CN109490701B - A power frequency series arc fault detection method - Google Patents

Abstract

The invention discloses a power frequency series arc fault detection method, which belongs to the field of circuit operation protection. The present invention obtains the spectrogram by collecting the current data in the circuit and performing fast Fourier transform on it; then, the amplitudes of the fundamental wave and the harmonics of each order are counted, and the ratio of the amplitude of the harmonics of each order to the amplitude of the fundamental wave is calculated. ; Finally, add the ratio of each harmonic amplitude to the fundamental wave amplitude into the built-in matrix, and then perform principal component analysis, that is, PCA calculation; derive the obtained principal component matrix, and compare it with the given threshold to determine the load category and At the same time, it also calculates the zero zone length of the obtained raw current data, and also compares it with the threshold value. When these two conditions are met at the same time, it can be considered that an arc fault has occurred. This method can identify different loads and has relatively high performance. High accuracy can effectively protect the circuit and ensure that users can use electricity safely.

Description

A power frequency series arc fault detection method

technical field

The invention relates to a power frequency series arc fault detection method, which belongs to the field of circuit operation protection.

Background technique

Arc is a self-sustaining discharge phenomenon that emits intense light and heat, and an arc fault occurs when an unintended arc occurs in a live line. Arc faults include series arc faults, parallel arc faults and compound arc faults. Among them, the fault currents of parallel arc faults and compound arc faults are very large and are easily disconnected by circuit breakers, but the current of series arc faults is small and difficult to detect. Once it happens, it will cause great loss of life and property.

According to the statistics of the Fire Department of the Ministry of Public Security, in 2016 alone, a total of 312,000 fires were reported nationwide, with 1,582 deaths, 1,065 injuries and direct property losses of 3.72 billion yuan. Electric fires caused by regulations, etc. accounted for 30.4% of the total, and in large fires, this proportion rose to 50%. Among these electrical fires, the fire caused by the fault arc is the most dangerous and the fire with the highest occurrence rate.

At present, the research on arc fault protection technology in the field of low-voltage power distribution in my country is still blank. There is no database of arc faults in my country. The market-oriented production of AFCI (arc fault circuit breaker) is still in its infancy. The low-voltage series arc fault detection technology has great application prospects.

SUMMARY OF THE INVENTION

The invention proposes a power frequency series arc fault detection method. The method collects the current in the circuit through a sensor, and then uses the fast Fourier transform to process the waveform data to obtain its spectrum information, and then counts the harmonics and fundamentals of each order. The amplitude ratio of the wave is calculated and the principal component analysis is performed to reduce the dimension to obtain new eigenvalues. At the same time, the proportion of the zero area is added as an auxiliary criterion to judge the load type and whether an arc fault has occurred.

The present invention adopts following technical scheme for solving its technical problem:

A power frequency series arc fault detection method, comprising the following steps:

Step S1, collecting current data in the circuit;

Step S2, obtain its harmonic amplitude value of each order and the ratio R _n of the harmonic amplitude value to the fundamental wave amplitude value from the collected current data;

Step S3, adding the obtained harmonic amplitude ratio R _n to the built-in data set for PCA calculation;

Step S4, derive the characteristic value of the principal component corresponding to R _n and compare it with the threshold value to judge the load type and operating state, if it is a fault state, proceed to the next step, if it is a normal state, then return to step S1;

Step S5, execute the zero zone auxiliary criterion;

Step S6, if the zero zone auxiliary criterion is satisfied, it is determined that an arc fault has occurred, and the load type is given.

In step S1, the current data in the circuit is collected in real time through the sensor, and the sampling frequency of the sensor is fs≥1kHz.

In step S2, the specific execution steps are as follows:

Step S2.1, perform FFT calculation on the collected current data;

Step S2.2, count the fundamental wave amplitude I _m1 and each harmonic amplitude I _mn , where n=2-10, and n is a positive integer;

其中n＝2-10，且n取正整数。Step S2.3, calculate the ratio of the amplitude of each harmonic to the amplitude of the fundamental wave

where n=2-10, and n is a positive integer.

In step S3, the specific execution steps are as follows:

Step S3.1, adding the obtained amplitude ratio to the built-in data set to obtain a new data set D;

Step S3.2, perform PCA calculation on the obtained data set D to obtain a principal component matrix;

Step S3.3, calculate the covariance matrix Y=Cov(D) of the data set D;

Step S3.4, calculate the eigenvectors and eigenvalues of the covariance matrix Y;

Step S3.5, arrange according to the size of the eigenvalues, and reserve 2-3 eigenvectors to form a new matrix U;

Step S3.6, calculate a new principal component matrix C=Y ^T D.

In step S4, the specific execution steps are as follows:

Step S4.1, derive the principal component eigenvalue C _m corresponding to R _n ;

Step S4.2, compare C ₁ -C _m , determine the load type and operating state, if it is a fault state, execute the next step, if it is a normal state, return to step S1;

In S5, the specific steps are as follows:

Step S5.1, obtain the absolute value of the collected waveform data;

Step S5.2, obtain the waveform maximum value I _max for the waveform after obtaining the absolute value;

Step S5.3, calculate the waveform threshold I _T =I _max /10;

Step _S5.4 , count the sum S2 of all sampling points below the threshold IT in the original current data _;

Step S5.5, calculate K=S ₂ /S, where S is the total number of sampling points of the original current data.

The specific process of step S6 is as follows. The specific judgment condition is whether the proportion K of the zero area is greater than or equal to 0.11. If the condition is met, it is judged that an arc fault has occurred and the load type is given according to the result of S4; if the condition is not met, then return to step S1 .

The beneficial effects of the present invention are as follows:

Based on the spectrum information of the current signal, the present invention performs principal component analysis and calculation on the current signal to obtain new eigenvalues, which can distinguish different loads and different operating states. The angle analyzes and calculates the current waveform, and establishes different thresholds to ensure the accuracy of the detection effect and ensure that the arc fault can be detected quickly and efficiently.

Description of drawings

Fig. 1 is the flow chart of the method of the present invention.

Figure 2 is a flow chart of harmonic and ratio calculation.

Figure 3 is a flow chart of PCA calculation.

Figure 4 is a flow chart of the zero-zone auxiliary criterion.

Fig. 5 is the current waveform obtained by a certain sampling under a certain load.

Figure 6 is a current waveform FFT spectrum diagram.

Detailed ways

The present invention is described in detail below in conjunction with the accompanying drawings:

As shown in Figure 1, the present invention proposes a power frequency series arc fault detection method by performing PCA (principal component analysis) calculation on the frequency spectrum of the current signal to obtain eigenvalues, and at the same time calculating the proportion of the zero area of the waveform. The method includes the following steps:

Step S1, collecting current data in the circuit;

In step S1, in order to ensure the accuracy of the fast Fourier transform result in step S2, according to the Nyquist sampling theorem, the sampling frequency should be greater than or equal to twice the highest frequency, and the sampling frequency of the sensor f _s ≥ 1 kHz.

Step S2, obtain its harmonic amplitude value of each order and the ratio R _n of the harmonic amplitude value to the fundamental wave amplitude value from the collected current data;

In step S2, the specific execution steps are shown in Figure 2:

Step S2.1, performing FFT (Fast Fourier Transform) calculation on the collected current data;

Step S2.2, the fundamental wave amplitude I _m1 and the harmonic amplitudes I _mn of each order are counted (n=2-10, n is a positive integer).

(n＝2-10，n取正整数)。Step S2.3, calculate the ratio of the amplitude of each harmonic to the amplitude of the fundamental wave

(n=2-10, n is a positive integer).

Step S3, adding the obtained harmonic amplitude ratio R _n to the built-in data set for PCA calculation;

In step S3, the specific execution steps are shown in Figure 3:

Step S3.1, adding the obtained amplitude ratio to the built-in data set to obtain a new data set D;

Step S3.2, perform PCA calculation on the obtained data set D to obtain a principal component matrix;

Step S3.3, calculate the covariance matrix Y=Cov(D) of the data set D;

Step S3.4, calculate the eigenvectors and eigenvalues of the covariance matrix Y;

Step S3.5, arrange according to the size of the eigenvalues, and reserve appropriate M eigenvectors to form a new matrix U;

Step S3.6, calculate a new principal component matrix C=Y ^T D, where T represents the transposition, that is, Y ^T is the transposition matrix of Y;

Step S4, derive the principal component characteristic value corresponding to R _n and compare it with the threshold value to judge the load type and operating state;

In step S4, the specific execution steps of threshold comparison are as follows:

Step S4.1, derive the principal component eigenvalue C _m corresponding to R _n ;

Step S4.2, compare C ₁ -C _m , determine the load type and operating state, if it is a fault state, execute the next step, if it is a normal state, return to step S1;

Step S5, execute the zero zone auxiliary criterion;

In step S5, the steps of the execution part of the zero-zone auxiliary criterion are shown in Figure 4:

Step S5.1, obtain the absolute value of the collected waveform data;

Step S5.2, obtain the waveform maximum value I _max for the waveform after obtaining the absolute value;

Step S5.3, calculate the waveform threshold I _T =I _max /10;

Step S5.4, count the sum _{S 2} of all sampling points below the threshold IT in the original current data;

Step S5.5, calculate K=S ₂ /S, where S is the total number of sampling points of the original current data;

Step S6, if the zero zone auxiliary criterion is satisfied, it can be determined that an arc fault has occurred, and the load type is given;

In step S6, the specific judgment condition is whether K is greater than or equal to 0.11. If the condition is met, it is judged that an arc fault has occurred and the load type is given according to the result of S4; if the condition is not met, return to step S1.

The method is described in detail below in conjunction with examples, but should not be construed as the scope of the above-mentioned subject matter of the present invention is limited to the following embodiments, and all technologies realized based on the content of the present invention belong to the scope of the present invention.

First collect the current data in the circuit, the sampling frequency is f _s =1.25MHz, the number of sampling cycles is 5, the number of sampling points is S = 2500, and the obtained waveform is shown in Figure 5.

Next, perform fast Fourier transform on the waveform data to obtain a spectrogram as shown in Figure 6, and export the spectrogram to a table as shown in Table 1.

Table 1. Spectrum results of sampled waveform after FFT

Frequency (Hz) Amplitude (A) Frequency (Hz) Amplitude (A) Frequency (Hz) Amplitude (A) Frequency (Hz) Amplitude (A) Frequency (Hz) Amplitude (A) 10 0.003168 110 0.004428 210 0.004287 310 0.004077 410 0.002127 20 0.002316 120 0.003433 220 0.002707 320 0.002879 420 0.001675 30 0.00116 130 0.002544 230 0.004547 330 0.003619 430 0.004582 40 0.378712 140 0.024081 240 0.026295 340 0.015104 440 0.015706 50 0.750856 150 0.044535 250 0.04308 350 0.024773 450 0.023847 60 0.372113 160 0.024322 260 0.022147 360 0.012833 460 0.011267 70 0.001746 170 0.002497 270 0.00218 370 0.000928 470 0.001188 80 0.00199 180 0.001823 280 0.001342 380 0.00092 480 0.001798 90 0.004317 190 0.003522 290 0.002418 390 0.002847 490 0.003321 100 0.004859 200 0.004194 300 0.003721 400 0.003915 500 0.003268

This table is used to count the amplitude of the fundamental wave and the amplitude of each harmonic, and the amplitude of each harmonic and the fundamental wave after statistics and the ratio of the amplitude of each harmonic to the fundamental wave are calculated as shown in the following table:

Table 2 Fundamental and harmonic amplitudes and ratios

For the convenience of calculation and export, the ratio R _{n is} added to the last row of the built-in data set to obtain a new data set D. The built-in data set is the sample data of different operating states of different loads. Some of the data are shown in the following table. The row is a sample, each column is a ratio of the amplitude of the harmonic to the fundamental wave, starting from the ratio of the second harmonic to the fundamental wave amplitude of 2/1, a total of 9 columns:

Table 3 Built-in datasets (parts)

2/1 3/1 4/1 5/1 6/1 7/1 8/1 9/1 10/1 0.004605 0.571571 0.000499 0.340038 0.00207 0.256996 0.001321 0.172801 0.000796 0.003671 0.571057 0.00114 0.339484 0.001544 0.262031 0.000998 0.172926 0.002288 0.003524 0.571288 0.000994 0.340813 0.002402 0.257808 0.000889 0.172095 0.003141 0.003117 0.576952 0.003047 0.340024 0.002761 0.258102 0.000527 0.163757 0.002829 0.002848 0.57335 0.001823 0.343117 0.001604 0.263931 0.002886 0.174907 0.001257 0.004722 0.567805 0.001068 0.345945 0.00287 0.259424 0.003018 0.177468 0.000514 3.15E-03 0.578271 0.001963 0.34479 0.00153 0.268077 0.001707 0.172359 0.003361 0.001736 0.59684 0.000837 0.32651 0.000723 0.266068 0.001978 0.145804 0.001597

First, the covariance matrix Y of the data set D is calculated. The specific calculation formula is as follows:

Among them, X _n is the column vector of the data set D, and its value is 1-9. The covariance matrix obtained after substituting the data for calculation is as follows:

Since there are 9 columns of vectors, the resulting covariance matrix is a 9×9 matrix.

The next step is to calculate the eigenvalues and eigenvectors of the covariance matrix and arrange them according to their size. The results are shown in the following table:

Table 4 Eigenvalues

Eigenvalues 0.30134 0.03501 0.01069 0.00899 0.00297 0.00106 0.00005 0.00028 0.00027

Table 5 Eigenvectors

1 2 3 4 5 6 7 8 9 -0.277 -0.394 0.330 -0.575 -0.436 0.221 0.095 0.236 0.161 -0.536 0.473 0.640 0.138 0.221 0.088 -0.004 -0.028 -0.063 -0.316 -0.430 0.031 -0.153 0.117 -0.385 -0.208 -0.555 -0.423 -0.421 0.297 -0.345 0.109 -0.545 -0.461 -0.041 0.277 -0.127 -0.311 -0.368 -0.066 0.216 0.395 -0.263 -0.248 0.320 0.573 -0.295 0.267 -0.387 -0.242 -0.019 0.209 0.076 -0.564 0.515 -0.283 -0.296 -0.145 0.356 0.067 0.141 0.803 -0.003 -0.127 -0.209 0.138 -0.394 -0.475 0.500 0.195 -0.015 0.372 -0.361 -0.228 -0.188 -0.179 0.398 -0.200 0.642 -0.484 -0.008 -0.187

In the next step, in order to ensure the accuracy of the judgment, three eigenvectors are reserved, that is, the first three columns form a new matrix U, then the principal component matrix C=Y ^T D, and the calculated results are shown in the following table:

Table 6 Principal Component Matrix

C₁ C₂ C₃ -0.56407 -0.46019 -0.08109 -0.5651 -0.46124 -0.07848 -0.56471 -0.46019 -0.07989 -0.56623 -0.46087 -0.08697 -0.56914 -0.46383 -0.07683 -0.5671 -0.46041 -0.07367 -0.57342 -0.46721 -0.0787 -0.56822 -0.46797 -0.10797 -0.57419 -0.46612 -0.08546 -0.57028 -0.46293 -0.07535 … … … -0.07987 -0.0491 0.00679

Finally, the eigenvalues of the output components corresponding to R _n are derived, that is, the eigenvalues C ₁ =-0.07987, C ₂ =-0.0491, and C ₃ =0.00679 of the last row. The threshold table is as follows:

Table 7 Threshold comparison table

It can be seen from the comparison that the waveform is the fault operation state of the resistive load, so the auxiliary criterion of zero zone is implemented.

The amplitude of the original current waveform is 0.78A, so the waveform threshold I _T =I _max /10=0.78/10=0.078A, the sum of the points whose absolute value is less than the waveform threshold I _T in the statistical raw current data S ₂ =422, calculate The proportion of the zero area K=S ₂ /S=422/2500=0.169, which is greater than the given threshold value of 0.11, so it is determined that an arc fault has occurred, and the load is a resistive load.

The above content is a further detailed description of the present invention in conjunction with specific embodiments, and it cannot be considered that the specific implementation of the present invention is limited to these descriptions. Those of ordinary skill in the technical field to which the present invention pertains should know that, without departing from the concept of the present invention, several simple deductions or substitutions can be made, which should be regarded as belonging to the protection scope of the present invention.

Claims (5)

1. A power frequency series arc fault detection method is characterized by comprising the following steps:

step S1, collecting current data in the circuit;

step S2, calculating the harmonic amplitude and the ratio R of the harmonic amplitude and the fundamental amplitude of the collected current data_n；

Step S3, obtaining the ratio R of the harmonic amplitude to the fundamental amplitude_nAdding a built-in data set to perform PCA calculation; the specific implementation steps are as follows:

step S3.1, obtaining the ratio R of the harmonic amplitude to the fundamental amplitude_nAdding the built-in data set to obtain a new data set D;

s3.2, carrying out PCA calculation on the obtained data set D to obtain a principal component matrix;

step S3.3, calculating a covariance matrix Y ═ cov (D) of the dataset D;

s3.4, calculating an eigenvector and an eigenvalue of the covariance matrix Y;

step S3.5, arranging according to the magnitude of the eigenvalue, and reserving M eigenvectors to form a new matrix U, wherein M is 2-3;

step S3.6, calculating a new principal component matrix C ═ UD;

step S4, derive R_nComparing the corresponding principal component characteristic value with a threshold value, judging the load type and the running state, if the load type and the running state are in a fault state, carrying out the next step, and if the load type and the running state are in a normal state, returning to the step S1;

step S5, executing zero zone auxiliary criterion; the specific steps are as follows:

s5.1, calculating an absolute value of the acquired waveform data;

step S5.2, the maximum value I of the waveform is obtained from the waveform with the absolute value obtained_max；

Step S5.3, calculating a waveform threshold I_T＝I_max/10；

Step S5.4, counting all the data of the original current which are lower than the current valueThreshold value I_TSum of sampling points S₂；

Step S5.5, calculate the zero zone ratio K ═ S₂S, wherein S is the total sampling point number of the original current data;

and step S6, if the zero zone auxiliary criterion is met, judging that the arc fault occurs, and giving the load type.

2. The power frequency series arc fault detection method according to claim 1, characterized in that: in step S1, current data in the circuit is collected in real time through a sensor, and the sampling frequency fs of the sensor is more than or equal to 1 kHz.

3. The power frequency series arc fault detection method according to claim 1, characterized in that: in step S2, the specific implementation steps are as follows:

s2.1, carrying out FFT calculation on the acquired current data;

step S2.2, statistics of fundamental wave amplitude I_m1And amplitude of each harmonic I_mnWherein n is 2-10, and n is a positive integer;

step S2.3, calculating the ratio of each harmonic amplitude to the fundamental amplitude

Wherein n is 2-10, and n is a positive integer.

4. The power frequency series arc fault detection method according to claim 1, characterized in that: in step S4, the specific steps are as follows:

step S4.1, derive R_nCorresponding principal component characteristic value C_m，m＝1-M；

Step S4.2, compare C₁-C_MThe load type and the operation state are determined, and if the load type and the operation state are in the failure state, the next step is executed, and if the load type and the operation state are in the normal state, the operation returns to the step 1.

5. The power frequency series arc fault detection method according to claim 1, characterized in that: step S6 is carried out in the following specific process, the specific judgment condition is whether the zero zone occupation ratio K is more than or equal to 0.11, if the condition is met, the arc fault is judged to occur, and the load type is given according to the result of S4; if the condition is not satisfied, the process returns to step S1.

Images