distance deltaiIs composed of
and calculating the local density rho and the distance delta of all the sample points, projecting the sample points to a two-dimensional plane formed by the local density rho and the distance delta to form a rho-delta image, and counting the centers and the types of the sample clusters through the rho-delta image.
And (3) assigning an attribute label (class1, class2 and …) to the cluster center, assigning other sample points to the cluster which is closest to the cluster center and has the sample point with the high density value, and obtaining a cluster analysis result.
In any of the above embodiments, preferably, the following adaptive clustering center decision method is adopted to automatically obtain the number of sample clusters and the center sample point, including:
calculating and normalizing the local density rho and the distance delta of each sample point;
calculating a clustering index gamma according to the normalized local density rho and the distance delta of the sample point and the following formula
The clustering index gamma is arranged in ascending order to obtain gamma ═ gamma1,γ2,γ3,...,γN]And index sequence l in original sampleN×1N is the total number of samples;
constructing N clustering index subsets: γ γ 1 ═ γ1]，γγ2＝[γ1,γ2],γγ3＝[γ1,γ2,γ3],…,γγN＝[γ1,γ2,…,γk,…,γN]；
Calculating standard deviation of each clustering index subset and rounding downwards to obtain [ sigma ═ sigma [ [ sigma ]1,σ2,...,σk,...,σN]；
Finding a non-zero value sigma of the standard deviation sigmakRemember [ sigma ]k-1,σk,...,σN]The total number is the number of cluster centers, [ l ]k-1,lk,…,lN]The index position of the cluster center in the original sample sequence.
In any of the above embodiments, preferably, the method further includes recording a sinusoidal voltage phase corresponding to a pulse occurrence time during the pulse extraction process, drawing a cluster center time domain single wave and a PRPD spectrogram of the cluster sample for each cluster of pulses according to a clustering result, and further determining a property to which a sample cluster signal belongs.
Compared with the prior art, the invention has the beneficial effects that:
1. according to the method for separating the partial discharge pulse of the cable under the variable frequency resonance, the pulse voltage signal is obtained during the voltage withstanding test, the partial discharge signal and the power supply interference signal are separated from the pulse voltage signal by utilizing the CFSFDP algorithm, and the quality of the insulation performance of the cable and the defects of the cable inside latency can be detected at the same time.
2. According to the cable partial discharge pulse separation method under the variable frequency resonance, the preset voltage threshold is set, and different characteristics of large amplitude difference of the variable frequency power supply and the PD pulse are utilized to carry out primary separation on pulse signals, so that subsequent clustering separation is realized, and calculation data are simplified.
3. According to the cable partial discharge pulse separation method under the variable frequency resonance, different types of pulses have characteristic quantities in different ranges and are separated in a clustering mode. Compared with the method of extracting the wave front and the wave tail area of the pulse waveform profile by adopting an optimal fitting mode, the characteristic quantity extracted by the method can be used for quantifying the waveform profile characteristic, and meanwhile, the calculation is simple and convenient and is high in speed.
4. The method for separating the partial discharge pulse of the cable under the variable frequency resonance further combines the time domain waveform and the PRPD spectrogram to realize the judgment of the partial discharge source. The separation of the single partial discharge source and the interference pulse under the frequency conversion series resonance is realized, and the separation of the multi-source discharge pulse under the background of the frequency conversion series resonance is further researched subsequently.
Additional aspects and advantages of the invention will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the invention.
Drawings
FIG. 1 is a schematic diagram of a variable frequency series resonance employed in the prior art of the present invention;
FIG. 2(a) shows signals collected by a high-frequency current sensor under a resonant voltage;
FIG. 2(b) shows signals collected by the high-frequency current sensor under the power frequency voltage;
FIG. 3(a) is a time domain waveform diagram of a single interference signal under resonance conditions;
FIG. 3(b) is a time domain waveform diagram of a single partial discharge signal at a power frequency voltage;
fig. 3(c) is a spectrum mean curve for counting 500 normalized interference signals;
fig. 3(d) is a spectrum mean curve of statistical 500 normalized partial discharge signals;
FIG. 4 is second order envelope extraction;
FIG. 5 is a diagram illustrating the pulse extraction results;
FIG. 6(a) is a diagram of an original sample artificially constructed with three clusters of two-dimensional normal distributions;
FIG. 6(b) is a normalized ρ - δ plot;
FIG. 6(c) is a diagram of clustering effect analysis;
FIG. 7(a) is a sequence diagram of a clustering index using an adaptive threshold strategy;
FIG. 7(b) is a sequence diagram of the standard deviation of the clustering index using the adaptive threshold strategy;
FIG. 8 is a schematic diagram of an experimental platform constructed according to the present invention;
FIG. 9 is a flow chart of a partial discharge pulse separation strategy under resonance conditions;
FIG. 10(a) is a three-dimensional spectrogram of the original characteristic quantity of the pulse separation result under the resonance condition of the present invention;
FIG. 10(b) is a sequence of clustering index subset standard deviations;
fig. 10(c) is a clustering result chart.
FIG. 11(a) is a time domain spectrum of a class of pulse signals;
FIG. 11(b) a time domain spectrum of another type of pulse signal;
FIG. 11(c) a PRPD spectrum of a class of pulse signals;
FIG. 11(d) the PRPD spectrum of another type of pulse signal.
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
In the present invention, unless otherwise expressly specified or limited, the terms "mounted," "connected," "secured," and the like are to be construed broadly and can, for example, be fixedly connected, detachably connected, or integrally connected; can be mechanically or electrically connected; they may be connected directly or indirectly through intervening media, or they may be interconnected between two elements. The specific meanings of the above terms in the present invention can be understood by those skilled in the art according to specific situations.
Referring to fig. 1, the present invention provides a technical solution: a method for separating partial discharge pulses of a cable under variable frequency resonance comprises the following steps:
s1, obtaining variable frequency type series resonance circuitThe original pulse-type voltage signal of (1); preferably, the method also comprises the steps of presetting a voltage threshold, eliminating interference voltage, and setting a voltage threshold U in the process of extracting the single-wave pulsethFiltering the original pulse-type voltage signal above the voltage threshold UthThe pulse waveform of (a) is considered to be an interference signal generated by the variable frequency power supply.
Fig. 1 is a schematic diagram of a conventional frequency-conversion series resonant system. Mainly comprises 4 parts of rectification, inversion, boosting and resonance. The inverter circuit modulation method is mainly SPWM modulation method. The control device is a full-control device IGBT. The working principle is as follows: the rectified and filtered direct current passes through an H bridge inverter circuit to obtain a sinusoidal voltage; the sine voltage is boosted to one side of the resonant circuit through the exciting transformer; by collecting the voltage and the current of the resonant circuit, the on-off state and the duty ratio of the IGBT are controlled, so that the frequency of the output voltage of the variable frequency power supply reaches the resonant frequency, and the voltage of a tested product reaches a preset value. The switching device in the inversion module acts for multiple times in the whole voltage period, and due to the fact that the switching device needs to cut off and conduct the voltage of 10-380V, pulse type overvoltage signals can be generated in a loop and are transmitted to a resonant loop through an excitation transformer or transmitted through a ground wire. If the partial discharge detection is carried out on the test article, the pulse signals can be collected by the sensor, and the partial discharge detection result is seriously interfered.
In order to compare the interference of the variable frequency power supply with the PD pulse characteristics, the partial discharge test is carried out on the 35kV short cable with metal particle defects at the terminal under a laboratory series resonance platform and a power frequency partial discharge-free test platform respectively. Fig. 2 shows signals acquired by a high frequency current sensor (HFCT) under two conditions (frequency fs ═ 80 MHz). Compared with the characteristics of partial discharge signals under power frequency, the interference pulse phase under the resonance condition is distributed in the whole sine wave period, the number of pulses in a single period is large, and the amplitude is large. When the resonance voltage is increased, the interference pulse amplitude of the variable frequency power supply is increased, the increase amplitude of the partial discharge pulse signal is smaller than that of the interference pulse, the distribution range of the pulse amplitude is further expanded, and even the difference is 1-2 orders of magnitude. If two orders of magnitude pulse signals are to be distinguished simultaneously, a higher requirement is put forward on the vertical resolution of the acquisition device. Considering the situation that the pulse amplitude exceeds the range of the acquisition device, corresponding protective measures are needed to avoid the damage of the acquisition device.
And further analyzing the characteristics of the interference pulse single-wave signal of the variable-frequency power supply based on the conditions. Fig. 3(a) and 3(b) are time domain waveform diagrams of a single interference signal and a single partial discharge signal under a power frequency voltage under a resonance condition, fig. 3(c) and 3(d) are respectively frequency spectrum mean curves for counting 500 normalized interference signals and partial discharge signals, and a shaded area represents the size of a frequency spectrum standard deviation. The interference pulse is characterized in that: the amplitude of the signal is large, oscillation begins to attenuate after one cycle of oscillation is increased, and a frequency spectrum shows that a double-frequency peak value exists and the standard deviation is large; the partial discharge pulse amplitude is small and has an exponential attenuation trend, a single frequency peak point is formed on a frequency spectrum, and the fluctuation amplitude is small, so that the single partial discharge waveform is more stable than the interference pulse of the variable frequency power supply. The comparison shows that the interference signal and the partial discharge signal have obvious time-frequency domain difference, and the characteristic quantity extracted from the interference signal and the partial discharge signal can obtain good classification effect.
S2, performing pulse edge search on the pulse type voltage signal, and analyzing the pulse type according to the search result; preferably, when the pulse edge search is performed on the pulse-type voltage signal, a pulse edge search method based on a second-order envelope curve is adopted, and the method includes the following steps:
s201, carrying out mean value removal and absolute value processing on an original pulse type voltage signal; let the original data sequence be XN×1，XiFor signal amplitude, N is the original data length.
S202, according to a preset search window and a preset search window, carrying out maximum value search on the processed sequence to obtain a sequence V 'consisting of all first-order maximum values'peak；
In order to eliminate partial oscillation of the first-order maximum value sequence, maximum value search is carried out on the first-order maximum value sequence to obtain a second-order maximum value sequence Vpeak；
S203, pair sequence VpeakLinear interpolation is carried out to obtain a second-order envelope line sequence V with the length of NN×1；
S204, searching for the pulse with the second-order envelope amplitude value higher than that of the processed pulseMaximum sequence of white noise levels for a model voltage signal
m is the number of maxima whose second-order envelope is higher than white noise level, and the sequence is used
Middle arbitrary point Vi′(i∈[1,2,…,m]) Respectively searching zero point of second-order envelope curve to left as starting point, recording its index as single-pulse starting point PsiSearching right for zero point of second-order envelope curve to record index as end point P of single pulseeiAnd obtaining a single-pulse start-stop index matrix: pIndex＝[Ps1,Pe1；Ps2,Pe2；L；Psm,Pem]。
Fig. 4 is a schematic diagram of second-order envelope extraction of a section of original signal, where the extracted second-order envelope is a unipolar pulse, and substantially covers the entire time domain pulse, and is used for edge search, so as to achieve the purpose of completely extracting a pulse signal. Fig. 5 shows the extracted 4 monopulse waveforms in the dashed box for a segment of the original signal pulse extraction result.
S3, extracting characteristic quantities of different types of pulses, and carrying out CFSFDP clustering analysis according to different ranges of the characteristic quantities;
further, when analyzing the pulse type according to the search result, the following method is adopted:
deleting repeated rows according to the obtained monopulse start-stop index matrix;
for the extracted pulse interval distance dpluseComparing the magnitude with a preset threshold value when the pulse interval d is two timespluse>When a threshold value is preset, two pulses are judged, and when the two pulses are separated by dpluseJudging the same pulse when the pulse is less than or equal to a preset threshold value;
and judging each pulse type according to the calculated pulse number.
Preferably, in any of the above embodiments, the pulse types are classified according to the magnitude of the signal amplitude, and include variable frequency power interference signals and PD pulse signals. The PD pulse signals are further classified according to source.
In any of the above embodiments, preferably, when extracting the feature quantities of the pulses of different types, a frequency domain equivalent bandwidth is selected as the frequency domain feature quantity, where the frequency domain equivalent bandwidth is calculated by using the following formula:
wherein s (f) is a fourier transform of the time domain signal.
Taking the extreme value sequence of the absolute value of the pulse signal (after mean value removal), and obtaining a first-order envelope line sequence V after linear interpolationN1×1N is the length of the pulse signal sequence, Vm is the maximum value of the first-order envelope curve, and the 1-norm A of the pulse envelope wave front is respectively definedsIs composed of
Wave tail 1-norm AeIs composed of
Different types of pulses have different ranges of characteristic quantities and are separated in a clustering mode. Compared with the prior art that the optimal fitting mode is adopted to extract the wave front and tail areas of the pulse waveform profile, the characteristic quantity extracted by the method can quantify the waveform profile characteristics, and meanwhile, the calculation is simple and convenient, and the speed is high.
The CFSFDP clustering analysis comprises the following steps:
forming the extracted characteristic quantities into a cluster data set D ═ { x ═ xi}，i＝1，…，N；
Computing any two data x in a clustered data set DiAnd xjA distance d betweenij＝dist(xi,xj) For any data point x in DiDefining local densityρiIs composed of
Distance deltaiIs composed of
And calculating the local density rho and the distance delta of all the sample points, projecting the sample points to a two-dimensional plane formed by the local density rho and the distance delta to form a rho-delta image, and counting the centers and the types of the sample clusters through the rho-delta image.
As shown in fig. 6(a), three clusters of original sample points and uniformly distributed noise points in two-dimensional normal distribution are artificially constructed, the horizontal and vertical coordinates are two-dimensional random variables x and y, and the number of sample points in each cluster is 500. The normalized ρ - δ graph is plotted in fig. 6(b), and it can be clearly seen that there are 3 outliers (circled by a dashed line) corresponding to the center points of the 3 types of sample clusters, respectively, fig. 6(c) shows the clustering effect, and the black dots are noise dots.
And (3) assigning an attribute label (class1, class2 and …) to the cluster center, assigning other sample points to the cluster which is closest to the cluster center and has the sample point with the high density value, and obtaining a cluster analysis result.
In any of the above embodiments, preferably, the following adaptive clustering center decision method is adopted to automatically obtain the number of sample clusters and the center sample point, including:
calculating and normalizing the local density rho and the distance delta of each sample point;
calculating a clustering index gamma according to the normalized local density rho and the distance delta of the sample point and the following formula
The clustering index gamma is arranged in ascending order to obtain gamma ═ gamma1,γ2,γ3,...,γN]And index sequence l in original sampleN×1N is the total number of samples;
constructing N clustering index subsets: γ γ 1 ═ γ1]，γγ2＝[γ1,γ2],γγ3＝[γ1,γ2,γ3],…,γγN＝[γ1,γ2,…,γk,…,γN]；
Calculating standard deviation of each clustering index subset and rounding downwards to obtain [ sigma ═ sigma [ [ sigma ]1,σ2,...,σk,...,σN]；
Finding a non-zero value sigma of the standard deviation sigmakRemember [ sigma ]k-1,σk,...,σN]The total number is the number of cluster centers, [ l ]k-1,lk,…,lN]The index position of the cluster center in the original sample sequence.
As shown in fig. 7, when the adaptive threshold strategy is used to perform adaptive cluster center decision on the sample in fig. 6(a), the larger the value of the cluster index γ determined by equation (6), the more likely it is a cluster center. Finally, the index of the cluster center point number of 3 and the cluster center point in the original sample sequence is obtained [49,1763,747 ]. The number of sample clusters and the central sample point can be automatically obtained through a self-adaptive clustering center decision algorithm.
And S4, separating the local discharge signal from the interference pulse of the variable frequency power supply according to the clustering analysis result.
In any of the above embodiments, preferably, the method further includes recording a sinusoidal voltage phase corresponding to a pulse occurrence time during the pulse extraction process, drawing a cluster center time domain single wave and a PRPD spectrogram of the cluster sample for each cluster of pulses according to a clustering result, and further determining a property to which a sample cluster signal belongs.
In order to verify the effectiveness of the partial discharge signal separation of the proposed method under the variable frequency resonance condition. An experimental platform as shown in fig. 8 was constructed herein. The model of the variable frequency power supply is SAMCO-vm5, the input voltage is 380V, and the capacity is 40 kW; the capacity of the excitation transformer is 40kVA, the output voltage is 2kV/6kV/20kV, and the output frequency is 30-300 Hz; the inductors are connected in series by 4 sections, and the total inductance is 400H; the withstand voltage of the capacitive voltage divider is 150kV, and the capacitance of the high-voltage arm is 1000 pF; the cable is YJL 26/37-1 × 95, the length is about 3m, the actually measured capacitance is 0.5nF, and a plurality of metal particles are arranged in the interface of the main insulation and the accessory of the cable; the bandwidth of HFCT-6 dB is 2.5-216 MHz, and the maximum sensitivity is 5.83 mV/mA; a broadband voltage follower is connected between the sensor and the acquisition device to prevent the acquisition card from being damaged by high-amplitude interference pulses, the sampling frequency of the PD acquisition device is 80MHz, the vertical resolution is 12bit, and the sampling depth is 6.4 MHz; the measured resonant frequency was about 186.8Hz and the resonant voltage was 23 kV.
In order to minimize the problem of unbalanced pulse signal samples (the number of PD signals is much smaller than that of interference signals), a threshold Uth may be set during the single-wave pulse extraction, and a pulse waveform above the threshold is considered as an interference signal generated by the variable frequency power supply. And most of interference is eliminated by adopting a threshold strategy, the redundancy of signals to be separated is reduced, and the requirement on the vertical resolution of the acquisition device is reduced. It is worth to say that Uth cannot be set too low, otherwise part of PD signals may be mistaken as interference signals and automatically filtered out, and the threshold Uth is selected to be 100mV, but at the same time, interference signals with amplitudes lower than Uth may still exist, and need to be further separated and identified by extracting feature quantities.
A flow chart of the partial discharge pulse separation strategy based on the variable frequency series resonance condition is shown in fig. 9. In order to further analyze the clustering result, the sinusoidal voltage phase corresponding to the pulse generation moment is recorded simultaneously in the pulse extraction process, and a clustering center time domain single wave and a PRPD spectrogram of the cluster of samples are drawn for each cluster of pulses according to the clustering result so as to judge the property of the sample cluster signal.
Based on a frequency conversion series resonance partial discharge detection platform built in a laboratory, 10 groups of original pulse sequences are collected, the time length of each group is 80ms, and the total number is about 150 sine wave periods. The 3-dimensional spectrogram of the extracted single-wave pulse characteristic quantity is shown in fig. 10(a), the clustering index subset standard deviation sequence obtained by adopting the self-adaptive CFSFDP method is shown in fig. 10(b), and the number of non-zero standard deviations is 1. The number of cluster centers is 2, which is 1987, 275 at the original sequence index position, as described in section 2.3.2. Fig. 10(c) is a diagram illustrating the clustering result, wherein the red (Class 2) and yellow (Class 1) signal clusters are two types of pulse signals, and the black dots are outliers. The experimental result shows that the acquired data at least comprises two types of pulse signals.
The PRPD spectrogram corresponding to the extracted sample center point and returned to the time domain waveform and the two clusters of pulse signals is shown in fig. 11. The time domain waveform shown in fig. 11(a) is similar to that shown in fig. 3(a), the spectrum of the cluster signal PRPD is distributed in the whole sinusoidal cycle, and the phase is relatively fixed, and the cluster signal can be determined as an interference signal in consideration of the periodic pulse signal generated by the IGBT acting at a fixed moment in a plurality of oscillation cycles; the time domain waveform shown in fig. 11(b) is similar to that shown in fig. 3(b), the spectrum of the cluster signal PRPD is in a three-quadrant asymmetric distribution in a sine cycle, and is consistent with the characteristics of the surface discharge partial discharge spectrum, and the cluster signal can be determined to be a partial discharge cluster signal.
In the description of the present invention, it is to be understood that the indicated orientations or positional relationships are based on the orientations or positional relationships shown in the drawings and are only for convenience in describing the present invention and simplifying the description, but are not intended to indicate or imply that the indicated devices or elements must have a particular orientation, be constructed and operated in a particular orientation, and are not to be construed as limiting the present invention.
In the description herein, references to the description of the term "one embodiment," "some embodiments," "an example," "a specific example," or "some examples," etc., mean that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, the schematic representations of the terms used above do not necessarily refer to the same embodiment or example. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.
Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
Claims (9)
1. A method for separating partial discharge pulses of a cable under variable frequency resonance is characterized by comprising the following steps:
acquiring an original pulse type voltage signal in the variable-frequency series resonant circuit;
searching pulse edges of the pulse type voltage signals, and analyzing the pulse type according to the search result;
extracting characteristic quantities of different types of pulses, and performing CFSFDP clustering analysis according to different ranges of the characteristic quantities;
and separating the local discharge signal from the interference pulse of the variable frequency power supply according to the clustering analysis result.
2. The method as claimed in claim 1, further comprising presetting a voltage threshold and removing the interference voltage, including performing a single-wave pulse extraction process
In, setting a voltage threshold value UthFiltering the original pulse-type voltage signal above the voltage threshold UthThe pulse waveform of (a) is considered to be an interference signal generated by the variable frequency power supply.
3. The method for separating partial discharge pulses of a cable under variable frequency resonance according to claim 1, wherein a pulse edge search method based on a second-order envelope is adopted when performing pulse edge search on the pulse-type voltage signal, and the method comprises the following steps:
carrying out mean value removal and absolute value processing on the original pulse voltage signal;
carrying out maximum value search on the processed sequence according to a preset search window to obtain a sequence V 'consisting of all first-order maximum values'peak；
Then carrying out maximum search on the first order maximum value sequence to obtain a second order maximum value sequence Vpeak；
For sequence VpeakLinear interpolation is carried out to obtain a second-order envelope line sequence V with the length of NN×1；
Searching a maximum value sequence V 'with the amplitude of the second-order envelope line higher than the white noise level of the processed pulse type voltage signal'm×1M is the number of maxima whose second-order envelope is higher than white noise level, and the sequence V'm×1Middle arbitrary point Vi′(i∈[1,2,…,m]) Respectively searching zero point of second-order envelope curve to left as starting point, recording its index as single-pulse starting point PsiSearching right for zero point of second-order envelope curve to record index as end point P of single pulseeiAnd obtaining a single-pulse start-stop index matrix: pIndex＝[Ps1,Pe1；Ps2,Pe2；L；Psm,Pem]。
4. The method for separating the partial discharge pulse of the cable under the variable frequency resonance as claimed in claim 3, wherein the following method is adopted when the pulse type is analyzed according to the search result:
deleting repeated rows according to the obtained monopulse start-stop index matrix;
for the extracted pulse interval distance dpluseComparing the magnitude with a preset threshold value when the pulse interval d is two timespluse>When a threshold value is preset, two pulses are judged, and when the two pulses are separated by dpluseJudging the same pulse when the pulse is less than or equal to a preset threshold value;
and judging each pulse type according to the calculated pulse number.
5. The method of claim 3, wherein the pulse types include variable frequency power supply interference signals and PD pulse signals according to the order of magnitude of the signal.
6. The method according to claim 3, wherein when extracting the characteristic quantities of the different types of pulses, the method selects a frequency domain equivalent bandwidth as the frequency domain characteristic quantity, and the frequency domain equivalent bandwidth is calculated by using the following formula:
wherein s (f) is a fourier transform of the time domain signal.
7. The method for cable partial discharge pulse separation at variable frequency resonance according to claim 3, wherein the performing CFSFDP clustering analysis comprises the following steps:
forming the extracted characteristic quantities into a cluster data set D ═ { x ═ xi}，i＝1，…，N；
Computing any two data x in a clustered data set DiAnd xjA distance d betweenij＝dist(xi,xj) For any data point x in DiDefining the local density piIs composed of
Distance deltaiIs composed of
calculating the local density rho and the distance delta of all sample points, projecting the sample points to a two-dimensional plane formed by the local density rho and the distance delta to form a rho-delta image, and counting the center and the number of types of sample clusters through the rho-delta image;
and (3) assigning an attribute label (class1, class2 and …) to the cluster center, assigning other sample points to the cluster which is closest to the cluster center and has the sample point with the high density value, and obtaining a cluster analysis result.
8. The method for separating partial discharge pulses of a cable under variable frequency resonance according to claim 7, wherein the following adaptive clustering center decision method is adopted to automatically obtain the number of sample clusters and the center sample point, and comprises:
calculating and normalizing the local density rho and the distance delta of each sample point;
calculating a clustering index gamma according to the normalized local density rho and the distance delta of the sample point and the following formula
The clustering index gamma is arranged in ascending order to obtain gamma ═ gamma1,γ2,γ3,...,γN]And index sequence l in original sampleN×1N is the total number of samples;
constructing N clustering index subsets: γ γ 1 ═ γ1]，γγ2＝[γ1,γ2],γγ3＝[γ1,γ2,γ3],…,γγN＝[γ1,γ2,…,γk,…,γN]；
Calculating standard deviation of each clustering index subset and rounding downwards to obtain [ sigma ═ sigma [ [ sigma ]1,σ2,...,σk,...,σN]；
Finding a non-zero value sigma of the standard deviation sigmakRemember [ sigma ]k-1,σk,...,σN]The total number is the number of cluster centers, [ l ]k-1,lk,…,lN]The index position of the cluster center in the original sample sequence.
9. The method for separating the partial discharge pulses of the cable under the variable frequency resonance as recited in claim 7, further comprising recording the sinusoidal voltage phase corresponding to the pulse generation time during the pulse extraction process, drawing a clustering center time domain single wave and a PRPD spectrogram of the cluster of samples for each cluster of pulses according to the clustering result, and further judging the property of the cluster of samples.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110604569.9A CN113608073A (en) | 2021-05-31 | 2021-05-31 | Cable partial discharge pulse separation method under variable frequency resonance |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110604569.9A CN113608073A (en) | 2021-05-31 | 2021-05-31 | Cable partial discharge pulse separation method under variable frequency resonance |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113608073A true CN113608073A (en) | 2021-11-05 |
Family
ID=78303430
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110604569.9A Pending CN113608073A (en) | 2021-05-31 | 2021-05-31 | Cable partial discharge pulse separation method under variable frequency resonance |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113608073A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114528888A (en) * | 2022-04-25 | 2022-05-24 | 广东玖智科技有限公司 | PPG signal clustering center acquisition method and device and PPG signal processing method and device |
| CN114545167A (en) * | 2022-02-23 | 2022-05-27 | 四川大学 | Cable terminal partial discharge pulse classification method based on t-SNE algorithm |
-
2021
- 2021-05-31 CN CN202110604569.9A patent/CN113608073A/en active Pending
Non-Patent Citations (1)
| Title |
|---|
| 黄永禄 等: "《基于改进CFSFDP算法的变频谐振下电缆局部放电脉冲分离方法》", 《高电压技术》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114545167A (en) * | 2022-02-23 | 2022-05-27 | 四川大学 | Cable terminal partial discharge pulse classification method based on t-SNE algorithm |
| CN114528888A (en) * | 2022-04-25 | 2022-05-24 | 广东玖智科技有限公司 | PPG signal clustering center acquisition method and device and PPG signal processing method and device |
| CN114528888B (en) * | 2022-04-25 | 2022-07-12 | 广东玖智科技有限公司 | PPG signal clustering center acquisition method and device and PPG signal processing method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109901031B (en) | Signal-to-noise separation method for partial discharge signal and information data processing terminal | |
| Contin et al. | Classification and separation of partial discharge signals by means of their auto-correlation function evaluation | |
| Contin et al. | Digital detection and fuzzy classification of partial discharge signals | |
| CA2918679C (en) | Pattern recognition method for partial discharge of three-phase cylinder type ultrahigh voltage gis | |
| Chan et al. | Time-frequency sparsity map on automatic partial discharge sources separation for power transformer condition assessment | |
| CN112014700B (en) | GIS insulator defect identification method and system based on partial discharge multi-information fusion | |
| CN113608073A (en) | Cable partial discharge pulse separation method under variable frequency resonance | |
| Seo et al. | Probabilistic wavelet transform for partial discharge measurement of transformer | |
| CN109102508B (en) | Method for identifying insulation defects of alternating current cable based on partial discharge image characteristics | |
| Pylarinos et al. | Impact of noise related waveforms on long term field leakage current measurements | |
| Ardila-Rey et al. | Separation techniques of partial discharges and electrical noise sources: A review of recent progress | |
| CN112763871A (en) | Partial discharge classification identification method | |
| CN113484706A (en) | Double-sensor detection method and system for partial discharge of cable under series resonance | |
| CN108108659B (en) | Island detection key feature extraction method based on empirical mode decomposition | |
| Chang et al. | Assessment of the insulation status aging in power cable joints using support vector machine | |
| CN104865508A (en) | Partial discharge recognition method based on data grouping quantification | |
| CN109324268B (en) | Power distribution network early fault detection method and device based on Bayesian inference | |
| CN111693829A (en) | Partial discharge noise and discharge distinguishing method for non-contact ultrasonic detection | |
| KR101961793B1 (en) | Apparatus and Method for eliminating noise | |
| Seo et al. | A novel signal extraction technique for online partial discharge (PD) measurement of transformers | |
| Kumar et al. | Classification of PD faults using features extraction and K-means clustering techniques | |
| CN116008735A (en) | Partial discharge signal extraction method and system based on density | |
| Hao et al. | A new method for automatic multiple partial discharge classification | |
| Kumar et al. | Performance Evaluation of AI-based Algorithms for Condition Assessment of Power Components | |
| Li et al. | A novel partial discharge pulse separation method for variable frequency resonant test |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20211105 |
|
| RJ01 | Rejection of invention patent application after publication |