CN104655991A - Power system fault matching method based on mutant point dejection combinational algorithm - Google Patents

Abstract

The invention provides a power system fault matching method based on a mutant point dejection combinational algorithm. The method comprises the following steps: randomly extracting one group of fault data at each of two ends of a line from a fault recording database in a prudential system; respectively calculating fault start points corresponding to two ends by using the mutant point dejection combinational algorithm; solving associated criterion parameter according to fault start points at two ends calculated in the step II; judging whether the fault data at two ends are matched or not according to triple judgment of data match, namely carrying out fault phase selection on two ends by applying a mutational volume current phase selection method to judge whether the fault types corresponding to data at two ends are matched or not; if so, judging whether load currents at two ends are matched or not according to the load currents and judging whether data at two ends are matched or not by using an initial fault current angle criterion; if so, showing that the triple criterion judgment results are matched; and finally judging whether the data at two ends are matched or not, and if not, judging that the data at two ends are not matched.

Description

Based on the electric power system fault data matching method of Singularity detection combinational algorithm

Technical field

The present invention relates to electric power system fault positioning field, particularly a kind of electric power system fault data matching method based on Singularity detection combinational algorithm.

Background technology

Electric power system fault data contain abundant information resources; be fully used for making it; current people are based on original fault information processing system, and maintenance data digging technology develops proterctive equipment fault information managing and analytic system (being called for short " guarantor's communication system ").Protecting communication system to read by the important information such as fault massage substation protective device, wave recording device of GPS precision time service system synchronization, complete the functions such as calculating, analysis, drawing, information output, providing decision information for running with relay protection personnel.

Localization of fault protects a critical function of communication system.At present, in transmission open acess technology, both-end distance measuring, owing to making full use of failure message, has the advantages such as distance accuracy is high, simple to operate, practical, is used widely in electric system.The electricity that the fault data that both-end distance measuring module in guarantor's communication system utilizes is faulty line two ends homogeneous fault and so-called matched data, and in fault recorder data storehouse, store the data of magnanimity, therefore must screen to make it mutually mate to fault data before both-end distance measuring.Generally, owing to there being GPS precision time service system to ensure data syn-chronization, the fault data that markers is identical is matched data.But in reality, external interference may make GPS precision time service system produce error, although there is scholar to propose to adopt High Precision Crystal Oscillator to carry out on-line monitoring and correction to gps clock, but further contemplate the error that the factors such as mutual inductor phase shift, hardware time delay and sampling rate difference are introduced, also also non-critical is identical for the matched data markers in guarantor's communication system.Therefore, when continuity failure appears in circuit, each fault data markers close to time, only carry out Data Matching according to markers and range finder module may be made to read corrupt data, causing trouble is located unsuccessfully then.Therefore, study a kind of new fault data matching process and make both-end distance measuring module no longer solely rely on GPS precision time service system when carrying out data screening, tool is of great significance.

Summary of the invention

The present invention mainly solves the technical matters that existing method exists, and designs a kind of electric power system fault data matching method based on Singularity detection combinational algorithm.

Technical scheme provided by the invention is a kind of electric power system fault data matching method based on Singularity detection combinational algorithm, comprises the following steps,

Step one, acquisition fault data, be included in random each one group of the fault data extracting certain circuit two ends in the fault recorder data storehouse protected in communication system, and note circuit one end is that m holds, and the other end is n end;

Step 2, utilize Singularity detection combinational algorithm to calculate the corresponding fault starting point in two ends respectively, comprise and following sub-step is performed respectively to the two ends of circuit,

Steps A 1, the fault data group of certain circuit one end extracted based on step one, generate baseband signal sequence, and implementation carries out phase-model transformation for using K transfer pair three-phase current:

$\left[\begin{array}{c}{i}_0\\ i_1\\ i_2\end{array}\right]=\left[\begin{array}{ccc}1& 1& 1\\ 1& 2& -3\\ 1& -3& 2\end{array}\right]\left[\begin{array}{c}{i}_a\\ i_b\\ i_c\end{array}\right]$

Wherein, i _a, i _b, i _cfor three-phase current, i ₀, i ₁, i ₂for corresponding three kinds of mold components, choose a kind of mold component as baseband signal, generate baseband signal sequence;

Steps A 2, utilization Bayesian Classification Arithmetic are classified to steps A 1 gained baseband signal sequence, find out suspicious catastrophe point;

Steps A 3, obtain the Sudden Changing Rate electric current of steps A 1 gained baseband signal sequence, use Sudden Changing Rate current algorithm to calculate current break point n _s0;

Steps A 4, utilization Singularity detection combinational algorithm determination fault starting point, comprise with steps A 3 gained n _s0for baseline configuration time window [n _s0-Δ n, n _s0], wherein Δ n is default window width, and utilize time window to filter out fault starting point from by the suspicious catastrophe point of steps A 2 gained, screening mode is as follows,

1) when only having a suspicious catastrophe point in time window, fault starting point n _ssuspicious catastrophe point for this reason;

2) when having two and above suspicious catastrophe point in time window, fault starting point n _sfor these suspicious catastrophe points average after round;

3) when there is no suspicious catastrophe point in time window, fault starting point n _sfor steps A 3 gained current break point n _s0;

Step 3, ask for related criteria parameter according to the two ends fault starting point calculated in step 2, and judge whether two ends fault data mates according to triple criterions of Data Matching, comprise following sub-step:

Step B1, with the corresponding fault starting point n in two ends _sfor benchmark, fourier algorithm is used to calculate the normal of each phase of two end datas and fault current and voltage phasor respectively;

Step B2, use Sudden Changing Rate electric current phase selection method to carry out Fault Phase Selection respectively to two ends, judge whether the fault type that two end datas are corresponding mates;

If the fault type that step B3 step B2 judged result is two end datas does not mate, then directly judge that two end datas do not mate, process ends; If coupling, calculate m and hold electric current with it is as follows,

Two ends are all benchmark phase with A phase, push away the electric current phasor of a cycle before using fourier algorithm to calculate the fault starting point of being held by step 2 gained m, obtain m and hold electric current push away electric current phasor and the voltage phasor of a cycle before using fourier algorithm to calculate the fault starting point of being held by step 2 gained n, obtain n respectively and hold electric current with n terminal voltage

Circuit distributed parameter model is used to hold electric current by n again with n terminal voltage calculate that corresponding m holds electric current

${\stackrel{\·}{I}}_m^{\′}=\frac{\mathrm{sh}(\mathrm{\γ}\×L)}{Z_c}\×{\stackrel{\·}{U}}_n-\mathrm{ch}(\mathrm{\γ}\×L)\×{\stackrel{\·}{I}}_n$

Wherein, γ and z _cbe respectively propagation constant and the characteristic impedance of circuit, L is line length;

Step B4, judge whether two ends load current mates according to load current criterion, if comprise | ρ-1|≤λ, then load current coupling; Otherwise load current does not mate, wherein, parameter λ is default threshold values;

Do not mate if step B5 step B4 judged result is two ends load current, then judge that two end datas do not mate, process ends; If coupling, then whether two end datas mate to utilize switching angle criterion to judge, if comprise parameter then switching angle coupling, otherwise switching angle does not mate, wherein, for the threshold values preset;

If step B6 step B5 judged result is switching angle coupling, illustrate that triple criterion result of determination is all mated, then finally judge two ends Data Matching, otherwise two end datas do not mate.

And steps A 3 uses the realization of Sudden Changing Rate current algorithm searching current break point as follows:

A) Sudden Changing Rate electric current Δ i (k) is calculated according to formula Δ i (k)=i (k)-i (k-N), wherein N is the sampling number of fault wave recording device one-period, the sampled point k place electric current that i (k) is baseband signal i;

B) according to detection criteria determination current break point, criterion is as follows,

$\left\{\begin{array}{c}\left|\mathrm{\ψ}\left(k\right)\right|>\mathrm{\ξ}\\ \mathrm{card}\left(A\right)\≥\mathrm{\β}\end{array}\right.$

Wherein, parameter a={ ψ (n) || ψ (n) | > ξ, k≤n < k+ α }, α and β, ξ are parameter preset, and first the k value meeting above-mentioned criterion is the catastrophe point that detected signal is corresponding.

The present invention is directed to electric power system fault design data and go out a kind of Singularity detection combinational algorithm, propose triple criterions of carrying out Data Matching in fault recorder data storehouse on this basis, whether circuit two ends fault data belongs to homogeneous fault to utilize triple criterion effectively to judge, namely whether data mate, thus ensure for the basic data of effectively carrying out providing of two ends of electric transmission line fault localization.Therefore, tool of the present invention has the following advantages: use the fault starting-tool point method based on jump-value of current to carry out fault data coupling, result of determination reliability is high, real-time, simple and practical, the more reliable electric power system fault data matching method of markers coupling can be carried out for the both-end distance measuring module of protecting in communication system provides a kind of than simple dependence GPS precision time service.

Accompanying drawing explanation

Fig. 1 is the both end power supplying system circuit diagram of the embodiment of the present invention for emulating.

Fig. 2 is the electric power system fault Data Matching decision flowchart of the embodiment of the present invention.

Fig. 3 is the Fault Phase Selection process flow diagram of the embodiment of the present invention.

Embodiment

Below by embodiment also by reference to the accompanying drawings, technical scheme of the present invention is described in further detail.

See Fig. 2, it is as follows that embodiments of the invention comprise step:

The step of step one, acquisition electric power system fault data: random each one group of the fault data extracting certain circuit two ends in the fault recorder data storehouse in guarantor's communication system.

See Fig. 1, circuit one end (left end) is m end, and the other end (right-hand member) is n end, E _m, E _nand Z _m, Z _nrepresent that m end and n hold voltage and the impedance of power supply respectively, L is line length, and d is the fault distance of m end.Each one group of the fault data of embodiment difference random selecting faulty line m side and n side in fault recorder data storehouse.

The step of step 2, detection failure starting point: utilize Singularity detection combinational algorithm to calculate the corresponding fault starting point of two end datas respectively, asks for related criteria parameter when carrying out fault data coupling for using triple criterion afterwards and does corresponding preparation.

The step 2 of embodiment specifically comprises and performs following sub-step based on corresponding failure data respectively to circuit two ends:

Steps A 1, the fault data group of certain circuit one end extracted based on step one, generate baseband signal sequence, and as sequence to be detected, implementation carries out phase-model transformation for using K transfer pair three-phase current:

$\left[\begin{array}{c}{i}_0\\ i_1\\ i_2\end{array}\right]=\left[\begin{array}{ccc}1& 1& 1\\ 1& 2& -3\\ 1& -3& 2\end{array}\right]\left[\begin{array}{c}{i}_a\\ i_b\\ i_c\end{array}\right]$

Wherein, i _a, i _b, i _cfor three-phase current, i ₀, i ₁, i ₂for corresponding three kinds of mold components, when specifically implementing, wherein any one mold component can be selected to carry out subsequent analysis.Embodiment chooses mold component i ₁=i _a+ 2i _b-3i _cas baseband signal, generate baseband signal sequence, carry out analysis hereinafter and calculating.

Steps A 2, utilization Bayesian Classification Arithmetic are classified to steps A 1 gained baseband signal sequence, find out " suspicious catastrophe point ".

Steps A 3, obtain the Sudden Changing Rate electric current of steps A 1 gained baseband signal sequence, use Sudden Changing Rate current algorithm to calculate current break point n _s0.

Steps A 4, utilization Singularity detection combinational algorithm determination fault starting point.

The steps A 4 of embodiment is with steps A 3 gained n _s0for baseline configuration time window [n _s0-Δ n, n _s0], wherein Δ n is default window width, and during concrete enforcement, those skilled in the art can preset value voluntarily.Time window is utilized to filter out fault starting point from " the suspicious catastrophe point " that drawn by steps A 2.Screening scheme is:

1) when only having one " suspicious catastrophe point " in time window, fault starting point n _snamely for this reason " suspicious catastrophe point ";

2) when having two and above " suspicious catastrophe point " in time window, fault starting point n _sfor these " suspicious catastrophe points " average after round;

3) when there is no " suspicious catastrophe point " in time window, fault starting point n _sfor steps A 3 gained current break point n _s0, n _s=n _s0.

During concrete enforcement, the Bayesian Classification Arithmetic that steps A 2 adopts can reference: Liu Mige, Li little Bin. based on the inflection point detection [J] of Bayes decision-making. computer utility, 2013,01:230-232. for the sake of ease of implementation, provides steps A 2 in embodiment to use being implemented as follows of " suspicious catastrophe point " in Bayesian Classification Arithmetic searching baseband signal sequence:

A) element in baseband signal sequence is divided into two classes: definition catastrophe point set ω ₁, prior probability is P (ω ₁); Not mutated set ω ₂, prior probability is P (ω ₂);

B) pre-service is carried out to baseband signal sequence and obtain sequence { z (k) }.

First derivative of current x (k) at each sampled point k place of baseband signal sequence is obtained:

$x\left(k\right)=\frac{d\left(i_1\left(k\right)\right)}{\mathrm{dt}}=\frac{i_1(k+1)-i_1(k-1)}{2T}$

Wherein, i ₁k () is baseband signal i ₁sampled point k place electric current.

Obtain Sudden Changing Rate y (k) of x (k) at each sampled point k place again:

y(k)＝[x(k)-x(k+1)] ²

Finally y (k) is normalized and obtains z (k):

$z\left(k\right)=\frac{y\left(k\right)-(1-\mathrm{\&delta;})\&CenterDot;\min \left(y\right)}{(1+\mathrm{\&delta;})\&CenterDot;\max \left(y\right)-(1-\mathrm{\&delta;})\&CenterDot;\min \left(y\right)}$

Wherein, maxy (k), miny (k) are respectively the maximum and least member in sequence { y (k) }, T is the sampling interval between neighbouring sample point, parameter δ is the positive number that numerical value is very little, and concrete those skilled in the art can the value of parameter preset δ voluntarily when implementing.

C) to class conditional probability P (z (k) | ω ₁) and P (z (k) | ω ₂) do to estimate as follows:

$\hat{P}\left(z\left(k\right)\right|{\mathrm{\&omega;}}_1)=\frac{{z\left(k\right)}^{\mathrm{\&chi;}}}{\underset{k=1}{\overset{\mathrm{\&Gamma;}}{\mathrm{\&Sigma;}}}z{\left(k\right)}^{\mathrm{\&chi;}}}$

$\hat{P}\left(z\left(k\right)\right|{\mathrm{\&omega;}}_2)=\frac{1-z{\left(k\right)}^{\mathrm{\&chi;}}}{\mathrm{\&Gamma;}-\underset{k=1}{\overset{\mathrm{\&Gamma;}}{\mathrm{\&Sigma;}}}z{\left(k\right)}^{\mathrm{\&chi;}}}$

Wherein, Γ is the element number of sequence { z (k) }, i.e. the sum of sampled point, 1≤k≤Γ; Parameter χ is just, concrete those skilled in the art can the value of parameter preset χ voluntarily when implementing; for class conditional probability P (z (k) | ω ₁) estimated value, for class conditional probability P (z (k) | ω ₂) estimated value.

D) principle of classification

Calculate classification error rate P (e)

$\begin{array}{c}P\left(e\right)=P(x\&Element;R_1,x\&Element;{\mathrm{\&omega;}}_2)+P(x\&Element;R_2,x\&Element;{\mathrm{\&omega;}}_1)\\ =\underset{R_1}{\&Integral;}P\left(x\right|{\mathrm{\&omega;}}_2)P\left({\mathrm{\&omega;}}_2\right)\mathrm{dx}+\underset{R_2}{\&Integral;}P\left(x\right|{\mathrm{\&omega;}}_1)P\left({\mathrm{\&omega;}}_1\right)\mathrm{dx}\\ =P\left({\mathrm{\&omega;}}_2\right)\underset{R_1}{\&Integral;}P\left(x\right|{\mathrm{\&omega;}}_2)\mathrm{dx}+P\left({\mathrm{\&omega;}}_1\right)\underset{R_2}{\&Integral;}P\left(x\right|{\mathrm{\&omega;}}_1)\mathrm{dx}\\ =P\left({\mathrm{\&omega;}}_2\right)P_2\left(e\right)+P\left({\mathrm{\&omega;}}_1\right)P_1\left(e\right)\end{array}$

Wherein, R ₁for the data area that catastrophe point in sequence { z (k) } is corresponding, R ₂for corresponding data area not mutated in sequence { z (k) }, P (x ∈ R ₁, x ∈ ω ₂) represent region R ₁in data judging be the probability of not mutated point, P (x ∈ R ₂, x ∈ ω ₁) represent region R ₂in data judging be the probability of catastrophe point, P (ω ₂) P ₂(e) probability for by Singularity detection being not mutated point (undetected), P (ω ₁) P ₁e () is for being detected as the probability of catastrophe point (flase drop), prior probability P (ω by not mutated point ₁) and P (ω ₂) be constant (all unknown).Calculating inference process can see Bayesian Classification Arithmetic pertinent literature.

In order at P ₂p is made during (e)=ε ₁e () is minimum, the Neyman-Pearson criterion in Bayesian decision theory can be used to classify.Parameter ε is a very little positive number, and concrete those skilled in the art can the value of parameter preset ε voluntarily when implementing.

E) determine " suspicious catastrophe point "

According to Neyman-Pearson criterion, have:

${\&Integral;}_{R_1^{*}}P\left(z\left(k\right)\right|{\mathrm{\&omega;}}_2)\mathrm{dx}=\mathrm{\&epsiv;}$

$\mathrm{\&mu;}=\frac{\hat{P}\left(z\left(k^{*}\right)\right|{\mathrm{\&omega;}}_1)}{\hat{P}\left(z\left(k^{*}\right)\right|{\mathrm{\&omega;}}_2)}=\frac{z{\left(k^{*}\right)}^{\mathrm{\&chi;}}\&CenterDot;[\mathrm{\&Gamma;}-\underset{k=1}{\overset{\mathrm{\&Gamma;}}{\mathrm{\&Sigma;}}}z{\left(k\right)}^{\mathrm{\&chi;}}]}{\underset{k=1}{\overset{\mathrm{\&Gamma;}}{\mathrm{\&Sigma;}}}z{\left(k\right)}^{\mathrm{\&chi;}}\&CenterDot;[1-z{\left(k^{*}\right)}^{\mathrm{\&chi;}}]}$

Wherein for region R ₁a certain subregion.

By element (1≤k≤N, N is the sampling number of one-period) obtains sequence according to order arrangement from small to large then construct new sequence { S (k) }, make wherein element according to inequality S (k ^*)≤ε < S (k ^*+ 1) the corresponding sequence number k of the element meeting inequality is obtained ^*, then obtain the value of parameter μ.

Last according to criterion then $z\left(k\right)=\left\{\begin{array}{c}{\mathrm{\&omega;}}_1\\ {\mathrm{\&omega;}}_2\end{array}\right.$ Carry out the classification of signal sampling point, wherein gather ω ₁in element be referred to as " suspicious catastrophe point ".

For the sake of ease of implementation, steps A 3 in embodiment is provided to use Sudden Changing Rate current algorithm to find being implemented as follows of current break point:

A) Sudden Changing Rate electric current Δ i (k) is calculated according to formula Δ i (k)=i (k)-i (k-N):

Wherein, i (k) the sampled point k place electric current that is baseband signal i.

Embodiment is according to formula Δ i ₁(k)=i ₁(k)-i ₁(k-N) Sudden Changing Rate electric current Δ i is calculated ₁(k), wherein N is the sampling number of fault wave recording device one-period;

B) according to detection criteria determination current break point, criterion is as follows:

$\left\{\begin{array}{c}\left|\mathrm{\&psi;}\left(k\right)\right|>\mathrm{\&xi;}\\ \mathrm{card}\left(A\right)\&GreaterEqual;\mathrm{\&beta;}\end{array}\right.$

Wherein, parameter a={ ψ (n) || ψ (n) | > ξ, k≤n < k+ α }, card (A) represents the element number in set A.Parameter alpha and β can preset, and concrete setting can be relevant with the sampling number N of one-period, generally desirable (capping when N is larger, removes the limit time less).First the k value meeting above-mentioned criterion is catastrophe point corresponding to detected signal.It is concrete that those skilled in the art can the value of parameter preset ξ voluntarily when implementing.

After respectively steps A 1 ~ A4 is performed to the m of circuit end and n end, obtain corresponding fault starting point.

Step 3, fault data coupling judges: ask for related criteria parameter according to the two ends corresponding failure starting point calculated in previous step and judge whether two ends fault data mates according to triple criterions of Data Matching.

The design of the present invention's triple criterion is as follows:

The step 3 of embodiment specifically comprises following sub-step:

Step B1, with the corresponding fault starting point n in two ends _sfor benchmark, fourier algorithm is used to calculate the normal of each phase of two end datas and fault current and voltage phasor respectively.

Step B2, use Sudden Changing Rate electric current phase selection method to carry out Fault Phase Selection respectively to two ends, judge whether the fault type that two end datas are corresponding mates, and this is fault type criterion, namely in subsequent step, if two ends Fault Phase Selection result is different, then directly judge that two end datas do not mate, process ends; If two ends Fault Phase Selection comes to the same thing, namely two ends fault type coupling, then must use load current criterion and switching angle criterion to do to judge further.During concrete enforcement, Fault Phase Selection realizes can reference: Yang Qixun. micro computer relay protection basis [M]. and Beijing: China Electric Power Publishing House, 2004.

The phase selection flow process that embodiment is carried out circuit either end as shown in Figure 3.In figure, be respectively the Sudden Changing Rate electric current of three-phase.

First according to the normal of each phase of this end data of step B1 gained and fault current and voltage phasor, the Sudden Changing Rate of calculating three kinds of phase differential currents then carry out correlation ratio comparatively with judgement:

When $|{\stackrel{\&CenterDot;}{I}}_B-{\stackrel{\&CenterDot;}{I}}_C|=\min \left\{\right|{\stackrel{\&CenterDot;}{I}}_A-{\stackrel{\&CenterDot;}{I}}_B|,|{\stackrel{\&CenterDot;}{I}}_B-{\stackrel{\&CenterDot;}{I}}_C|,|{\stackrel{\&CenterDot;}{I}}_C-{\stackrel{\&CenterDot;}{I}}_A\left|\right\},$ Then judge $|{\stackrel{\&CenterDot;}{I}}_B-{\stackrel{\&CenterDot;}{I}}_C|<<|{\stackrel{\&CenterDot;}{I}}_C-{\stackrel{\&CenterDot;}{I}}_A|$ Whether set up,

If be false, enter judgement operation X;

If set up and then judge further whether set up, if set up, judge that fault is as A phase ground connection, if be false, enter judgement operation X.

Described judgement operation X, comprises and first judging whether set up, if set up, then judge further whether set up, if set up, judge that fault is as AB line to line fault, otherwise be judged to be AB line to line fault ground connection; If be false, same judgement further whether set up, if set up, judge that fault is as AC line to line fault, otherwise be judged to be AC line to line fault ground connection.

When $|{\stackrel{\&CenterDot;}{I}}_A-{\stackrel{\&CenterDot;}{I}}_B|=\min \left\{\right|{\stackrel{\&CenterDot;}{I}}_A-{\stackrel{\&CenterDot;}{I}}_B|,|{\stackrel{\&CenterDot;}{I}}_B-{\stackrel{\&CenterDot;}{I}}_C|,|{\stackrel{\&CenterDot;}{I}}_C-{\stackrel{\&CenterDot;}{I}}_A\left|\right\}$ Or $|{\stackrel{\&CenterDot;}{I}}_C-{\stackrel{\&CenterDot;}{I}}_A|=\min \left\{\right|{\stackrel{\&CenterDot;}{I}}_A-{\stackrel{\&CenterDot;}{I}}_B|,|{\stackrel{\&CenterDot;}{I}}_B-{\stackrel{\&CenterDot;}{I}}_C|,|{\stackrel{\&CenterDot;}{I}}_C-{\stackrel{\&CenterDot;}{I}}_A\left|\right\}$ Time determination step with $|{\stackrel{\&CenterDot;}{I}}_A-{\stackrel{\&CenterDot;}{I}}_B|=\min \left\{\right|{\stackrel{\&CenterDot;}{I}}_A-{\stackrel{\&CenterDot;}{I}}_B|,|{\stackrel{\&CenterDot;}{I}}_B-{\stackrel{\&CenterDot;}{I}}_C|,|{\stackrel{\&CenterDot;}{I}}_C-{\stackrel{\&CenterDot;}{I}}_A\left|\right\}$ Time similar, it will not go into details in the present invention, omits in Fig. 3.

If the fault type that step B3 step B2 judged result is two end datas does not mate, then directly judge that two end datas do not mate, process ends; If two ends fault type mates, then the m calculated according to the m end data fault starting point detected in step 2 holds electric current electric current is held with the m that the n end data fault starting point detected according to step 2 calculates offside (be set to m end, represent with subscript m) electric current is calculated comprising the electricity using circuit distributed parameter model by circuit side (be set to n to hold, represent with subscript n)

Two ends are all benchmark phase with A phase, use fourier algorithm to calculate by the fault starting point n detected _s1(m holds corresponding fault starting point n _s), n _s2(n holds corresponding fault starting point n _s) front (the i.e. sampled point n pushing away a cycle _s1-N and n _s2-N is corresponding) electric current and voltage phasor, comprise the electric current phasor pushing away a cycle before using fourier algorithm to calculate the fault starting point of being held by step 2 gained m, obtain m and hold electric current push away electric current phasor and the voltage phasor of a cycle before using fourier algorithm to calculate the fault starting point of being held by step 2 gained n, obtain n respectively and hold electric current with n terminal voltage circuit distributed parameter model is used to calculate that corresponding m holds electric current by n end data again

${\stackrel{\&CenterDot;}{I}}_m^{\&prime;}=\frac{\mathrm{sh}(\mathrm{\&gamma;}\&times;L)}{Z_c}\&times;{\stackrel{\&CenterDot;}{U}}_n-\mathrm{ch}(\mathrm{\&gamma;}\&times;L)\&times;{\stackrel{\&CenterDot;}{I}}_n$

Wherein, γ and z _cbe respectively propagation constant and the characteristic impedance of circuit, L is line length.

Step B4, judge whether two ends load current mates according to load current criterion.Load current criterion is: if | ρ-1|≤λ, then load current coupling, otherwise load current does not mate.Wherein, parameter λ is threshold values, I _m, I ' _mbe with effective value.It is concrete that those skilled in the art can the value of pre-set threshold value λ voluntarily when implementing.

Do not mate if step B5 step B4 judged result is two ends load current, then judge that two end datas do not mate, process ends; If two ends load current mates, then whether two end datas mate to utilize switching angle criterion to judge.Switching angle criterion is: if parameter then switching angle coupling, otherwise switching angle does not mate.Wherein, for threshold values, concrete those skilled in the art can pre-set threshold value voluntarily when implementing value.

If step B6 step B5 judged result is switching angle coupling, then illustrate that triple criterion result of determination is all mated, then finally judge two ends Data Matching, otherwise finally still judge that two end datas do not mate.

Specific embodiment described herein is only to the explanation for example of the present invention's spirit.Those skilled in the art can make various amendment or supplement or adopt similar mode to substitute to described specific embodiment, but can't depart from spirit of the present invention or surmount the scope that appended claims defines.

Claims (2)

1., based on an electric power system fault data matching method for Singularity detection combinational algorithm, it is characterized in that: comprise the following steps,

Step one, acquisition fault data, be included in random each one group of the fault data extracting certain circuit two ends in the fault recorder data storehouse protected in communication system, and note circuit one end is that m holds, and the other end is n end;

Step 2, utilize Singularity detection combinational algorithm to calculate the corresponding fault starting point in two ends respectively, comprise and following sub-step is performed respectively to the two ends of circuit,

Steps A 1, the fault data group of certain circuit one end extracted based on step one, generate baseband signal sequence, and implementation carries out phase-model transformation for using K transfer pair three-phase current:

$\left[\begin{array}{c}{i}_0\\ i_1\\ i_2\end{array}\right]=\left[\begin{array}{ccc}1& 1& 1\\ 1& 2& -3\\ 1& -3& 2\end{array}\right]\left[\begin{array}{c}{i}_a\\ i_b\\ i_c\end{array}\right]$

Wherein, i _a, i _b, i _cfor three-phase current, i ₀, i ₁, i ₂for corresponding three kinds of mold components, choose a kind of mold component as baseband signal, generate baseband signal sequence;

Steps A 2, utilization Bayesian Classification Arithmetic are classified to steps A 1 gained baseband signal sequence, find out suspicious catastrophe point;

Steps A 3, obtain the Sudden Changing Rate electric current of steps A 1 gained baseband signal sequence, use Sudden Changing Rate current algorithm to calculate current break point n _s0;

Steps A 4, utilization Singularity detection combinational algorithm determination fault starting point, comprise with steps A 3 gained n _s0for baseline configuration time window [n _s0-Δ n, n _s0], wherein Δ n is default window width, and utilize time window to filter out fault starting point from by the suspicious catastrophe point of steps A 2 gained, screening mode is as follows,

1) when only having a suspicious catastrophe point in time window, fault starting point n _ssuspicious catastrophe point for this reason;

2) when having two and above suspicious catastrophe point in time window, fault starting point n _sfor these suspicious catastrophe points average after round;

3) when there is no suspicious catastrophe point in time window, fault starting point n _sfor steps A 3 gained current break point n _s0;

Step 3, ask for related criteria parameter according to the two ends fault starting point calculated in step 2, and judge whether two ends fault data mates according to triple criterions of Data Matching, comprise following sub-step:

Step B1, with the corresponding fault starting point n in two ends _sfor benchmark, fourier algorithm is used to calculate the normal of each phase of two end datas and fault current and voltage phasor respectively;

Step B2, according to step B1 acquired results, use Sudden Changing Rate electric current phase selection method to carry out Fault Phase Selection respectively to two ends, judge whether the fault type that two end datas are corresponding mates;

If the fault type that step B3 step B2 judged result is two end datas does not mate, then directly judge that two end datas do not mate, process ends; If coupling, calculate m and hold electric current with it is as follows,

Two ends are all benchmark phase with A phase, push away the electric current phasor of a cycle before using fourier algorithm to calculate the fault starting point of being held by step 2 gained m, obtain m and hold electric current push away electric current phasor and the voltage phasor of a cycle before using fourier algorithm to calculate the fault starting point of being held by step 2 gained n, obtain n respectively and hold electric current with n terminal voltage

Circuit distributed parameter model is used to hold electric current by n again with n terminal voltage calculate that corresponding m holds electric current

${\stackrel{.}{I}}_m^{\&prime;}=\frac{\mathrm{sh}(\mathrm{\&gamma;}\&times;L)}{Z_c}\&times;{\stackrel{.}{U}}_n-\mathrm{ch}(\mathrm{\&gamma;}\&times;L)\&times;{\stackrel{.}{I}}_n$

Wherein, γ and z _cbe respectively propagation constant and the characteristic impedance of circuit, L is line length;

Step B4, judge whether two ends load current mates according to load current criterion, if comprise | ρ-1|≤λ, then load current coupling; Otherwise load current does not mate, wherein, parameter λ is default threshold values;

Do not mate if step B5 step B4 judged result is two ends load current, then judge that two end datas do not mate, process ends; If coupling, then whether two end datas mate to utilize switching angle criterion to judge, if comprise parameter then switching angle coupling, otherwise switching angle does not mate, wherein, for the threshold values preset;

If step B6 step B5 judged result is switching angle coupling, illustrate that triple criterion result of determination is all mated, then finally judge two ends Data Matching, otherwise two end datas do not mate.

2. according to claim 1 based on the electric power system fault data matching method of Singularity detection combinational algorithm, it is characterized in that: steps A 3 uses the realization of Sudden Changing Rate current algorithm searching current break point as follows:

A) Sudden Changing Rate electric current Δ i (k) is calculated according to formula Δ i (k)=i (k)-i (k-N), wherein N is the sampling number of fault wave recording device one-period, the sampled point k place electric current that i (k) is baseband signal i;

B) according to detection criteria determination current break point, criterion is as follows,

$\left\{\begin{array}{c}\left|\mathrm{\&psi;}\left(k\right)\right|>\mathrm{\&xi;}\\ \mathrm{card}\left(A\right)\&GreaterEqual;\mathrm{\&beta;}\end{array}\right.$

Wherein, parameter a={ ψ (n) || ψ (n) | > ξ, k≤n < k+ α }, α and β, ξ are parameter preset, and first the k value meeting above-mentioned criterion is the catastrophe point that detected signal is corresponding.