A network-constructing direct-drive wind power system oscillation suppression method based on impedance matching

Abstract

The application relates to an impedance-matching-based network-connection direct-drive wind power system oscillation suppression method and belongs to the field of network-connection wind power system oscillation suppression. The method comprises the following steps: building a network-side virtual synchronous machine control structure and establishing a sequence impedance model based on a static natural coordinate system; and based on the established sequence impedance model, a strategy based on impedance remodeling is constructed to suppress the oscillation of the network-connection direct-drive wind power system. The application can effectively suppress the oscillation and provides reliable technical support for safe operation of the network-connection direct-drive wind power system.

Description

Technical Field

[0001] This invention belongs to the field of oscillation suppression in grid-connected wind power systems, and specifically relates to an oscillation suppression method for grid-connected direct-drive wind power systems based on impedance matching. Background Technology

[0002] To address the issues of inertia deficiency and insufficient frequency response in renewable energy grid-connected systems, Virtual Synchronous Generator (VSG) technology has been widely applied in grid-connected wind power systems due to its ability to simulate the external characteristics of synchronous generators and actively provide inertial and voltage / frequency support. The IEEE Task Force working group first proposed the VSG concept in 1997, pointing out that converters can simulate the power droop characteristics of synchronous generators through control loops. Professor Qingchang Zhong of the University of Liverpool further developed the "Synchronverter" strategy, establishing an equivalent relationship between the rotor motion equations of the converter and the synchronous generator, achieving accurate simulation of the external characteristics of synchronous generators by power electronic equipment. Subsequent research has shown that VSG can effectively improve the dynamic performance of wind power grid-connected systems.

[0003] However, in actual engineering, the grid-side converter (GSC) of grid-connected direct-drive permanent magnet wind turbines (D-PMSG) often adopts voltage-current dual closed-loop control. If the parameters are not configured properly (such as the current loop integral coefficient being too high), it can easily lead to dynamic response mismatch in the control loop, causing low-frequency (0-10Hz) oscillations. Although this type of oscillation has a different frequency range than subsynchronous oscillation (SSO), it is essentially a resonance problem caused by the interaction between "converter control and grid impedance". At best, it can lead to excessive total harmonic distortion (THD) of the grid connection current; at worst, it can cause the wind turbine to disconnect from the grid and the grid frequency to drop sharply.

[0004] While significant progress has been made in research on oscillation suppression in wind power systems, a lack of specificity remains a concern: Regarding the target of suppression, existing methods primarily focus on the subsynchronous frequency band, paying less attention to low-frequency (around 10Hz) resonance caused by grid-side dual-loop parameter mismatch. In terms of suppression strategies, current approaches mostly employ "full-band adjustment," and existing technologies have not deeply analyzed the resonant coupling mechanism between dual-loop parameter mismatch and grid impedance in direct-drive wind turbine VSG control. Summary of the Invention

[0005] The purpose of this invention is to provide a method for suppressing oscillations in grid-connected direct-drive wind power systems based on impedance matching, addressing the low-frequency oscillation problem caused by improper dual-loop parameters of the grid-side converter when using VSG.

[0006] To achieve the above objectives, the technical solution of the present invention is: a method for suppressing oscillations in a grid-connected direct-drive wind power system based on impedance matching, comprising:

[0007] A virtual synchronous machine control structure on the grid side is constructed, and a sequence impedance model is established based on the stationary natural coordinate system.

[0008] Based on the established sequence impedance model, a strategy based on impedance reshaping is constructed to suppress oscillations in grid-type direct-drive wind power systems.

[0009] Furthermore, the grid-side virtual synchronous machine control structure adopts a grid-type control permanent magnet direct-drive wind turbine grid-connected system, including a wind turbine, a machine-side converter MSC, a grid-side converter GSC, a filter circuit, a transmission line, and an AC power grid connected in sequence; the machine-side converter MSC operates in rectification mode to realize wind turbine control, and the grid-side converter GSC operates in inverter mode to transmit electrical energy to the AC power grid through the filter circuit and transmission line; the machine-side converter MSC and the grid-side converter GSC are connected through a DC capacitor.

[0010] Furthermore, a sequence impedance model is established based on the stationary natural coordinate system, as follows:

[0011] The sequence impedance modeling method involves injecting a positive disturbance with frequency f at the grid connection point PCC. p The positive-sequence disturbance voltage; specifically, the AC variable with frequency f is the fundamental frequency and the AC variable with frequency f is the positive-sequence disturbance voltage. p The small-signal components are transformed from the time domain to the frequency domain. The frequency domain expression of the product of the two variables is solved using the frequency domain convolution theorem, yielding the small-signal voltage and current values ​​at each frequency at the measurement point. Ignoring the relatively small proportion of higher-order small-signal terms, the final value is determined by the positive disturbance frequency f. p The positive sequence impedance formula is derived from the voltage-current signal ratio:

[0012]

[0013] It is virtual impedance. This represents the combined transfer function from the power outer loop to the Park transform and then to the voltage-current dual loop. L represents the combined coupling gain from the power outer loop to the Park converter and then to the voltage feedback. f R f and C f Let be the inductance, resistance, and capacitance of the filter circuit, respectively, and s be the Laplace operator.

[0014] Furthermore, the impedance reshaping strategy is a targeted impedance reshaping strategy based on adaptive frequency tracking.

[0015] Furthermore, the targeted impedance reshaping strategy based on adaptive frequency tracking dynamically adjusts the center frequency band of the suppression channel by identifying the dominant oscillation frequency of the system in real time, thereby achieving adaptive suppression of multimodal broadband oscillations.

[0016] Furthermore, the targeted impedance reshaping strategy based on adaptive frequency tracking includes:

[0017] Real-time frequency detection based on improved sliding DFT: Identifying the dominant oscillation frequency in grid-connected current. ;

[0018] Phase-shift-free oscillation component extraction: based on the detected Dynamically configure bandpass filter parameters to extract oscillating components without introducing phase lag;

[0019] Targeted impedance reshaping channel: The extracted oscillation component is injected into the current loop reference value through a designed transfer function to reshape the impedance of a predetermined frequency band.

[0020] Furthermore, based on the improved sliding DFT, real-time frequency detection is implemented using the sliding discrete Fourier transform (SDFT), specifically as follows:

[0021] set up Let N be the sampling sequence of the grid connection point current, and N be the sampling window length at any given time. The spectral value of the kth harmonic It can be obtained through recursion using the following formula:

[0022]

[0023] The spectral value of the kth harmonic is 𝑋 𝑘 [𝑛] via the spectrum X from the previous time step k [n-1] multiplied by a phase factor In addition to the sampling points that newly enter the window The contribution, minus the sampling points removed from the window. The contributions were recognized;

[0024] Subsequently, the energy at each frequency point was calculated. ;

[0025] The system's current dominant oscillation frequency Determined by the frequency point with the highest energy:

[0026]

[0027] in, This represents the system's sampling frequency.

[0028] Furthermore, the extraction of phase-shift-free oscillation components is specifically implemented as follows:

[0029] To achieve the dominant oscillation frequency Accurate filtering of nearby components is achieved using a fourth-order Chebyshev type I IIR band-stop filter, which is composed of two identical second-order band-stop filters cascaded together. The transfer function of a single second-order band-stop filter is:

[0030]

[0031] Where s is the Laplace operator, The center angular frequency is updated in real time; The quality factor of the filter determines the stopband width, and its value is based on the impedance bandwidth to be reshaped; the desired stopband bandwidth is set as... ,but Determined by the following formula:

[0032]

[0033] Ultimately, the transfer function of the fourth-order bandstop filter is the product of two second-order sections:

[0034]

[0035] To implement this in a digital controller, the analog transfer function needs to be... Discretization into digital filters Using the bilinear transformation method, the mapping relationship is as follows:

[0036]

[0037] in Let z be the sampling period of the control system, where z is a complex variable in the discrete time domain, and z⁻¹ represents the unit delay of one sampling period;

[0038] After discretization, the group delay of the digital filter is obtained through simulation or computation tools. Subsequently, a depth of [value] is set in the forward path. The delay unit for the original signal Alignment is performed; the final oscillation component is obtained from the following formula:

[0039]

[0040] Original signal The input center frequency is The output obtained from the band-stop filter, Original signal The input delay time is The output obtained from the delay unit.

[0041] Furthermore, the targeted impedance reshaping channel is specifically implemented as follows:

[0042] Extracted phase-shift-free oscillation components At the reference value of the injected current inner loop, the control law is designed as follows:

[0043]

[0044] in, It is a remodeling controller to be designed. and These are the d-axis current reference value and q-axis current reference value input to the inner current loop before impedance reshaping, respectively. and These are the d-axis current reference value and q-axis current reference value input to the inner current loop after impedance reshaping, respectively;

[0045] To achieve inductive impedance reshaping to offset capacitive negative resistance, A leading phase needs to be provided; its transfer function is designed as follows:

[0046]

[0047] The s in the numerator provides the required lead phase, and the denominator is a second-order system with a center frequency set to . Gain The damping ratio ζ determines the effective bandwidth of the reshaping effect, which in turn determines the magnitude of the virtual impedance.

[0048] Further design a parameter adaptive mechanism to monitor the grid connection point voltage and current in real time, and estimate the cascaded amplitude margin of the current system through an online frequency sweep method. and phase margin To adjust .

[0049] Furthermore, regarding the estimated phase margin of the current system... If it deviates from the expected value Then adjust according to the following formula :

[0050]

[0051] Where α is the learning rate, and α is the actual phase margin at the current time n. Below expectations At that time, slowly increase the gain for the next time step. To increase damping; conversely, to decrease gain at the next moment. To avoid causing excessive interference to the system, This represents the gain at the current moment.

[0052] Compared with the prior art, the present invention has the following beneficial effects: the present invention can effectively suppress oscillations and provide reliable technical support for the safe operation of grid-type direct-drive wind power systems. Attached Figure Description

[0053] Figure 1 This is the topology of a grid-connected system for permanent magnet direct-drive wind turbines.

[0054] Figure 2 This refers to the system topology and control structure.

[0055] Figure 3 The frequency sweep results and impedance model are shown.

[0056] Figure 4 This is a normal impedance diagram of the power grid and the virtual synchronous generator (VSG).

[0057] Figure 5 This refers to the oscillation of active power.

[0058] Figure 6 This is an FFT analysis of the oscillation.

[0059] Figure 7 This is an impedance diagram of the power grid and the virtual synchronous generator (VSG) during oscillation.

[0060] Figure 8 This is the overall control architecture after impedance reshaping.

[0061] Figure 9 This is a flowchart of the real-time frequency detection process based on SDFT.

[0062] Figure 10 This is a block diagram illustrating the principle of extracting phase-shift-free oscillation components.

[0063] Figure 11 This is a schematic diagram of the injection point for the targeted impedance reshaping channel. Detailed Implementation

[0064] The technical solution of the present invention will now be described in detail with reference to the accompanying drawings.

[0065] This invention provides a method for oscillation suppression in grid-connected direct-drive wind power systems based on impedance matching, comprising:

[0066] A virtual synchronous machine control structure on the grid side is constructed, and a sequence impedance model is established based on the stationary natural coordinate system.

[0067] Based on the established sequence impedance model, a strategy based on impedance reshaping is constructed to suppress oscillations in grid-type direct-drive wind power systems.

[0068] The following is a detailed implementation process of the present invention.

[0069] This invention addresses the low-frequency oscillation problem caused by improper dual-loop parameters of the grid-side converter in grid-connected direct-drive wind power systems employing VSGs. The core research approach is "impedance reshaping." First, a grid-side VSG control structure is constructed, treating the direct-drive wind turbine as an equivalent DC current source. A sequence impedance model is established based on a stationary natural coordinate system, and its accuracy is verified through frequency sweep experiments. Second, the oscillation mechanism of dual-loop parameter mismatch is analyzed in depth, clarifying the intrinsic relationship between the excessive phase lag in the low-frequency band caused by the high integral coefficient of the current loop and the resonance formed by the grid's inductive impedance. Finally, two targeted suppression strategies are proposed, and their effectiveness in suppressing low-frequency oscillations and improving system stability is verified through simulation. This research provides technical support for the stable operation of grid-connected direct-drive wind power systems in complex grid environments and offers a new analytical perspective and solution for suppressing low-frequency oscillations in power electronics-dominated systems.

[0070] 1. Control structure and sequence impedance modeling of the grid-side virtual synchronous machine

[0071] 1.1 Control Structure of Grid-Side Virtual Synchronizer

[0072] The research object of this invention is a grid-connected system of permanent magnet direct-drive wind turbines using grid-based control. Its system topology is shown in Figure 1, mainly consisting of wind turbines, a turbine-side converter (MSC), a grid-side converter (GSC), filter circuits, power lines, and an AC power grid. The turbine-side converter operates in rectification mode to control the wind turbine, while the grid-side converter operates in inverter mode, transmitting electrical energy to the power grid via filter circuits and line impedance.

[0073] The turbine-side converter and grid-side converter of a direct-drive wind turbine are connected via a DC capacitor. This DC capacitor decouples the turbine and grid sides, minimizing the impact of instantaneous power fluctuations on the grid side. Furthermore, in actual operation, wind speed changes are slow, and the turbine has significant inertia, making it equivalent to a constant power source compared to grid-side energy changes. Therefore, the turbine side of the direct-drive wind power system can be equated to a DC current source, while the grid-side converter employs Virtual Synchronous Control (VSG), a typical grid control method. The simplified system topology and grid-side control structure are as follows: Figure 2 As shown, its control loop consists of an instantaneous power calculation loop, a power outer loop, and a voltage generation loop, in sequence. Among these, V... dc e is the voltage across the DC-side capacitor. sa e sb and e sc L is the three-phase voltage at the outlet of an infinite power grid. g and R g The equivalent impedance of the transmission line can be varied to change the short-circuit ratio of the power grid, simulating power grid environments of different intensities. L f R f and C fThese are the inductor, resistor, and capacitor of the filter circuit, respectively. oabc i is the three-phase voltage of the power grid at the grid connection point. oabc V represents the three-phase current at the grid connection point. iabc i represents the three-phase voltage at the inverter output. iabc ω represents the three-phase current at the inverter output. N It is the reference angular frequency, which is usually set to the angular frequency at 50Hz.

[0074] 1.2 Sequence Impedance Modeling

[0075] This invention employs a sequence impedance modeling method in a static natural coordinate system. The most widely used virtual synchronous control fP control method and droop-type EQ control method are used in the sequence impedance modeling process. The frequency and voltage control elements are shown in the following equations:

[0076]

[0077] in and These are the active power droop coefficient and the reactive power droop coefficient, respectively. and These are the actual and reference values ​​of active power, respectively. and These are the actual and reference values ​​of reactive power, respectively. and These are the reference and actual values ​​of the angular frequency, respectively. J and are the damping and inertia coefficients, respectively. It is a given value of voltage. and These are the amplitude and phase angle generated by the fP loop and the EQ loop, respectively.

[0078] Traditional virtual impedance control involves introducing current feedback into the voltage loop to dynamically adjust the voltage reference value. and These are the d-axis and q-axis components of the inverter output current, respectively. and These are the d-axis and q-axis components of the voltage outer loop reference value, respectively. and These represent the d-axis and q-axis components of the voltage generated by the power outer loop, respectively. and These are the d-axis and q-axis components of the resistance, respectively. and These are the d-axis and q-axis components of the reactance, respectively. Transient components are ignored in the dq-axis reference coordinate system, i.e.:

[0079]

[0080] The sequence impedance modeling method involves injecting a frequency f at the PCC point (connection point) between the grid-connected inverter and the grid. p The positive-sequence disturbance voltage, with its amplitude typically set between 1% and 5% of the power frequency voltage, will not cause significant fluctuations in the system and thus will not alter its stable operating point. First, the nonlinear element is addressed by first converting the AC variable with frequency f to the fundamental frequency. p The small-signal components are transformed from the time domain to the frequency domain, thus converting the AC variable into a DC variable in the frequency domain, facilitating convolution operations. Then, the frequency domain expression of the product term of the two variables can be solved using the frequency domain convolution theorem, obtaining the small-signal voltage and current values ​​at various frequencies at the measurement point. Depending on the required modeling accuracy, higher-order small-signal terms with a smaller proportion are appropriately ignored, and finally, the value is obtained from the positive disturbance frequency f. p The positive sequence impedance formula is derived from the voltage-current signal ratio:

[0081]

[0082] The above is a simplified result of the positive sequence impedance. It is virtual impedance. This represents the combined transfer function from the power outer loop to the Park transform and then to the voltage-current dual loop. This represents the combined coupling gain from the power outer loop to the Park transform and then to the voltage feedback. If the expression is not integrated, it will be filled with many intermediate variables, making the formula too complicated and causing inconvenience for analysis.

[0083] Impedance modeling and frequency sweep verification were performed. At the harmonic injection point, the amplitude was set to 0.03Un, and the frequency f... p The disturbance voltage ranges from 2.5 Hz to 1000 Hz. Figure 3 It can be seen that the frequency sweep results are basically consistent with the impedance model, which verifies the accuracy of the sequence impedance modeling.

[0084] 1.3 Oscillation Mechanism Analysis

[0085] When a grid-type converter uses dual closed-loop control, improper parameter settings can cause a mismatch in the dynamic response between the outer voltage loop and the inner current loop, which in turn can lead to oscillations. Figure 4 As can be seen, under normal parameters, the phase of the system impedance under VSG control is always greater than 0, that is, the system impedance is inductive, and the grid impedance is also inductive. At this time, the maximum phase difference between the two is about 60 degrees.

[0086] When the dual closed-loop parameters are incorrect, such as the current loop integral coefficient being too high, adjusting the voltage loop proportional coefficient, voltage loop integral coefficient, and current loop proportional coefficient to 1, and adjusting the current loop integral coefficient to 50, will cause the system to oscillate. Figure 5 It is an oscillation of active power.

[0087] FFT analysis of the grid connection point current revealed that the oscillation frequency was concentrated around 10Hz. Figure 6 As shown. Now observe the impedance of the VSG and the power grid. Figure 7 At this point, fluctuations also occurred in some regions of the VSG impedance, especially in the area around 10Hz, where severe fluctuations occurred. Furthermore, most of the region below 10Hz was in negative phase, meaning the system was capacitive. This demonstrates that changes in the double closed-loop parameters affect the system impedance characteristics. Consequently, the capacitive VSG impedance is prone to resonance with the inductive power grid, leading to oscillations. Figure 7 Within the range in the middle frame, you can see that the two curves in this range of the amplitude-frequency curve are relatively close and have many intersection points. In the corresponding phase-frequency curve below, the VSG curve has many abrupt changes, some even abruptly changing to below -150 degrees of phase. At this time, the grid phase is more than 30 degrees. When the phase difference between the two exceeds 180 degrees, it will cause instability in the system.

[0088]

[0089] The phase lag in the voltage-current dual closed loop is mainly caused by the phase lag in the current loop, which consists of three parts. Indicates the phase of the controlled object. Indicates the phase of the PI controller. Indicates the phase delay; substituting the data, the total phase lag of the current loop can be calculated to be -141.5°. Figure 7 They are quite close. This is the actual value of the angular frequency. It is the proportional coefficient of the inner current loop. It is the integral coefficient of the inner current loop. It is the sampling period.

[0090] When a control system (especially the current loop) generates significant phase lag in the low-frequency range, from an impedance perspective, this is equivalent to connecting a capacitor in parallel at the VSG output. The equivalent capacitance can be derived from the impedance magnitude:

[0091]

[0092] Consider the equivalent capacitance C of the VSG eq and grid inductance L g By substituting into the formula below, we can find that the resonant frequency is around 10Hz. The theoretical derivation and simulation are consistent. It can be seen that the cause of the oscillation is the resonance between the VSG equivalent capacitance and the inductive grid impedance.

[0093]

[0094] The root cause of the oscillation after adjusting the above parameters is the improper current loop parameters, especially the excessively high current loop integral coefficient, which severely reduces the closed-loop bandwidth. This leads to excessive phase lag accumulation in the 0 to 10 Hz frequency band, causing the system phase margin to become negative. In terms of impedance interaction, this reshapes the characteristics of the VSG output impedance in this frequency band, making it easier for it to form a harmful resonant circuit with the inductive mains impedance.

[0095] 2. Targeted Impedance Reshaping Strategy Based on Adaptive Frequency Tracking

[0096] Traditional impedance reshaping strategies based on fixed-band filters suffer significant reductions or even failure in suppression effectiveness when grid impedance or operating point changes. To address this issue, this invention proposes an Adaptive Frequency-Tracking Targeted Impedance Reshaping (AFT-IR) strategy. This strategy dynamically adjusts the center band of the suppression channel by identifying the dominant oscillation frequency of the system in real time, achieving adaptive suppression of multimode broadband oscillations.

[0097] 2.1 Overall Control Architecture

[0098] The overall control architecture of the proposed AFT-IR strategy is as follows: Figure 8 As shown, it mainly consists of three core modules:

[0099] (1) Real-time frequency detection module based on improved sliding DFT: used to identify the dominant oscillation frequency in the grid connection point current. .

[0100] (2) Phase-off-shift-free oscillation component extraction module: Based on the detected... Dynamically configure bandpass filter parameters to accurately extract oscillation components without introducing phase lag.

[0101] (3) Targeted impedance reshaping channel module: The extracted oscillation component is injected into the current loop reference value through a carefully designed transfer function to achieve precise reshaping of impedance in a specific frequency band.

[0102] The following section will provide a detailed modeling and design analysis of each module.

[0103] 2.2 Real-time frequency detection module based on improved sliding DFT

[0104] Rapid and accurate detection of the dominant oscillation frequency This is a prerequisite for the adaptive strategy. To achieve rapid tracking of time-varying oscillation frequencies, this strategy employs the computationally efficient Sliding Discrete Fourier Transform (SDFT) algorithm. This algorithm updates the spectrum recursively within each sampling period, avoiding the massive computational burden and time delay of the traditional Fast Fourier Transform (FFT), making it particularly suitable for online real-time analysis.

[0105] The algorithm principle is as follows: Let Let N be the sampling sequence of the grid-connected point current, and N be the sampling window length. At any given time... The spectral value of the kth harmonic It can be obtained recursively using the following formula:

[0106]

[0107] The spectral value of the kth harmonic is 𝑋 𝑘 [𝑛] via the spectrum X from the previous time step k [n-1] multiplied by a phase factor In addition to the sampling points that newly enter the window The contribution, minus the sampling points removed from the window. The contribution is obtained; thus, there is no need to perform a complete N-point FFT calculation, which greatly improves computational efficiency.

[0108] Subsequently, the energy at each frequency point was calculated. ;

[0109] The system's current dominant oscillation frequency Determined by the frequency point with the highest energy:

[0110]

[0111] in, This is the system's sampling frequency. This module can output accurate data in real time with an extremely short delay of several sampling periods. This provides a frequency reference for subsequent processes. The implementation process is as follows: Figure 9 As shown.

[0112] 2.3 Extraction of Oscillation Components Without Phase Shift

[0113] After obtaining the oscillation frequency, the frequency component needs to be extracted from the grid-connected point voltage signal without distortion. If a traditional Infinite Impulse Response (IIR) bandpass filter is used directly, its inherent nonlinear phase response will distort the oscillation waveform, causing incorrect phase compensation in the feedback control based on this signal, which in turn degrades system stability. To solve this problem, this invention designs a delay-aligned bandstop synthesis structure, the core of which is the precise design of a bandstop filter (BRF).

[0114] To achieve the dominant oscillation frequency Accurate filtering of nearby components is achieved using a fourth-order Chebyshev type I IIR band-stop filter, consisting of two identical second-order band-stop filters cascaded together, to obtain a steeper stopband attenuation. The transfer function of a single second-order band-stop filter is:

[0115]

[0116] Where s is the Laplace operator, The center angular frequency is updated in real time; The quality factor of the filter determines the stopband width, and its value is based on the impedance bandwidth to be reshaped; the desired stopband bandwidth is set as... (Unit: Hz) Determined by the following formula:

[0117]

[0118] To ensure sufficient suppression while maintaining system robustness, it is usually set to... Ultimately, the transfer function of the fourth-order bandstop filter is the product of two second-order sections:

[0119]

[0120] Through cascade design, while ensuring While achieving deep notch filtering (theoretically zero gain), it also achieves sharper frequency selectivity.

[0121] To implement this in a digital controller, the analog transfer function needs to be... Discretization into digital filters This invention employs the bilinear transformation method, and its mapping relationship is as follows:

[0122]

[0123] in Let z be the sampling period of the control system, where z is a complex variable in the discrete time domain, and z⁻¹ represents the unit delay of one sampling period;

[0124] After discretization, the group delay of the digital filter is obtained through simulation or computation tools. (Unit: number of sampling periods), then, a depth of [missing information] is set in the forward path. The delay unit for the original signal Alignment is performed; the final oscillation component is obtained from the following formula:

[0125]

[0126] Workflow diagram as follows Figure 10 As shown, the original signal The input is fed into the delay unit, and its delay time is... It is precisely configured to have the same delay as the band-stop filter group in the other path. The original signal is then fed into another path with a center frequency of... The band-stop filter, its output To filter out the background signal that causes oscillations, the subtractor performs the operation and outputs the dominant oscillation frequency.

[0127] The core advantage of this structure lies in ensuring that the two signals performing the subtraction operation are synchronized in time through aligned delays. This is despite the oscillating component in the final output. There is a fixed transmission delay. However, it has an approximately linear phase response within the passband, meaning it has no phase distortion. This ensures that the subsequent controller receives a "morphologically accurate" oscillating signal, laying a solid foundation for generating the correct suppression signal.

[0128] 2.4 Design of Targeted Impedance Reshaping Channel

[0129] Extracted phase-shift-free oscillation components It is fed into the impedance reshaping channel, the purpose of which is to change the inverter's impedance. Near the output impedance, inject damping to eliminate negative resistance characteristics.

[0130] Extracted phase-shift-free oscillation components At the reference value of the inner loop of the injected current, such as Figure 11 As shown, the control law is designed as follows:

[0131]

[0132] in, It is a remodeling controller to be designed. and These are the d-axis current reference value and q-axis current reference value input to the inner current loop before impedance reshaping, respectively. and These are the d-axis current reference value and q-axis current reference value input to the inner current loop after impedance reshaping, respectively;

[0133] To achieve inductive impedance reshaping to offset capacitive negative resistance, A leading phase needs to be provided; its transfer function is designed as follows:

[0134]

[0135] The s in the numerator provides the required lead phase (approximate differential property), and the denominator is a second-order system with its center frequency set to... Gain Determine the magnitude of the virtual impedance. The larger the value, the stronger the reshaping effect; its value can be determined by the stability margin. The damping ratio ζ determines the effective bandwidth of the reshaping effect, and is usually taken as 0.5 to 1.0 to ensure a certain bandwidth while avoiding excessive peaks.

[0136] The controller generates peak gain and maximum lead phase at a certain point, thereby accurately and powerfully reshaping the inverter output impedance phase in this frequency band. It can significantly improve the phase, completely destroy the phase condition that resonates with the grid impedance, and fundamentally suppress oscillation.

[0137] To ensure the proposed strategy maintains optimal performance across a wide operating range, a parameter adaptive mechanism was designed. The controller monitors the grid connection point voltage and current in real time and estimates the cascaded gain margin of the current system using an online frequency sweep method. and phase margin To adjust With phase margin For example, if it deviates from the expected value Then adjust according to the following formula :

[0138]

[0139] Where α is the learning rate, a small positive number (e.g., 0.01~0.1), representing the actual phase margin at the current time n. Below expectations At that time, slowly increase the gain for the next time step. To increase damping; conversely, to decrease gain at the next moment. To avoid causing excessive interference to the system, This represents the gain at the current moment.

[0140] The three modules of this strategy are interconnected, forming an intelligent closed-loop control system of "perception-decision-execution". Through frequency adaptation and phase distortion-free signal extraction, it achieves accurate and robust suppression of multimodal oscillations, which has significant advantages in adaptability, stability and dynamic performance compared with traditional fixed-parameter suppression strategies.

[0141] The above are preferred embodiments of the present invention. Any changes made to the technical solution of the present invention that do not exceed the scope of the technical solution of the present invention shall fall within the protection scope of the present invention.

Claims

1. A method for suppressing oscillation of a network-configuration direct-drive wind power system based on impedance matching, characterized in that, Comprise: The grid-side virtual synchronous machine control structure is built, and a sequence impedance model is established based on a stationary natural coordinate system; Based on the established sequence impedance model, a strategy based on impedance reshaping is constructed to suppress the oscillation of the grid-connection type direct-driven wind power system; The strategy based on impedance reshaping is a targeted impedance reshaping strategy based on adaptive frequency tracking, comprising: Real-time frequency detection based on improved sliding DFT: discriminating the dominant oscillation frequency in grid point currents ; Phase-free oscillation component extraction: based on the detected , dynamically configure the bandpass filter parameters to extract the oscillation component without introducing phase lag; The targeted impedance reshaping channel: the extracted oscillation component is injected into the current loop reference value through a designed transfer function, so as to reshape the impedance in the predetermined frequency band; Real-time frequency detection based on improved sliding DFT is implemented by using sliding discrete Fourier transform (SDFT), and the specific implementation manner is as follows: Let is the sampling sequence of the grid-connected point current, N is the sampling window length, at any time , the spectral value of the kth harmonic is recursively obtained by the following formula: The spectral value X of the kth harmonic 𝑘 [ n ] is obtained by multiplying the spectral value X of the previous time instant k [ n - 1 ] by a phase factor , adding the contribution of the sample points newly entering the window , and subtracting the contribution of the sample points moving out of the window . Subsequently, the energy of each frequency point is calculated ; the current dominant oscillation frequency of the system determined from the frequency bin with the highest energy wherein, is the sampling frequency of the system.

2. The impedance matching based network configuration direct drive wind power system oscillation suppression method according to claim 1, characterized in that, The grid-side virtual synchronous machine control structure adopts a grid-connection type control permanent magnet direct-driven wind turbine, which comprises a wind turbine, a machine-side converter (MSC), a grid-side converter (GSC), a filter circuit, a power transmission line and an alternating current power grid connected in sequence; the machine-side converter MSC operates in a rectification mode to realize wind turbine control, and the grid-side converter GSC operates in an inversion mode to transmit electric energy to the alternating current power grid through the filter circuit and the power transmission line; the machine-side converter MSC and the grid-side converter GSC are connected through a direct current capacitor.

3. The impedance matching based network configuration direct drive wind power system oscillation suppression method according to claim 2, characterized in that, The sequence impedance model is established based on the stationary natural coordinate system, and the specific implementation manner is as follows: The sequence impedance modeling method is to inject a positive sequence disturbance voltage with a positive disturbance frequency of f p at a point of common coupling (PCC); specifically, an alternating current variable with a frequency of a fundamental frequency and a small signal component with a frequency of f p are transformed from a time domain to a frequency domain, a frequency domain expression of a product item of the two variables is solved by a frequency domain convolution theorem, and small signal voltage and current values at each frequency in a measurement point are obtained; high-order small signal items with a proportion less than a preset value are ignored, and finally a positive sequence impedance formula is obtained from a voltage current signal ratio of the positive disturbance frequency f p . is a virtual impedance, represents the overall transfer function from the power outer loop to the Park transformation and then to the voltage and current double loop, represents the overall coupling gain from the power outer loop to the Park transformation and then to the voltage feedback, L f , R f , and C f are the inductance, resistance, and capacitance of the filter circuit, respectively, and s is the Laplace operator. 4.The impedance matching based network configuration direct-drive wind power system oscillation suppression method of claim 1, wherein, The targeted impedance reshaping strategy based on adaptive frequency tracking dynamically adjusts the center frequency band of the suppression channel by identifying the dominant oscillation frequency of the system in real time, so as to realize adaptive suppression of multi-modal wide-frequency oscillation.

5. The impedance matching based network configuration direct drive wind power system oscillation suppression method of claim 1, wherein, Phase offset-free oscillation component extraction, the specific implementation manner is as follows: To achieve the accurate filtering of the component near the dominant oscillation frequency A fourth-order Chebyshev type I IIR band-stop filter with two identical second-order band-stop filter cascaded is used, and the transfer function of the single second-order band-stop filter is: where s is the Laplacian operator, is the real-time updated center angular frequency; is the quality factor of the filter, which determines the width of the stop band, and its value is determined by the required reshaped impedance frequency band range; set the desired stop band bandwidth as then is determined by the following formula: Finally, the transfer function of the fourth-order band-stop filter is the product of two second-order nodes: To implement in a digital controller, the analog transfer function is discretized into a digital filter using a bilinear transformation method whose mapping relationship is: wherein is the sampling period of the control system, z is a complex variable in the discrete-time domain, and z-1 represents a unit delay of one sampling period; After discretization, the group delay of the digital filter is obtained by simulation or calculation tools Then, a delay of depth is set in the forward path to align the original signal The final oscillation component is given by is the original signal is the output of a bandpass filter with a center frequency of is the original signal is the output of a delay with a delay time of is the original signal is the output of a delay with a delay time of 6. The impedance matching based network configuration direct drive wind power system oscillation suppression method according to claim 5, characterized in that, The targeted impedance reshaping channel, the specific implementation manner is as follows: extracted phase-free oscillation component The control law is designed as follows at the reference value of the injected current inner loop: wherein, is a remodelling controller to be designed, and are d-axis and q-axis current reference values input to the current inner loop before impedance remodelling, respectively, and are d-axis and q-axis current reference values input to the current inner loop after impedance remodelling, respectively. To achieve a perceptual impedance reshaping to counteract the capacitive negative resistance, A lead phase is required; its transfer function is designed as: The s on the numerator provides the desired lead phase, and the denominator is a second order system with a center frequency set to , and a gain that determines the size of the virtual impedance, and the damping ratio ζ determines the effective bandwidth of the reshaping action. Further design parameter adaptive mechanism, real-time monitoring and grid point voltage and current, through the online frequency sweep method, estimate the current system cascade amplitude margin and phase margin , to adjust .

7. The impedance matching based network configuration direct drive wind power system oscillation suppression method according to claim 6, characterized in that, For the estimated current system phase margin , if it deviates from the expected value , adjust it according to the following formula : wherein a is a learning rate, and when the actual phase margin at the current time n is lower than the expected value , the gain at the next time is slowly increased to enhance the damping; otherwise, the gain at the next time is decreased to avoid excessive disturbance to the system, is the gain at the current time.

Images