Broadband oscillation suppression method for wind power through flexible direct current power transmission grid-connected system

Abstract

The invention discloses a broadband oscillation suppression method of a wind power flexible direct current transmission grid-connected system, and relates to the technical field of broadband oscillation suppression of a power system, and the method comprises the steps: firstly, recognizing the frequency information of broadband oscillation according to a multi-frequency oscillation frequency online recognition link; then, designing a multi-band wave trap by adopting a windowing orthogonal matching pursuit (OMP) algorithm; and finally, providing a real-time center frequency for the multi-frequency wave trap according to a frequency identification link so as to realize broadband oscillation self-adaptive suppression. According to the method, the time-varying oscillation frequency can be effectively tracked through an online identification algorithm, so that the method adapts to oscillation scenes under different system working conditions. The windowing OMP method provided by the invention can avoid the problem that the coefficient of the wave trap needs to be redesigned when the center frequency of the wave trap changes, reduces the calculation amount, improves the real-time performance of the algorithm, and achieves the self-adaptive suppression of broadband oscillation.

Description

Technical Field

[0001] The present invention relates to the technical field of broadband oscillation suppression in power systems, and in particular to a method for suppressing broadband oscillation in a wind power grid-connected system via flexible direct current transmission. Background Art

[0002] As the penetration rate of renewable energy continues to rise, flexible direct current (HVDC) transmission technology has become a key solution for large-scale wind power transmission. In actual system operation, broadband oscillations frequently occur due to the interaction between wind turbines and the flexible direct current (HVDC) transmission system. Several typical oscillation cases have occurred globally. For example, the wind power system in Guyuan, Hebei Province, my country, experienced 3-12 Hz oscillations; the Shanghai Nanhui wind power grid-connected flexible direct current (HVDC) system experienced 20-30 Hz oscillations; the Yunnan Luxi back-to-back flexible direct current (HVDC) project experienced oscillations of approximately 1270 Hz; and the North Sea flexible direct current project in Germany experienced 250-350 Hz oscillations. Broadband oscillations exhibit multimodal characteristics and significant frequency-varying properties. Broadband oscillations are extremely harmful, not only damaging key components of generators but also triggering cascading grid disconnections at renewable energy sites, seriously threatening the safe and stable operation of power systems. Therefore, research on methods to suppress broadband oscillations is crucial for ensuring the stable operation of power systems.

[0003] Notch technology has the advantage of filtering out oscillation components without affecting system damping, and its parameters are easy to design. However, the current method based on fixed notch filters has the disadvantage that when the actual system operating conditions vary over a wide range, the fixed notch filter suppression strategy formulated for a specific scenario is difficult to adapt to the time-varying oscillation frequency. Among the existing improvement schemes, adaptive notch technology combined with online identification algorithms can improve the suppression effect, but it can only suppress a single oscillation mode and cannot cope with the coexistence of multiple frequency oscillation modes. Because broadband oscillations have the characteristics of multi-mode coexistence and dynamic changes with the system operating conditions, conventional suppression methods often find it difficult to maintain a stable control effect. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to overcome the deficiencies of the prior art and provide a method for suppressing broadband oscillations in a wind power grid-connected system via flexible direct current transmission. The present invention can effectively suppress broadband oscillations of multiple oscillation modes.

[0005] The present invention adopts the following technical solutions to solve the above technical problems:

[0006] A method for suppressing broadband oscillations in a wind power grid-connected system via flexible direct current transmission proposed in the present invention includes:

[0007] The online identification link of broadband oscillation frequency is established by using the all-phase fast Fourier transform (ApFFT) method based on Hanning window dual-spectral interpolation.

[0008] Design a multi-frequency notch filter using the windowed OMP algorithm;

[0009] According to the online identification link of broadband oscillation frequency, the real-time center frequency is provided for the multi-frequency notch filter to realize broadband oscillation adaptive suppression.

[0010] As a further optimization scheme of the broadband oscillation suppression method for a wind power grid-connected system via flexible direct current transmission described in the present invention, the broadband oscillation frequency online identification step includes:

[0011] Collect the voltage signal of the system grid connection point and update the signal sequence; every time t s The voltage signal is collected once, and the broadband oscillation parameter estimation is performed on the signal sequence using the Hanning window bispectral line interpolation ApFFT at N points. The system is a wind power grid-connected system via flexible direct current transmission. The frequency f of the broadband oscillation is i and amplitude A i The calculation formula is:

[0012]

[0013] Among them, k m , k0 are the maximum spectral line position of the Fourier spectrum and the spectral line position corresponding to the actual frequency, N is the number of sampling points, W(·) represents the windowed spectrum function, y1 and y2 are the amplitudes corresponding to the maximum spectral line and the amplitudes corresponding to the second maximum spectral line, a is k0 and k m The deviation, f s is the sampling frequency;

[0014] Determine the calculated amplitude A of different broadband oscillations i Whether the ratio of the amplitude of the multi-frequency notch to the fundamental frequency amplitude A0 exceeds the set threshold ε, if it exceeds the threshold, the broadband oscillation frequency is input to the multi-frequency notch filter as the center frequency of the multi-frequency notch filter.

[0015] As a further optimization scheme for the broadband oscillation suppression method of a wind power grid-connected system via flexible direct current transmission described in the present invention, a method for designing a multi-frequency notch filter based on a windowed OMP algorithm includes:

[0016] (1) Input multiple notch point sets ω i , stopband bandwidth Δω i and the passband ripple α of the multi-frequency notch filter as the initial parameters of the OMP algorithm;

[0017] (2) Calculate δ based on the relationship between the passband ripple δ of the ideal prototype notch filter and the passband ripple α of the multi-frequency notch filter in formula (3), and use this δ as a constraint condition for the OMP algorithm;

[0018]

[0019] Where r is the number of notch frequencies;

[0020] (3) Execute the windowed OMP algorithm as follows:

[0021] i) Initialize the residual phasor r (0) =F; F is the ideal prototype notch filter;

[0022] ii) calculating the coefficient matrix of the least squares solution, which is the coefficient matrix of the single-frequency notch filter for iteration 1;

[0023] iii) performing k iterations to obtain sparse single-frequency notch filter coefficients;

[0024] iv) Adding a Chebyshev window to optimize the notch depth and passband ripple of the single-frequency notch filter;

[0025] v) Calculate the error ε of the kth iteration (k) , the calculation method is as follows:

[0026]

[0027] Among them, ||·||2 is the l2 norm, is the k-th iteration windowed notch filter, F d (e jω ) is an ideal notch filter, e is the Euler number, j is the imaginary unit, and ω is the notch angular frequency;

[0028] If ε (k) Output single-frequency notch filter coefficient when <δ Among them, f n is the nth single-frequency notch filter coefficient; otherwise, return to step i) and re-execute the OMP algorithm;

[0029] (4) According to the linear relationship between the single-frequency notch filter coefficient and the multi-frequency notch filter coefficient, the multi-frequency notch filter coefficient is calculated. And the corresponding multi-frequency notch filter H(e jω ); where h n is the nth multi-frequency notch filter coefficient.

[0030] Compared with the prior art, the present invention adopts the above technical solution and has the following technical effects:

[0031] (1) The present invention firstly adopts the dual-spectrum interpolation ApFFT method based on Hanning window to establish the online identification link of multi-frequency oscillation frequency; secondly, the multi-frequency notch filter is designed by the windowed OMP algorithm; finally, the frequency identification link is used to provide the multi-frequency notch filter with a real-time center frequency to achieve broadband oscillation adaptive suppression;

[0032] (2) This method can not only effectively track the time-varying oscillation frequency through the online identification algorithm, thus adapting to the oscillation scenarios under different system working conditions, but also the windowed OMP algorithm proposed in this method can avoid the problem of redesigning the notch filter coefficients when the notch filter center frequency changes, thus reducing the computational complexity, improving the real-time performance of the algorithm, and realizing adaptive suppression of broadband oscillations. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 Flowchart of the windowed OMP algorithm used in the specific implementation of the present invention;

[0034] Figure 2 Flowchart of the suppression method of the adaptive multi-frequency notch filter in a specific embodiment of the present invention. DETAILED DESCRIPTION

[0035] In order to make the objectives, technical solutions and advantages of the present invention more clear, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0036] First, the voltage signal of the system grid connection point is collected and the signal sequence is updated. s The voltage signal is collected once, and the broadband oscillation parameters are estimated based on the Hanning window dual spectrum interpolation ApFFT at N points on the signal sequence. The frequency and amplitude calculation formula of the broadband oscillation is:

[0037]

[0038] Among them, k m , k0 are the maximum spectral line position of the Fourier spectrum and the spectral line position corresponding to the actual frequency, N is the number of sampling points, W(·) represents the windowed spectrum function, y1, y2 are the amplitude corresponding to the maximum spectral line and the amplitude corresponding to the second maximum spectral line, a is k0 and k m The deviation, f s is the sampling frequency.

[0039] Determine the calculated amplitude A of different broadband oscillations i Whether the ratio of the amplitude of the multi-frequency notch to the fundamental frequency amplitude A0 exceeds the set threshold ε, if it exceeds the threshold, the broadband oscillation frequency is input to the multi-frequency notch filter as the center frequency of the multi-frequency notch filter.

[0040] Secondly, the windowed OMP algorithm is used to design the multi-frequency notch filter, as shown in the attached figure. Figure 1 shown.

[0041] 1) Input multiple notch point sets ω i , stopband bandwidth Δω i and the passband ripple α of the multi-frequency notch filter as the initial parameters of the OMP algorithm;

[0042] 2) Calculate δ based on the relationship between the passband ripple δ of the single-frequency notch filter and the passband ripple α of the multi-frequency notch filter in equation (3), and use this δ as a constraint condition for the OMP algorithm.

[0043]

[0044] Where r is the number of notch frequencies;

[0045] 3) Execute the windowed OMP algorithm, the steps include:

[0046] i) Initialize the residual phasor r (0) =F; F is the ideal prototype notch filter;

[0047] ii) calculating the coefficient matrix of the least squares solution, which is the coefficient matrix of the single-frequency notch filter for iteration 1;

[0048] iii) performing k iterations to obtain sparse single-frequency notch filter coefficients;

[0049] iv) Adding a Chebyshev window to optimize the notch depth and passband ripple of the single-frequency notch filter;

[0050] v) Calculate the error ε of the kth iteration (k) , the calculation method is as follows:

[0051]

[0052] Among them, ||·||2 is the l2 norm, is the k-th iteration windowed notch filter, F d (e jω ) is an ideal notch filter, e is the Euler number, j is the imaginary unit, and ω is the notch angular frequency;

[0053] If ε (k) Output single-frequency notch filter coefficient when <δ Among them, f n is the nth single-frequency notch filter coefficient; otherwise, return to step i) and re-execute the OMP algorithm;

[0054] 4) According to the linear relationship between the single-frequency notch filter coefficient and the multi-frequency notch filter coefficient, the multi-frequency notch filter coefficient is calculated And the corresponding multi-frequency notch filter H(e jω ); where h n is the nth multi-frequency notch filter coefficient;

[0055] If the center frequency of the multi-frequency notch filter shifts, it is only necessary to perform the coefficient correction process in step (4). This notch filter coefficient adjustment mechanism completely avoids the iterative calculation step and significantly reduces the computational complexity during the sparse reconstruction of the multi-frequency notch filter, making this technical solution particularly suitable for online applications of broadband oscillation suppression.

[0056] Finally, according to the frequency identification link, the real-time center frequency is provided for the multi-frequency notch filter to realize broadband oscillation adaptive suppression, as shown in the attached figure. Figure 2 shown.

[0057] Use t s =1s sampling time window, sampling frequency is 2000Hz, and threshold ε is set to 2%. During data processing, frequency identification is performed using a sliding time window with a step size of 0.015s. When the Hanning window dual-line interpolation ApFFT algorithm detects that the threshold ε is greater than 2%, the corresponding frequency is input to the multi-frequency notch filter as the center frequency. After the multi-frequency notch filter enters the center frequency, the windowed OMP algorithm is used to design a multi-frequency notch filter that meets the requirements. The multi-frequency notch filter is then re-introduced into the system. When the oscillation is suppressed and the multi-frequency notch filter is removed, the broadband oscillation suppression process ends.

[0058] The above description is only a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by a technician familiar with this technical field within the technical scope disclosed by the present invention should be covered by the scope of protection of the present invention.

Claims

1. A method for suppressing broadband oscillations in a wind power grid-connected system via flexible direct current transmission, characterized in that: include: The online identification link of broadband oscillation frequency is established by using the all-phase fast Fourier transform (ApFFT) method based on Hanning window dual-spectral interpolation. Design a multi-frequency notch filter using the windowed OMP algorithm; According to the online identification link of broadband oscillation frequency, the real-time center frequency is provided for the multi-frequency notch filter to realize broadband oscillation adaptive suppression.

2. The method for suppressing broadband oscillations in a wind power grid-connected system via flexible direct current transmission according to claim 1, characterized in that: The online identification of broadband oscillation frequency includes: Collect the voltage signal of the system grid connection point and update the signal sequence; every time t s The voltage signal is collected once, and the broadband oscillation parameter estimation is performed on the signal sequence using the Hanning window bispectral line interpolation ApFFT at N points. The system is a wind power grid-connected system via flexible direct current transmission. The frequency f of the broadband oscillation is i and amplitude A i The calculation formula is: Among them, k m , k0 are the maximum spectral line position of the Fourier spectrum and the spectral line position corresponding to the actual frequency, N is the number of sampling points, W(·) represents the windowed spectrum function, y1 and y2 are the amplitudes corresponding to the maximum spectral line and the amplitudes corresponding to the second maximum spectral line, a is k0 and k m The deviation, f s is the sampling frequency; Determine the calculated amplitude A of different broadband oscillations i Whether the ratio of the amplitude of the multi-frequency notch to the fundamental frequency amplitude A0 exceeds the set threshold ε, if it exceeds the threshold, the broadband oscillation frequency is input to the multi-frequency notch filter as the center frequency of the multi-frequency notch filter.

3. The method for suppressing broadband oscillations in a wind power grid-connected system via flexible direct current transmission according to claim 1, characterized in that: The method for designing a multi-frequency notch filter based on the windowed OMP algorithm includes: (1) Input multiple notch point sets ω i , stopband bandwidth Δω i and the passband ripple α of the multi-frequency notch filter as the initial parameters of the OMP algorithm; (2) Calculate δ based on the relationship between the passband ripple δ of the ideal prototype notch filter and the passband ripple α of the multi-frequency notch filter in formula (3), and use this δ as a constraint condition for the OMP algorithm; Where r is the number of notch frequencies; (3) Execute the windowed OMP algorithm as follows: i) Initialize the residual phasor r (0) =F; F is the ideal prototype notch filter; ii) calculating the coefficient matrix of the least squares solution, which is the coefficient matrix of the single-frequency notch filter for iteration 1; iii) performing k iterations to obtain sparse single-frequency notch filter coefficients; iv) Adding a Chebyshev window to optimize the notch depth and passband ripple of the single-frequency notch filter; v) Calculate the error ε of the kth iteration (k) , the calculation method is as follows: Among them, ||·||2 is the l2 norm, is the k-th iteration windowed notch filter, F d (e jω ) is an ideal notch filter, e is the Euler number, j is the imaginary unit, and ω is the notch angular frequency; If ε (k) Output single-frequency notch filter coefficient when <δ Among them, f n is the nth single-frequency notch filter coefficient; otherwise, return to step i) and re-execute the OMP algorithm; (4) According to the linear relationship between the single-frequency notch filter coefficient and the multi-frequency notch filter coefficient, the multi-frequency notch filter coefficient is calculated. And the corresponding multi-frequency notch filter H(e jω ); where h n is the nth multi-frequency notch filter coefficient.

Images