CN109212391B - Take into account the cable local discharge signal processing and localization method of DISCHARGE PULSES EXTRACTION and denoising - Google Patents
Take into account the cable local discharge signal processing and localization method of DISCHARGE PULSES EXTRACTION and denoising Download PDFInfo
- Publication number
- CN109212391B CN109212391B CN201811077524.5A CN201811077524A CN109212391B CN 109212391 B CN109212391 B CN 109212391B CN 201811077524 A CN201811077524 A CN 201811077524A CN 109212391 B CN109212391 B CN 109212391B
- Authority
- CN
- China
- Prior art keywords
- partial discharge
- signal
- matrix
- time window
- max
- 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.)
- Active
Links
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
- G01R31/1263—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 of solid or fluid materials, e.g. insulation films, bulk material; of semiconductors or LV electronic components or parts; of cable, line or wire insulation
- G01R31/1272—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 of solid or fluid materials, e.g. insulation films, bulk material; of semiconductors or LV electronic components or parts; of cable, line or wire insulation of cable, line or wire insulation, e.g. using partial discharge measurements
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Testing Relating To Insulation (AREA)
Abstract
The invention discloses a kind of signal processing of partial discharge methods and power cable partial discharge positioning method for taking into account DISCHARGE PULSES EXTRACTION and signal denoising, the signal processing of partial discharge method uses the data processing order of " first extracting; rear denoising ", first with singular value decomposition, optimal unusual threshold value is compared with maximum singular value, it realizes quick, the accurate positioning to partial discharge pulse, and partial discharge pulse is extracted, singular value reconstruct is recycled further to denoise the local pulse signal of extraction.The present invention is based on singular value decomposition-Reconstruction Method, it can be achieved that remaining to the effective inhibition for effectively inhibiting white regeneration Partial discharge signal white noise under the conditions of compared with low signal-to-noise ratio, while combining 1 layer scattering Wavelet Denoising Method, further suppresses High-frequency Interference.The accurate positioning, it can be achieved that power cable partial discharge source (i.e. shelf depreciation point) is extracted to partial discharge pulse based on above-mentioned signal processing of partial discharge method.
Description
Technical field
Power cable Partial Discharge Detecting Technology of the present invention field is related to partial discharge signal wave analysis processing technique,
It is fixed more particularly to a kind of signal processing of partial discharge method for taking into account DISCHARGE PULSES EXTRACTION and signal denoising and power cable shelf depreciation
Position method.
Background technique
Shelf depreciation (partial discharge, PD) (abbreviation partial discharge) is that power cable insulation the performance deteriorated occurs
One of form, while being also one of the main reason for power cable insulation further deteriorates and leads to failure of insulation.By to electricity
Power cable carries out Partial Discharge Detection and analysis, can diagnose for power cable insulation and provide reliable judging basis with monitoring.
Traditional Partial discharge signal processing method mainly includes two major classes: 1) feature based on phase distribution mode statistical spectrogram
Extracting method, this method count the local discharge signal (abbreviation Partial discharge signal) of a period of time interior, several power frequency period first, draw
Shelf depreciation amplitude is with the X-Y scheme of phase change, i.e. shelf depreciation phase distribution (phase resolved partial out
Discharge, PRPD) figure, in order to more intuitively reflect local discharge characteristic, then from PRPD extract electric discharge operating frequency phase φ,
Discharge capacity q and discharge time n constitutes two dimension or three-dimensional spectrum (shelf depreciation fingerprint chromatogram), and by fingerprint chromatogram, extracts and use
In the characteristic quantity of characterization shelf depreciation spectrogram, such as degree of skewness Sk, kurtosis ku, degree of asymmetry cc etc.；2) based on the spy of pulse
Extracting method is levied, this method usually utilizes partial discharge pulse to extract the characteristic quantity for characterizing local discharge signal feature, such as
Equivalent time width T, equivalent bandwidth F etc., different type Partial discharge signal can effectively be identified by extracting different characteristic quantity and
Effective assessment to power cable insulation state can be achieved.However, in above two Partial discharge signal processing method, with partial discharge arteries and veins
It purges with and is taken as to obtain the basis of characteristic quantity, but in general, partial discharge DISCHARGE PULSES EXTRACTION is built upon Partial discharge signal denoising base
On plinth, i.e., " first denoising, rear extraction ", therefore denoising result directly affects the validity of DISCHARGE PULSES EXTRACTION.
Application No. is CN201611094903.6, a kind of entitled " pulse suitable for high-frequency current Partial Discharge Detection
The Chinese invention patent of extracting method ", provide it is a kind of based on wavelet decomposition-reconstruct partial discharge denoising method, then using from
Dynamic threshold calculations extract pulse burst.Due to live Partial discharge signal complexity, only by the peak values of coefficient of wavelet decomposition at different levels with
Missing inspection, erroneous detection can inevitably occur for the method that the ratio between virtual value determines signal type.Partial discharge pulse is chosen by automatic threshold
Method is only applicable to the higher situation of signal-to-noise ratio, and when partial discharge pulse biggish there are amplitude difference, the lesser part of amplitude
Pulse is easier to be identified as noise.
Summary of the invention
Extract that validity is poor, is influenced etc. to ask by denoising result for the local pulse currently based on " first denoising, then extract "
Topic, the present invention provides a kind of signal processing of partial discharge methods for taking into account DISCHARGE PULSES EXTRACTION and signal denoising, can effectively extract
Local pulse, even if still there is stronger white noise rejection ability in low signal ratio.
Another object of the present invention is intended to provide a kind of power cable partial discharge positioning method.
The present invention is primarily based on random in short-term using the local discharge signal data processing policy of " first extracting, rear denoising "
Singular value decomposition and in short-term singular value decomposition are realized to the quick of pulse, accurate positioning and are extracted, and singular value is further utilized
Decomposition-reconstruct method is denoised, and local signal white noise is made to be effectively suppressed.
The DISCHARGE PULSES EXTRACTION and the signal processing of partial discharge method of signal denoising provided by the invention taken into account includes following step
It is rapid:
(1) Partial discharge signal is filtered, obtains Partial discharge signal sequence s (n), n=1,2,3 ..., N0, N0It is whole
A Partial discharge signal sequence length；
(2) it is calculated by the white noise standard deviation η of Partial discharge signal sequence s (n) and determines the unusual threshold value of Partial discharge signal global optimum
τ*；
It (3) is T using lengthWTime window intercept Partial discharge signal section y (n) from the partial signal sequence s (n), n=1,
2,3 ..., N, N are Partial discharge signal sequence length in time window；Matrix is constructed using the Partial discharge signal in the signal segment, and to structure
The matrix built carries out singular value decomposition, and the maximum singular value σ of Partial discharge signal in this section is calculatedmax；By σmaxWith step (2)
In the unusual threshold tau * of global optimum that is calculated compare, if σmax≥τ*, which is recorded as pulse and is searched
Rope initial time ti, i indicate extract umber of pulse, enter step (4)；If σmax<τ*, time window is along the sliding of the direction of sequence ends
One step first sets step-length, intercepts local signal section again using time window, rebuilds matrix, calculates σmax, with τ*Compare,
Circulation is until σmax≥τ*；
(4) the Partial discharge signal data in time window are rebuild into matrix, and singular value point is carried out to the matrix of building
Solution, is calculated the unusual estimation threshold value of local optimum of Partial discharge signal in this sectionBy optimal unusual estimation threshold valueWith most
Big singular value δmaxCarry out following compare:
If1. time window utilizes time window weight along the sliding second setting step-length that moves a step in the direction that sequence originates
New interception local signal section, rebuilds matrix, calculates δmax, withCompare, circulation untilThe time window is risen
Moment beginning is recorded as the starting t of partial discharge DISCHARGE PULSES EXTRACTIONs；2. time window is back to pulse search initial time ti, and by the time
The sliding third setting step-length that moves a step in the direction of windowsill sequence ends, is intercepted local signal section again using time window, rebuild
Matrix calculates δmax, withCompare, circulation untilThe time window end time is recorded as partial discharge DISCHARGE PULSES EXTRACTION
Finish time te；Extract ts-teThe Partial discharge signal data at moment are i-th of partial discharge pulse signal；
IfThe time window initial time is recorded as to the starting t of partial discharge DISCHARGE PULSES EXTRACTIONs, and by time window edge
The sliding third setting step-length that moves a step in the direction of sequence ends, is intercepted local signal section again using time window, rebuilds square
Battle array calculates δmax, withCompare, circulation untilThe time window end time is recorded as partial discharge DISCHARGE PULSES EXTRACTION
Finish time te；Extract ts-teThe Partial discharge signal data at moment are i-th of partial discharge pulse signal；
(5) singular value decomposition-reconstruct is carried out to i-th of partial discharge pulse signal that step (4) is extracted, after obtaining denoising
I-th of partial discharge pulse signal；
(6) after i-th of partial discharge pulse signal after being denoised, return step (3)-(5) are searched for and are extracted next
Non- partial discharge pulse train is all set to zero, completion office until searching original Partial discharge signal end by partial discharge pulse signal
Put DISCHARGE PULSES EXTRACTION and signal denoising.
The above-mentioned signal processing of partial discharge method for taking into account DISCHARGE PULSES EXTRACTION and signal denoising in step (1), utilizes high pass
Filter is filtered Partial discharge signal, to filter out low-frequency interference signal.The cutoff frequency range of the high-pass filter
Are as follows: 50kHz~100kHz.The Partial discharge signal handled by high pass filter filters further reaches acquisition dress through coaxial cable
It sets to obtain Partial discharge signal sequence s (n), the sample rate of the acquisition device is at least 50MS/s, to guarantee that single partial discharge pulse has
More data point is convenient for subsequent analysis.
The above-mentioned signal processing of partial discharge method for taking into account DISCHARGE PULSES EXTRACTION and signal denoising determines noise by the following method
Standard deviation and the calculating unusual threshold value of global optimum:
(21) determine the white noise standard deviation η of Partial discharge signal sequence s (n): using wavelet transformation (such as db4, db6 or
Db8 small echo etc.) wavelet decomposition is carried out to Partial discharge signal sequence s (n), by first layer wavelet details coefficient w1,j1 × j matrix is constructed,
J=N0/ 2, withFor approximate evaluation white noise standard deviation, wherein median refers to matrix element w1,j
Absolute value take median；
(22) the unusual threshold tau of Partial discharge signal global optimum is calculated as follows*,In formulaβ=L0/K0, K0=N0-L0+1；L0Value range be usually N0/20
~N0/2。
The above-mentioned signal processing of partial discharge method for taking into account DISCHARGE PULSES EXTRACTION and signal denoising is small using db4, db6 or db8
Wave etc. carries out wavelet decomposition to Partial discharge signal sequence s (n).The present invention carries out several layers of wavelet decompositions to Partial discharge signal using small echo
There is no particular/special requirement, regardless of carrying out several layers of wavelet decompositions, the present invention only takes first layer wavelet details coefficient.
The above-mentioned signal processing of partial discharge method for taking into account DISCHARGE PULSES EXTRACTION and signal denoising determines maximum odd by the following method
Different value and partial discharge pulse search POS INT point:
It (31) is T using lengthWTime window intercept Partial discharge signal segment y (n) from the partial signal sequence s (n), n=
1,2,3 ..., N, N are Partial discharge signal sequence length in time window；
(32) Hankel matrix A is constructedL×K:
In formula, K=N-L+1, L=N/3；
(33) calculating matrix Y=AX, X are the random matrix X=randn (K, 2) of K × 2, and by Y orthonormal,
Obtain the rank orthogonal basis of L × 2 W1；
(34) calculating matrix B=W1 T·A；
(35) according to B=U Σ VTBy matrix B singular value decomposition, in formula, U and V are respectively that 2 × 2 peacekeeping K × K dimension is orthogonal
Matrix；It is diagonal matrix, diagonal element is that the singular value of matrix B arranges in descending order, σ1It is as maximum odd
Different value σmax；
(36) by σmaxWith the unusual threshold tau of global optimum being calculated in step (2)*Compare, if σmax≥τ*, when by this
Between window initial time be recorded as pulse search initial time ti, enter step (4)；I indicates to extract umber of pulse；If σmax<τ*, when
Between windowsill sequence ends the sliding first setting step-length that moves a step in direction, intercept the return of local signal section again using time window and hold
Row (31)-(36) rebuilds matrix, calculates σmax, with τ*Compare, circulation is until σmax≥τ*。
The above-mentioned signal processing of partial discharge method for taking into account DISCHARGE PULSES EXTRACTION and signal denoising determines i-th by the following method
Partial discharge pulse signal:
(41) the Partial discharge signal data in time window are rebuild into Hankel matrix Z:
In formula, K=N-L+1, L=N/3；
(42) according to Z=C Σ ' DTThe Hankel matrix Z of step (41) building is subjected to singular value decomposition, Z=C Σ ' DT,
C and D is respectively L × L peacekeeping K × K dimension orthogonal matrix in formula；Σ '=diag (δ1,δ2,…,δp) (p=min (L, K)) be pair
Angular moment battle array, diagonal element are that the singular value of matrix Z arranges in descending order；Amount of orientation δ=(δ1,δ2,...,δm), m is non-in Σ '
The number and m≤p of neutral element are the order of matrix Z；δ1For maximum singular value δmax；
(43) withFor the unusual estimation threshold value of local optimum of data in data frame, ω in formula
(β)≈0.56β3-0.95β2+ 1.82 β+1.43, β=L/K, median, which refers to, takes median to the absolute value of matrix delta；
(44) by optimal unusual estimation threshold valueWith maximum singular value δmaxCompare,
If1. time window utilizes time window weight along the sliding second setting step-length that moves a step in the direction that sequence originates
New interception local signal section, which returns, executes (41)-(44), rebuilds matrix, calculates δmax, withCompare, circulation untilThe time window initial time is recorded as to the starting t of partial discharge DISCHARGE PULSES EXTRACTIONs；2. time window is back to pulse to search
Rope initial time ti, and time window is cut along the sliding third setting step-length that moves a step in the direction of sequence ends using time window again
It takes local signal section to return and executes (41)-(44), rebuild matrix, calculate δmax, withCompare, circulation untilThe time window end time is recorded as to the finish time t of partial discharge DISCHARGE PULSES EXTRACTIONe；Extract ts-teThe partial discharge at moment
Signal data is i-th of partial discharge pulse signal；
IfThe time window initial time is recorded as to the starting t of partial discharge DISCHARGE PULSES EXTRACTIONs, and by time window edge
The sliding third setting step-length that moves a step in the direction of sequence ends, is intercepted local signal section again using time window and returns to execution (41)-
(44), matrix is rebuild, δ is calculatedmax, withCompare, circulation untilThe time window end time is recorded as
The finish time t of partial discharge DISCHARGE PULSES EXTRACTIONe；Extract ts-teThe Partial discharge signal data at moment are i-th of partial discharge pulse signal.
The above-mentioned signal processing of partial discharge method for taking into account DISCHARGE PULSES EXTRACTION and signal denoising, carries out singular value by the following method
Decomposition-reconstruct:
(51) according to ts-teI-th of partial discharge pulse data signal at moment constructs Hankel matrix Z':
In formula, K'=N'-L'+1, L'=N'/3；N' is ts-teThe Partial discharge signal sequence length at moment；
(52) according to Z '=C ' Σ " D 'TThe Hankel matrix Z' progress singular value decomposition that step (51) are constructed, Z '=
C′Σ″D′T, C' and D' is respectively L' × L' peacekeeping K' × K' dimension orthogonal matrix in formula；Σ "=diag (δ '1,δ'2,…, δ'p')
(p'=min (L', K')) is diagonal matrix, and diagonal element is that the singular value of matrix Z' arranges in descending order；Amount of orientation δ '=
(δ′1,δ′2,...,δ′m′), the number and m'≤p' of nonzero element in m' Σ " are the order of matrix Z'；
(53) withFor the unusual estimation threshold value of local optimum of data in data frame, ω in formula
(β′)≈0.56β′3-0.95β′2+ 1.82 β '+1.43, β '=L'/K', median refer to matrix delta ' absolute value take middle position
Number；
(54) by singular value matrix Σ "=diag (δ '1,δ'2,…,δ'p') in be lower thanSingular value be set as zero, obtain
Singular value matrix Σ " ' after to denoising；According to Z "=C ' Σ " ' D 'TThe Hankel matrix Z " of signal after being denoised；
(55) for the matrix Z " obtained after reconstruct, to t in the way of being averageds-teThe Partial discharge signal at moment carries out
Reconstruct
I-th of partial discharge pulse signal after being denoised in turn and its position in entire original signal sequence.
The above-mentioned signal processing of partial discharge method for taking into account DISCHARGE PULSES EXTRACTION and signal denoising, Partial discharge signal extracts in order to balance
Validity and extraction efficiency, T in step (31)WTake 1~2 μ s, first setting N/4~N/2 data of step-length in step (36)
Point, step (42) second set step-length as 2~5 data points, and third sets step-length as 2~5 data points.
The above-mentioned signal processing of partial discharge method for taking into account DISCHARGE PULSES EXTRACTION and signal denoising, the partial discharge pulse signal extracted
Be primarily present following three kinds of situations: (1) the partial discharge pulse signal that white noise is effectively suppressed, (2) there are the obvious higher-orders of oscillation
The partial discharge pulse signal of noise, (3) denoising result show as the noise signal of the higher-order of oscillation.In order to further suppress partial discharge arteries and veins
It rushes the higher-order of oscillation noise in signal, after step (6), 1 layer scattering is further utilized to extracted partial discharge pulse signal
Wavelet transformation carries out threshold denoising to detail coefficients, which is the conventional means that this field has disclosed, referring to Shen
It please be number for partial discharge signal denoising method and Zhang Bo based on wavelet adaptive threshold disclosed in CN201310473326.1
Deng disclosed Wavelet Denoising Method to research (Zhang Bo, Liu Chengguo, Xu Zhong, the soughing of the wind in forest trees for improving GIS ultrasound local discharge signal discrimination
Research [J] the electrical engineering journal of Wavelet Denoising Method to raising GIS ultrasound local discharge signal discrimination, 2017,12 (11):
41-45).After singular value decomposition-discrete wavelet joint denoising, the high fdrequency component in signal extracted effectively is pressed down
System；In addition, for the noise signal for being mistaken for partial discharge pulse signal, since high fdrequency component has obtained obvious inhibition, after denoising its
Local optimum singular value threshold value when maximum singular value will be significantly less than denoising, so reject only include high-frequency noise signal
Sequence；To the partial discharge pulse containing high-frequency noise, high-frequency noise is further suppressed；It is most of for partial discharge pulse signal
Energy is retained, and the maximum singular value variation of denoising front and back less, is achieved in and the further of partial discharge pulse is extracted and gone
It makes an uproar.
Invention further provides a kind of power cable partial discharge positioning methods, at the local discharge signal
Reason method extracts local pulse signal, extract two partial discharge pulse signal peak values for being no more than setting time interval and to it is corresponding when
Carve t1,t2, foundationDetermine position of the partial discharge source on power cable, d is that partial discharge source is tested apart from power cable
The distance at end, l are power cable total length, and v is pulse propagation velocity.
Compared with prior art, the signal processing of partial discharge provided by the invention for taking into account DISCHARGE PULSES EXTRACTION and signal denoising
Method has the advantages that following very prominent and advantageous effects:
1, the present invention is to Partial discharge signal, using the data processing order of " first extracting, rear to denoise ", first with singular value point
Solution, optimal unusual threshold value is compared with maximum singular value, realizes quick, accurate positioning to partial discharge pulse, and to partial discharge
Pulse extracts, and recycles singular value reconstruct further to denoise the local pulse signal of extraction, effectively avoids low amplitude value office
Pulse signal is put directly to be filtered out.
2, the present invention is in terms of the DISCHARGE PULSES EXTRACTION to Partial discharge signal, based on random singular value decomposition and global optimum in short-term
Unusual threshold value realizes the quick positioning of partial discharge pulse signal search starting point；Simultaneously using singular value decomposition in short-term and based on the time
The method of the unusual threshold estimation of the local optimum of data carries out porch search, it can be achieved that partial discharge pulse signal originates in window
With the accurate positioning of final position, effective extraction to partial discharge pulse signal is completed.
3, the present invention passes through the singular value decomposition-of the unusual threshold estimation of local optimum in terms of the denoising to Partial discharge signal
Reconstruct, even if being still able to achieve effective inhibition of Partial discharge signal white noise under the conditions of compared with low signal-to-noise ratio.
4, the present invention also has good expansion, combines 1 layer scattering wavelet transformation, carries out threshold value to detail coefficients and goes
It makes an uproar, the High-frequency Interference in local signal after removal white noise can be further suppressed.
5, the present invention is by may be implemented electricity to the precise positioning of partial discharge pulse signal on power cable and effectively extraction
The accurate positioning in power cable partial discharge source (i.e. shelf depreciation point), to ensure that power cable operation provides reliable basis.
Detailed description of the invention
Fig. 1 is the signal processing of partial discharge method flow schematic diagram that the present invention takes into account DISCHARGE PULSES EXTRACTION and signal denoising.
Fig. 2 is the Partial discharge signal drawn in the embodiment of the present invention 1；Wherein (a) is original Partial discharge signal under power frequency, (b) is
Partial discharge signal after denoising, (c) the partial discharge pulse signal to extract.
Fig. 3 is the Partial discharge signal drawn in the embodiment of the present invention 2；Wherein (a) is original Partial discharge signal under oscillation wave, (b)
For the Partial discharge signal after denoising.
Fig. 4 is that the partial discharge source drawn in embodiment 2 positions schematic diagram.
Specific embodiment
Provide the embodiment of the present invention below with reference to attached drawing, and by embodiment to technical solution of the present invention carry out into
Clear, the complete explanation of one step.Obviously, the embodiment is only a part of the embodiments of the present invention, rather than whole realities
Apply example.Based on the content of present invention, those of ordinary skill in the art obtained institute without making creative work
There are other embodiments, belongs to the range that the present invention is protected.
Embodiment 1
The research object that the present embodiment is directed to is the cold of preset longitudinal tool marks defect (long 100mm, width 0.2mm, depth 1mm)
Contracting cable termination, cable model YJV22-8.7/15 will design defective cable and test partial discharge under 15kV power-frequency voltage
Signal, and carry out DISCHARGE PULSES EXTRACTION and denoising.
The signal processing of partial discharge method provided in this embodiment for taking into account DISCHARGE PULSES EXTRACTION and signal denoising, such as Fig. 1 institute
Show, comprising the following steps:
S1 is filtered Partial discharge signal, obtains Partial discharge signal sequence s (n), n=1,2,3 ..., N0, N0It is entire
Partial discharge signal sequence length.
The present embodiment uses High Frequency Current Sensor to obtain the Partial discharge signal on cable, the Partial discharge signal warp of acquisition first
High-pass filter is exported to acquisition device.The cutoff frequency range for the high-pass filter that the present embodiment uses are as follows: 50kHz~
100kHz.The acquisition device model Tektronix TDS7104 that the present embodiment uses, sample rate are set as 50MS/s.From adopting
The original Partial discharge signal sequence s (n) of 2ms duration is chosen in the original Partial discharge signal sequence that acquisition means obtain (such as Fig. 2 (a) institute
Show) carry out partial discharge DISCHARGE PULSES EXTRACTION and denoising, n=1,2,3 ..., N0, N0=105For entire Partial discharge signal sequence length.
S2 is calculated by the white noise standard deviation η of Partial discharge signal sequence s (n) and is determined the unusual threshold value of Partial discharge signal global optimum
τ*:
S21 determines the white noise standard deviation of Partial discharge signal sequence s (n): being played a game using small echo (morther wavelet selection ' db8 ')
Discharge signal sequence s (n) carries out wavelet decomposition, takes first layer wavelet details coefficient w1,j, and utilize first layer wavelet details coefficient
Construct 1 × j matrix, j=N0/ 2, withFor approximate evaluation white noise standard deviation, wherein median refers to
To matrix element w1,jAbsolute value take median；
S22 calculates the unusual threshold tau of Partial discharge signal global optimum as follows*,In formulaβ=L0/K0, K0=N0-L0+1；L0=N0/ 10, N0=105。
S3 obtains maximum singular value using random singular value decomposition, determines partial discharge pulse search POS INT point:
S31 is T using lengthWThe time window of (1.5 μ s) intercepts Partial discharge signal segment y from partial signal sequence s (n)
(n), n=1,2,3 ..., N, N=76 are Partial discharge signal sequence length in time window；
S32 constructs Hankel matrix AL×K:
In formula, L=25, K=N-L+1=52；
S33 calculating matrix Y=AX, X are the rank random matrix of K × 2 X=constructed using randn function in matlab
Randn (52,2), and matrix Y orthonormal is obtained into the rank orthogonal basis of L × 2 W using orth function in matlab1；
S34 calculating matrix B=W1 T·A；
S35 is according to B=U Σ VTBy matrix B singular value decomposition, in formula, U and V are respectively that 2 × 2 peacekeepings 52 × 52 dimension is orthogonal
Matrix；It is diagonal matrix, diagonal element is that the singular value of matrix B arranges in descending order, σ1It is as maximum odd
Different value σmax；
S36 is by σmaxWith the unusual threshold tau of global optimum being calculated in step S2*Compare, if σmax≥τ*, by the time
Window initial time is recorded as pulse search initial time ti, enter step S4；I indicates to extract umber of pulse, every to have extracted an arteries and veins
Signal is rushed, umber of pulse is extracted and is increased by 1, carry out the extraction of next pulse signal；If σmax<τ*, time window is along sequence ends
Direction cunning move a step the first setting step-length (38 data points), intercepted again using time window local signal section return execute
S31-S36 rebuilds matrix, calculates σmax, compared with τ *, circulation is until σmax≥τ*。
S4 executes side search, obtains i-th of partial discharge pulse signal:
Partial discharge signal data in time window are constructed Hankel matrix Z by S41:
In formula, L=25, K=N-L+1=52；
S42 is according to Z=C Σ ' DTThe Hankel matrix Z that step S41 is constructed carries out singular value decomposition, Z=C Σ ' DT, formula
Middle C and D is respectively 25 × 25 peacekeepings 52 × 52 dimension orthogonal matrix；Σ '=diag (δ1,δ2,…, δp) (p=25) be to angular moment
Battle array, diagonal element are that the singular value of matrix Z arranges in descending order；Amount of orientation δ=(δ1,δ2,...,δm), m is non-zero entry in Σ '
The number and m≤p of element are the order of matrix Z；δ1For maximum singular value δmax；
S43 withFor the unusual estimation threshold value of local optimum of data in data frame, ω in formula
(β)≈0.56β3-0.95β2+ 1.82 β+1.43, β=L/K, median, which refers to, takes median to the absolute value of matrix delta；
S44 is by optimal unusual estimation threshold valueWith maximum singular value δmaxCompare:
IfNeed to complete following two step: 1. time window is along sliding second setting that moves a step in the direction that sequence originates
Step-length (3 data points) is intercepted local signal section again using time window and returns to execution S41-S44, rebuild matrix, count
Calculate δmax, withCompare, circulation untilThe time window initial time is recorded as to the starting t of partial discharge DISCHARGE PULSES EXTRACTIONs；
2. time window is back to pulse search initial time ti, and time window is set along the sliding third that moves a step in the direction of sequence ends
Fixed step size (3 data points) is intercepted local signal section again using time window and returns to execution S41-S44, rebuilds matrix,
Calculate δmax, withCompare, circulation untilThe time window end time is recorded as to the end of partial discharge DISCHARGE PULSES EXTRACTION
Moment te；Extract ts-teThe Partial discharge signal data at moment are i-th partial discharge pulse signal；
IfTime window default is searched for since the initial position of Partial discharge signal sequence, which is originated
Moment is recorded as the starting t of partial discharge DISCHARGE PULSES EXTRACTIONs, and by time window along the sliding third setting step that moves a step in the direction of sequence ends
Long (3 data points) is intercepted local signal section again using time window and returns to execution S41-S44, rebuild matrix, calculate
δmax, withCompare, circulation untilThe time window end time is recorded as to the finish time of partial discharge DISCHARGE PULSES EXTRACTION
te；Extract ts-teThe Partial discharge signal data at moment are i-th of partial discharge pulse signal.
S5 carries out singular value decomposition-reconstruct to i-th of partial discharge pulse signal that step S4 is extracted, after being denoised
I-th of partial discharge pulse signal:
S51 is according to ts-teI-th of partial discharge pulse data signal at moment constructs Hankel matrix Z':
In formula, K'=N'-L'+1, L'=N'/3；N' is ts-teThe Partial discharge signal sequence length at moment；
S52 is according to Z '=C ' Σ " D 'TThe Hankel matrix Z' that step S51 is constructed carries out singular value decomposition, Z '=C '
Σ″D′T, C' and D' is respectively L' × L' peacekeeping K' × K' dimension orthogonal matrix in formula；Σ "=diag (δ '1,δ'2,…, δ'p')
(p'=min (L', K')) is diagonal matrix, and diagonal element is that the singular value of matrix Z' arranges in descending order；Amount of orientation δ '=
(δ′1,δ′2,...,δ′m′), the number and m'≤p' of nonzero element in m' Σ " are the order of matrix Z'；
S53 withFor the unusual estimation threshold value of local optimum of data in data frame, ω in formula
(β′)≈0.56β′3-0.95β′2+ 1.82 β '+1.43, β '=L'/K', median refer to matrix delta ' absolute value take middle position
Number；
S54 is by singular value matrix Σ "=diag (δ '1,δ'2,…,δ'p') in be lower thanSingular value be set as zero, obtain
Singular value matrix Σ " ' after denoising；The Hankel matrix Z " of signal after being denoised according to Z "=C ' Σ " ' D ' T；
S55 is for obtained matrix Z " after reconstruct, to t in the way of being averageds-teThe Partial discharge signal at moment carries out weight
Structure
I-th of partial discharge pulse signal after being denoised in turn and its position in entire original signal sequence are such as schemed
Shown in 2 (c).
S6 denoised after i-th of partial discharge pulse signal after return step S3-S5 search for and extract next partial discharge
Pulse (i.e. i+1 partial discharge pulse signal), until original Partial discharge signal end is searched, non-partial discharge pulse train is whole
It is set as zero, completes partial discharge DISCHARGE PULSES EXTRACTION and signal denoising.
S7 carries out a layer scattering Wavelet Denoising Method to the partial discharge pulse signal of said extracted, filters out High-frequency Interference, finally obtains
Partial discharge signal such as Fig. 2 (b) shown in, it can be seen from the figure that the white noise in original Partial discharge signal has obtained effective inhibition,
Low amplitude value partial discharge pulse signal also accurately is extracted out simultaneously.
Embodiment 2
The research object that the present embodiment is directed to is length 498m 10kV XLPE cable (away from test lead 249m transition joint
Locate default semi-conductive layer and overlap bad defect).Defective cable will be designed and test Partial discharge signal under oscillation wave, and carried out
Defect location is realized in DISCHARGE PULSES EXTRACTION and denoising.
The signal processing of partial discharge method provided in this embodiment for taking into account DISCHARGE PULSES EXTRACTION and signal denoising, such as Fig. 1 institute
Show, comprising the following steps:
S1 is filtered Partial discharge signal, obtains Partial discharge signal sequence s (n), n=1,2,3 ..., N0, N0It is entire
Partial discharge signal sequence length.
The present embodiment is first under oscillation wave condition (ceiling voltage is no more than 20kV, frequency 67Hz), using high-frequency electrical
Flow sensor obtains the Partial discharge signal on cable, and the Partial discharge signal of acquisition is exported through high-pass filter to acquisition device.This implementation
The cutoff frequency range for the high-pass filter that example uses are as follows: 50kHz~100kHz.The acquisition device model that the present embodiment uses
For Tektronix TDS7104, sample rate 50MS/s.6ms is chosen from the original Partial discharge signal sequence that acquisition device obtains
Original Partial discharge signal sequence s (n) (shown in such as Fig. 3 (a)) the progress partial discharge DISCHARGE PULSES EXTRACTION of duration and denoising, n=1,2,
3,…,N0, N0=3 × 105For entire Partial discharge signal sequence length.
S2 is calculated by the white noise standard deviation η of Partial discharge signal sequence s (n) and is determined the unusual threshold value of Partial discharge signal global optimum
τ*:
S21 determines the white noise standard deviation of Partial discharge signal sequence s (n): being played a game using small echo (morther wavelet selection ' db8 ')
Discharge signal sequence s (n) carries out wavelet decomposition, takes first layer wavelet details coefficient w1,j, and utilize first layer wavelet details coefficient
Construct 1 × j matrix, j=N0/ 2, withFor approximate evaluation white noise standard deviation, wherein median refers to
To matrix element w1,jAbsolute value take median；
S22 calculates the unusual threshold tau of Partial discharge signal global optimum as follows*,In formulaβ=L0/K0, K0=N0-L0+1；L0=N0/ 10, N0=3 × 105。
S3 obtains maximum singular value using random singular value decomposition, determines partial discharge pulse search POS INT point:
S31 is T using lengthWThe time window of (1.5 μ s) intercepts Partial discharge signal segment y from partial signal sequence s (n)
(n), n=1,2,3 ..., N, N=76 are Partial discharge signal sequence length in time window；
S32 constructs Hankel matrix AL×K:
In formula, L=25, K=N-L+1=52；
S33 calculating matrix Y=AX, X are the rank random matrix of K × 2 X=constructed using randn function in matlab
Randn (52,2), and matrix Y orthonormal is obtained into the rank orthogonal basis of L × 2 W using orth function in matlab1；
S34 calculating matrix B=W1 T·A；
S35 is according to B=U Σ VTBy matrix B singular value decomposition, in formula, U and V are respectively that 2 × 2 peacekeepings 52 × 52 dimension is orthogonal
Matrix；It is diagonal matrix, diagonal element is that the singular value of matrix B arranges in descending order, σ1It is as maximum odd
Different value σmax；
S36 is by σmaxWith the unusual threshold tau of global optimum being calculated in step S2*Compare, if σmax≥τ*, by the time
Window initial time is recorded as pulse search initial time ti, enter step S4；I indicates to extract umber of pulse, every to have extracted an arteries and veins
Signal is rushed, umber of pulse is extracted and is increased by 1, carry out the extraction of next pulse signal；If σmax<τ*, time window is along sequence ends
Direction cunning move a step the first setting step-length (38 data points), intercepted again using time window local signal section return execute
S31-S36 rebuilds matrix, calculates σmax, with τ*Compare, circulation is until σmax≥τ*。
S4 executes side search, obtains i-th of partial discharge pulse signal:
Partial discharge signal data in time window are constructed Hankel matrix Z by S41:
In formula, L=25, K=N-L+1=52；
S42 is according to Z=C Σ ' DTThe Hankel matrix Z that step S41 is constructed carries out singular value decomposition, Z=C Σ ' DT, formula
Middle C and D is respectively 25 × 25 peacekeepings 52 × 52 dimension orthogonal matrix；Σ '=diag (δ1,δ2,…, δp) (p=25) be to angular moment
Battle array, diagonal element are that the singular value of matrix Z arranges in descending order；Amount of orientation δ=(δ1,δ2,...,δm), m is non-zero entry in Σ '
The number and m≤p of element are the order of matrix Z；δ1For maximum singular value δmax；
S43 withFor the unusual estimation threshold value of local optimum of data in data frame, ω in formula
(β)≈0.56β3-0.95β2+ 1.82 β+1.43, β=L/K, median, which refers to, takes median to the absolute value of matrix delta；
S44 is by optimal unusual estimation threshold valueWith maximum singular value δmaxCompare:
IfNeed to complete following two step: 1. time window is along sliding second setting that moves a step in the direction that sequence originates
Step-length (3 data points) is intercepted local signal section again using time window and returns to execution S41-S44, rebuild matrix, count
Calculate δmax, withCompare, circulation untilThe time window initial time is recorded as to the starting t of partial discharge DISCHARGE PULSES EXTRACTIONs；
2. time window is back to pulse search initial time ti, and time window is set along the sliding third that moves a step in the direction of sequence ends
Fixed step size (3 data points) is intercepted local signal section again using time window and returns to execution S41-S44, rebuilds matrix,
Calculate δmax, withCompare, circulation untilThe time window end time is recorded as to the end of partial discharge DISCHARGE PULSES EXTRACTION
Moment te；Extract ts-teThe Partial discharge signal data at moment are i-th partial discharge pulse signal；
IfTime window default is searched for since the initial position of Partial discharge signal sequence, which is risen
Moment beginning is recorded as the starting t of partial discharge DISCHARGE PULSES EXTRACTIONs, and by time window along the sliding third setting that moves a step in the direction of sequence ends
Step-length (3 data points) is intercepted local signal section again using time window and returns to execution S41-S44, rebuild matrix, count
Calculate δmax, withCompare, circulation untilAt the end of the time window end time is recorded as partial discharge DISCHARGE PULSES EXTRACTION
Carve te；Extract ts-teThe Partial discharge signal data at moment are i-th of partial discharge pulse signal.
S5 carries out singular value decomposition-reconstruct to i-th of partial discharge pulse signal that step S4 is extracted, after being denoised
I-th of partial discharge pulse signal:
S51 is according to ts-teI-th of partial discharge pulse data signal at moment constructs Hankel matrix Z':
In formula, K'=N'-L'+1, L'=N'/3；N' is ts-teThe Partial discharge signal sequence length at moment；
S52 is according to Z '=C ' Σ " D 'TThe Hankel matrix Z' that step S51 is constructed carries out singular value decomposition, Z '=C '
Σ″D′T, C' and D' is respectively L' × L' peacekeeping K' × K' dimension orthogonal matrix in formula；Σ "=diag (δ '1,δ'2,…, δ'p')
(p'=min (L', K')) is diagonal matrix, and diagonal element is that the singular value of matrix Z' arranges in descending order；Amount of orientation δ '=
(δ′1,δ′2,...,δ′m′), the number and m'≤p' of nonzero element in m' Σ " are the order of matrix Z'；
S53 withFor the unusual estimation threshold value of local optimum of data in data frame, ω in formula
(β′)≈0.56β′3-0.95β′2+ 1.82 β '+1.43, β '=L'/K', median refer to matrix delta ' absolute value take middle position
Number；
S54 is by singular value matrix Σ "=diag (δ '1,δ'2,…,δ'p') in be lower thanSingular value be set as zero, obtain
Singular value matrix Σ " ' after denoising；According to Z "=C ' Σ " ' D 'TThe Hankel matrix Z " of signal after being denoised；
S55 is for obtained matrix Z " after reconstruct, to t in the way of being averageds-teThe Partial discharge signal at moment carries out weight
Structure
I-th of partial discharge pulse signal after being denoised in turn and its position in entire original signal sequence.
S6 denoised after i-th of partial discharge pulse signal after return step S3-S5 search for and extract next partial discharge
Pulse (i.e. i+1 partial discharge pulse signal), until original Partial discharge signal end is searched, non-partial discharge pulse train is whole
It is set as zero, completes partial discharge DISCHARGE PULSES EXTRACTION and signal denoising.
S7 carries out a layer scattering Wavelet Denoising Method to the pulse signal of putting of said extracted, filters out High-frequency Interference, finally obtained
Shown in Partial discharge signal such as Fig. 3 (b), extracted partial discharge pulse signal enlarged drawing is corresponded to for dashed box in Partial discharge signal figure below.
It can be seen from the figure that the white noise in original Partial discharge signal has obtained effective inhibition, while low amplitude value partial discharge pulse signal
Accurately it is extracted out.
To the Partial discharge signal after denoising obtained in step S7, extraction time interval is no more than two partial discharge pulses of 6 μ s
Signal peak and corresponding moment t1,t2.By calibration, pulse propagation velocity is v=163.23m/ μ s, is calculate by the following formula partial discharge
Source is away from test lead distanceL is total cable length.Repeatedly being calculated can obtain after being averaged, partial discharge source be located at away from
From (as shown in Figure 4) at test lead 247.17m, it coincide substantially with initial preset defective locations (away from test lead 249m).Therefore,
The signal processing of partial discharge method provided through the invention can be accurately positioned and effectively extract partial discharge pulse signal, in turn
The accurate position for obtaining partial discharge source.
Claims (10)
1. a kind of signal processing of partial discharge method for taking into account DISCHARGE PULSES EXTRACTION and signal denoising, it is characterised in that including following step
It is rapid:
(1) Partial discharge signal is filtered, obtains Partial discharge signal sequence s (n), n=1,2,3 ..., N0, N0For entire partial discharge
Signal sequence length；
(2) it is calculated by the white noise standard deviation η of Partial discharge signal sequence s (n) and determines the unusual threshold tau of Partial discharge signal global optimum*；
It (3) is T using lengthWTime window intercept Partial discharge signal section y (n) from the partial signal sequence s (n), n=1,2,3 ...,
N, N are Partial discharge signal sequence length in time window；Matrix is constructed using the Partial discharge signal in the signal segment, and to the matrix of building
Singular value decomposition is carried out, the maximum singular value σ of Partial discharge signal in this section is calculatedmax；By σmaxIt is calculated with step (2)
The unusual threshold tau of global optimum*Compare, if σmax≥τ*, which is recorded as pulse search initial time ti,
I indicates extraction umber of pulse, enters step (4)；If σmax<τ*, time window is along sliding the first setting step that moves a step in the direction of sequence ends
It is long, it intercepts local signal section again using time window, rebuilds matrix, calculate σmax, with τ*Compare, circulation is until σmax≥τ*；
(4) the Partial discharge signal data in time window are rebuild into matrix, and singular value decomposition is carried out to the matrix of building, calculated
Obtain the unusual estimation threshold value of local optimum of Partial discharge signal in this sectionBy the unusual estimation threshold value of local optimumIt is unusual with maximum
Value δmaxCarry out following compare:
If1. time window is intercepted along the sliding second setting step-length that moves a step in the direction that sequence originates using time window again
Local signal section rebuilds matrix, calculates δmax, withCompare, circulation untilThe time window initial time is remembered
Record is the starting t of partial discharge DISCHARGE PULSES EXTRACTIONs；2. time window is back to pulse search initial time ti, and by time window along sequence end
The sliding third setting step-length that moves a step in direction only, is intercepted local signal section again using time window, rebuilds matrix, calculated
δmax, withCompare, circulation untilThe time window end time is recorded as to the finish time of partial discharge DISCHARGE PULSES EXTRACTION
te；Extract ts-teThe Partial discharge signal data at moment are i-th of partial discharge pulse signal；
IfThe time window initial time is recorded as to the starting t of partial discharge DISCHARGE PULSES EXTRACTIONs, and by time window along sequence end
The sliding third setting step-length that moves a step in direction only, is intercepted local signal section again using time window, rebuilds matrix, calculated
δmax, withCompare, circulation untilThe time window end time is recorded as to the finish time of partial discharge DISCHARGE PULSES EXTRACTION
te；Extract ts-teThe Partial discharge signal data at moment are i-th of partial discharge pulse signal；
(5) singular value decomposition and reconstruct are carried out to i-th of partial discharge pulse signal that step (4) is extracted, i-th after being denoised
A partial discharge pulse signal；
(6) after i-th of partial discharge pulse signal after being denoised, return step (3)-(5) are searched for and extract next partial discharge
Non- partial discharge pulse train is all set to zero, completes partial discharge pulse by pulse signal until searching original Partial discharge signal end
Extraction and signal denoising.
2. taking into account the signal processing of partial discharge method of DISCHARGE PULSES EXTRACTION and signal denoising according to claim 1, feature exists
In, by the following method determine noise criteria difference and calculate the unusual threshold value of global optimum:
(21) determine the white noise standard deviation η of Partial discharge signal sequence s (n): using wavelet transformation to Partial discharge signal sequence s (n) into
Row wavelet decomposition, by first layer wavelet details coefficient w1,jConstruct 1 × j matrix, j=N0/ 2, withIt is close
Like estimation white noise standard deviation, wherein median refers to matrix element w1,jAbsolute value take median；
(22) the unusual threshold tau of Partial discharge signal global optimum is calculated as follows*,In formulaβ=L0/K0, K0=N0-L0+1；L0Value range be usually N0/20
~N0/2。
3. taking into account the signal processing of partial discharge method of DISCHARGE PULSES EXTRACTION and signal denoising according to claim 1, feature exists
In determining maximum singular value and partial discharge pulse search POS INT point by the following method:
It (31) is T using lengthWTime window intercept Partial discharge signal segment y (n) from the partial signal sequence s (n), n=1,2,
3 ..., N, N are Partial discharge signal sequence length in time window；
(32) Hankel matrix A is constructedL×K:
In formula, K=N-L+1, L=N/3；
(33) calculating matrix Y=AX, X are the random matrix X=randn (K, 2) of K × 2, and by Y orthonormal, obtain L
× 2 rank orthogonal basis W1；
(34) calculating matrix B=W1 T·A；
(35) according to B=U Σ VTBy matrix B singular value decomposition, in formula, U and V are respectively 2 × 2 peacekeeping K × K dimension orthogonal matrix；It is diagonal matrix, diagonal element б1、б2It is arranged in descending order for the singular value of matrix B, wherein σ1It is maximum odd
Different value σmax；
(36) by σmaxWith the unusual threshold tau of global optimum being calculated in step (2)*Compare, if σmax≥τ*, by the time window
Initial time is recorded as pulse search initial time ti, enter step (4)；I indicates to extract umber of pulse；If σmax<τ*, time window edge
The sliding first setting step-length that moves a step in the direction of sequence ends, is intercepted local signal section again using time window and returns to execution (31)-
(36), matrix is rebuild, σ is calculatedmax, with τ*Compare, circulation is until σmax≥τ*。
4. taking into account the signal processing of partial discharge method of DISCHARGE PULSES EXTRACTION and signal denoising according to claim 1, feature exists
In, by the following method determine i-th of partial discharge pulse signal:
(41) the Partial discharge signal data in time window are rebuild into Hankel matrix Z:
In formula, K=N-L+1, L=N/3；
(42) according to Z=C Σ ' DTThe Hankel matrix Z of step (41) building is subjected to singular value decomposition, Z=C Σ ' DT, C in formula
It is respectively L × L peacekeeping K × K dimension orthogonal matrix with D；Σ '=diag (δ1,δ2,…,δp) (p=min (L, K)) be diagonal matrix,
Its diagonal element is that the singular value of matrix Z arranges in descending order；Amount of orientation δ=(δ1,δ2,...,δm), m is nonzero element in Σ '
Number and m≤p are the order of matrix Z；δ1For maximum singular value δmax；
(43) withFor the unusual estimation threshold value of local optimum of data in data frame, ω (β) ≈ in formula
0.56β3-0.95β2+ 1.82 β+1.43, β=L/K, median, which refers to, takes median to the absolute value of matrix delta；
(44) by optimal unusual estimation threshold valueWith maximum singular value δmaxCompare,
If1. time window is intercepted along the sliding second setting step-length that moves a step in the direction that sequence originates using time window again
Local signal section, which returns, executes (41)-(44), rebuilds matrix, calculates δmax, withCompare, circulation untilIt will
The time window initial time is recorded as the starting t of partial discharge DISCHARGE PULSES EXTRACTIONs；2. time window is back to pulse search initial time ti,
And time window is intercepted local signal section again using time window and returned along the sliding third setting step-length that moves a step in the direction of sequence ends
Receipt row (41)-(44) rebuild matrix, calculate δmax, withCompare, circulation untilThe time window is terminated
Moment is recorded as the finish time t of partial discharge DISCHARGE PULSES EXTRACTIONe；Extract ts-teThe Partial discharge signal data at moment are i-th of partial discharge arteries and veins
Rush signal；
IfThe time window initial time is recorded as to the starting t of partial discharge DISCHARGE PULSES EXTRACTIONs, and by time window along sequence end
The sliding third setting step-length that moves a step in direction only, is intercepted local signal section again using time window and returns to execution (41)-(44), weight
New building matrix, calculates δmax, withCompare, circulation untilThe time window end time is recorded as partial discharge pulse
The finish time t of extractione；Extract ts-teThe Partial discharge signal data at moment are i-th of partial discharge pulse signal.
5. taking into account the signal processing of partial discharge method of DISCHARGE PULSES EXTRACTION and signal denoising according to claim 1, feature exists
In progress singular value decomposition and reconstruct by the following method:
(51) according to ts-teI-th of partial discharge pulse data signal at moment constructs Hankel matrix Z':
In formula, K'=N'-L'+1, L'=N'/3；N' is ts-teThe Partial discharge signal sequence length at moment；
(52) according to Z '=C ' Σ " D 'TThe Hankel matrix Z' of step (51) building is subjected to singular value decomposition, Z '=C ' Σ " D
′T, C' and D' is respectively L' × L' peacekeeping K' × K' dimension orthogonal matrix in formula；Σ "=diag (δ '1,δ'2,…,δ'p') (p'=
Min (L', K')) it is diagonal matrix, diagonal element is that the singular value of matrix Z' arranges in descending order；Amount of orientation δ '=(δ '1,δ
′2,...,δ′m′), the number and m'≤p' of nonzero element in m' Σ " are the order of matrix Z'；
(53) withFor the unusual estimation threshold value of local optimum of data in data frame, ω (β ') ≈ in formula
0.56β′3-0.95β′2+ 1.82 β '+1.43, β '=L'/K', median refer to matrix delta ' absolute value take median；
(54) by singular value matrix Σ "=diag (δ '1,δ'2,…,δ'p') in be lower thanSingular value be set as zero, denoised
Singular value matrix Σ " ' afterwards；According to Z "=C ' Σ " ' D 'TThe Hankel matrix Z " of signal after being denoised；
(55) for the matrix Z " obtained after reconstruct, to t in the way of being averageds-teThe Partial discharge signal at moment is reconstructed
I-th of partial discharge pulse signal after being denoised in turn and its position in entire original signal sequence.
6. taking into account the signal processing of partial discharge method of DISCHARGE PULSES EXTRACTION and signal denoising according to claim 1, feature exists
In step (1), Partial discharge signal is filtered using high-pass filter, the Partial discharge signal after filtering processing is through coaxial electrical
Cable reaches acquisition device and obtains Partial discharge signal sequence s (n), the cutoff frequency range of the high-pass filter be 50kHz~
100kHz；The sample rate of the acquisition device is at least 50MS/s.
7. taking into account the signal processing of partial discharge method of DISCHARGE PULSES EXTRACTION and signal denoising according to claim 1, feature exists
In using db4, db6 or db8 small echo to Partial discharge signal sequence s (n) progress wavelet decomposition.
8. believing according to claim 1 to the shelf depreciation for taking into account DISCHARGE PULSES EXTRACTION and signal denoising described in 7 any one claims
Number processing method, it is characterised in that the TWTake 1~2 μ s, described first setting N/4~N/2 data point of step-length, the second setting
Step-length is 2~5 data points, and third sets step-length as 2~5 data points.
9. taking into account the signal processing of partial discharge method of DISCHARGE PULSES EXTRACTION and signal denoising according to claim 8, feature exists
In further utilizing 1 layer scattering Wavelet Denoising Method to extracted partial discharge pulse signal after step (6).
10. a kind of power cable partial discharge positioning method, it is characterised in that use claim 1 to 9 any claim institute
It states signal processing of partial discharge method and extracts local pulse signal, extract the two partial discharge pulses letter for being no more than setting time interval
Number peak value and corresponding moment t1,t2, foundationDetermine position of the partial discharge source on power cable, d be partial discharge source away from
With a distance from power cable test lead, l is power cable total length, and v is pulse propagation velocity.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811077524.5A CN109212391B (en) | 2018-09-15 | 2018-09-15 | Take into account the cable local discharge signal processing and localization method of DISCHARGE PULSES EXTRACTION and denoising |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811077524.5A CN109212391B (en) | 2018-09-15 | 2018-09-15 | Take into account the cable local discharge signal processing and localization method of DISCHARGE PULSES EXTRACTION and denoising |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109212391A CN109212391A (en) | 2019-01-15 |
| CN109212391B true CN109212391B (en) | 2019-11-08 |
Family
ID=64983898
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811077524.5A Active CN109212391B (en) | 2018-09-15 | 2018-09-15 | Take into account the cable local discharge signal processing and localization method of DISCHARGE PULSES EXTRACTION and denoising |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109212391B (en) |
Families Citing this family (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109991520A (en) * | 2019-03-19 | 2019-07-09 | 中国电力科学研究院有限公司 | A kind of cable oscillation wave partial discharge detecting system velocity of wave New calculating method |
| CN111308285B (en) * | 2020-03-03 | 2021-04-13 | 西南交通大学 | Narrow-band interference noise reduction method |
| CN111723677B (en) * | 2020-05-27 | 2022-11-04 | 中国电建集团华东勘测设计研究院有限公司 | Wavelet denoising method based on self-adaptive threshold |
| CN112433135B (en) * | 2020-12-09 | 2021-11-30 | 四川大学 | Partial discharge signal positioning method based on Manton kernel integral least square |
| CN113221615A (en) * | 2020-12-31 | 2021-08-06 | 中国石油化工股份有限公司 | Partial discharge pulse extraction method based on noise reduction clustering |
| CN113283297A (en) * | 2021-04-20 | 2021-08-20 | 上海电力大学 | Method for estimating partial discharge signal time based on singular value decomposition and improved wavelet packet |
| CN113433437B (en) * | 2021-06-28 | 2022-09-06 | 华北电力大学 | Method and system for extracting discharge current pulse |
| CN114325256A (en) * | 2021-11-25 | 2022-04-12 | 中国电力科学研究院有限公司 | Power equipment partial discharge identification method, system, equipment and storage medium |
| CN114252738B (en) * | 2021-12-24 | 2022-10-18 | 华中科技大学 | Cable joint typical defect partial discharge characterization feature mining and pattern recognition method |
| CN118050800B (en) * | 2024-04-15 | 2024-06-25 | 西南交通大学 | Pulse signal processing method and system based on baseline correction |
Family Cites Families (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR101570292B1 (en) * | 2014-05-23 | 2015-11-19 | 성균관대학교산학협력단 | Apparatus and method for detecting series arc fault in low voltage dc power distribution system |
| CN106019090B (en) * | 2016-05-11 | 2018-10-02 | 西安西热节能技术有限公司 | Shelf depreciation electromagnetic wave signal energy feature extraction method |
| CN107037327A (en) * | 2016-10-09 | 2017-08-11 | 中国电力科学研究院 | Partial discharges fault judges feature extracting method and decision method |
| CN108446632B (en) * | 2018-03-20 | 2022-05-03 | 珠海华网科技有限责任公司 | Partial discharge pulse edge searching and partial discharge confirming method |
-
2018
- 2018-09-15 CN CN201811077524.5A patent/CN109212391B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN109212391A (en) | 2019-01-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109212391B (en) | Take into account the cable local discharge signal processing and localization method of DISCHARGE PULSES EXTRACTION and denoising | |
| CN102680860B (en) | Automatic fault-point locating method for traveling-wave based fault location of high-voltage electric power lines | |
| CN105548832B (en) | High voltage power cable fault recognition method | |
| CN109557429B (en) | GIS partial discharge fault detection method based on improved wavelet threshold denoising | |
| CN103257306B (en) | Method for diagnosing direct current partial discharging insulation state of converter transformer and measurement system | |
| CN106771905B (en) | A kind of DISCHARGE PULSES EXTRACTION method suitable for high-frequency current Partial Discharge Detection | |
| CN106771922B (en) | A kind of high-tension electricity system of detecting partial discharge in equipment and Recognition of Partial Discharge | |
| CN104730424B (en) | Cable local discharge localization method based on auto-correlation-Wavelet Modulus Maxima analysis | |
| WO2009003361A1 (en) | An online partial discharge monitoring system of generator and method thereof | |
| CN103532901B (en) | A kind of time synchronization method and system for power line carrier communication | |
| CN114019309B (en) | Cable defect positioning method based on frequency domain reflection technology | |
| CN109799434A (en) | PD Pattern Recognition system and method | |
| CN101710166B (en) | Method for monitoring partial discharge of power cable connector on line | |
| CN107390097A (en) | A kind of acoustoelectric combined shelf depreciation simulation detection system of GIS and its detection method | |
| CN102323518A (en) | Method for identifying local discharge signal based on spectral kurtosis | |
| CN105021957A (en) | Power cable accessory fault identification method and system | |
| CN113514735A (en) | Anti-interference high-voltage transformer bushing partial discharge online monitoring method | |
| CN116106690A (en) | Method for improving cable defect time domain recovery waveform characteristics based on low-frequency stitching and application | |
| Sun et al. | Application of wavelet soft threshold denoising algorithm based on EMD decomposition in vibration signals | |
| CN111157853A (en) | Method and system for identifying discharge state of power transmission line | |
| CN114113943A (en) | Transformer partial discharge detection system, method and equipment based on current and ultrasonic signals | |
| CN104991171B (en) | GIS partial discharge frequency dividing failure collection of illustrative plates method for drafting based on ultra-high frequency signal | |
| CN109596354A (en) | Band-pass filtering method based on the identification of adaptive resonance frequency band | |
| CN104977602A (en) | Control method and apparatus for earthquake data acquisition construction | |
| CN209746080U (en) | Multi-band partial discharge detection device for cable accessory partial discharge ground electric wave |
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 | ||
| CB03 | Change of inventor or designer information | ||
| CB03 | Change of inventor or designer information |
Inventor after: Zhou Kai Inventor after: Li Zerui Inventor after: Huang Yonglu Inventor after: Xie Min Inventor after: Zhao Shilin Inventor after: Zhu Guangya Inventor after: Ran Li Inventor before: Zhou Kai Inventor before: Huang Yonglu Inventor before: Xie Min Inventor before: Zhao Shilin Inventor before: Li Zerui Inventor before: Zhu Guangya Inventor before: Ran Li |
|
| GR01 | Patent grant | ||
| GR01 | Patent grant |