VECTOR ANALYSIS CALCULATION-BASED ARC CROSSTALK SIGNAL IDENTIFICATION METHOD

Abstract

A vector analysis calculation-based arc crosstalk signal identification method. A new sampling circuit manner is proposed in the method, wherein a current signal is sampled on zero and live lines, and the signal is converted into two digital signals with a sampling rate of 200 MHz by means of a dual-channel ADC, and the digital signals are sent to a hardware digital signal processing unit. Five pass bands are selected to perform band-pass filtering on the two signals separately. Time-sharing processing and vector analysis are performed on the filtered signals, and the amplitude ratio and fluctuation characteristics of two resistor terminal voltages, as well as the phase difference between shunt resistor and inductor terminal voltage signals are extracted as crosstalk feature quantities. According to a zero-crossing signal, a system segments the feature quantities extracted by a hardware processing module and sends same to a neural network for classification and determination.

Claims

1-9. (canceled)

10. A sampling circuit, comprising: a first resistor R.sub.1, a second resistor R.sub.2, a third resistor R.sub.3, an inductor L and a capacitor C, wherein the inductor L is connected in series between a first measurement point and a second measurement point in a live line, the third resistor R.sub.3 is connected in parallel with the inductor L, the first resistor R.sub.1 is connected in series between the first measuring point and a third measuring point, the second resistor R.sub.2 is connected in series between the second measuring point and the third measuring point, the third measuring point is connected to a neutral line via the capacitor C, and a load is connected between the live line and the neutral line; and in performing sampling, a voltage signal of the inductor between the first measuring point and the second measuring point and a voltage signal of the first resistor between the first measuring point and the third measuring point are collected respectively to analyze whether an arc fault occurs in the sampled line.

11. A method for identifying an arc crosstalk signal based on vector analysis calculation, comprising: performing sampling, by the sampling circuit according to claim 1, to obtain a voltage signal of an inductor and voltage signal of a first resistor; calculating a voltage signals of a second resistor based on the voltage signal of the inductor and the voltage signal of the first resistor, determining the voltage signal of the first resistor as a first sampling signal, determining the voltage signal of the second resistor as a second sampling signal, and determining the voltage signal of the inductor as a third sampling signal; and determining whether a signal in a sampled line is an arc signal or a crosstalk signal based on an amplitude ratio feature vector, an amplitude fluctuation feature vector, an phase difference feature vector and/or a phase fluctuation feature vector, wherein the amplitude ratio feature vector is equal to a ratio of a voltage effective value of the first sampling signal to a voltage effective value of the second sampling signal, the amplitude fluctuation feature vector represents a magnitude relationship between a voltage amplitude fluctuation of the first sampling signal and a voltage amplitude fluctuation of the second sampling signal, the phase difference feature vector is equal to a difference between a phase of the first sampling signal and a phase of the third sampling signal, and the phase fluctuation feature vector is equal to a difference between a phase fluctuation of the first sampling signal and a phase fluctuation of the third sampling signal.

12. The method for identifying an arc crosstalk signal based on vector analysis calculation according to claim 11, wherein the determining whether a signal in a sampled line is an arc signal or a crosstalk signal based on an amplitude ratio feature vector, an amplitude fluctuation feature vector, an phase difference feature vector and a phase fluctuation feature vector comprises: determining, by a trained neural network model, whether the signal in the sampled line is the arc signal or the crosstalk signal based on an amplitude ratio feature vector, an amplitude fluctuation feature vector, an phase difference feature vector and a phase fluctuation feature vector.

13. The method for identifying an arc crosstalk signal based on vector analysis calculation according to claim 11, wherein before determining whether the signal in the sampled line is the arc signal or the crosstalk signal, the method further comprises: performing analog-to-digital conversion on the voltage signal of the inductor and the voltage signal of the first resistor; filtering, by several band-pass digital filters, the voltage signal of the inductor after analog-to-digital conversion and the voltage signal of the first resistor after analog-to-digital conversion; and performing adaptive gain adjustment on the filtered voltage signal of the inductor and the filtered voltage signal of the first resistor based on amplitudes of the filtered voltage signal of the inductor and the filtered voltage signal of the first resistor, wherein a gain multiple for the voltage signal of the inductor in a frequency band is same as a gain multiple for the voltage signal of the first resistor in the frequency band.

14. The method for identifying an arc crosstalk signal based on vector analysis calculation according to claim 11, wherein the determining whether a signal in a sampled line is an arc signal or a crosstalk signal based on an amplitude ratio feature vector comprises: determining that the signal in the sampled line is the arc signal in a case that the ratio of the voltage effective value of the first sampling signal to the voltage effective value of the second sampling signal is less than a predetermined threshold, and determining that the signal in the sampled line is the crosstalk signal in a case that the ratio of the voltage effective value of the first sampling signal to the voltage effective value of the second sampling signal is greater than a predetermined threshold.

15. The method for identifying an arc crosstalk signal based on vector analysis calculation according to claim 11, wherein the determining whether a signal in a sampled line is an arc signal or a crosstalk signal based on an amplitude fluctuation feature vector comprises: determining that the signal in the sampled line is the arc signal in a case that the voltage amplitude fluctuation of the first sampling signal is less than the voltage amplitude fluctuation of the second sampling signal, and determining that the signal in the sampled line is the crosstalk signal in a case that the voltage amplitude fluctuation of the first sampling signal is greater than the voltage amplitude fluctuation of the second sampling signal.

16. The method for identifying an arc crosstalk signal based on vector analysis calculation according to claim 11, wherein the determining whether a signal in a sampled line is an arc signal or a crosstalk signal based on a phase difference feature vector comprises: determining that the signal in the sampled line is the arc signal in a case that the difference between the phase of the first sampling signal and the phase of the third sampling signal is greater than 0° and ranges from 0° to 180°, and determining that the signal in the sampled line is the crosstalk signal in a case that the difference between the phase of the first sampling signal and the phase of the third sampling signal is less than 0° and ranges from −90° to 0°.

17. The method for identifying an arc crosstalk signal based on vector analysis calculation according to claim 11, wherein the determining whether a signal in a sampled line is an arc signal or a crosstalk signal based on the phase fluctuation feature vector comprises: determining that the signal in the sampled line is the arc signal in a case that the phase of the first sampling signal is ahead of the phase of the third sampling signal, and determining that the signal in the sampled line is the crosstalk signal in a case that the phase of the first sampling signal lags behind the phase of the third sampling signal.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The present disclosure is further described below in combination with the accompanying drawings.

[0021] FIG. 1 shows a flow chart of a crosstalk feature extraction system according to the present disclosure;

[0022] FIG. 2 shows an equivalent circuit analysis in a case that an arc occurs in a main circuit under an ideal condition;

[0023] FIG. 3 shows an equivalent circuit analysis in a case that an arc occurs in a bypass circuit under an ideal condition;

[0024] FIG. 4 shows variation curves of a ratio of an amplitude of {dot over (U)}.sub.1 and an amplitude of {dot over (U)}.sub.2 with variations of a value of an inductor in a sampling circuit in a case that an arc occurs in a bypass circuit and variation curves of a ratio of an amplitude of {dot over (U)}.sub.1 and an amplitude of {dot over (U)}.sub.2 with variations of a value of an inductor in a sampling circuit in a case that an arc occurs in a main circuit;

[0025] FIG. 5 shows variation curves of a difference of a phase of {dot over (U)}.sub.1 and a phase of {dot over (U)}.sub.2 with variations of a value of an inductor in a sampling circuit in a case that an arc occurs in a bypass circuit and variation curves of a difference of a phase of {dot over (U)}.sub.1 and a phase of {dot over (U)}.sub.2 with variations of a value of an inductor in a sampling circuit in a case that an arc occurs in a main circuit;

[0026] FIG. 6 is a flow chart of calculating a ratio of amplitudes according to the present disclosure;

[0027] FIG. 7 shows a flow chart of calculating a difference between phases according to the present disclosure;

[0028] FIG. 8 shows a flow chart of calculating amplitude fluctuation according to the present disclosure; and

[0029] FIG. 9 shows a flow chart of calculating phase fluctuation according to the present disclosure.

DETAILED DESCRIPTION OF THE EMBODIMENTS

[0030] Hereinafter, a method for identifying an arc crosstalk according to the present disclosure is described in combination with FIGS. 1 to 9.

[0031] FIG. 1 shows a process of identifying features of a crosstalk signal based on vector analysis calculation according to the present disclosure. As shown in FIG. 1, an adopted sampling circuit includes: a first resistor R.sub.1, a second resistor R.sub.2 a third resistor R.sub.3 an inductor L and a capacitor C. The inductor C is connected in series between a first measurement point 1 and a second measurement point 2 in a live line, the third resistor R.sub.3 is connected in parallel with the inductor C, the first resistor R.sub.1 is connected in series between the first measuring point 1 and a third measuring point 3, the second resistor R.sub.2 is connected in series between the second measuring point 2 and the third measuring point 3, the third measuring point 3 is connected to a neutral line via the capacitor C, and a load is connected between the live line and the neutral line. In performing sampling, a voltage signal of the inductor between the first measuring point 1 and the second measuring point 2 and a voltage signal of the first resistor between the first measuring point 1 and the third measuring point 3 are collected respectively to analyze whether an arc fault occurs in the sampled line. The system performs operations mainly based on processing of a hardware digital signal processing system, including the following steps 1 to 5.

[0032] In step 1, a Y-type sampling circuit according to the present disclosure is constructed, AD sampling, with a sampling rate up to 200 MHz, is performed on signals of the resistor R1 and the inductor L to obtain, then the digital signals y.sub.R1 (n) and y.sub.L(n) are transmitted to the hardware digital signal processing system for performing vector analysis in real time, and then an amplitude ratio feature vector, an amplitude fluctuation feature vector, a phase difference feature vector and a phase fluctuation feature vector are calculated.

[0033] In step 2, the signals y.sub.R1(n) and y.sub.L(n) are respectively filtered by band-pass digital filters. Each of the filters may be designed as a 64-order filter. Pass frequency bands include a frequency band ranging from 5 MHz to 10 MHz, a frequency band ranging from 15 MHz to 20 MHz, a frequency band ranging from 25 MHz to 30 MHz, a frequency band ranging from 35 MHz to 40 MHz, and a frequency band ranging from 45 MHz to 50 MHz. h(n) represents a unit impulse response of a digital filter. The filtered signal is expressed as:

[00009]$\{\begin{array}{c}{y}_{R1\_\mathrm{FIR}}\left(n\right)=y_{R1}\left(n\right)*h\left(n\right)=\stackrel{+\infty }{\underset{k=-\infty }{.Math.}}h\left(k\right).Math.y_{R1}\left(n-k\right)\\ y_{L\_\mathrm{FIR}}\left(n\right)=y_L\left(n\right)*h\left(n\right)=\stackrel{+\infty }{\underset{k=-\infty }{.Math.}}h\left(k\right).Math.y_L\left(n-k\right)\end{array}$

[0034] After performing filtering, adaptive gain adjustment may be performed based on the amplitude of the filtered signal, amplifying a weak arc signal, preventing a digital signal overflow, and thereby ensuring the reliability of extracting features of the arc crosstalk.

[0035] In step 3, time-sharing processing is performed on the filtered data from two channels, vector analysis is performed on the signal within every 20 us to calculate an amplitude ratio feature vector and a phase difference feature vector according to the present disclosure.

[0036] The amplitude ratio feature vector is calculated by obtaining a ratio of effective values of amplitudes of the signals of the resistors R.sub.1 and R.sub.2 in each of time periods. As shown in the flow chart of the system processing in FIG. 6, signal sequences of the resistor R.sub.1 and the inductor L are respectively set as:

[00010]$\{\begin{array}{c}{y}_{R1\_\mathrm{FIR}}=\left\{a_1,a_2,.Math.,a_N\right\}\\ y_{L\_\mathrm{FIR}}=\left\{b_1,b_2,.Math.,b_N\right\}\end{array}$

[0037] Based on the Y-type circuit, the resistor R.sub.1 resistor R.sub.2, and inductor L form a closed-loop triangle. Based on a Kirchhoff s Voltage rule, a signal sequence of the resistor R.sub.2 may be calculated as the following expression:

y.sub.R2_FIR={b.sub.1−a.sub.1, b.sub.2−a.sub.2, . . . ,b.sub.N−a.sub.N}

[0038] (1) Squaring operation is respectively performed on the signal sequence of the resistor R.sub.1 and on the signal sequence of the resistor R.sub.2 by using the following equations:

[00011]$\{\begin{array}{c}{y}_{R1\_\mathrm{square}}=\left\{a_1^2,a_2^2,.Math.,a_N^2\right\}\\ y_{R2\_\mathrm{square}}=\left\{{\left(b_1-a_1\right)}^2,{\left(b_2-a_2\right)}^2,.Math.,{\left(b_N-a_N\right)}^2\right\}\end{array}$

[0039] (2) Average operation is respectively performed on the squares of two signal sequences by using the following equations:

[00012]$\{\begin{array}{c}{\stackrel{\_}{y}}_{R1\_\mathrm{square}}=\frac{1}{N}.Math.\left\{a_1^2+a_2^2+.Math.+a_N^2\right\}\\ {\stackrel{\_}{y}}_{R2\_\mathrm{square}}=\frac{1}{N}.Math.\left[{\left(b_1-a_1\right)}^2+{\left(b_2-a_2\right)}^2+.Math.+{\left(b_N-a_N\right)}^2\right]\end{array}$

[0040] (3) Square root operation is respectively performed on average values of sums of the squares of the two signals by using the following equations:

[00013]$\{\begin{array}{c}{\stackrel{\_}{y}}_{R1\_\mathrm{Val}}=\sqrt{\frac{1}{N}.Math.\left\{a_1^2+a_2^2+.Math.+a_N^2\right\}}\\ {\stackrel{\_}{y}}_{R2\_\mathrm{val}}=\sqrt{\frac{1}{N}.Math.\left[{\left(b_1-a_1\right)}^2+{\left(b_2-a_2\right)}^2+.Math.+{\left(b_N-a_N\right)}^2\right]}\end{array}$

[0041] (4) A ration of amplitudes of the signals of the resistor R.sub.1 and the resistor R.sub.2 is calculated by using the following equation:

[00014]$\mathrm{Amp\_R}=\frac{{\stackrel{\_}{y}}_{R1\_\mathrm{Val}}}{{\stackrel{\_}{y}}_{R2\_\mathrm{Val}}}$

[0042] By using the above algorithm, an amplitude ratio feature vector of a signal on each of channels, including Amp_R.sub.1, Amp_R.sub.2, Amp_R.sub.3, Amp_R.sub.4, Amp_R.sub.5, are respectively calculated.

[0043] The amplitude fluctuation difference feature vector is calculated by obtaining a ratio of absolute values of a difference between an amplitude of a signals of the resistor R.sub.1 and an amplitude of a signals of the resistor R.sub.2 in each of time period. As shown in the flow chart of the system processing in FIG. 8, signal sequences of the resistor R.sub.1 and the inductor L are respectively set as:

[00015]$\{\begin{array}{c}{y}_{R1\_\mathrm{FIR}}=\left\{a_1,a_2,.Math.,a_N\right\}\\ y_{L\_\mathrm{FIR}}=\left\{b_1,b_2,.Math.,b_N\right\}\end{array}$

[0044] Based on the Y-type circuit, the resistor R.sub.1, resistor R.sub.2 and inductor L form a closed-loop triangle. Based on a Kirchhoff s Voltage rule, a signal sequence of the resistor R.sub.2 may be calculated as the following expression:

{(b.sub.1−a.sub.1), (b.sub.2−a.sub.2), . . . , (b.sub.N−a.sub.N)}

[0045] (1) An absolute value of differences of a signal of the resistor R.sub.1 and an absolute value of differences a signal of the resistor R2 are respectively calculated by using the following equations:

[00016]$\{\begin{array}{c}{y}_{R1\_\mathrm{AMP}}=\left\{.Math.a_2-a_1.Math.,.Math.a_3-a_2.Math.,.Math.,.Math.a_N-a_{N-1}.Math.\right\}\\ y_{R2\_\mathrm{AMP}}=\left\{.Math.\left(b_2-a_2\right)-\left(b_1-a_1\right).Math.,.Math.\left(b_3-a_3\right)-\left(b_2-a_2\right).Math.,.Math.,.Math.\left(b_N-a_N\right)-\left(b_{N-1}-a_{N-1}\right).Math.\right\}\end{array}$

[0046] (2) Statistics is performed on the amplitude of the signal of the resistor R1 and the amplitude of the signal of the resistor R2 in each of time periods. The absolute value of differences of the signal of the resistor R.sub.1 and the absolute value of differences of the signal of the resistor R2 are outputted as feature vectors. For data within every 20 ms, a set of feature vectors are outputted and sent to a neural network.

[0047] The phase fluctuation feature vector is calculated by obtaining a difference between a phase of a signal of the resistor R and a phase of a signal of the inductor L in each of time periods. As shown in the flow chart of the system processing in FIG. 9, signal sequences of the resistor R and the inductor L are respectively set as:

[00017]$\{\begin{array}{c}{y}_{R\_\mathrm{FIR}}=\left\{a_2-a_1,a_3-a_2,.Math.,a_N-a_{N-1}\right\}\\ y_{L\_\mathrm{FIR}}=\left\{b_2-b_1,b_3-b_2,.Math.,b_N-b_{N-1}\right\}\end{array}$

[0048] (1) A differential sequence of the signal of the resistor R and a differential sequence of the signal of the inductor L are respectively calculated by using the following equations:

[00018]$\{\begin{array}{c}{y}_{R\_\mathrm{FIR}}=\left\{a_2-a_1,a_3-a_2,.Math.,a_N-a_{N-1}\right\}\\ y_{L\_\mathrm{FIR}}=\left\{b_2-b_1,b_3-b_2,.Math.,b_N-b_{N-1}\right\}\end{array}$

[0049] (2) Statistics is performed on the fluctuation of the signal of the resistor and the fluctuation of the signal of the inductor in each of time periods. A feature vector is outputted in a case that R1*L<0. For data within every 20 ms, a set of feature quantities are outputted and sent to a neural network.

[0050] The phase difference feature vector is calculated by obtaining a difference between a phase of a signal of the resistor R and a phase of a signal of the inductor L in each of time periods. As shown in the flow chart of the system processing in FIG. 7, signal sequences of the resistor R and the inductor L are respectively set as:

[00019]$\{\begin{array}{c}{y}_{R\_\mathrm{FIR}}=\left\{a_1,a_2,.Math.,a_N\right\}\\ y_{L\_\mathrm{FIR}}=\left\{b_1,b_2,.Math.,b_N\right\}\end{array}$

[0051] (1) A frequency f.sub.0 equal to 30 Mhz in a pass frequency band of the filter is determined as a reference frequency, and a standard complex signal sequence at this frequency point is calculated and expressed as:

y.sub.e{e.sup.jω.sup.0.sup.1, e.sup.jω.sup.0.sup.2, . . . , e.sup.jω.sup.0.sup.N}

[0052] (2) The signal sequence of the resistor R.sub.1 and the signal sequence of inductor L are respectively multiplied with the standard complex signal sequence, improving the signal-to-noise ratio of the signal at the frequency to be measured, and thereby greatly improves the effect of extracting phase features of weak signals. The calculation results are as follows:

[00020]$\{\begin{array}{c}{y}_{R1\_\mathrm{product}}=\left\{a_1.Math.e^{j{\omega }_0^1},a_2.Math.e^{j{\omega }_0^2},.Math.,a_N.Math.e^{j{\omega }_0^N}\right\}\\ y_{L\_\mathrm{product}}=\left\{b_1.Math.e^{j{\omega }_0^1},b_2.Math.e^{j{\omega }_0^2},.Math.,b_N.Math.e^{j{\omega }_0^N}\right\}\end{array}$

[0053] (3) Integral operation is performed on the obtained product sequence. Based on a principle of converting discrete signal integral to summation, the following integral results are obtained:

[00021]$\{\begin{array}{c}{D}_1=\underset{n=1}{\overset{N}{.Math.}}{a}_n.Math.e^{j{\omega }_0^n}\\ D_2=\underset{n=1}{\overset{N}{.Math.}}{b}_n.Math.e^{j{\omega }_0^n}\end{array}$

[0054] (4) Based on the integral results, initial phases of the two signal sequences at the reference frequency point in the time period are calculated by using the follows equations:

[00022]$\{\begin{array}{c}{\phi }_{R1\_0}=\arctan \left(\frac{\mathrm{Re}\left(D_1\right)}{\mathrm{Im}\left(D_1\right)}\right)\\ {\phi }_{L\_0}=\arctan \left(\frac{\mathrm{Re}\left(D_2\right)}{\mathrm{Im}\left(D_2\right)}\right)\end{array}$

[0055] (5) A difference between the initial phases is calculated to obtain the phase difference feature vector of the two signals at the frequency f.sub.0 of 30 Mhz in a first vector analysis:

Δφ.sub.0=φR.sub.1_0−φ.sub.L_0

[0056] (6) In the pass frequency bands, a reference frequency f.sub.0 is respectively selected as f.sub.1, f.sub.2, f.sub.3, f.sub.4 and f.sub.5., phase differences Δφ.sub.1, Δφ.sub.2, Δφ.sub.3, Δφ.sub.4 and Δφ.sub.5, respectively corresponding to the channels, between a phase of a signal of a resistor R.sub.1 and a phase of a signal of an inductor L in the time period is calculated by performing the above steps (1) to (5).

[0057] In step 4, a hardware digital signal processing unit performs vector analysis on the two sampling signals, outputs an amplitude ratio feature value, an amplitude fluctuation feature value, a phase difference feature value and a phase fluctuation feature value in real time, and then sends all the feature values to a MCU system for statistical processing.

[0058] In step 5, the system monitors a hardware zero-crossing detection circuit, statistics is performed on the amplitude ratio feature vector, the amplitude fluctuation feature vector, the phase difference feature vector and the phase fluctuation feature vector each time a zero-crossing signal arrives, and splices the feature values in a half wave to obtain a 10*500 feature matrix. Then, the feature matrix is inputted to a neural network for calculation to obtain an identification result.

[0059] Compared with the conventional devices, with which only a fault arc signal is detected and an arc in a main circuit cannot be distinguished from an arc in a bypass circuit, in the method for identifying an arc crosstalk signal according to the present disclosure, vector analysis is performed on high-frequency components of an arc signal with respect to amplitude and phase, targeted processing is performed on high-frequency features of the arc signal, accurately identifying the arc crosstalk signal, achieving strong anti-interference ability in various complex power consumption environments, and thereby achieving a stable and reliable identification result.

[0060] The embodiments described above are only preferred embodiments of the present disclosure. For those skilled in the art, several improvements and changes may be made without departing from the principle of the present disclosure. Any modification, equivalent replacement, improvement and the like made within the spirit and principle of the present disclosure should be included in the protection of the present disclosure.