Characteristic current detection method and system
By filtering out DC and fundamental frequency signals, downward shifting and adaptive quantile noise estimation, the characteristic current is detected, which solves the problem of low demodulation accuracy of feature codewords caused by spectrum leakage and time error, and achieves higher detection accuracy and time accuracy.
Patent Information
- Application Number
- CN202411772650.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-04
- Publication Date
- 2025-03-14
AI Technical Summary
In the prior art, due to the large time error of spectrum leakage and detection, the demodulation accuracy of the feature code word is low.
By filtering out the DC and fundamental frequency signals in the ADC sampled data containing characteristic current buffered in real-time segmentation buffers, shifting the frequency downward to convert the characteristic current to the fundamental frequency, denoising the fundamental amplitude signal using adaptive quantile noise estimation, and obtaining the characteristic codewords through point-by-point judgment.
The detection error caused by frequency asynchronous sampling is reduced, the accuracy and time accuracy of feature code word demodulation are improved, and the problem of low feature code word demodulation accuracy under the influence of spectrum leakage and time error is solved.
Smart Images
Figure CN119619613A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of low voltage power distribution system topology identification, and in particular relates to a characteristic current detection method and system. Background Art
[0002] The characteristic current carries the specific frequency of the characteristic code. By analyzing these characteristic currents, we can accurately identify the topological relationship (connection relationship, current transmission path, etc.) between the nodes in the power system, helping us to accurately locate the fault point and determine the fault type. Moreover, compared with the traditional power system topology analysis method, the characteristic current has the advantages of not requiring asymptotic impedance algorithm and disturbance signal injection, and can more efficiently identify the topological relationship of the substation, effectively ensuring the lean management of the substation. The characteristic current detection system needs to set the effective sampling frequency and single sampling time according to the frequency characteristics of the signal, and the sampling rate is often not an integer multiple of the detection frequency, which will cause the spectrum to leak and the detection time error is in the unit of single sampling time, affecting the demodulation accuracy of the characteristic codeword. Summary of the invention
[0003] The object of the present invention is to provide a characteristic current detection method and system to solve the problem in the prior art that the characteristic codeword demodulation accuracy is low due to spectrum leakage and large detection time error.
[0004] To solve the above technical problems, the present invention provides a characteristic current detection method, comprising the following steps: after filtering out the DC and baseband signals in the ADC sampling data containing the characteristic current in the real-time segmented cache, down-shifting the frequency according to the frequency of the characteristic current to convert the characteristic current to the baseband, filtering out the signals other than the baseband in the frequency-shifted signal to obtain the characteristic current signal of the baseband and obtaining the fundamental amplitude signal according to the characteristic current signal, denoising the amplitude signal using adaptive quantile noise estimation, comparing the amplitude of each sampling point of the denoised signal with a threshold value, adding 1 to the first count if it is greater than or equal to the threshold value, otherwise adding 1 to the second count, if the first count in the signal segment corresponding to a single codeword in the denoised signal is greater than or equal to the sampling point setting value, the codeword corresponding to the signal segment is 1, if the second count is greater than or equal to the sampling point setting value, the codeword corresponding to the signal segment is 0, and the setting value is the number of sampling points corresponding to the single codeword.
[0005] Furthermore, the adaptive quantile noise estimation is an adaptive quantile noise estimation improved by using noise update control, and the method of denoising the amplitude signal by using the adaptive quantile noise estimation includes:
[0006] The amplitude signal is set to the frame number signal data to calculate the average value as the initial value of the quantile noise estimation quatile(0);
[0007]
[0008] R(n)=Rmax(n)-Rmin(n)
[0009]
[0010]
[0011] Wherein, quatile(n) is the quantile noise estimation value of the nth sampling point; R(n) is the real-time estimation range of the sampling point; Rmax(n) is the maximum estimation value; Rmin(n) is the minimum estimation value; Rmax(0) and Rmin(0) are determined according to the value of the first sampling point; α and β are smoothing factors used to control the estimation step; r(n) is the amplitude of the nth sampling point of the fundamental amplitude signal; p is the quantile probability; λ = 1 / L, L is the number of sampling points of the ADC sampling data containing the characteristic current in the real-time segmented buffer;
[0012] datamean(n)=datamean(n-1)+r(n)-r(n-M+1)
[0013]
[0014] In the formula, datamean(n) is the sliding average; th is the set noise judgment threshold; qnoise(n) is the final update noise, and M is the speed of signal change response;
[0015]
[0016] Where P(H1|r(n)) is the characteristic current signal probability, q(n) is the signal probability determined based on the signal-to-noise ratio; ε th is the signal-to-noise ratio decision threshold; σ(n) is the a posteriori signal-to-noise ratio, ε(n) is the a priori signal-to-noise ratio, and x(n) is the unknown denoised signal;
[0017] noise(n)=λ1*noise(n-1)+(1-λ1)*P(H1|r(n))*noise(n-1)+P(H0|r(n))*r(n)
[0018]
[0019] x(n)=g(n)*r(n)
[0020] Where λ1 is the noise estimation smoothing coefficient; x(n) is the denoised signal, P(H0|r(n)) is the noise existence probability, g(n) is the gain value, and noise(n) is the estimated noise after correction.
[0021] Furthermore, the direct current and baseband signals in the ADC sampling data containing the characteristic current in real-time segmented buffer are filtered out by a low-pass filter; the low-pass filter includes a low-pass filter calculated by a matlab method.
[0022] Furthermore, the signal other than the fundamental frequency in the frequency-shifted signal is filtered out by a high-pass filter to obtain a characteristic current signal of the fundamental frequency, and the high-pass filter includes a high-pass filter calculated by a matlab method.
[0023] Furthermore, the method for obtaining the fundamental amplitude signal according to the characteristic current signal includes:
[0024] According to the real part and the imaginary part of the characteristic current signal, the real signal r′(t) of the characteristic current signal is obtained:
[0025]
[0026] Where real(t) is the real part of the complex signal after low-pass filtering, imag(t) is the imaginary part of the complex signal after low-pass filtering; w0 is the fundamental angular frequency; θ0 is the initial phase of the fundamental wave; A1 and A2 are the characteristic current fundamental frequency signal and fundamental wave signal amplitude respectively;
[0027] The fundamental amplitude signal r(n) is obtained by taking twice the fundamental angular frequency as the window length of each sliding average and performing sliding average on r′(t).
[0028] In order to solve the above technical problems, the present invention further provides a characteristic current detection system, including a processor, and the processor is used to execute a computer program to implement the steps of the characteristic current detection method.
[0029] The beneficial effects of the above technical solution are as follows: the present invention provides a new characteristic current detection method, which performs high-pass filtering and downward frequency shifting on the characteristic current, so that the detection of the characteristic current only needs to detect one frequency point (i.e., the frequency point of the base frequency) instead of two symmetrical frequency points on the opposite sides of the frequency of the characteristic current after intermodulation. Therefore, it is possible to ensure that the sampling rate of the ADC sampling is an integer multiple of the frequency to be detected, thereby reducing the subsequent detection error caused by asynchronous sampling of the frequency of the characteristic current; the fundamental amplitude signal obtained by the characteristic current signal of the base frequency is further denoised to obtain the amplitude signal of the characteristic current that is less affected by noise; on this basis, the denoised The signal is judged point by point according to the sampling points. In the signal segment corresponding to a codeword, if there are more sampling points with larger amplitude, it is judged as codeword 1, and if there are more sampling points with smaller amplitude, it is judged as codeword 0. In this way, the specific codeword of the characteristic code carried by the characteristic current can be obtained for decoding to obtain corresponding information. Since the method of obtaining the characteristic code in the present invention is to judge each sampling point within a single sampling time (that is, the total length of each ADC sampling data corresponding to the segmented cache), the time error of the detection is usually only a few sampling points. Compared with the method in the prior art in which the time error is based on the single sampling time, the detection time accuracy is higher, thereby solving the problem of low demodulation accuracy of the characteristic codeword in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS
[0030] Figure 1 1 is a flow chart of a characteristic current detection method according to an embodiment of the characteristic current detection method of the present invention;
[0031] Figure 2 : is a characteristic current noise modulation waveform diagram of an embodiment of a characteristic current detection method of the present invention;
[0032] Figure 3 is a characteristic current denoising gain waveform diagram of an embodiment of a characteristic current detection method of the present invention;
[0033] Figure 4 It is a characteristic current denoising and demodulation signal waveform diagram of an embodiment of the characteristic current detection method of the present invention. DETAILED DESCRIPTION
[0034] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention is further described in detail below with reference to the accompanying drawings and embodiments.
[0035] Characteristic current detection method embodiment:
[0036] A characteristic current detection method of the present embodiment pre-processes the characteristic current, filters out DC and baseband signals and converts them into baseband signals, so that the detection of the characteristic current only needs to detect one frequency point, filters out signals other than the baseband to obtain a characteristic current signal of the baseband, and obtains a fundamental amplitude signal based on the characteristic current signal, performs adaptive quantile noise estimation on the fundamental amplitude signal to remove noise, and makes point-by-point judgments on the denoised signal according to sampling points to improve the spectrum detection accuracy and reduce the detection time.
[0037] Reference Figure 1 , then the characteristic current detection method includes:
[0038] After filtering out the DC and baseband signals in the ADC sampling data containing the characteristic current in the real-time segmented buffer, the characteristic current is shifted downward according to the frequency of the characteristic current to convert the characteristic current to the baseband, and the signals other than the baseband in the frequency-shifted signal are filtered out to obtain the characteristic current signal of the baseband, and the fundamental amplitude signal is obtained according to the characteristic current signal. The amplitude signal is denoised by using adaptive quantile noise estimation, and the amplitude of each sampling point of the denoised signal is compared with the threshold value. If it is greater than or equal to the threshold value, the first count is increased by 1, otherwise the second count is increased by 1. If the first count in the signal segment corresponding to a single codeword in the denoised signal is greater than or equal to the sampling point setting value, the codeword corresponding to the signal segment is 1, and if the second count is greater than or equal to the sampling point setting value, the codeword corresponding to the signal segment is 0, and the setting value is the number of sampling points corresponding to the single codeword.
[0039] In this embodiment, by high-pass filtering and downward frequency shifting the characteristic current, only one frequency point (i.e., the frequency point of the baseband) needs to be detected for the detection of the characteristic current, rather than two symmetrical frequency points on opposite sides of the frequency of the characteristic current after intermodulation. Therefore, it is possible to ensure that the sampling rate of the ADC sampling is an integer multiple of the frequency to be detected, thereby reducing the subsequent detection error caused by asynchronous sampling of the frequency of the characteristic current; the fundamental amplitude signal obtained by the characteristic current signal of the baseband is further denoised to obtain the amplitude signal of the characteristic current that is less affected by noise; on this basis, the denoised signal is judged point by point according to the sampling points, In a signal segment corresponding to a codeword, if there are more sampling points with larger amplitudes, it is judged as codeword 1, and if there are more sampling points with smaller amplitudes, it is judged as codeword 0. Thus, the specific codeword of the characteristic code carried by the characteristic current can be obtained for decoding to obtain corresponding information; since the method of obtaining the characteristic code in the present invention is to make point-by-point judgments on each sampling point within a single sampling duration (i.e., the total length of each segment of ADC sampling data corresponding to the segmented cache), the time error of the detection is usually only a few sampling points. Compared with the method in the prior art in which the time error is in units of a single sampling duration, the detection time accuracy is higher, thereby solving the problem of low demodulation accuracy of the characteristic codeword in the prior art.
[0040] Specifically, the real-time segmented cache measures the ADC sampling data, the sampling rate is fs, the window length is L, and L is determined according to the processing speed. The signal to be detected at the receiving end is obtained by intermodulation of the characteristic current and the fundamental signal. The mathematical expression of the received signal is:
[0041] r(t)=[A0+A1*cos(wt+θ1)+A3*cos(3wt+θ3)]*A2*cos(w0t+θ0)
[0042] Wherein, w is the characteristic current angular frequency, w0 is the fundamental angular frequency, θ0 is the fundamental initial phase, A0, A1, A2, A3 are the amplitudes corresponding to the frequencies of different characteristic current signal components, θ1, θ3 are the initial phases corresponding to the signals of different component frequencies.
[0043] Specifically, taking the sampling data using a noisy current signal with a signal-to-noise ratio of 10 dB as an example, Figure 2 As shown, take fs=5000 Hz, segment length L=250, characteristic current frequency f=833.3 Hz, then the sampling signal contains frequencies such as DC, f0, f+f0, f-f0, f+3f0, f-3f0, etc.
[0044] In this embodiment, a low-pass filter is used to filter out the DC and baseband signals in the ADC sampling data containing the characteristic current in real-time segmented buffer; the low-pass filter includes a low-pass filter calculated by the matlab method. In other embodiments, other filters or other filtering methods can be used to achieve the effect of filtering out DC and baseband signals. The collected signal is subjected to high-pass filtering to filter out DC and baseband signals, the transition band is 100hz-600hz, the passband gain and stopband attenuation are 0.01dB and 40dB respectively, and a 27-order high-pass filter is calculated by the matlab ripple method, and the signal after the high-pass is point-multiplied by e -i2πf / f s , i∈[0,…,(L-1)], realize the downward frequency shift f, move the (f+f0) and (f-f0) frequency points to f0 and -f0 respectively, and convert the frequency to be detected from f to f0, thereby reducing the subsequent detection error caused by asynchronous sampling of the frequency f. In other embodiments, other filters or other filtering methods can be used to achieve the effect of filtering out DC and baseband signals.
[0045] Since the real and imaginary parts of the complex signal are:
[0046]
[0047] In this embodiment, the method for obtaining the fundamental amplitude signal according to the characteristic current signal includes:
[0048] According to the real part and the imaginary part of the characteristic current signal, the real signal r′(t) of the characteristic current signal is obtained:
[0049]
[0050] Where real(t) is the real part of the complex signal after low-pass filtering, imag(t) is the imaginary part of the complex signal after low-pass filtering; w0 is the fundamental angular frequency; θ0 is the initial phase of the fundamental wave; A1 and A2 are the amplitudes of the characteristic current fundamental frequency signal and the fundamental wave signal respectively;
[0051] The sliding average window length of each sliding average is twice the fundamental wave angular frequency, and the fundamental wave amplitude signal r(n) is obtained by sliding average of r′(t). The sliding average method belongs to the prior art and will not be described here.
[0052] In this embodiment, the characteristic current signal of the fundamental frequency is obtained by filtering out the signals other than the fundamental frequency in the frequency-shifted signal through a high-pass filter, and the high-pass filter includes a high-pass filter calculated by a matlab method. In other embodiments, other filters or other filtering methods can be used to achieve the effect of filtering out signals other than the fundamental frequency. Specifically, the filter index in the high-pass filter is set to be consistent with the low-pass filter, and a 28-order filter is calculated to filter out signals other than the fundamental frequency.
[0053] In this embodiment, the adaptive quantile noise estimation is an adaptive quantile noise estimation improved by using noise update control. The method of denoising the amplitude signal by using the adaptive quantile noise estimation includes:
[0054] The amplitude signal is set to the frame number signal data to calculate the average value as the initial value of the quantile noise estimation quatile(0);
[0055]
[0056] R(n)=Rmax(n)-Rmin(n)
[0057]
[0058] Wherein, quatile(n) is the quantile noise estimation value of the nth sampling point; R(n) is the real-time estimation range of the sampling point; Rmax(n) is the maximum estimation value; Rmin(n) is the minimum estimation value; Rmax(0) and Rmin(0) are determined according to (in this embodiment, Rmax(0) and Rmin(0) are directly taken as the value of the first sampling point, and in other embodiments, the value of the first sampling point can be adaptively processed to obtain Rmax(0) and Rmin(0)); α and β are smoothing factors for controlling the estimation step; r(n) is the amplitude of the nth sampling point of the fundamental amplitude signal; p is the quantile probability; λ=1 / L, L is the number of sampling points of the ADC sampling data containing the characteristic current in the real-time segmented buffer.
[0059] Specifically, in this embodiment, the frame number signal is set to take the first 10 frame signals.
[0060] Specifically, in order to avoid the noise deviation caused by the quantile estimation principle when the codewords are continuous 1 or 0, the adaptive quantile noise estimation in this embodiment is an adaptive quantile noise estimation after the adaptive quantile noise estimation in the prior art is improved, and the improvement means is noise update control. After the noise update control improvement, the final updated noise is obtained as follows:
[0061] datamean(n)=datamean(n-1)+r(n)-r(n-M+1)
[0062]
[0063] Where datamean(n) is the sliding average; th is the set noise judgment threshold; qnoise(n) is the final updated noise, and M is the speed of signal change response; when receiving continuous codewords 1 or 0, the range of quantile estimation will slowly converge to the vicinity of the signal or noise, and the estimated quatile(n) is greatly different from the true value of the noise. At this time, the estimated value quatile(n) is equivalent to the sliding average value datamean(n). A larger th value is used to control the noise estimation value not to be updated in this case. For example, in this embodiment, the judgment threshold is th=0.75.
[0064]
[0065]
[0066] Where P(H1|r(n)) is the characteristic current signal probability, q(n) is the signal probability determined based on the signal-to-noise ratio; ε th is the signal-to-noise ratio decision threshold; σ(n) is the posterior signal-to-noise ratio, ε(n) is the prior signal-to-noise ratio, and x(n) is the unknown denoising signal.
[0067] noise(n)=λ1*noise(n-1)+(1-λ1)*P(H1|r(n))*noise(n-1)+P(H0|r(n))*r(n)
[0068]
[0069] x(n)=g(n)*r(n)
[0070] Where λ1 is the noise smoothing coefficient; x(n) is the denoised signal, P(H0|r(n)) is the probability of noise existence, g(n) is the gain value, and noise(n) is the estimated noise after correction.
[0071] Specifically, the gain g(n) for removing noise and the denoised signal x(n) are respectively as follows: Figure 3 and Figure 4 shown.
[0072] Specifically, x(n) is judged point by point and compared with the fixed threshold value Lvl (that is, the amplitude of each sampling point of the denoised signal is compared with the threshold value) as follows:
[0073]
[0074] Cnt1 is the first count, and Cnt0 is the second count; x(n) here represents the signal segment corresponding to a certain codeword; when Cnt1>=Tbit (sampling point setting value), BitCode=[BitCode 1], that is, the codeword corresponding to the signal segment is 1; when Cnt0>=Tbit, BitCode=[BitCode 0], that is, the codeword corresponding to the signal segment is 0; the first count and the second count will be cleared to restart the counting before starting to count the signal segment corresponding to the next codeword; after receiving all the codewords, the BitCode is compared with the known feature codeword to complete the detection. The time detection error of this method is only a few sampling points, which is the time detection accuracy that the frequency domain detection method cannot achieve, and the calculation amount is also lower than the FFT transformation.
[0075] Characteristic current detection system embodiment:
[0076] The present invention also provides a characteristic current detection system, including a processor, the processor is used to execute a computer program to implement the steps of the characteristic current detection method. The specific process, principle and effect of the characteristic current detection system have been described in detail in the embodiment of the characteristic current detection method, and will not be repeated here.
[0077] Specific implementation methods are given above, but the present invention is not limited to the described implementation methods. The basic idea of the present invention lies in the above basic scheme. For ordinary technicians in this field, it does not take creative work to design various deformed models, formulas, and parameters according to the teachings of the present invention. Changes, modifications, substitutions, and variations of the implementation methods without departing from the principles and spirit of the present invention still fall within the scope of protection of the present invention.
Claims
1. A characteristic current detection method, characterized in that: The method comprises the following steps: after filtering out the direct current and baseband signals in the ADC sampling data containing the characteristic current in the real-time segmented buffer, shifting the frequency downward according to the frequency of the characteristic current to convert the characteristic current to the baseband, filtering out the signals other than the baseband in the frequency-shifted signal to obtain the characteristic current signal of the baseband, and obtaining the fundamental amplitude signal according to the characteristic current signal, denoising the amplitude signal by using the adaptive quantile noise estimation, comparing the amplitude of each sampling point of the denoised signal with the threshold value, adding 1 to the first count if it is greater than or equal to the threshold value, otherwise adding 1 to the second count, if the first count in the signal segment corresponding to a single codeword in the denoised signal is greater than or equal to the sampling point setting value, the codeword corresponding to the signal segment is 1, if the second count is greater than or equal to the sampling point setting value, the codeword corresponding to the signal segment is 0, and the setting value is the number of sampling points corresponding to the single codeword.
2. The characteristic current detection method according to claim 1, characterized in that: The adaptive quantile noise estimation is an adaptive quantile noise estimation improved by noise update control. The method of denoising the amplitude signal by using the adaptive quantile noise estimation includes: The amplitude signal is set to the frame number signal data to calculate the average value as the initial value of the quantile noise estimation quatile(0); R(n)=Rmax(n)-Rmin(n) Where quatile(n) is the quantile noise estimate of the nth sampling point; R(n) is the real-time estimation range of the sampling point; Rmax(n) is the maximum estimated value; Rmin(n) is the minimum estimated value; Rmax(0) and Rmin(0) are determined based on the value of the first sampling point; α and β are smoothing factors used to control the estimation step; r(n) is the amplitude of the nth sampling point of the fundamental amplitude signal; p is the quantile probability; λ = 1 / L, L is the number of sampling points of the ADC sampling data containing the characteristic current in the real-time segmented buffer; datamean(n)=datamean(n-1)+r(n)-r(n-M+1) In the formula, datamean(n) is the sliding average; th is the set noise judgment threshold; qnoise(n) is the final update noise, and M is the speed of signal change response; Where P(H1|r(n)) is the characteristic current signal probability, q(n) is the signal probability determined based on the signal-to-noise ratio; ε th is the signal-to-noise ratio decision threshold; σ(n) is the a posteriori signal-to-noise ratio, ε(n) is the a priori signal-to-noise ratio, and x(n) is the unknown denoised signal; noise(n)=λ1*noise(n-1)+(1-λ1)*P(H1|r(n))*noise(n-1)+P(H0|r(n))*r(n) x(n)=g(n)*r(n) Where λ1 is the noise estimation smoothing coefficient; x(n) is the denoised signal, P(H0|r(n)) is the noise existence probability, g(n) is the gain value, and noise(n) is the estimated noise after correction.
3. The characteristic current detection method according to claim 1, characterized in that: The direct current and baseband signals in the ADC sampling data containing the characteristic current in real-time segmented buffer are filtered out by a low-pass filter; the low-pass filter includes a low-pass filter calculated by a matlab method.
4. The characteristic current detection method according to claim 3, characterized in that: The signal other than the fundamental frequency in the frequency-shifted signal is filtered out by a high-pass filter to obtain a characteristic current signal of the fundamental frequency, wherein the high-pass filter includes a high-pass filter calculated by a matlab method.
5. The characteristic current detection method according to claim 1, characterized in that: The method for obtaining the fundamental amplitude signal according to the characteristic current signal includes: According to the real part and the imaginary part of the characteristic current signal, the real signal r′(t) of the characteristic current signal is obtained: Where real(t) is the real part of the complex signal after low-pass filtering, imag(t) is the imaginary part of the complex signal after low-pass filtering; w0 is the fundamental angular frequency; θ0 is the initial phase of the fundamental wave; A1 and A2 are the characteristic current fundamental frequency signal and fundamental wave signal amplitude respectively; The fundamental amplitude signal r(n) is obtained by taking twice the fundamental angular frequency as the window length of each sliding average and performing sliding average on r′(t).
6. A characteristic current detection system, comprising a processor, characterized in that: The processor is used to execute a computer program to implement the steps of the characteristic current detection method according to any one of claims 1 to 5.