CN118152942A - Ultrahigh frequency partial discharge identification method based on shape, energy and statistical characteristics - Google Patents
Ultrahigh frequency partial discharge identification method based on shape, energy and statistical characteristics Download PDFInfo
- Publication number
- CN118152942A CN118152942A CN202410212708.7A CN202410212708A CN118152942A CN 118152942 A CN118152942 A CN 118152942A CN 202410212708 A CN202410212708 A CN 202410212708A CN 118152942 A CN118152942 A CN 118152942A
- Authority
- CN
- China
- Prior art keywords
- partial discharge
- ultrahigh frequency
- energy
- shape
- frequency partial
- 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
- 238000000034 method Methods 0.000 title claims abstract description 50
- 238000004364 calculation method Methods 0.000 claims abstract description 40
- 238000012545 processing Methods 0.000 claims abstract description 9
- 238000000605 extraction Methods 0.000 claims description 61
- 230000000737 periodic effect Effects 0.000 claims description 21
- 238000004422 calculation algorithm Methods 0.000 claims description 20
- 230000005856 abnormality Effects 0.000 claims description 6
- 238000007635 classification algorithm Methods 0.000 claims description 5
- 230000002159 abnormal effect Effects 0.000 claims description 3
- 230000035945 sensitivity Effects 0.000 abstract description 2
- 230000003595 spectral effect Effects 0.000 description 36
- 238000001228 spectrum Methods 0.000 description 35
- 238000005070 sampling Methods 0.000 description 15
- 230000006870 function Effects 0.000 description 10
- 238000009825 accumulation Methods 0.000 description 6
- 238000007599 discharging Methods 0.000 description 6
- 238000012360 testing method Methods 0.000 description 6
- 230000007774 longterm Effects 0.000 description 4
- 238000012706 support-vector machine Methods 0.000 description 4
- 238000004590 computer program Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000013528 artificial neural network Methods 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000003066 decision tree Methods 0.000 description 2
- 238000001514 detection method Methods 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 239000002184 metal Substances 0.000 description 1
- 229910052751 metal Inorganic materials 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000007619 statistical method Methods 0.000 description 1
Landscapes
- Testing Relating To Insulation (AREA)
Abstract
The invention discloses an ultrahigh frequency partial discharge identification method based on shape, energy and statistical characteristics, which comprises the following steps of calibrating the characteristic based on the shape, the energy and the statistical ultrahigh frequency partial discharge: the method comprises the steps of reading ultrahigh frequency partial discharge data, processing the data, extracting and calibrating ultrahigh frequency partial discharge characteristic quantity; characteristic calculation, fault anomaly type identification and positioning are carried out on the ultrahigh frequency partial discharge: the method comprises the steps of collecting ultrahigh frequency partial discharge signals and processing the signals; extracting the characteristic quantity of ultrahigh frequency partial discharge; normal operation data and interference data logic discrimination; and (3) identifying and positioning the ultrahigh frequency partial discharge type and logically reasoning the ultrahigh frequency partial discharge characteristic quantity. The method is suitable for the ultrahigh frequency partial discharge identification of the primary equipment of the power grid, the existing partial discharge site is found in the early stage of the short circuit fault, the partial discharge is prevented from further evolving into the power grid accident, and the automation level is improved; the time domain and the frequency characteristics of the ultrahigh frequency partial discharge are respectively extracted based on three different dimensions of shape, energy and statistics, and different objects have different characteristic sensitivities and are wide in application range; the invention adopts three different logics of AND logic, three-out-of-two logic and three-out-of-one logic, and has high flexibility.
Description
Technical Field
The invention relates to an ultrahigh frequency partial discharge type identification method, in particular to an ultrahigh frequency partial discharge identification method based on shape, energy and statistical characteristics.
Background
Transformers, GIS (gas insulated switchgear ), GILs (gas insulated metal enclosed transmission lines, gas Insulated Line), switchgear etc. are the main devices of the power grid, the safe operation of the devices is related to the reliability level of the power system, if a fault occurs, it will cause a local or even large-area power outage. With the continuous increase of the number of substations, the operation and maintenance cost of the substation equipment is higher and higher, and the online monitoring requirement of the substation main equipment is more and more urgent.
The current power grid scale is continuously increased, partial discharge faults such as a transformer, a GIS (gas insulated switchgear), a GIL (gas insulated switchgear), a switch cabinet and the like are frequent, if the partial discharge faults are not handled in time, the partial discharge faults can be developed into short circuit faults, and power failure of the power grid can be caused by equipment faults, so that the partial discharge of the transformer, the GIS, the GIL, the switch cabinet and the like is particularly important to be timely found and identified, and the partial discharge is characterized by wider frequency band, wherein the partial discharge of an ultra-high frequency band (with the frequency of 300-3000 MHz) is a key important characteristic.
The method for extracting the characteristic parameters of the ultra-high frequency partial discharge detection atlas and the abnormality detection method of the application number 201710866621.1 are described, wherein the characteristic of the partial discharge mode is extracted by adopting a statistical method, a characteristic vector is constructed to represent the statistical distribution characteristic of the atlas, and the method is used as the classification basis of the partial discharge category and focuses on extracting the statistical characteristic based on the discharge quantity to identify the ultra-high frequency discharge type; the method and the device for calculating the time difference of the ultrahigh frequency partial discharge signal based on short-time energy of application number 202310293838.3 are used for describing the method and the device for calculating the time difference of the ultrahigh frequency partial discharge signal based on short-time energy, wherein the method and the device are used for converting a voltage waveform of the short-time energy signal into an energy accumulation curve, determining that the ultrahigh frequency partial discharge signal is larger than an inflection point generated by background noise on the energy accumulation curve, and determining the inflection point as the starting moment of the occurrence of the ultrahigh frequency partial discharge signal, wherein the emphasis is on noise reduction treatment and the identification of the inflection point, so that partial discharge is identified.
The problem of low accuracy or missing identification exists in the current adoption of a single method for ultrahigh frequency partial discharge.
Disclosure of Invention
The invention aims to: the invention aims to provide an ultrahigh frequency partial discharge identification method based on shape, energy and statistical characteristics, which improves the identification accuracy through the identification of a comprehensive method and avoids missed identification and erroneous identification.
The technical scheme is as follows: the invention comprises the following steps: based on shape, energy and statistics, ultra-high frequency partial discharge characteristic calibration: the method comprises the steps of reading ultrahigh frequency partial discharge data, processing signals, extracting ultrahigh frequency partial discharge characteristic quantity and calibrating; characteristic calculation, fault anomaly type identification and positioning are carried out on the ultrahigh frequency partial discharge: the method comprises the steps of collecting ultrahigh frequency partial discharge signals and processing the signals; extracting the characteristic quantity of ultrahigh frequency partial discharge; normal operation data and interference data logic discrimination; and (3) identifying and positioning the ultrahigh frequency partial discharge type and logically reasoning the ultrahigh frequency partial discharge characteristic quantity.
The ultrahigh frequency partial discharge feature quantity extraction comprises feature quantity extraction based on shape, feature quantity extraction based on energy and feature quantity extraction based on statistics.
The feature quantity extraction based on the shape comprises time domain feature extraction based on the shape, frequency domain feature extraction based on the shape and period extraction based on the shape.
The energy-based feature quantity extraction comprises energy-based time domain feature extraction, energy-based frequency domain feature extraction and energy-based period extraction.
The feature quantity extraction based on statistics comprises time domain feature extraction based on statistics and frequency domain feature extraction based on statistics.
The ultrahigh frequency partial discharge characteristic quantity calibration comprises a time domain characteristic calibration based on a shape, a frequency domain characteristic calibration based on a shape, a period calibration based on a shape, an energy-based time domain characteristic calibration, an energy-based frequency domain characteristic calibration, an energy-based period calibration, a time domain characteristic calibration based on statistics and a frequency domain characteristic calibration based on statistics.
The ultrahigh frequency partial discharge type identification and positioning are specifically as follows: and carrying out ultrahigh frequency partial discharge type identification on the shape-based time domain features, the shape-based frequency domain features, the shape-based periodic features, the energy-based time domain features, the energy-based frequency domain features, the energy-based periodic features and the statistics-based time domain features by adopting an error comparison algorithm, a similarity algorithm and a classification algorithm, positioning according to the sizes of the ultrahigh frequency signals acquired at different positions, and acquiring the ultrahigh frequency signals at different positions after time synchronization to carry out spatial positioning.
The ultrahigh frequency partial discharge fault abnormality judgment logic comprises a logic module based on shape characteristics, a logic module based on energy characteristics and a logic module based on statistical characteristics.
The ultrahigh frequency partial discharge fault abnormality judgment logic adopts three different logic judgment modes, namely AND logic, three-out-of-two logic and three-out-of-one logic.
The ultrahigh frequency partial discharge recognition device based on the shape, the energy and the statistical characteristics comprises an ultrahigh frequency partial discharge signal acquisition module, an ultrahigh frequency partial discharge signal denoising processing module, an ultrahigh frequency partial discharge characteristic quantity extraction module, an ultrahigh frequency partial discharge characteristic quantity discrimination algorithm module and an ultrahigh frequency partial discharge characteristic quantity logic reasoning module.
The beneficial effects are that: the method is suitable for the ultrahigh frequency partial discharge identification of the primary equipment of the power grid, the existing partial discharge site is found in the early stage of the short circuit fault, the partial discharge is prevented from further evolving into the power grid accident, and the automation level is improved; the time domain and the frequency characteristics of the ultrahigh frequency partial discharge are respectively extracted based on three different dimensions of shape, energy and statistics, and different objects have different characteristic sensitivities and are wide in application range; the accuracy of identification is improved through the discrimination of the comprehensive method, and missed discrimination and erroneous discrimination are avoided; the invention adopts three different logics of AND logic, three-out-of-two logic and three-out-of-one logic, and has high flexibility.
Drawings
FIG. 1 is a flow chart of the calibration of the ultra-high frequency partial discharge characteristics based on shape, energy and statistics;
FIG. 2 is a flow chart of the ultrahigh frequency partial discharge feature calculation, fault anomaly type identification and positioning of the invention;
FIG. 3 is a logic diagram (AND logic condition) of the ultrahigh frequency partial discharge fault exception judgment of the present invention;
FIG. 4 is a logic diagram (three-out-of-two logic condition) of the ultrahigh frequency partial discharge fault exception determination of the present invention;
Fig. 5 is a logic diagram (three-out logic condition) of fault exception determination for partial discharge at very high frequencies according to the present invention.
Detailed Description
The invention is further described below with reference to the accompanying drawings.
As shown in fig. 1 and 2, the ultrahigh frequency partial discharge identification method based on shape, energy and statistical characteristics of the invention comprises the following steps: based on the shape, the energy and the statistics, the ultrahigh frequency partial discharge characteristic calibration, the ultrahigh frequency partial discharge characteristic calculation and the fault abnormal type identification and positioning are performed.
The ultra-high frequency partial discharge characteristic calibration based on the shape, the energy and the statistics specifically comprises the following steps:
s1, reading an ultrahigh frequency partial discharge data file of a site and an experiment
And reading n groups of ultrahigh frequency partial discharge data files acquired on site and acquired through experiments, wherein the data files comprise normal data, abnormal data and interference data, and calibrating different types of data and different partial discharge types of data respectively.
S2, denoising the ultrahigh frequency partial discharge signal: white noise is filtered out by adopting a denoising algorithm.
S3, extracting and calibrating the ultrahigh frequency partial discharge characteristic quantity, extracting the ultrahigh frequency partial discharge characteristic quantity by adopting the following different characteristic extraction methods, and carrying out data calibration according to the same type of data of different objects, wherein the data calibration comprises time domain characteristic calibration based on a shape, frequency domain characteristic calibration based on a shape, periodic calibration based on a shape, time domain characteristic calibration based on energy, frequency domain characteristic calibration based on energy, periodic calibration based on energy, time domain characteristic calibration based on statistics and frequency domain characteristic calibration based on statistics. The method specifically comprises the following steps:
s31, extracting feature quantity based on the shape; the method specifically comprises the following steps:
s311, shape-based time domain feature extraction
The shape feature extraction of the time domain includes, but is not limited to, envelope, direct current bias value, peak-to-peak value, short-time average amplitude, long-time average amplitude, variance, skewness and kurtosis, wherein the calculation formula of the envelope is as follows:
the curve function is known:
y＝f(x)
The normal family of curves is represented as follows:
where (X, Y) is the varying coordinates of the normal, (X, Y) is the point on the curve, and Y is a function of X; y' is the first derivative of y and y "is the second derivative of y.
The normal family envelope satisfies the set of equations:
and solving the parameter equation of the envelope curve:
substituting y 'and y',
Obtaining parameter equation of envelope curve
S312, shape-based frequency feature extraction
The shape feature extraction of the frequency comprises, but is not limited to, the number of frequency peaks, a first spectral peak center frequency value, a first spectral peak amplitude value, a second spectral peak center frequency value, a second spectral peak amplitude value, a third spectral peak center frequency value, a third spectral peak amplitude value, a first power spectrum density, a second power spectrum density, a third power spectrum and a third power spectrum density, wherein a frequency peak threshold is set, statistics of the frequency peak threshold is the frequency peak, the center frequency value and the amplitude are obtained by adopting fast Fourier transform, and three frequency center values with the largest amplitude are recorded. The power spectrum and the power spectrum density calculation formula are respectively as follows:
Wherein, P f is the power spectrum, DFT is the fourier transform function to calculate the amplitude of the frequency f, N is the number of samples of the signal, the above is the power spectrum of each sample at a certain frequency, and the calculation result is multiplied by the number of samples of the signal in a unit time, so as to calculate the power spectrum in a unit time.
Where P d is the power spectral density, P f is the power spectrum, DFT is the Fourier transform function to calculate the magnitude of the frequency f, and N is the number of samples of the signal.The sum of the power spectra for each frequency is also equal to the/>, in the time domainI.e. accumulation of the sampled data, x (i) is the sampled data, the total power spectrum of each sample is calculated, and the calculated result is multiplied by the number of samples of the signal in unit time, so that the total power spectrum in unit time can be calculated.
S313, periodic feature extraction based on shape
The shape-based periodic calculation formula is as follows:
When the errors of T 1 and T 2 are within the allowable range (the error range can be set, the default value is 1%), T is the calculation period, and the calculation result errors of the time-domain-based shape T 0 starting time and the time-domain-based shape T 0 +t starting time peak-to-peak value, the short-time average amplitude and the long-time average amplitude are within a certain range, and the calculation result errors of the frequency-domain-based shape T 0 starting time and the frequency peak value of T 0 +t, the first spectral peak center frequency value and the first spectral peak amplitude are within the allowable range (the error range can be set, and the default value is 1%).
S32, extracting energy-based characteristic quantity; the method specifically comprises the following steps:
S321, energy-based time domain feature extraction
The time domain energy feature extraction includes, but is not limited to, long-term average energy, short-term average energy, root mean square, amplitude accumulation sum, discharge time and the like, wherein the long-term average energy is an energy average value calculated for a longer period of time, the time length can be set, the short-term average energy is an energy average value calculated for a shorter time meeting a discharge triggering condition, the discharge triggering condition is that the discharge electric quantity is greater than a certain threshold value, and the discharge time is a continuous discharge time with the energy value greater than a certain threshold value, and the calculation formula of the discharge time is as follows:
when x (t 1)x(t1)≥Setting,t1 is the discharge start time
When x (t 2)x(t2)≤Setting,t2 is the discharge end time
tdischarge＝t2-t1
Where t 1 is the discharge start time, t 2 is the discharge end time, and t discharge is the discharge time.
S322, extracting frequency characteristic quantity based on energy
The energy characteristic extraction of the frequency includes, but is not limited to, a first spectral peak energy value, a second spectral peak energy value, a third spectral peak energy value, etc., and the spectral peak energy calculation formulas are as follows:
Ef＝DFT(x)·DFT(x)
Wherein E f is a spectral peak energy value, x is sampling data, representing sampling data in a period of sampling time, DFT is a Fourier transform function to calculate the amplitude of the frequency f, for example, the sampling data x takes a long window for a long time, the calculated result is a long-time spectral peak energy value, for example, the sampling data x takes a short window for a long time, and the calculated result is a short-time spectral peak energy value.
S323, periodic characteristic quantity extraction based on energy
The energy-based periodic calculation formula is as follows:
T1＝DFT(x)·DFT(x)
T2＝DFT(x+T)·DFT(x+T)
When the errors of the first, second and third spectral peak energy values T 1 and T 2 are within the allowable range (the error range can be set, the default value is 1%), T is the calculation period, the sampling start time of the x data is T 0, and the sampling start time of the x+t data is T 0 +t.
S33, extracting feature quantity based on statistics; the method specifically comprises the following steps:
S331, time domain feature extraction based on statistics
The time domain statistical feature extraction includes, but is not limited to, initial phase angle of positive and negative half-wave discharge, positive and negative half-wave discharge duration, number of positive and negative half-wave discharge pulses, short-time zero-crossing rate, etc. The phase angle of the positive and negative half waves for starting discharging is a discharge quantity larger than a certain set value, meanwhile, the discharge duration is larger than a set threshold (the set threshold can be set), meanwhile, the discharge duration is larger than an allowable error range (the error range can be set, the default value is 1%), namely, the start time of discharging is judged, the time of the start time of discharging from the phase angle 0 degree signal is converted into a phase angle, namely, the phase angle of the discharging is the phase angle of the initial discharging, and the phase angles of the positive and negative half waves for starting discharging are calculated respectively. Wherein the calculation formula of the initial phase angle of discharge is as follows:
When x (t 1)≥Setting_dischage,t1 is the discharge start time
When x (t 2)≤Setting_dischage,t2 is the discharge end time
T discharge＝t2-t1,tdischarge is the sustain discharge time
T discharge is greater than or equal to setting_pulse, and the continuous discharge time is greater than the set setting_pulse
T 0 is the signal zero crossing time, T cycle signal period.
S332, frequency characteristic quantity extraction based on statistics
The statistical feature extraction of the frequency domain includes, but is not limited to, power spectrum band statistics, frequency spectrum band information entropy statistics, and the like, wherein a calculation formula based on the power spectrum band information entropy is as follows:
Wherein E k is the band energy value, P k is the band power spectrum, and F n is the power spectrum band information entropy.
The ultrahigh frequency partial discharge characteristic calculation, fault anomaly type identification and positioning specifically comprise the following steps:
S1, collecting an ultrahigh frequency partial discharge signal: and the ultrahigh frequency partial discharge data of the primary equipment of the power grid are collected on site.
S2, denoising the ultrahigh frequency partial discharge signal: and filtering out white noise of the ultrahigh frequency partial discharge data of the power grid primary equipment acquired on site.
S3, extracting ultrahigh frequency partial discharge characteristic quantity: as described above.
S4, normal operation data logic discrimination: and (5) calibrating and judging whether the normal operation data belong to the normal operation data according to the normal operation data of different objects.
S5, interference data logic discrimination: and calibrating and judging whether the interference data belong to the interference data according to different interference data of different objects.
S6, identifying and positioning the ultrahigh frequency partial discharge type: the method comprises the steps of performing ultrahigh frequency partial discharge type identification on submodules such as a shape-based time domain feature, a shape-based frequency domain feature, a shape-based periodic feature, an energy-based time domain feature, an energy-based frequency domain feature, an energy-based periodic feature, a statistics-based time domain feature and the like by adopting an error comparison algorithm, a similarity algorithm and a classification algorithm, positioning according to the sizes of ultrahigh frequency signals acquired at different positions, and performing space positioning according to the acquired ultrahigh frequency signals at different positions after time synchronization.
The error comparison algorithm specifically comprises the following steps: and calibrating the calculation result data, comparing the field or test acquired ultrahigh frequency partial discharge data with the calibration data, and judging the ultrahigh frequency partial discharge according to the error range of the calibration data by adopting an error algorithm.
The similar algorithm is: and (3) calibrating the calculation result data, comparing the field or test acquired ultrahigh frequency partial discharge data with the calibration data, adopting a similar algorithm, and judging the ultrahigh frequency partial discharge when the error of the calculation result of the similar algorithm is within the allowable range.
The classification algorithm is as follows: and (3) calibrating the calculation result data, comparing the field or test acquired ultrahigh frequency partial discharge data with the calibration data, and adopting a classification method including but not limited to a linear discriminant method, a distance discriminant method, a Bayesian classifier, a decision tree, a neural network ANN, a Support Vector Machine (SVM) and other classification methods, wherein the ultrahigh frequency partial discharge is judged to be the type when the classification conditions of the classification method are met.
S7, ultrahigh frequency partial discharge characteristic quantity logic reasoning
The ultrahigh frequency partial discharge fault abnormality judgment logic comprises three logic modules, namely a logic module based on shape characteristics, a logic module based on energy characteristics and a logic module based on statistical characteristics. The shape feature-based logic module comprises a shape-based time domain feature condition, a shape-based frequency feature condition and a shape feature-based periodic condition, the energy feature-based logic module comprises an energy-based time domain feature condition, an energy-based frequency feature condition and an energy feature-based periodic condition, and the statistics feature-based logic module comprises a statistics-based time domain feature condition and a statistics-based frequency feature condition.
As shown in fig. 3, the strong logic discrimination conditions are: the logic modules based on the shape characteristics, the logic modules based on the energy characteristics and the logic modules based on the statistics characteristics adopt AND logic conditions for judgment, namely, the three modules meet the conditions and judge that the partial discharge exists.
As shown in fig. 4, the partial strength logic discrimination conditions are: and judging the conditions of the three-out-of-two logic based on the shape feature logic module, the energy feature logic module and the statistical feature logic module, namely judging that the three modules have two or more modules which meet the conditions as partial discharge.
As shown in fig. 5, the weak logic discrimination conditions are: and judging whether any one of the three modules meets the condition, namely, judging that the three modules are partial discharge, namely, judging that one module, two modules or three modules meet the condition, namely, judging that the three modules are all partial discharge.
The ultrahigh frequency partial discharge type identification device based on the shape characteristics, the energy characteristics and the statistical characteristics comprises the following modules:
Module 1: the ultrahigh frequency partial discharge signal acquisition module;
the main hardware circuit for the ultra-high frequency partial discharge signal acquisition comprises an ultra-high frequency receiving antenna, a signal amplifier, a filter, an A/D analog-to-digital conversion module and the like.
The main software for collecting the ultrahigh frequency partial discharge signals processes analog-to-digital conversion to convert analog quantity into digital quantity, and the sampling rate must be high enough to ensure that the analog quantity of the ultrahigh frequency signals is not distorted.
Module 2: the ultra-high frequency partial discharge signal denoising processing module;
the noise of the ultrahigh frequency partial discharge signal is removed by adopting a software algorithm, so that white noise is removed as much as possible, and an effective high frequency partial discharge characteristic signal is left.
Module 3: the ultrahigh frequency partial discharge characteristic quantity extraction module;
3.1 shape-based time-Domain feature extraction
The shape feature extraction of the time domain includes, but is not limited to, envelope, direct current bias value, peak-to-peak value, short-time average amplitude, long-time average amplitude, variance, skewness and kurtosis, wherein the calculation formula of the envelope is as follows:
the curve function is known:
y＝f(x)
the normal family of this curve is expressed by the equation:
here, (X, Y) is the varying coordinates of the normal, (X, Y) is the point on the curve, and Y is a function of X. y' is the first derivative of y and y "is the second derivative of y.
The normal family envelope satisfies the set of equations:
and solving the parameter equation of the envelope curve:
substituting y 'and y',
Obtaining parameter equation of envelope curve
3.2 Shape-based frequency feature extraction
The shape feature extraction of the frequency comprises, but is not limited to, the number of frequency peaks, a first spectral peak center frequency value, a first spectral peak amplitude value, a second spectral peak center frequency value, a second spectral peak amplitude value, a third spectral peak center frequency value, a third spectral peak amplitude value, a first power spectrum density, a second power spectrum density, a third power spectrum and a third power spectrum density, wherein a frequency peak threshold is set, statistics of the frequency peak threshold is the frequency peak, the statistics of the frequency peak threshold is larger than the frequency peak value, the center frequency value and the amplitude are obtained by adopting fast Fourier transform, three frequency center values with the largest amplitude are recorded, and the power spectrum density calculation formulas are as follows:
Wherein, P f is the power spectrum, DFT is the fourier transform function to calculate the amplitude of the frequency f, N is the number of samples of the signal, the above is the power spectrum of each sample at a certain frequency, and the calculation result is multiplied by the number of samples of the signal in a unit time, so as to calculate the power spectrum in a unit time.
Where P d is the power spectral density, P f is the power spectrum, DFT is the Fourier transform function to calculate the magnitude of the frequency f, and N is the number of samples of the signal.
The sum of the power spectra for each frequency is also equal to the/>, in the time domainI.e. accumulation of the sampled data, x (i) is the sampled data, the total power spectrum of each sample is calculated, and the calculated result is multiplied by the number of samples of the signal in unit time, so that the total power spectrum in unit time can be calculated.
3.3 Shape-based periodic feature extraction
The shape-based periodic calculation formula is as follows:
When the errors of the T 1 and the T 2 are within a certain range, T is a calculation period, and the calculation result errors of the time-domain-based shape T 0, the time-domain-based shape T 0 +T, the short-time average amplitude and the long-time average amplitude are required to be simultaneously satisfied within a certain range, and the calculation result errors of the frequency-domain-based shape T 0, the frequency peak number of the T 0 +T, the first spectral peak center frequency value and the first spectral peak amplitude are required to be simultaneously satisfied within a certain range.
3.4 Energy-based time-Domain feature extraction
The time domain energy feature extraction includes, but is not limited to, long-term average energy, short-term average energy, root mean square, amplitude accumulation sum, discharge time and the like, wherein the long-term average energy is an energy average value calculated for a longer period of time, the time length can be set, the short-term average energy is an energy average value calculated for a shorter time meeting a discharge triggering condition, the discharge triggering condition is that the discharge electric quantity is greater than a certain threshold value, and the discharge time is a continuous discharge time with the energy value greater than a certain threshold value, and the calculation formula of the discharge time is as follows:
when x (t 1)x(t1)≥Setting,t1 is the discharge start time
When x (t 2)x(t2)≤Setting,t2 is the discharge end time
tdischarge＝t2-t1
In the above formula, t 1 is the discharge start time, t 2 is the discharge end time, and t discharge is the discharge time.
3.5 Energy-based frequency feature extraction
The energy characteristic extraction of the frequency includes, but is not limited to, a first spectral peak energy value, a second spectral peak energy value, a third spectral peak energy value, etc., and the spectral peak energy calculation formulas are as follows:
Ef＝DFT(x)·DFT(x)
Wherein E f is a spectral peak energy value, x is sampling data, representing sampling data in a period of sampling time, DFT is a Fourier transform function to calculate the amplitude of the frequency f, for example, the sampling data x takes a long window for a long time, the calculated result is a long-time spectral peak energy value, for example, the sampling data x takes a short window for a long time, and the calculated result is a short-time spectral peak energy value.
3.6 Energy-based periodic feature extraction
The energy-based periodic calculation formula is as follows:
T1＝DFT(x)·DFT(x)
T2＝DFT(x+T)·DFT(x+T)
When the errors of the first spectral peak energy value, the second spectral peak energy value and the third spectral peak energy value T 1 and T 2 are all within a certain range, T is the calculation period, the sampling start time of x data is T 0, and the sampling start time of x+T data is T 0 +T.
3.7 Statistics-based time Domain feature extraction
The time domain statistical feature extraction includes, but is not limited to, initial phase angle of positive and negative half-wave discharge, positive and negative half-wave discharge duration, number of positive and negative half-wave discharge pulses, short-time zero-crossing rate, etc. The phase angle of the positive and negative half waves for starting discharge is the discharge quantity which is larger than a certain set value, meanwhile, the discharge duration is longer than a certain set value, namely, the start time of the discharge is judged as the start time of the discharge, the time of the start time of the discharge from the phase angle 0 degree signal is converted into the phase angle, namely, the phase angle of the discharge is the initial phase angle of the discharge, and the initial phase angles of the positive and negative half waves for starting the discharge are calculated respectively. Wherein the calculation formula of the initial phase angle of discharge is as follows:
When x (t 1)≥Setting_dischage,t1 is the discharge start time
When x (t 2)≤Setting_dischage,t2 is the discharge end time
T discharge＝t2-t1,tdischarge is the sustain discharge time
T discharge is greater than or equal to setting_pulse, and the continuous discharge time is greater than the set setting_pulse
T 0 is the signal zero crossing time, T cycle signal period
3.8 Statistical-based frequency feature extraction
The statistical feature extraction of the frequency domain includes, but is not limited to, power spectrum band statistics, frequency spectrum band information entropy statistics, and the like, wherein a calculation formula based on the power spectrum band information entropy is as follows:
Wherein E k is the band energy value, P k is the band power spectrum, and F n is the power spectrum band information entropy.
Module 4: the ultrahigh frequency partial discharge characteristic quantity judgment algorithm module;
4.1 error comparison algorithm
And calibrating the calculation result data, comparing the field or test acquired ultrahigh frequency partial discharge data with the calibration data, and judging the ultrahigh frequency partial discharge according to the error range of the calibration data by adopting an error algorithm.
4.2 Similarity Algorithm
And (3) calibrating the calculation result data, comparing the field or test acquired ultrahigh frequency partial discharge data with the calibration data, adopting a similar algorithm, and judging the ultrahigh frequency partial discharge when the error of the calculation result of the similar algorithm is within the allowable range.
4.3 Classification Algorithm
And (3) calibrating the calculation result data, comparing the field or test acquired ultrahigh frequency partial discharge data with the calibration data, and adopting a classification method including but not limited to a linear discriminant method, a distance discriminant method, a Bayesian classifier, a decision tree, a neural network ANN, a Support Vector Machine (SVM) and other classification methods, wherein the ultrahigh frequency partial discharge is judged to be the type when the classification conditions of the classification method are met.
Module 5: the ultrahigh frequency partial discharge characteristic quantity logic reasoning module;
The ultrahigh frequency partial discharge fault abnormality judgment logic comprises three logic modules, namely a shape feature-based logic module, an energy feature-based logic module and a statistics feature-based logic module, wherein the shape feature-based logic module comprises a shape-based time domain feature condition, a shape-based frequency feature condition and a shape feature-based periodic condition, the energy feature-based logic module comprises an energy-based time domain feature condition, an energy-based frequency feature condition and an energy feature-based periodic condition, and the statistics feature-based logic module comprises a statistics-based time domain feature condition and a statistics-based frequency feature condition.
4.1 Strong logic discriminant Condition
The logic modules based on the shape characteristics, the logic modules based on the energy characteristics and the logic modules based on the statistics characteristics adopt AND logic conditions for judgment, namely, the three modules meet the conditions and judge that the partial discharge exists.
4.2 Logical discrimination conditions for stronger
And judging the conditions of the three-out-of-two logic based on the shape feature logic module, the energy feature logic module and the statistical feature logic module, namely judging that the three modules have two or more modules which meet the conditions as partial discharge.
4.3 Weak logic discrimination conditions
And judging whether any one of the three modules meets the condition, namely, judging that the three modules are partial discharge, namely, judging that one module, two modules or three modules meet the condition, namely, judging that the three modules are all partial discharge.
The invention also provides a computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the steps of the above method when executing the computer program.
The invention also provides a computer readable storage medium having stored thereon a computer program which when executed by a processor performs the steps of the method described above.
Claims (10)
1. The ultrahigh frequency partial discharge identification method based on the shape, the energy and the statistical characteristics is characterized by comprising the following steps of: based on shape, energy and statistics, ultra-high frequency partial discharge characteristic calibration: the method comprises the steps of reading ultrahigh frequency partial discharge data, processing signals, extracting ultrahigh frequency partial discharge characteristic quantity and calibrating;
Characteristic calculation, fault anomaly type identification and positioning are carried out on the ultrahigh frequency partial discharge: the method comprises the steps of collecting ultrahigh frequency partial discharge signals and processing the signals; extracting the characteristic quantity of ultrahigh frequency partial discharge; normal operation data and interference data logic discrimination; and (3) identifying and positioning the ultrahigh frequency partial discharge type and logically reasoning the ultrahigh frequency partial discharge characteristic quantity.
2. The method for identifying partial discharge of ultrahigh frequency based on shape, energy and statistical characteristics according to claim 1, wherein the feature extraction of partial discharge of ultrahigh frequency comprises feature extraction based on shape, feature extraction based on energy and feature extraction based on statistics.
3. The method for identifying ultrahigh frequency partial discharge based on shape, energy and statistical characteristics according to claim 2, wherein the feature quantity extraction based on shape comprises time domain feature extraction based on shape, frequency domain feature extraction based on shape and period extraction based on shape.
4. The method for identifying ultrahigh frequency partial discharge based on shape, energy and statistical characteristics according to claim 2, wherein the feature quantity extraction based on energy comprises time domain feature extraction based on energy, frequency domain feature extraction based on energy and periodic extraction based on energy.
5. The method for identifying ultrahigh frequency partial discharge based on shape, energy and statistical characteristics according to claim 2, wherein the feature quantity extraction based on statistics comprises time domain feature extraction based on statistics and frequency domain feature extraction based on statistics.
6. The method for identifying partial discharge of ultrahigh frequency based on shape, energy and statistical characteristics according to claim 1, wherein the calibration of the characteristic quantity of partial discharge of ultrahigh frequency comprises calibration of time domain based on shape, calibration of frequency domain based on shape, calibration of period based on shape, calibration of time domain based on energy, calibration of frequency domain based on energy, calibration of period based on energy, calibration of time domain based on statistics and calibration of frequency domain based on statistics.
7. The method for identifying the partial discharge of the ultrahigh frequency based on the shape, the energy and the statistical characteristics according to claim 1, wherein the identifying and positioning of the partial discharge type of the ultrahigh frequency is specifically as follows: and carrying out ultrahigh frequency partial discharge type identification on the shape-based time domain features, the shape-based frequency domain features, the shape-based periodic features, the energy-based time domain features, the energy-based frequency domain features, the energy-based periodic features and the statistics-based time domain features by adopting an error comparison algorithm, a similarity algorithm and a classification algorithm, positioning according to the sizes of the ultrahigh frequency signals acquired at different positions, and acquiring the ultrahigh frequency signals at different positions after time synchronization to carry out spatial positioning.
8. The method for identifying partial discharge of an ultrahigh frequency based on shape, energy and statistical characteristics according to claim 7, wherein the ultrahigh frequency partial discharge fault abnormality judgment logic comprises a logic module based on shape characteristics, a logic module based on energy characteristics and a logic module based on statistical characteristics.
9. The method for identifying partial discharge of ultrahigh frequency based on shape, energy and statistical characteristics according to claim 8, wherein the abnormal judgment logic of partial discharge fault of ultrahigh frequency adopts three different logic judgment of AND logic, two-out-of-three logic and one-out-of-three logic.
10. The ultrahigh frequency partial discharge recognition device based on the shape, the energy and the statistical characteristics is characterized by comprising an ultrahigh frequency partial discharge signal acquisition module, an ultrahigh frequency partial discharge signal denoising processing module, an ultrahigh frequency partial discharge characteristic quantity extraction module, an ultrahigh frequency partial discharge characteristic quantity discrimination algorithm module and an ultrahigh frequency partial discharge characteristic quantity logic reasoning module.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410212708.7A CN118152942A (en) | 2024-02-27 | 2024-02-27 | Ultrahigh frequency partial discharge identification method based on shape, energy and statistical characteristics |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410212708.7A CN118152942A (en) | 2024-02-27 | 2024-02-27 | Ultrahigh frequency partial discharge identification method based on shape, energy and statistical characteristics |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN118152942A true CN118152942A (en) | 2024-06-07 |
Family
ID=91287787
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410212708.7A Pending CN118152942A (en) | 2024-02-27 | 2024-02-27 | Ultrahigh frequency partial discharge identification method based on shape, energy and statistical characteristics |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN118152942A (en) |
-
2024
- 2024-02-27 CN CN202410212708.7A patent/CN118152942A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Pinpart et al. | Differentiating between partial discharge sources using envelope comparison of ultra-high-frequency signals | |
| CN103197218B (en) | A kind of high-voltage cable insulation defect partial discharge electrification detection diagnostic method | |
| Li et al. | A generic waveform abnormality detection method for utility equipment condition monitoring | |
| CN102841297B (en) | Self-adaptive partial-discharge monitoring method and device for sampling frequency | |
| CN109029959B (en) | Method for detecting mechanical state of transformer winding | |
| Cai et al. | Intelligent DC arc-fault detection of solar PV power generation system via optimized VMD-based signal processing and PSO–SVM classifier | |
| CN109443190B (en) | Transient traveling wave-based transformer winding deformation online monitoring method and device | |
| CN114584069B (en) | Photovoltaic direct-current arc detection method and system | |
| CN114325256A (en) | Power equipment partial discharge identification method, system, equipment and storage medium | |
| Fatama et al. | A multi feature based islanding classification technique for distributed generation systems | |
| CN108594156B (en) | Improved current transformer saturation characteristic identification method | |
| CN116643163A (en) | Remote on-line monitoring system of vacuum circuit breaker | |
| Matz et al. | Automated power quality monitoring system for on-line detection and classification of disturbances | |
| CN115616330B (en) | Power transmission line multiple lightning identification method and system based on waveform similarity | |
| CN115808600B (en) | DC arc detection method based on mathematical morphology and pattern recognition | |
| CN118152942A (en) | Ultrahigh frequency partial discharge identification method based on shape, energy and statistical characteristics | |
| CN115561575A (en) | Method for distinguishing electrical abnormal state of offshore wind farm based on multi-dimensional matrix profile | |
| CN109917245B (en) | Ultrasonic detection partial discharge signal mode identification method considering phase difference | |
| Zeng et al. | High impedance fault detection in distribution network based on S-transform and average singular entropy | |
| Behzadi et al. | Identification of combined power quality disturbances in the presence of distributed generations using variational mode decomposition and K-nearest neighbors classifier | |
| CN111537796A (en) | Ultrahigh harmonic measurement method based on fixed-frequency asynchronous sampling | |
| CN116626454B (en) | Oil paper insulation UHF partial discharge signal anti-interference identification and positioning method and device based on correction time-frequency clustering | |
| CN115616332B (en) | AC power transmission line lightning stroke interference identification method and system based on extension fusion | |
| Song et al. | DC Fault Detection method for DC grids based on frequency domain voltage characteristics | |
| CN114062833B (en) | Single-phase earth fault insulation degradation identification method based on fuzzy inference system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |