A low-frequency signal fast frequency identification method and system based on a segmented time window

Abstract

This invention discloses a method and system for rapid frequency identification of low-frequency signals based on segmented time windows. The method involves real-time monitoring of the three-phase voltage and current of a power line, real-time calculation of the line's active power, and upon receiving a command to initiate low-frequency identification, applying a 3Hz low-pass filter to the line's active power. It then uses a combination of cycle number calculation and segmented time window FFT to calculate historical data within one cycle, ensuring accurate identification of the oscillation frequency and amplitude within two cycles of the low-frequency oscillation. This identification method and system solves the problem of inaccurate identification due to the large span of the low-frequency band, providing a time margin for the next stage of low-frequency control or protection.

Description

Technical Field

[0001] This invention belongs to the field of power system stability and control technology, and more specifically, relates to a method and system for rapid and high-precision identification of low-frequency band frequency and amplitude using segmented time windows. Background Technology

[0002] In recent years, the rapid development of power electronics technology for new energy generators has accelerated the transformation of the power system into a new type of power system. However, while this transformation is reshaping the energy supply pattern, it has also disrupted the stable balance of the traditional power system, and the problem of broadband oscillations has gradually become prominent. Among them, low-frequency oscillations, due to their characteristics of long propagation distance and harm throughout the entire "generation-transmission-distribution-consumption" chain, have led to an increasingly strong demand for rapid and accurate identification.

[0003] The low-frequency band ranges from 0.1 to 2.5 Hz. Currently, the online identification methods for low-frequency oscillations in electrical equipment commonly used include traditional linear and nonlinear identification methods, such as small-disturbance stability analysis, FFT (Fast Fourier Transform), and wavelet transform.

[0004] The above methods have achieved the identification of low-frequency oscillations of the monitored object to a certain extent, but they are limited by problems such as computational complexity, data length requirements, and spectral leakage. Furthermore, the single-cycle method has relatively high requirements for waveform standardization. In summary, it is difficult to achieve accurate identification of frequency and amplitude across the entire 0.1-2.5Hz frequency band within two cycles by relying on only one detection method.

[0005] With the development of new energy sources in recent years, broadband oscillations have become increasingly prominent. Low-frequency oscillations within broadband oscillations, in particular, have received special attention due to their long oscillation periods, difficulty in identification, long propagation distances, and potential limitations on power transmission capacity. Therefore, there is an urgent need for a method to quickly and accurately identify low-frequency oscillations, enabling accurate identification at the onset of oscillations and providing technical support for subsequent monitoring, early warning, control, and protection. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a method and system for rapid frequency identification of low-frequency signals based on segmented time windows. By rapidly identifying the frequency and amplitude of low-frequency signals with small disturbances, it provides favorable conditions for subsequent monitoring, early warning, and protection control, indirectly improving the power quality of the power grid and ensuring the safe and stable operation of the system.

[0007] The present invention adopts the following technical solution. The first aspect of the present invention provides a method for fast frequency identification of low-frequency signals based on a segmented time window, comprising the following steps: The low-frequency band is divided into multiple bands, and different downsampling factors and corresponding time windows are configured for each band. The three-phase voltage and current of the monitoring line are acquired, the three-phase fundamental active power of the monitoring line is calculated, and the active power signal is filtered to obtain the filtered active power signal. Based on the filtered active power signal, FFT calculation is performed to identify a rough estimate of the oscillation frequency; Based on the frequency band to which the rough estimate of the oscillation frequency belongs, select the corresponding time window, perform FFT calculation on the filtered active power signal, and obtain the preliminary estimate of the oscillation frequency and the active power signal after eliminating the bias. Using the same time window, perform FFT calculation again on the active power signal after offset removal to obtain accurate low-frequency oscillation frequency and amplitude.

[0008] Preferably, dividing the low-frequency band into multiple frequency bands and configuring different downsampling factors and corresponding time windows for each frequency band includes: Set a segmented FFT time window, including: dividing the low-frequency band into segments. N Each frequency band has a set downsampling factor, with larger downsampling intervals for low-frequency bands and smaller downsampling intervals for high-frequency bands, as expressed by the following formula: FFT win_N ={ WIN i | i = 1, 2, …, N , WIN i > WIN i+1} In the formula: FFT win_N Denotes the set of downsampling factors, with N One element, WIN i Indicates the first i The downsampling factor, i.e. the th downsampling factor i Downsampling factor for each frequency band, WIN i ∈ FFT win_N , i = 1, 2, …, N ; Based on the downsampling factor for each frequency band, the corresponding time window for each frequency band is calculated, as expressed by the following formula:

[0009] In the formula: WIN _T ( i ) indicates the first i A time window; N FFT This indicates the number of data points used in the FFT. F S This indicates the original sampling rate of the three-phase voltage and current of the monitored line.

[0010] Preferably, the frequency range of 0.1~2.5Hz is divided into 6 frequency bands using 0.25, 0.35, 0.85, 1.35, and 1.85 as dividing nodes; The raw sampling rate of the three-phase voltage and current of the monitoring line F S =9600 Hz ; Total number of data points used in each FFT N FFT = 1024.

[0011] Preferably, the steps of acquiring the three-phase voltage and current of the monitoring line, calculating the three-phase fundamental active power of the monitoring line, filtering it, and obtaining the filtered active power signal include: At the set sampling frequency F S Obtain the three-phase voltage and current of the monitored line, and calculate the three-phase fundamental active power of the monitored line. P j ; To set the cutoff frequency f cut-off The filtering stage affects the three-phase fundamental active power. P j Filtering is performed to obtain the filtered active power signal. P Filter ; The filtering stage is a low-pass filter with a total order of 4.

[0012] Preferably, the filtering stage includes two cascaded second-order IIR low-pass filters, namely a first IIR low-pass filter and a second IIR low-pass filter. Filtering stage, first IIR Low-pass filter and second IIR The transfer function of a low-pass filter is expressed by the following formula:

[0013]

[0014]

[0015] In the formula: H ( z ), H 1( z )and H 2( z ) represent the filtering stage and the first stage, respectively. IIR Low-pass filter and second IIR The transfer function of a low-pass filter; b 10 , b 11 and b 12 Indicates first IIR low-pass filter FIR Partial coefficients, b 20 , b 21 and b 22 Indicates the second IIR low-pass filter FIR Partial coefficients; a 10 , a 11 and a 12 Indicates first IIR low-pass filter IIR Partial coefficients, a 20 , a 21 and a 22 Indicates the second IIR low-pass filter IIR Partial coefficients; z -1 This indicates the unit delay operator.

[0016] Preferably, the step of performing FFT calculations based on the filtered active power signal to identify a rough estimate of the oscillation frequency includes: The filtered active power signal is processed using a multi-cycle method. P Filter Perform cycle number identification, where, in the set... N sampling Within each sampling period, the filtered active power signal P Filter Historical data is used to calculate peaks and troughs. When two consecutive peaks or troughs appear, it is considered that the low-frequency oscillation has completed one cycle. At this point, a set length is used to calculate the peaks and troughs.WIN _ typical of FFT Roughly perform time window FFT Calculate and identify a rough estimate of the oscillation frequency. f rough .

[0017] Preferably, the set N sampling Each sampling period is 1.5 sampling periods; Set length as WIN _ typical of FFT The time window is 12.8. s .

[0018] Preferably, the step of selecting the corresponding time window based on the frequency band to which the rough estimate of the oscillation frequency belongs, and performing FFT calculation on the filtered active power signal to obtain the preliminary estimate of the oscillation frequency and the active power signal after bias elimination includes: Based on a rough estimate of the oscillation frequency f rough Frequency band k , k = 1, 2, …, N Obtain the corresponding downsampling factor WIN k and corresponding FFT Time window WIN _ T ( k ); choose FFT Time window WIN _ T ( k ) the filtered active power signal P Filter conduct FFT Calculate and obtain the maximum value of the oscillation waveform. P MAX Minimum value P MIN and preliminary estimate of oscillation frequency f initial ; Based on the maximum value of the oscillation waveform P MAX Minimum value P MIN The active power signal after offset elimination is calculated and expressed by the following formula: P Z = P Filter — ( P MAX +P MIN ) / 2 In the formula: P Z This represents the active power signal after offset removal.

[0019] The second aspect of the present invention provides a fast frequency identification system for low-frequency signals based on a segmented time window, which runs a fast frequency identification method for low-frequency signals based on a segmented time window according to the first aspect, including: an analog signal acquisition stage, an instruction receiving stage, an active power calculation stage, a filtering stage, a low-frequency signal identification stage, and an identification result output stage. The analog quantity acquisition stage connects to the three-phase voltage and three-phase current of the monitoring line, at a set sampling frequency. F S Acquire the three-phase voltage and current of the monitored line; The instruction receiving stage determines whether to begin low-frequency signal identification by receiving a start low-frequency identification instruction from the main control unit. The active power calculation step is used to calculate the three-phase fundamental active power to obtain the three-phase fundamental active power of the monitored line. Pj ; The filtering stage includes two cascaded second-order IIR low-pass filters, namely the first IIR low-pass filter and the second IIR low-pass filter. The low-frequency signal identification step performs FFT calculation based on the filtered active power signal to identify a rough estimate of the oscillation frequency. According to the frequency band to which the rough estimate of the oscillation frequency belongs, the corresponding time window is selected, and FFT calculation is performed on the filtered active power signal to obtain a preliminary estimate of the oscillation frequency. The same time window is used for the active power signal after bias elimination, and FFT calculation is performed again on the active power signal after bias elimination to obtain an accurate low-frequency oscillation frequency and amplitude. The identification result output stage sends the accurate low-frequency oscillation frequency and amplitude as the system identification result, i.e., telemetry, to the main control unit.

[0020] Preferably, the communication cycle between the main control unit and the low-frequency signal fast frequency identification system of the segmented time window is 20Hz; The frequency band identified by the low-frequency signal identification step is 0.1Hz to 2.5Hz.

[0021] Compared with the prior art, the beneficial effects of the present invention include at least the following: This invention proposes a method and system for fast frequency identification of low-frequency signals based on segmented time windows. Compared with existing low-frequency oscillation identification schemes, it can accurately identify low-frequency oscillations with small disturbances within two cycles of detecting low-frequency oscillations in the monitored line, providing strong technical support for the monitoring, control and protection of low-frequency oscillations, and has extremely important practical significance. Attached Figure Description

[0022] Figure 1 is a flowchart illustrating a method and system for fast frequency identification of low-frequency signals based on a segmented time window, according to an embodiment of the present invention. Figure 2 is a flowchart of a fast and high-precision frequency identification method for low-frequency bands provided according to an embodiment of the present invention; Figure 3 is an active power waveform diagram used to verify the validity of this application according to an embodiment of the present invention. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of this invention. The described embodiments are merely some embodiments of this invention, and not all embodiments. Based on the spirit of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of this invention.

[0024] like Figure 1 , 2 As shown, Embodiment 1 of the present invention provides a fast frequency identification method for low-frequency signals based on segmented time windows. This method calculates the active power of the line in real time by monitoring the three-phase voltage and three-phase current of the line. After receiving the command to start low-frequency identification, the active power of the line is low-pass filtered at a specific cutoff frequency. By combining the calculation of the number of cycles with the segmented time window FFT, historical data within one cycle is calculated to ensure accurate identification of the oscillation frequency and amplitude within two cycles of low-frequency oscillation. The identification method solves the problem of inaccurate identification due to the large span of the low-frequency band, and provides a time margin for the control or protection of low frequencies in the next stage.

[0025] Specifically, relying on a combination of multi-cycle wave analysis and piecewise time-window FFT, the present invention provides a method for rapid and high-precision identification of low-frequency oscillations, comprising the following steps: Step 1: Divide the low-frequency band into multiple frequency bands, and configure different downsampling factors and corresponding time windows for each frequency band.

[0026] Preferably, but not limitingly, step 1 specifically includes: Step 1.1: Set the segmented FFT time window, including: dividing the 0.1~2.5Hz range into segments according to frequency range. N Each frequency band is the one mentioned N Each frequency band within the frequency band is assigned a downsampling factor, with larger downsampling intervals for low-frequency bands and smaller downsampling intervals for high-frequency bands, as expressed by the following formula: FFT win_N ={ WIN i | i = 1, 2, …, N , WIN i > WIN i+1} In the formula: FFT win_N Denotes the set of downsampling factors, with N One element, WIN i Indicates the first i The downsampling factor, i.e. the th downsampling factor i Downsampling factor for each frequency band, WIN i ∈ FFT win_N , i = 1, 2, …, N .

[0027] Further, but not restrictively, using 0.25, 0.35, 0.85, 1.35, and 1.85 as dividing points, the 0.1~2.5Hz frequency range is divided into 6 frequency bands, i.e. N = 6, and configure downsampling factors corresponding to 6 frequency bands. The set of downsampling factors is represented by the following formula: FFT win_6 ={ WIN 1, WIN 2, WIN 3, WIN 4, WIN 5, WIN 6}={120, 100, 48, 32, 24, 12}.

[0028] Table 1. Frequency range and interval frequency table for each segment

[0029] Step 1.2: Based on the downsampling factor of each frequency band, calculate the corresponding time window for each frequency band, expressed by the following formula:

[0030] In the formula: WIN _ T ( i ) indicates the first i The first time window, i.e. the first i The time window corresponding to each frequency band; N FFT This indicates the number of data points used in the FFT. F S This indicates the raw sampling rate of the three-phase voltage and current of the monitored line, preferably but not limited to [specific values]. F S =9600 Hz .

[0031] Further preferred, but not restrictive, is that the total number of data points used in each FFT is 1024, i.e. N FFT =1024, therefore, the 6 frequency bands correspond to 6 time windows. WIN _ T ( i They are respectively: WIN _ T (1) = 12.8 s ; WIN _ T (2) = 10.67 s ; WIN _ T (3) = 5.12 s ; WIN _ T (4) = 1.28 s ; WIN _ T (5) = 2.56 s ; WIN _ T (6) = 3.41 s .

[0032] Step 2: Obtain the three-phase voltage and current of the monitored line, calculate the three-phase fundamental active power of the line, filter it, and obtain the filtered active power signal.

[0033] Preferably, but not limitingly, step 2 specifically includes: Step 2.1: At the set sampling frequency F SObtain the three-phase voltage and current of the monitored line, and calculate the three-phase fundamental active power of the line. P j It can be expressed by the following formula:

[0034] In the formula: P j This represents the three-phase fundamental active power, measured in kW. U A , U B and U C These represent the three-phase voltage amplitudes, in units of V; I A , I B and I C These represent the amplitudes of the three-phase currents, in A. Re This indicates taking the real part.

[0035] It's understandable that the 1 / 2 in the formula means that the "complex phase amplitude" of the sinusoidal quantity is the peak value, not the effective value, while active power calculation requires the effective value. Converting the peak value to the effective value requires dividing by 1 / 2. Therefore, the actual single-phase active power is 1 / 2 of the peak product.

[0036] Further preferred, but not limiting, sampling frequency F S It is 9600Hz.

[0037] Step 2.2: After receiving the low-frequency identification command from the main control unit, the command receiving stage sets the cutoff frequency. f cut-off The filtering stage affects the three-phase fundamental active power. P j Filtering is performed to obtain the filtered active power signal. P Filter .

[0038] Further preferably, but not limitingly, the filtering stage comprises two cascaded second-order IIR low-pass filters (Infinite Impulse Response Filters), namely the first IIR low-pass filter and the second IIR low-pass filter, that is, the filtering stage is equivalent to a low-pass filter with a total order of 4. The cutoff frequency of the filtering stage... f cut-offThe preferred frequency is 3Hz, but not limited to 0.000997408035.

[0039] It is worth noting that this embodiment provides an extremely precise optimal filter gain. The reason for providing this level of precision is that, for a low-pass filter, the coefficient is proportional to the bandwidth; the narrower the bandwidth, the smaller the coefficient. In this embodiment, the analog sampling frequency... F S The Hz frequency is 9600Hz. To design a 3Hz low-pass filter based on this data, the filter coefficients must be smaller to achieve a steep cutoff characteristic. More preferably, when implementing this invention in an embedded system of a relay protection device, to ensure the accuracy of filter identification and the stability of the filter, a double-precision data type is used to define the filter gain and sensing coefficients. This allows for precise storage of the calculated minimum coefficients, avoiding accumulated errors.

[0040] Specifically, the filtering stage, the first IIR Low-pass filter and second IIR The transfer function of a low-pass filter is expressed by the following formula:

[0041]

[0042]

[0043] In the formula: H ( z ), H 1( z )and H 2( z ) represent the filtering stage and the first stage, respectively. IIR Low-pass filter and second IIR The transfer function of a low-pass filter; b 10 , b 11 and b 12 Indicates first IIR low-pass filter FIR Partial coefficients, b 20 , b 21 and b 22 Indicates the second IIR low-pass filter FIR The forward coefficients, also known as the partial coefficients, mainly control the zero-point and passband gain of the filter. a 10 , a 11 and a 12 Indicates first IIR low-pass filter IIR Partial coefficients, a 20 , a 21 and a 22 Indicates the second IIR low-pass filter IIR The feedback coefficients, also known as the partial coefficients, mainly control the poles and stability of the filter. z -1 This represents the unit delay operator, which in a discrete-time system is equivalent to delaying by one sampling period.

[0044] Further optimization, but not limitation, is made regarding the filtering stage and the first... IIR Low-pass filter and second IIR The transfer function of a low-pass filter is expressed by the following formula:

[0045]

[0046] In the formula: H ( z ), H 1( z )and H 2( z ) represent the filtering stage and the first stage, respectively. IIR Low-pass filter and second IIR The transfer function of a low-pass filter.

[0047] Step 3: Based on the filtered active power signal, perform FFT calculation to identify a rough estimate of the oscillation frequency.

[0048] Preferably, but not limitingly, step 3 specifically includes: The filtered active power signal is processed using a multi-cycle method. P Filter Perform cycle number identification; specifically, within the set... N sampling Within each sampling period, the filtered active power signal P Filter Historical data is used to calculate peaks and troughs. When two consecutive peaks or troughs appear, it is considered that the low-frequency oscillation has completed one cycle. At this point, a set length is used to calculate the peaks and troughs. WIN_ typical of FFT Roughly perform time window FFT Calculate and identify a rough estimate of the oscillation frequency. f rough .

[0049] Preferably, but not limitingly, the set N sampling Each sampling period is 1.5 sampling periods, that is N sampling =1.5; set length as WIN _ typical of FFT The time window is 12.8. s ,Right now WIN _ typical =12.8 s .

[0050] Step 4: Based on the frequency band to which the rough estimate of the oscillation frequency belongs, select the corresponding time window, perform FFT calculation on the filtered active power signal, and obtain the preliminary estimate of the oscillation frequency and the active power signal after eliminating the bias.

[0051] Preferably, but not limitingly, step 4 specifically includes: Step 4.1: Rough estimation based on oscillation frequency f rough Frequency band k , k = 1, 2, …, N Obtain the corresponding downsampling factor WIN k and corresponding FFT Time window WIN _ T ( k ).

[0052] Step 4.2: Select FFT Time window WIN _ T ( k ) the filtered active power signal P Filter conduct FFT Calculate and obtain the maximum value of the oscillation waveform. P MAX Minimum value P MIN and preliminary estimate of oscillation frequency f initial .

[0053] Step 4.3: Based on the maximum value of the oscillation waveform P MAX Minimum valueP MIN The active power signal after offset elimination is calculated and expressed by the following formula: P Z = P Filter — ( P MAX + P MIN ) / 2 In the formula: P Z This represents the active power signal after offset removal.

[0054] Step 5: Using the same time window, analyze the active power signal after offset elimination. P Z Perform FFT calculations again to obtain the accurate low-frequency oscillation frequency and amplitude.

[0055] Specifically, choosing FFT Time window WIN _ T ( k For the active power signal after offset elimination P Z conduct FFT Calculate and obtain the accurate low-frequency oscillation frequency. f low and amplitude P low This enables rapid and high-precision identification of low-frequency oscillations.

[0056] Embodiment 2 of the present invention provides a system for fast frequency identification of low-frequency signals based on segmented time windows, which runs a fast frequency identification method for low-frequency signals based on segmented time windows as described in Embodiment 1, including: a main control unit, an analog signal acquisition stage, an instruction receiving stage, an active power calculation stage, a filtering stage, a low-frequency signal identification stage, and an identification result output stage.

[0057] The analog quantity acquisition link is connected to the three-phase voltage and three-phase current of the monitoring line, and the three-phase voltage and current of the monitoring line are acquired at a set sampling frequency FS. The instruction receiving link determines whether to start low-frequency signal identification by receiving the start low-frequency identification instruction issued by the main control unit, i.e., remote signaling status. The communication cycle between the main control unit and the system is 20Hz. The active power calculation step is used to calculate the three-phase fundamental active power to obtain the three-phase fundamental active power Pj of the monitored line. The filtering stage includes two cascaded second-order IIR low-pass filters, namely the first IIR low-pass filter and the second IIR low-pass filter. In other words, the filtering stage is equivalent to a low-pass filter with a total order of 4. The low-frequency signal identification process identifies a frequency range of 0.1Hz to 2.5Hz. Based on the filtered active power signal, an FFT calculation is performed to identify a rough estimate of the oscillation frequency. According to the frequency band to which the rough estimate of the oscillation frequency belongs, the corresponding time window is selected, and an FFT calculation is performed on the filtered active power signal to obtain a preliminary estimate of the oscillation frequency. The same time window is used for the active power signal after bias elimination, and an FFT calculation is performed again on the active power signal after bias elimination to obtain an accurate low-frequency oscillation frequency and amplitude. The identification result output stage sends the accurate low-frequency oscillation frequency and amplitude as the system identification result, i.e., telemetry, to the main control unit.

[0058] Embodiment 3 of the present invention provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is loaded onto the processor, it implements a method for fast frequency identification of low-frequency signals based on a segmented time window as described in Embodiment 1.

[0059] Embodiment 4 of the present invention provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements a method and system for fast frequency identification of low-frequency signals based on a segmented time window, as described in Embodiment 1.

[0060] To more clearly illustrate the outstanding substantive features of this invention and the significant progress it brings to the prior art, an application example of implementing this invention is described below.

[0061] The method for fast frequency identification of low-frequency signals based on segmented time windows includes the following steps: Step 1: Using 0.25, 0.35, 0.85, 1.35, and 1.85 as dividing nodes, divide the range from 0.1 to 2.5. Hz The frequency range is divided into 6 bands, and a downsampling factor is configured for each of the 6 bands. The set of downsampling factors is represented by the following formula: FFT win_6 ={ WIN 1, WIN 2, WIN 3, WIN 4, WIN 5, WIN 6}={120, 100, 48, 32, 24, 12}.

[0062] Table 2 Frequency range and interval frequency table for each segment

[0063] Based on the downsampling factor for each frequency band, the corresponding time window for each frequency band is calculated, as expressed by the following formula:

[0064] In the formula: WIN _ T ( i ) indicates the first i The first time window, i.e. the first i The time window corresponding to each frequency band; N FFT This indicates the number of data points used in the FFT. N FFT = 1024; F S This indicates the raw sampling rate of the three-phase voltage and current of the monitored line. F S =9600 Hz .

[0065] Therefore, there are 6 time windows corresponding to the 6 frequency bands. WIN _ T ( i They are respectively: WIN _ T (1) = 12.8 s ; WIN _ T (2) = 10.67 s ; WIN _ T (3) = 5.12 s ; WIN _ T (4) = 1.28 s ; WIN _ T (5) = 2.56 s ; WIN _ T (6) = 3.41 s .

[0066] Step 2, with the set value of 9600 Hz sampling frequency F SObtain the three-phase voltage and current of the monitored line, and calculate the three-phase fundamental active power of the monitored line. P j It can be expressed by the following formula:

[0067] In the formula: P j This represents the three-phase fundamental active power, measured in kilowatts. U A , U B and U C These represent the three-phase voltage amplitudes, in volts. I A , I B and I C These represent the amplitudes of the three-phase currents, in amperes. Re This indicates taking the real part.

[0068] After receiving the low-frequency identification command from the main control unit, the command receiving stage sets the cutoff frequency to 3. Hz The filtering stage affects the three-phase fundamental active power. P j Filtering is performed to obtain the filtered active power signal. P Filter ,like Figure 3 As shown.

[0069] Among them, the filtering stage, the first IIR Low-pass filter and second IIR The transfer function of a low-pass filter is expressed by the following formula:

[0070]

[0071] In the formula: H ( z ), H 1( z )and H 2( z ) represent the filtering stage and the first stage, respectively. IIR Low-pass filter and second IIR The transfer function of a low-pass filter.

[0072] Step 3: Within the set 1.5 sampling periods, process the filtered active power signal using a multi-cycle method. P Filter Frequency identification was performed, and a frequency of 1 was identified when a second consecutive peak appeared; at this point, a 12.8s time window was used for coarse analysis. FFT The calculated low-frequency value was 1.6. Hz .

[0073] Step 4, rough estimate of oscillation frequency: 1.6 Hz ∈ [1.35 Hz 1.85 Hz This corresponds to frequency band 5, and the corresponding downsampling factor is obtained. WIN 5 = 24 and the corresponding FFT Time window WIN _ T (5) = 2.56 s ; choose FFT Time window WIN _ T (5) = 2.56 s The filtered active power signal P Filter conduct FFT Calculate and obtain the maximum value of the oscillation waveform. P MAX = 252kW, minimum value P MIN = Preliminary estimate of 248kW and oscillation frequency f initial =1.5625Hz; The active power signal after offset elimination is then calculated and expressed by the following formula: P Z = P Filter — ( P MAX + P MIN ) / 2 In the formula: P Z This represents the active power signal after offset removal.

[0074] Step 5, use the same time window, i.e., select FFT Time window WIN _ T (5) The active power signal after offset elimination P Z conduct FFTCalculate and obtain the accurate low-frequency oscillation frequency. f low = 1.5625 Hz and amplitude P low = 2 kW This enables rapid and high-precision identification of low-frequency oscillations.

[0075] Through the above calculations, low-frequency identification was achieved within two cycles, and the low-frequency oscillation frequency was obtained as 1.5625. Hz The actual oscillation frequency is 1.56. Hz The relative error is only 0.16%. This achieves rapid and high-precision identification of low-frequency oscillations.

[0076] It is worth noting that in the embodiments of the present invention, "steps + numbers" is only an expression for clearly describing the specific implementation of the method for fast frequency identification of low-frequency signals based on segmented time windows, and is not an absolute restriction on the order of the steps. Under the guidance of the core concept of the present invention, changing the order of these steps to obtain the same or similar technical effects all fall within the scope of the present invention.

[0077] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.

Claims

1. A method for fast frequency identification of low-frequency signals based on piecewise time windows, characterized in that, Includes the following steps: The low-frequency band is divided into multiple bands, and different downsampling factors and corresponding time windows are configured for each band. The three-phase voltage and current of the monitoring line are acquired, the three-phase fundamental active power of the monitoring line is calculated, and the active power signal is filtered to obtain the filtered active power signal. Based on the filtered active power signal, FFT calculation is performed to identify a rough estimate of the oscillation frequency; Based on the frequency band to which the rough estimate of the oscillation frequency belongs, select the corresponding time window, perform FFT calculation on the filtered active power signal, and obtain the preliminary estimate of the oscillation frequency and the active power signal after eliminating the bias. Using the same time window, perform FFT calculation again on the active power signal after offset removal to obtain accurate low-frequency oscillation frequency and amplitude.

2. The method for fast frequency identification of low-frequency signals based on piecewise time windows according to claim 1, characterized in that: The step of dividing the low-frequency band into multiple frequency bands and configuring different downsampling factors and corresponding time windows for each frequency band includes: Set a segmented FFT time window, including: dividing the low-frequency band range into... N Each frequency band has a set downsampling factor, with larger downsampling intervals for low-frequency bands and smaller downsampling intervals for high-frequency bands, as expressed by the following formula: FFT win_N ={ WIN i | i = 1, 2, …, N ， WIN i > WIN i+1} In the formula: FFT win_N Denotes the set of downsampling factors, with N One element, WIN i Indicates the first i One downsampling factor, WIN i ∈ FFT win_N , i = 1, 2, …, N ; Based on the downsampling factor for each frequency band, the corresponding time window for each frequency band is calculated, as expressed by the following formula: In the formula: WIN _ T ( i ) indicates the first i A time window, N FFT This indicates the number of data points used in the FFT. F S This indicates the original sampling rate of the three-phase voltage and current of the monitored line.

3. The method for fast frequency identification of low-frequency signals based on piecewise time windows according to claim 2, characterized in that: Using 0.25, 0.35, 0.85, 1.35, and 1.85 as dividing points, the frequency range of 0.1~2.5Hz is divided into 6 frequency bands. The raw sampling rate of the three-phase voltage and current of the monitoring line F S =9600 Hz ; Total number of data points used in each FFT N FFT = 1024.

4. A method for fast frequency identification of low-frequency signals based on piecewise time windows according to any one of claims 1 to 3, characterized in that: The process of acquiring the three-phase voltage and current of the monitoring line, calculating the three-phase fundamental active power of the monitoring line, filtering it, and obtaining the filtered active power signal includes: At the set sampling frequency F S Obtain the three-phase voltage and current of the monitored line, and calculate the three-phase fundamental active power of the monitored line. P j ; To set the cutoff frequency f cut-off The filtering stage affects the three-phase fundamental active power. P j Filtering is performed to obtain the filtered active power signal. P Filter ; The filtering stage is a low-pass filter with a total order of 4.

5. The method for fast frequency identification of low-frequency signals based on segmented time windows according to claim 4, characterized in that: The filtering stage includes two cascaded second-order IIR low-pass filters, namely the first IIR low-pass filter and the second IIR low-pass filter. The transfer functions of the filtering stage, the first IIR low-pass filter, and the second IIR low-pass filter are expressed by the following formula: In the formula: H ( z ), H 1( z )and H 2( z Let f(x) represent the transfer functions of the filtering stage, the first IIR low-pass filter, and the second IIR low-pass filter, respectively. b 10 , b 11 and b 12 Indicates first IIR low-pass filter FIR Partial coefficients, b 20 , b 21 and b 22 This represents the FIR coefficients of the second IIR low-pass filter; a 10 , a 11 and a 12 Indicates first IIR low-pass filter IIR Partial coefficients, a 20 , a 21 and a 22 This represents the IIR coefficients of the second IIR low-pass filter; z -1 This indicates the unit delay operator.

6. A method for fast frequency identification of low-frequency signals based on piecewise time windows according to any one of claims 1 to 3, characterized in that: The rough estimate of the oscillation frequency obtained by performing FFT calculation based on the filtered active power signal includes: The filtered active power signal is processed using a multi-cycle method. P Filter Frequency number identification is performed, where, within the set... N sampling Within each sampling period, the filtered active power signal P Filter Historical data is used to calculate peaks and troughs. When two consecutive peaks or troughs appear, it is considered that the low-frequency oscillation has completed one cycle. At this point, a set length is used to calculate the extreme points. WIN _ typical of FFT Roughly perform time window FFT Calculate and identify a rough estimate of the oscillation frequency. f rough .

7. The method for fast frequency identification of low-frequency signals based on piecewise time windows according to claim 6, characterized in that: Set N sampling Each sampling period is 1.5 sampling periods; Set length as WIN _ typical of FFT The time window is 12.

8. s .

8. A method for fast frequency identification of low-frequency signals based on piecewise time windows according to any one of claims 1 to 3, characterized in that: The process of selecting the corresponding time window based on the rough estimate of the oscillation frequency to perform FFT calculation on the filtered active power signal to obtain the preliminary estimate of the oscillation frequency and the active power signal after bias elimination includes: Based on a rough estimate of the oscillation frequency f rough Frequency band k , k = 1, 2, …, N Obtain the corresponding downsampling factor WIN k and corresponding FFT Time window WIN _ T ( k ); choose FFT Time window WIN _ T ( k ) the filtered active power signal P Filter conduct FFT Calculate and obtain the maximum value of the oscillation waveform. P MAX Minimum value P MIN and preliminary estimate of oscillation frequency f initial ; Based on the maximum value of the oscillation waveform P MAX Minimum value P MIN The active power signal after offset elimination is calculated and expressed by the following formula: P Z = P Filter — ( P MAX + P MIN ) / 2 In the formula: P Z This represents the active power signal after offset removal.

9. A fast frequency identification system for low-frequency signals based on piecewise time windows, comprising a fast frequency identification method for low-frequency signals based on piecewise time windows according to any one of claims 1 to 8, characterized in that, include: Analog signal acquisition stage, command reception stage, active power calculation stage, filtering stage, low frequency signal identification stage, identification result output stage; The analog quantity acquisition stage connects to the three-phase voltage and three-phase current of the monitoring line, at a set sampling frequency. F S Acquire the three-phase voltage and current of the monitored line; The instruction receiving stage determines whether to begin low-frequency signal identification by receiving a low-frequency identification start instruction issued by the main control unit. The active power calculation step is used to calculate the three-phase fundamental active power to obtain the three-phase fundamental active power of the monitored line. Pj ; The filtering stage includes two cascaded second-order IIR low-pass filters, namely the first IIR low-pass filter and the second IIR low-pass filter. The low-frequency signal identification step performs FFT calculation based on the filtered active power signal to identify a rough estimate of the oscillation frequency. According to the frequency band to which the rough estimate of the oscillation frequency belongs, the corresponding time window is selected, and FFT calculation is performed on the filtered active power signal to obtain a preliminary estimate of the oscillation frequency. The same time window is used for the active power signal after bias elimination, and FFT calculation is performed again on the active power signal after bias elimination to obtain an accurate low-frequency oscillation frequency and amplitude. The identification result output stage sends the accurate low-frequency oscillation frequency and amplitude as the system identification result, i.e., telemetry, to the main control unit.

10. A fast frequency identification system for low-frequency signals based on a segmented time window according to claim 9, characterized in that: The communication cycle between the main control unit and the low-frequency signal fast frequency identification system of the segmented time window is 20Hz. The frequency band identified by the low-frequency signal identification step is 0.1Hz to 2.5Hz.

Images