CN104764963A - Alternating-current arc fault detecting method - Google Patents

Abstract

The invention discloses an alternating-current arc fault detecting method. The alternating-current arc fault detecting method comprises the following steps that firstly, voltage signals or current signals in a loop to be detected are collected through a mutual inductor, the collected voltage signals or current signals are filtered and conditioned, and analog-digital conversion is conducted on the filtered and conditioned signals; secondly, methods for processing the voltage signals or current signals obtained in the first step are designed; thirdly, any one method or the combination of two or three methods is selected from the methods obtained in the second step and used for processing the voltage signals or current signals, so that a judgment result is obtained. The alternating-current arc fault detecting method has the advantages that implementation is easy, the detection speed is high, response is sensitive, and the accuracy is high.

Description

Alternating current arc fault detection method

Technical field

The present invention relates to alternating current arc field of fault detection, particularly a kind of alternating current arc fault detection method.

Background technology

Along with socioeconomic development, in the building such as residential building, factory, market, the kind of consumer and quantity are on the increase.Because electric wiring, consumer internal wiring are with for a long time and are carried or overlond running, cause electric wire insulation layer aging even damaged gradually.Easily produce fault electric arc in the position of insulation damages, if can not Timeliness coverage taking appropriate measures, fault electric arc be easy to cause fire or blast.In addition, on aircraft except the fault electric arc caused by insulation ag(e)ing, the wiring that the mechanical vibration in aircraft flight cause loosens also can produce fault electric arc, serious threat the people's lives and property safety.Traditional protective device can only play a protective role to circuit when excess current or short circuit, and the electric current of fault electric arc is in most cases less than the specified limit value of its action, probably can not react.Therefore, also need in distribution system to set up arc fault detection and protective device, to reach the object stoping arc fault to cause fire or blast, protection the people's lives and property safety.

Existing fault arc detection method is generally from following three aspects:

One, temporal signatures.

Method: comprise that voltage or current amplitude reduce, positive-negative half-cycle is asymmetric, zero stop the feature such as the appearance in district and the sudden change of waveform slope.

Shortcoming: voltage or current amplitude also can change when switching different loads.The asymmetric phenomenon of positive-negative half-cycle can be subject to the impact of material molten degree between the conductor of trouble spot two.The width that current zero stops district can be subject to the impact of the conditions such as plasmoid between electrode, clearance distance size, angle of impedance and change.The slope of voltage or current waveform stops the impact of district's duration length and material surface degree of oxidation by current zero.But plasmoid in reality between arc fault point two conductor, gap length are unfixed, in a lot of situation, loadtype neither be changeless.Therefore temporal signatures is adopted to be difficult to detect arc fault exactly.

Two, the amplitude of high-frequency signal

Method: to the voltage collected or current signal through high-pass filtering process, using filtered signal as analog-to-digital input signal.If the amplitude of high-frequency signal exceedes a certain threshold value and its duration exceedes another threshold value, then increase the value of counter.If exceed default threshold value in the value of official hour inside counting device, think and create arc fault.

Shortcoming: the amplitude of high band pulse is comparatively obvious by the impact of certain loads type and size of current, and the threshold value of high band pulse needs to adjust for inhomogeneous load.In the complicated distribution that load often changes, this method is difficult to ensure its accuracy.

Three, wavelet transformation combines with BP neural network

Method: the eigenwert being obtained the wavelet transformation of the multiple load of many groups by test, using the training sample of these eigenwerts as neural network, after neural network convergence, just can be used in the microprocessor.Using the energy of detail signal two-layer before wavelet transformation as eigenwert, by eigenwert input neural network, arc fault whether there are to occur the output valve of neural network and threshold value multilevel iudge.

Shortcoming: BP neural network needs to be trained by a large amount of test figures, just can be transplanted in processor after treating neural network convergence and use.If the load type that training sample is insufficient or sample packages contains is comprehensive not, the accuracy of this detection algorithm can be caused to reduce.And the workload in early stage greatly will be increased by the train samples that test acquisition is abundant.

Summary of the invention

The object of the invention is to solve the problem, devising a kind of alternating current arc fault detection method.

Realizing above-mentioned purpose technical scheme of the present invention is that a kind of alternating current arc fault detection method, comprises the following steps:

A, gather voltage signal in loop to be detected or current signal by mutual inductor, and filtering and conditioning are carried out to the voltage signal collected or current signal, by the signal input analog-to-digital converter after filtering and conditioning, the output of analog to digital converter is discrete voltage signal or current signal;

The method of the voltage signal that b, design treatment are obtained by step a or current signal, the method for this process comprises:

Method one, voltage signal or current signal carried out to wavelet transformation and decompose and obtain approximate signal and detail signal, come whether occurred arc fault in decision circuit by the contrast of detail signal and threshold value;

Method two, the frequency spectrum that windowing DFT algorithm obtains signal is done to voltage signal or current signal, the harmonic component calculating certain specific exponent number, relative to fundametal compoment proportion, comes whether occurred arc fault in decision circuit by the contrast of this ratio and threshold value;

Method three, the frequency spectrum that windowing DFT algorithm obtains signal is done to voltage signal or current signal, from frequency spectrum, choose several carry out computing according to the algorithm preset, come whether occurred arc fault in decision circuit by the contrast of operation result and threshold value;

The voltage signal that in c, selecting step b, the combination of any one method or any two methods or the common combination of three methods obtain step a or current signal process, thus obtain result of determination.

Method one particular content in described step b is as follows:

Non-linear form filtering process is carried out to the voltage signal obtained in step a or current signal;

Wavelet transform is done to filtered signal, obtains the detail signal d of certain one deck;

Calculate the effective value of d:, wherein Level is the exponent number (level of resolution of d self) of this layer, for the discrete of detail signal d is counted.

D is compared with threshold value thr1.If D≤thr1, the method is thought does not have arc fault to produce in circuit; If D>thr1, the method judges to create arc fault in circuit.

Method two particular content in described step b is as follows:

The frequency spectrum that windowed DFT obtains signal is done to signal.The amplitude at frequency f (i) place is designated as y (f (i)), wherein i=1,2,3 ... n.N is the discrete point number that this frequency spectrum comprises.

Calculate the harmonic component of certain specific exponent number: wherein j=1,2,3 ..., F is the frequency of AC power, and the span of δ is 0 ~ 20.

Calculate the fundametal compoment in frequency spectrum: wherein F is the frequency of AC power, and the span of δ is 0 ~ 20.

Calculate the accounting of the relative fundametal compoment of harmonic component of certain specific exponent number: η _j=, j=1,2,3 ...

By η _jcompare with threshold value thr2.If η _j≤ thr2 then the method thinks do not have arc fault to produce in circuit; If η _j>thr2 then the method judges to create arc fault in circuit.

Method three particular content in described step b is as follows:

The frequency spectrum that windowed DFT obtains signal is done to signal.The amplitude at frequency f (i) place is designated as y (f (i)), wherein i=1,2,3 ... n.N is the discrete point number that this frequency spectrum comprises.

Optional m point, wherein m≤n from frequency spectrum.The frequency of this m some correspondence is denoted as f (p successively ₁), f (p ₂), f (p ₃) ... f (p _m), corresponding amplitude is followed successively by y (f (p ₁)), y (f (p ₂)), y (f (p ₃)) ... y (f (p _m)).

Computing is according to the following formula put by the m selected: $A\left(f\left(j\right)\right)=\frac{{\log }_{\mathrm{\α}}y\left(f\left(j\right)\right)}{\mathrm{\β}}+\mathrm{\γ}.$ Wherein the span of j is p ₁, p ₂, p ₃p _m, α > 0 and α ≠ 1, β ≠ 0,

Averaged: $\mathrm{result}=\frac{1}{m}*\underset{j=p_1}{\overset{p_m}{\mathrm{\Σ}}}A\left(f\left(j\right)\right).$

Result is compared with threshold value thr3.If result≤thr3, then the method is thought does not have arc fault to produce in circuit; If result>thr3, then the method judges to create arc fault in circuit.

Utilize the alternating current arc fault detection method of Technical Design of the present invention, detection method of the present invention is easy to hardware implementing, and contributes to reducing hard-wired cost; This detection method operand is little, can shorten operation time, and detection speed is fast, and be quick on the draw, accuracy is high; This detection method can detect the arc fault of multiple loadtype well, and the setting of threshold value does not need to change with the change of loadtype; This detection method applied widely, all applicable to the power supply of the multi-frequency scopes such as 50Hz, 400Hz; Such as Fig. 6 is the detail signal curve of certain one deck that wavelet transform obtains in method one.The left-half of curve is the detail signal of circuit when normally working, and right half part is detail signal when there is arc fault in circuit.After there is arc fault as seen from Figure 6, the amplitude of this layer of detail signal obviously increases, and the method accurately can detect arc fault.For 2 ~ 7 subharmonic, table 1 lists Application way two and calculates the relative first-harmonic accounting of harmonic wave.From data in table, compared with when not having an arc fault, occur that arc fault time-harmonic wave accounting obviously increases, therefore the method effectively can detect arc fault.The comparison diagram of normal and fault under the 50Hz supply frequency that in Fig. 7, (a) calculates for using method three, (b) be under 400Hz supply frequency normally and the comparison diagram of fault.In figure, blue curve is result of calculation when not having arc fault, result of calculation when red curve is arc fault.This detection method improves the accuracy of arc fault detection technique, improves the reliability of relay protection system.

Accompanying drawing explanation

Fig. 1 is the detection algorithm process flow diagram of the embodiment 1 of alternating current arc fault detection method of the present invention;

Fig. 2 is the detection algorithm process flow diagram of the embodiment 2 of alternating current arc fault detection method of the present invention;

Fig. 3 is the detection algorithm process flow diagram of the embodiment 3 of alternating current arc fault detection method of the present invention;

Fig. 4 is the detection algorithm process flow diagram of the embodiment 4 of alternating current arc fault detection method of the present invention;

Fig. 5 is the detection algorithm process flow diagram of the embodiment 5 of alternating current arc fault detection method of the present invention;

Fig. 6 is the detail signal curve of certain one deck that in alternating current arc fault detection method one of the present invention, wavelet transform obtains;

Table 1 is the accounting of the relative first-harmonic of 2 ~ 7 subharmonic that alternating current arc fault detection method two of the present invention calculates;

To be alternating current arc fault detection method three of the present invention normal for 50Hz, 400Hz two kinds of supply frequencies and the result of calculation comparison diagram of fault for Fig. 7.

Embodiment

In order to make those skilled in the art person understand technical scheme of the present invention better, below in conjunction with accompanying drawing, the present invention is specifically described,

Embodiment 1

As shown in Figure 1, the voltage that the mode adopting three kinds of methods to combine obtains step a or current signal process, and concrete treatment step is as follows:

Step one, system initialization, arranging supply frequency according to actual conditions is F, arranges the threshold value n_thr of cycle index.

Step 2, to counter C1, C2, C3 zero setting, to cycle index n_cycle zero setting;

Step 3, collection voltage or current signal;

Step 4, filtering and conditioning are carried out to signal;

Step 5, by filtering and conditioning after signal input analog-to-digital converter, the output of analog to digital converter is discrete voltage signal or current signal;

Step 6, the signal after analog to digital conversion to be processed signal respectively by following two branch roads:

First branch road: method one, wavelet transformation

1, non-linear form filtering process is carried out to signal;

2, wavelet transform is done to filtered signal, obtain the detail signal d of certain one deck;

3, the effective value of d is calculated:, wherein Level is the exponent number (level of resolution of d self) of this layer, for the discrete of detail signal d is counted;

4, D is compared with threshold value thr1.If D≤thr1, the method is thought does not have arc fault to produce in circuit; If D>thr1, the method judges to create arc fault in circuit, adds one with seasonal counter C1.

Second branch road:

First the frequency spectrum that windowed DFT obtains signal is done to signal.The amplitude at frequency f (i) place is designated as y (f (i)), wherein i=1,2,3 ... n.N is the discrete point number that this frequency spectrum comprises.

Then the frequency spectrum obtained is processed according to following two methods respectively:

Method two, harmonic wave accounting

1, the harmonic component of certain specific exponent number is calculated: wherein j=1,2,3 ..., F is the frequency of AC power, and the span of δ is 0 ~ 20.

2, the fundametal compoment in frequency spectrum is calculated: wherein F is the frequency of AC power, and the span of δ is 0 ~ 20.

3, the accounting of the relative fundametal compoment of harmonic component of certain specific exponent number is calculated: η _j=, j=1,2,3 ...

4, by η _jcompare with threshold value thr2.If η _j≤ thr2 then the method thinks do not have arc fault to produce in circuit; If η _j>thr2 then the method judges to create arc fault in circuit, adds one with seasonal counter C2.

Method three, spectrum analysis

1, optional m point, wherein m≤n from frequency spectrum.The frequency of this m some correspondence is denoted as f (p successively ₁), f (p ₂), f (p ₃) ... f (p _m), corresponding amplitude is followed successively by y (f (p ₁)), y (f (p ₂)), y (f (p ₃)) ... y (f (p _m)).

2, computing is according to the following formula put by the m selected: $A\left(f\left(j\right)\right)=\frac{{\log }_{\mathrm{\α}}y\left(f\left(j\right)\right)}{\mathrm{\β}}+\mathrm{\γ}.$ Wherein the span of j is p ₁, p ₂, p ₃p _m, α > 0 and α ≠ 1, β ≠ 0,

3, averaged: $\mathrm{result}=\frac{1}{m}*\underset{j=p_1}{\overset{p_m}{\mathrm{\Σ}}}A\left(f\left(j\right)\right).$

4, result is compared with threshold value thr3.If result≤thr3, then the method is thought does not have arc fault to produce in circuit; If result>thr3, then the method judges to create arc fault in circuit, makes counter C3 add one.

Step 7,

If C1, C2, C3 are zero, forward step 3 to;

If have in C1, C2, C3 one non-vanishing, forward step 8 to.

Step 8,

If w1*C1+w2*C2+w3*C3 >=n_thr*3* μ sets up, then judge to create arc fault in circuit, send alerting signal and chopping signal.

If w1*C1+w2*C2+w3*C3<n_thr*3* μ sets up, then turn to step 9.

Wherein w1, w2, w3 and μ are any real number in 0 ~ 1 scope.

Step 9,

n_cycle＝n_cycle+1。

Step 10,

Judge whether n_cycle=n_thr sets up.If be false, jump to step 3; If n_cycle=n_thr, jump to step 2.

Embodiment 2

As shown in Figure 2, the voltage that the mode adopting method two and method 3 two kinds of methods to combine obtains step a or current signal process, and concrete treatment step is as follows:

Step one, system initialization, arranging supply frequency according to actual conditions is F, arranges the threshold value n_thr of cycle index.

Step 2, to counter C2, C3 zero setting, to cycle index n_cycle zero setting;

Step 3, collection voltage or current signal;

Step 4, filtering and conditioning are carried out to signal;

Step 5, by filtering and conditioning after signal input analog-to-digital converter, the output of analog to digital converter is discrete voltage signal or current signal;

Step 6, windowed DFT is done to signal, obtain the frequency spectrum of signal.The amplitude at frequency f (i) place is designated as y (f (i)), wherein i=1,2,3 ... n.N is the discrete point number that this frequency spectrum comprises.Then the frequency spectrum obtained is processed according to following two methods respectively:

Method two, even-order harmonic accounting

1, the harmonic component of certain specific exponent number is calculated: wherein j=1,2,3 ..., F is the frequency of AC power, and the span of δ is 0 ~ 20.

2, the fundametal compoment in frequency spectrum: wherein F is the frequency of AC power, and the span of δ is 0 ~ 20.

3, the accounting of the relative fundametal compoment of harmonic component of certain specific exponent number is calculated: η _j=, j=1,2,3 ...

4, by η _jcompare with threshold value thr2.If η _j≤ thr2 then the method thinks do not have arc fault to produce in circuit; If η _j>thr2 then the method judges to create arc fault in circuit, adds one with seasonal counter C2.

Method three, spectrum analysis

1, optional m point, wherein m≤n from frequency spectrum.The frequency of this m some correspondence is denoted as f (p successively ₁), f (p ₂), f (p ₃) ... f (p _m), corresponding amplitude is followed successively by y (f (p ₁)), y (f (p ₂)), y (f (p ₃)) ... y (f (p _m)).

2, computing is according to the following formula put by the m selected: $A\left(f\left(j\right)\right)=\frac{{\log }_{\mathrm{\&alpha;}}y\left(f\left(j\right)\right)}{\mathrm{\&beta;}}+\mathrm{\&gamma;}.$ Wherein the span of j is p ₁, p ₂, p ₃p _m, α > 0 and α ≠ 1, β ≠ 0,

3, averaged: $\mathrm{result}=\frac{1}{m}*\underset{j=p_1}{\overset{p_m}{\mathrm{\&Sigma;}}}A\left(f\left(j\right)\right).$

4, result is compared with threshold value thr3.If result≤thr3, then the method is thought does not have arc fault to produce in circuit; If result>thr3, then the method judges to create arc fault in circuit, makes counter C3 add one.

Step 7,

If C2, C3 are zero, forward step 3 to;

If have in C2, C3 one non-vanishing, forward step 8 to.

Step 8,

If w2*C2+w3*C3 >=n_thr*2* μ sets up, then judge to create arc fault in circuit, send alerting signal and chopping signal.

If w2*C2+w3*C3<n_thr*2* μ sets up, then turn to step 9.

Wherein w2, w3 and μ are any real number in 0 ~ 1 scope.

Step 9,

n_cycle＝n_cycle+1。

Step 10,

Judge whether n_cycle=n_thr sets up.If be false, jump to step 3; If n_cycle=n_thr, jump to step 2.

Embodiment 3

As shown in Figure 3, the voltage that employing method a pair step a obtains or current signal process, and concrete treatment step is as follows:

Step one, system initialization, arranging supply frequency according to actual conditions is F, arranges the threshold value n_thr of cycle index.

Step 2, to counter C1 zero setting, to cycle index n_cycle zero setting;

Step 3, collection voltage or current signal;

Step 4, filtering and conditioning are carried out to signal;

Step 5, by filtering and conditioning after signal input analog-to-digital converter, the output of analog to digital converter is discrete voltage signal or current signal;

Step 6, the signal after analog to digital conversion to be processed by method a pair signal:

1, non-linear form filtering process is carried out to signal;

2, wavelet transform is done to filtered signal, obtain the detail signal d of certain one deck;

3, the effective value of d is calculated:, wherein Level is the exponent number (level of resolution of d self) of this layer;

4, D is compared with threshold value thr1.If D≤thr1, the method is thought does not have arc fault to produce in circuit; If D>thr1, the method judges to create arc fault in circuit, adds one with seasonal counter C1.

Step 7,

If C1 is zero, forward step 3 to;

If C1 is non-vanishing, forward step 8 to.

Step 8,

If C1 >=n_thr* μ sets up, then judge to create arc fault in circuit, send alerting signal and chopping signal.

If C1<n_thr* μ sets up, then turn to step 9.

Wherein μ is any real number in 0 ~ 1 scope.

Step 9,

n_cycle＝n_cycle+1。

Step 10,

Judge whether n_cycle=n_thr sets up.If be false, jump to step 3; If n_cycle=n_thr, jump to step 2.

Embodiment 4

As shown in Figure 4, the voltage adopting method two to obtain step a or current signal process, and concrete treatment step is as follows:

Step one, system initialization, arranging supply frequency according to actual conditions is F, arranges the threshold value n_thr of cycle index.

Step 2, to counter C2 zero setting, to cycle index n_cycle zero setting;

Step 3, collection voltage or current signal;

Step 4, filtering and conditioning are carried out to signal;

Step 5, by filtering and conditioning after signal input analog-to-digital converter, the output of analog to digital converter is discrete voltage signal or current signal;

Step 6, the frequency spectrum that windowed DFT obtains signal is done to the signal after analog to digital conversion.The amplitude at frequency f (i) place is designated as y (f (i)), wherein i=1,2,3 ... n.N is the discrete point number that this frequency spectrum comprises.Then processed by method two:

1, the harmonic component of certain specific exponent number is calculated: wherein j=1,2,3 ..., F is the frequency of AC power, and the span of δ is 0 ~ 20.

2, the fundametal compoment in frequency spectrum is calculated: wherein F is the frequency of AC power, and the span of δ is 0 ~ 20.

3, the accounting of the relative fundametal compoment of harmonic component of certain specific exponent number is calculated: η _j=, j=1,2,3 ...

4, by η _jcompare with threshold value thr2.If η _j≤ thr2 then the method thinks do not have arc fault to produce in circuit; If η _j>thr2 then the method judges to create arc fault in circuit, adds one with seasonal counter C2.

Step 7,

If C2 is zero, forward step 3 to;

If C2 is non-vanishing, forward step 8 to.

Step 8,

If C2 >=n_thr* μ sets up, then judge to create arc fault in circuit, send alerting signal and chopping signal.

If C2<n_thr* μ sets up, then turn to step 9.

Wherein μ is any real number in 0 ~ 1 scope.

Step 9,

n_cycle＝n_cycle+1。

Step 10,

Judge whether n_cycle=n_thr sets up.If be false, jump to step 3; If n_cycle=n_thr, jump to step 2.

Embodiment 5

As shown in Figure 5, the voltage that employing method three couples of step a obtain or current signal process, and concrete treatment step is as follows:

Step one, system initialization, arranging supply frequency according to actual conditions is F, arranges the threshold value n_thr of cycle index.

Step 2, to counter C3 zero setting, to cycle index n_cycle zero setting;

Step 3, collection voltage or current signal;

Step 4, filtering and conditioning are carried out to signal;

Step 5, by filtering and conditioning after signal input analog-to-digital converter, the output of analog to digital converter is discrete voltage signal or current signal;

Step 6, windowed DFT is done to the signal after analog to digital conversion, obtain the frequency spectrum of signal.The amplitude at frequency f (i) place is designated as y (f (i)), wherein i=1,2,3 ... n.N is the discrete point number that this frequency spectrum comprises.Then the frequency spectrum obtained is processed according to method three:

1, optional m point, wherein m≤n from frequency spectrum.The frequency of this m some correspondence is denoted as f (p successively ₁), f (p ₂), f (p ₃) ... f (p _m), corresponding amplitude is followed successively by y (f (p ₁)), y (f (p ₂)), y (f (p ₃)) ... y (f (p _m)).

2, computing is according to the following formula put by the m selected: $A\left(f\left(j\right)\right)=\frac{{\log }_{\mathrm{\&alpha;}}y\left(f\left(j\right)\right)}{\mathrm{\&beta;}}+\mathrm{\&gamma;}.$ Wherein the span of j is p ₁, p ₂, p ₃p _m, α > 0 and α ≠ 1, β ≠ 0,

3, averaged: $\mathrm{result}=\frac{1}{m}*\underset{j=p_1}{\overset{p_m}{\mathrm{\&Sigma;}}}A\left(f\left(j\right)\right).$

4, result is compared with threshold value thr3.If result≤thr3, then the method is thought does not have arc fault to produce in circuit; If result>thr3, then the method judges to create arc fault in circuit, makes counter C3 add one.

Step 7,

If C3 is zero, forward step 3 to;

If C3 is non-vanishing, forward step 8 to.

Step 8,

If C3 >=n_thr* μ sets up, then judge to create arc fault in circuit, send alerting signal and chopping signal.

If C3<n_thr* μ sets up, then turn to step 9.

Wherein μ is any real number in 0 ~ 1 scope.

Step 9,

n_cycle＝n_cycle+1。

Step 10,

Judge whether n_cycle=n_thr sets up.If be false, jump to step 3; If n_cycle=n_thr, jump to step 2.

Technique scheme only embodies the optimal technical scheme of technical solution of the present invention, and those skilled in the art all embody principle of the present invention to some variations that wherein some part may be made, and belong within protection scope of the present invention.

Claims (4)

1. an alternating current arc fault detection method, is characterized in that, comprises the following steps:

A, gather voltage signal in loop to be detected or current signal by mutual inductor, and filtering and conditioning are carried out to the voltage signal collected or current signal, by the signal input analog-to-digital converter after filtering and conditioning, the output of analog to digital converter is discrete voltage signal or current signal;

The method of the voltage signal that b, design treatment are obtained by step a or current signal, this disposal route comprises:

Method one, voltage signal or current signal carried out to wavelet transformation and decompose and obtain approximate signal and detail signal, come whether occurred arc fault in decision circuit by the contrast of detail signal and threshold value;

Method two, the frequency spectrum that windowing DFT algorithm obtains signal is done to voltage signal or current signal, the harmonic component calculating certain specific exponent number, relative to fundametal compoment proportion, comes whether occurred arc fault in decision circuit by the contrast of this ratio and threshold value;

Method three, the frequency spectrum that windowing DFT algorithm obtains signal is done to voltage signal or current signal, from frequency spectrum, choose several carry out computing according to the algorithm preset, come whether occurred arc fault in decision circuit by the contrast of operation result and threshold value;

The voltage signal that in c, selecting step b, the combination of any one method or any two methods or the common combination of three methods obtain step a or current signal process, thus obtain result of determination.

2. alternating current arc fault detection method according to claim 1, is characterized in that, method one particular content in described step b is as follows:

Non-linear form filtering process is carried out to the voltage signal obtained in step a or current signal;

Wavelet transform is done to filtered signal, obtains the detail signal d of certain one deck;

Calculate the effective value of d:, wherein Level is the exponent number (level of resolution of d self) of this layer, for the discrete of detail signal d is counted;

D is compared with threshold value thr1.If D≤thr1, the method is thought does not have arc fault to produce in circuit; If D>thr1, the method judges to create arc fault in circuit.

3. alternating current arc fault detection method according to claim 1, is characterized in that, the method two particular content in described step b is as follows:

The frequency spectrum that windowed DFT obtains signal is done to signal.The amplitude at frequency f (i) place is designated as y (f (i)), wherein i=1,2,3 ... n.N is the discrete point number that this frequency spectrum comprises.

Calculate the harmonic component of certain specific exponent number: wherein j=1,2,3 ..., F is the frequency of AC power, and the span of δ is 0 ~ 20.

Calculate the fundametal compoment in frequency spectrum: wherein F is the frequency of AC power, and the span of δ is 0 ~ 20.

Calculate the accounting of the relative fundametal compoment of harmonic component of certain specific exponent number: η _j=, j=1,2,3 ...

By η _jcompare with threshold value thr2.If η _j≤ thr2 then the method thinks do not have arc fault to produce in circuit; If η _j>thr2 then the method judges to create arc fault in circuit.

4. alternating current arc fault detection method according to claim 1, is characterized in that, method three particular content in described step b is as follows:

The frequency spectrum that windowed DFT obtains signal is done to signal.The amplitude at frequency f (i) place is designated as y (f (i)), wherein i=1,2,3 ... n.N is the discrete point number that this frequency spectrum comprises.

Optional m point, wherein m≤n from frequency spectrum.The frequency of this m some correspondence is denoted as f (p successively ₁), f (p ₂), f (p ₃) ... f (p _m), corresponding amplitude is followed successively by y (f (p ₁)), y (f (p ₂)), y (f (p ₃)) ... y (f (p _m)).

Computing is according to the following formula put by the m selected:

$A\left(f\left(j\right)\right)=\frac{{\log }_{\mathrm{\&alpha;}}y\left(f\left(j\right)\right)}{\mathrm{\&beta;}}+\mathrm{\&gamma;}.$ Wherein the span of j is p ₁, p ₂, p ₃p _m, α > 0 and α ≠ 1, β ≠ 0,

Averaged: $\mathrm{result}=\frac{1}{m}*\underset{j=p_1}{\overset{p_m}{\mathrm{\&Sigma;}}}A\left(f\left(j\right)\right).$

Result is compared with threshold value thr3.If result≤thr3, then the method is thought does not have arc fault to produce in circuit; If result>thr3, then the method judges to create arc fault in circuit.