Network construction energy storage type STATCOM (SVG) low-frequency oscillation suppression method based on adaptive filter

Abstract

The invention relates to the field of network construction type power systems, in particular to a network construction energy storage type STATCOM (SVG) low-frequency oscillation suppression method based on an adaptive filter. A network construction type energy storage STATCOM (SVG) can provide inertia and frequency support for a system, so that the stability, the electric energy quality and the access capability of renewable energy sources of a power system are improved, but the problem of low-frequency oscillation still exists during grid connection, and the problem of low-frequency oscillation of different frequencies possibly occurring during grid connection is solved. According to the low-frequency oscillation suppression method based on the self-adaptive filter, by monitoring an unstable low-frequency mode, accurately estimating the low-frequency oscillation frequency and self-adaptively adjusting filter parameters, low-frequency oscillation of different frequencies of a network-building energy storage type STATCOM (SVG) is suppressed, and the grid-connected operation stability of the network-building energy storage type STATCOM (SVG) is improved.

Description

Technical Field

[0001] The present invention relates to the field of grid-type power systems and is an adaptive low-frequency oscillation suppression method for a grid-type energy storage STATCOM. Background Art

[0002] The static synchronous compensator (STATCOM) is a dynamic reactive power compensation device based on power electronics technology. It was first proposed in the 1980s and widely used in the 1990s. The energy storage STATCOM uses fully controlled switching devices (IGBTs) to form a self-commutated inverter, and combines small-capacity energy storage elements to achieve fast and continuous reactive power and active power compensation, such as Figure 1 Its main functions include improving grid voltage transient stability, suppressing bus voltage fluctuations and improving power quality.

[0003] However, traditional energy storage STATCOMs have limitations, such as limited response speed during grid faults and an inability to provide system inertia support. This has led to the emergence of grid-connected STATCOMs, which aim to mimic the characteristics of traditional synchronous generators through power electronics to provide voltage, frequency, and short-circuit current support for the grid. Grid-connected STATCOMs utilize a power synchronization control mechanism to transform them into voltage sources, capable of independently establishing AC voltage amplitude and phase without relying on the external grid. However, low-frequency oscillations still exist when connected to the grid.

[0004] Therefore, in order to solve the low-frequency oscillation problem, the present invention designs an adaptive low-frequency oscillation suppression method for a grid-type energy storage STATCOM, which suppresses low-frequency oscillations of different frequencies by adaptively adjusting filter parameters. Summary of the Invention

[0005] The grid-connected energy storage STATCOM can not only provide inertia and frequency support for the power grid, but also avoid the influence of the phase-locked loop on the grid-connected stability of the system, solving the problems of low system inertia and negative resistance effect easily caused by the phase-locked loop controller. However, there is still an urgent need to further reduce the low-frequency oscillation problem of its grid connection. Therefore, the present invention designs a low-frequency oscillation suppression method based on an adaptive filter to suppress the low-frequency oscillation of the grid-connected energy storage STATCOM. First, the adaptive filter is installed in the frequency error path of the VSG control structure, and then the low-frequency oscillation information is obtained based on the compaction technology approximate projection subspace tracking algorithm, the oscillation frequency and oscillation state are judged, and then the filter weight coefficients W1 and W2 are adaptively adjusted to filter out the low-frequency oscillation component.

[0006] The technical solution of the present invention is: a method for suppressing low-frequency oscillation of a networked energy storage type STATCOM based on an adaptive filter.

[0007] The beneficial effects of the present invention are: suppressing the low-frequency oscillation of the grid-connected energy storage type STATCOM, and the suppression effect is not affected by the operating conditions and the time-varying oscillation frequency.

[0008] The specific content is elaborated as follows:

[0009] 1. Installation location of adaptive filter:

[0010] The principle of adaptive filter to suppress low frequency oscillation is to filter out the transmission of low frequency oscillation current in the control loop of network energy storage type STATCOM, which is represented by H. When low frequency oscillation occurs in the system, the low frequency component of the control loop of network energy storage type STATCOM is introduced by its output current, which in turn affects the active power output. Therefore, the adaptive filter is installed in the frequency error path of VSG control. Figure 2 shown.

[0011] 2. Low-frequency oscillation signal extraction method based on approximate projection subspace tracking based on compaction technology:

[0012] The function of the low-frequency signal estimator is to monitor unstable low-frequency modes and accurately estimate the low-frequency oscillation frequency. It is installed at the grid-connected exit of the STATCOM energy storage type. The specific steps of the projected subspace tracking method using the compaction technology are as follows:

[0013] Firstly, the output current of the grid-type energy storage STATCOM is used as the input signal of the low-frequency signal estimator and represented as the superposition of several sinusoidal signals with attenuation coefficients and noise.

[0014]

[0015] Where: r is the number of complex sinusoidal components in the sampled signal, T s is the sampling period, B k 、 τ k 、f k are the amplitude, initial phase, attenuation coefficient and frequency of the kth sinusoidal component respectively; ω(n) is white noise.

[0016] The sampling data vector x(t) at time t is expressed as:

[0017] x(t)=[x(t),x(t+1),x(t+2),...,x(t+m-1)] T (2)

[0018] m is the window length

[0019] Define the unconstrained cost function

[0020]

[0021] Where: W(t) is the parameter matrix of m×r order introduced to construct the objective function; x(t) is the current vector output at time t, β is the forgetting factor, and c(t) is the parameter matrix introduced to construct the objective function.

[0022] When the cost function reaches the global minimum, W(t) approaches the eigenvector of the signal subspace S in c(t), thereby estimating the relevant oscillation information in the signal. W(t) is solved iteratively using the recursive least squares method. By deleting the first and last rows of the matrix W(t), we can obtain submatrices W1 and W2. Thus, the signal subspace S is expressed as

[0023] S=(W1) + W2 (5)

[0024] Where (.) + is the pseudo-inverse matrix.

[0025] By calculating the eigenvalue λ of the signal subspace S k The frequency and attenuation coefficient of the sinusoidal component in the signal can be estimated:

[0026]

[0027] τ k =(ln|λ k |) / T s (7)

[0028] τ k Greater than 0 indicates that the system diverges with a frequency of f k If the value is less than 0, it means that the system has sufficient damping at the corresponding frequency. To avoid unnecessary malfunction of the device, it is necessary to extract a low-frequency oscillation signal with a frequency range of 4-46Hz from the obtained oscillation signal.

[0029] 3. Adaptive filter design

[0030] When the attenuation coefficient τ is judged to be greater than 0, the weight coefficient is adaptively adjusted so that the characteristics of the filter can change with the changes of harmonics and noise. t (k) is the signal to be filtered at time t, e t (k) is the filtered signal at time t, according to f n Generate a sine and cosine signal with the same frequency, and generate a weighted signal y through the linear combination of weight coefficients W1 and W2 t (k), the weighted signal frequency is f n , the amplitude and phase are controlled by the weight coefficients, and the weight coefficients are updated using the recursive least squares algorithm, such as Figure 3 shown.

[0031] Subscript t corresponds to time t, k represents the iteration relationship, and the weight vector W t (k) The update formula is:

[0032] W t (k) = W t (k-1)+C t (k)ξ(k) (8)

[0033] The prior estimation error is:

[0034] ξ(k)=x t (k)-W t (k-1)e t (k) (9)

[0035] The gain vector is:

[0036]

[0037] The inverse correlation matrix is:

[0038] Z(k)=R -1 (k) = λ -1 [Z(k-1)-C t (k)e t (k)Z(k-1)] (11)

[0039] λ is the forgetting factor, which indicates the stickiness to the previous iteration information. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 Circuit diagram of energy storage STATCOM for VSG control network;

[0041] Figure 2 For the VSG control structure, the adaptive filter H is added in the frequency error path;

[0042] Figure 3 This is the block diagram of the weight coefficient control principle of the adaptive filter H. DETAILED DESCRIPTION

[0043] When the system is running stably, f er =f1, the error is 0, the filter does not work at this time. When low-frequency oscillation occurs, the adaptive suppression measures are as follows:

[0044] Step 1: Install a low-frequency signal estimator at the grid-connected outlet of the grid-connected energy storage STATCOM and decompose the input signal into a number of sinusoidal signals for superposition;

[0045] Step 2: Construct sampling data, set the window length m to 100, set the sampling frequency to 1000 Hz, and update the sampling data vector every 0.1 seconds;

[0046] Step 3: construct a cost function, estimate the signal subspace, and solve the eigenvector of the signal subspace;

[0047] Step 4, delete the first and last rows of the signal subspace matrix, and then update the signal subspace;

[0048] Step 5: By calculating the eigenvalues of the signal subspace, estimate the frequency and attenuation coefficient of the sinusoidal component in the signal. If the attenuation coefficient is greater than 0, it indicates that the system has divergent oscillations and needs to be suppressed. If the attenuation coefficient is less than 0, it means that the system has sufficient damping in the corresponding frequency band and no suppression is required.

[0049] Step 6: Extract the low-frequency oscillation signal in the frequency range of 4-46 Hz to ensure that the device does not malfunction.

[0050] Step 7: Start the adaptive filter and generate sine and cosine signals with the same frequency as the signal to be filtered. Then, generate a weighted signal through the linear combination of weight coefficients. The weight coefficients are updated using the recursive least squares method, and the forgetting factor is set to 0.998. Then, the designed adaptive filter is added to the VSG control structure to suppress low-frequency oscillations.

[0051] The above steps are used to realize the suppression of low-frequency oscillation by the adaptive filter.

Claims

1. A method for suppressing low-frequency oscillation of a grid-connected energy storage STATCOM (SVG) based on an adaptive filter, characterized in that The details include: 1) A method for suppressing low-frequency oscillations of grid-connected energy storage STATCOMs using a VSG control strategy; 2) The added adaptive controller is an adaptive filter, which is located in the frequency error path of the VSG control of the grid-connected energy storage STATCOM; 3) A low-frequency oscillation information extraction strategy based on a compaction technique approximating the projected subspace tracking algorithm; 4) Adaptively adjust filter parameters based on low-frequency oscillation information.

2. The grid-connected energy storage STATCOM according to claim 1, characterized in that: The main circuit 1) adopts a star-shaped cascaded H-bridge inverter, where each phase bridge arm is composed of multiple H-bridge modules connected in series, and the energy storage unit is connected to the capacitor of each H-bridge through a DC / DC converter. However, the existing invention does not have a method for suppressing low-frequency oscillation of a grid-type energy storage STATCOM.

3. The adaptive filter according to claim 1, wherein The filter added in the error path in step 2) can effectively filter out the low-frequency component of the active power, thereby reducing the risk of low-frequency oscillation, which is represented by H.

4. The low frequency oscillation information extraction strategy based on the compaction technique approximate projection subspace tracking algorithm according to claim 1 is characterized in that: The above 3) determines the oscillation frequency by estimating the frequency of the sinusoidal quantity in the signal, and determines whether it is divergent based on the attenuation coefficient.

5. The adaptive adjustment filter coefficient according to claim 1, characterized in that: 4) The adaptive algorithm adjusts the amplitude and phase weight coefficients in real time W 1 and W 2 to filter out the corresponding low-frequency oscillation component, so that the frequency of the signal to be filtered is constantly approaching the base frequency. Even if the oscillation frequency changes, the low-frequency oscillation can be well suppressed.

Images