Power grid impedance on-line identification method and equipment for network-forming SVG (static var generator), and medium

Abstract

The invention discloses a power grid impedance online identification method and equipment for a network-forming SVG (static var generator), and a medium. The method is executed during grid-connected operation of the SVG, and aims to obtain key basic parameters for evaluating the strength of the power grid in real time on the premise of not interrupting normal operation of the SVG and not increasing extra hardware. The method comprises the following steps of: firstly, applying a specific excitation signal under the condition of not influencing the original adjusting function of the SVG, and synchronously acquiring voltage and current response data of a grid-connected point; secondly, a specially designed signal processing algorithm is adopted, fundamental wave equivalent impedance of the power grid is extracted or estimated from the collected data, and the algorithm can effectively suppress interference caused by switching harmonic waves of the SVG and background noise of the power grid; and finally, outputting the fundamental wave equivalent impedance of the power grid as a basic parameter for accurately reflecting the strength of the power grid in real time.

Description

Technical Field

[0001] This invention relates to the field of power system control technology, and in particular to a method, device and medium for online identification of grid impedance for grid-type SVG (Static Var Generator). Background Technology

[0002] With the large-scale grid connection of new energy power generation, the degree of power electronics in the power system is constantly deepening, and the inertia and damping characteristics of the power system are continuously declining, posing a severe challenge to the stable operation of the power grid. Grid-based control technology, by enabling power electronic converters to simulate the operating characteristics of synchronous generators, can provide the necessary voltage and frequency support for the power grid, becoming one of the key technologies for improving the stability of high-proportion new energy power systems. As a typical grid-based device, the grid-based SVG, through virtual synchronous machine control technology, can exhibit inertia, damping, and voltage regulation characteristics similar to those of a synchronous generator.

[0003] However, existing grid-based SVG virtual synchronous machine control strategies generally suffer from a key drawback: their core control parameters (such as the virtual inertia parameter J and the damping parameter D) are usually preset fixed values. The strength of the actual power grid (usually characterized by indicators such as the short-circuit ratio) changes dynamically due to factors such as load variations, faults, and fluctuations in renewable energy output. Fixed-parameter control strategies struggle to achieve optimal performance across all operating conditions: under strong grid conditions, excessive virtual inertia can delay the system's dynamic response; while under weak grid conditions, insufficient virtual inertia may lead to frequency instability or power oscillations. This contradiction limits the adaptability of grid-based SVGs to a wide range of grid strengths.

[0004] To address the above problems, existing technologies have proposed some approaches to adaptive parameter adjustment, but these typically suffer from the following shortcomings: Limitations of power grid strength sensing methods: Existing methods mostly rely on offline calculations, power grid dispatch data, or external measurement devices to obtain power grid strength information, making it difficult to achieve real-time, online, continuous, and cost-effective sensing. In particular, there is a lack of an online impedance / strength identification method that can be seamlessly integrated into the grid-connected SVG's own control system, without additional hardware, and can be completed synchronously during its normal grid-connected operation.

[0005] The parameter adjustment strategy is crude: Existing adaptive strategies mostly use simple segmentation or table lookup methods, and parameter adjustment is prone to jumps, which can easily cause control system shocks at switching points. Moreover, the adjustment rules fail to fully reflect the dynamic matching relationship that should be satisfied between the power grid strength and the controller parameters.

[0006] Insufficient coordination with control mode switching: Networked SVG typically has multiple operating modes, such as network following mode and network construction mode. Existing technologies fail to organically integrate adaptive parameter adjustment with the adaptive and smooth switching of operating modes. Mode switching often relies on simple threshold triggering, and the lack of coordinated optimization between parameter configuration and switching timing during the switching process can easily lead to transient shocks, affecting the stability of the switching process and system safety.

[0007] Therefore, there is an urgent need for a grid-based SVG integrated control scheme that integrates real-time online sensing of grid strength, intelligent parameter mapping and smooth adaptive adjustment, and seamless switching of cooperative operation modes, so as to comprehensively improve its stability, adaptability and support capabilities in complex and ever-changing grid environments. Summary of the Invention

[0008] To at least solve one of the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: According to a first aspect of the present invention, an online method for identifying grid impedance for a grid-connected SVG is provided. This method is performed during grid-connected operation of the SVG to obtain fundamental parameters for assessing grid strength. The method includes the following steps: S100, without interrupting the normal regulation function of the grid-type SVG and without relying on additional hardware, performs an impedance measurement operation to obtain an equivalent impedance parameter for characterizing the strength of the power frequency grid; the impedance measurement operation includes: applying or using an electrical disturbance as an excitation, synchronously collecting voltage response data and current response data caused by the electrical disturbance at the grid connection point, and then calculating or estimating the equivalent impedance parameter based on the response data.

[0009] S110, the equivalent impedance parameter is output as a basic parameter for real-time assessment of power grid strength. According to a second aspect of the invention, an electronic device is provided, including a processor and a memory; the processor executes the steps of the method described in the first aspect of the invention by calling programs or instructions stored in the memory.

[0010] According to a third aspect of the present invention, a computer-readable storage medium is provided that stores a program or instructions that cause a computer to perform the steps of the method described in the first aspect of the present invention.

[0011] The present invention has at least the following beneficial effects: 1. Achieved highly integrated and low-cost online reporting: Without relying on offline data, external dedicated measurement equipment or power grid dispatch communication, it directly utilizes the control system and sensors of the grid-connected SVG itself to synchronously complete impedance measurement during its normal grid-connected operation, realizing deep integration of sensing functions and significantly reducing system complexity and cost.

[0012] 2. Provides real-time, continuous and reliable intensity sensing data: This method can be executed continuously with a control cycle (millisecond level) at the power electronic device level, and outputs the equivalent impedance or short-circuit ratio information of the power grid in real time, providing an accurate and timely data foundation for upper-level adaptive control. Moreover, the reliability of the identification results is guaranteed by a proprietary anti-interference algorithm under interference such as the switching noise of the grid-type SVG itself.

[0013] 3. Possesses engineering practicality and user-friendliness: Whether it is the careful design of the disturbance amplitude and frequency in the active disturbance method or the intelligent excitation judgment mechanism in the passive analysis method, both ensure that the impact on the power grid power quality is minimal, meeting the practical engineering requirements of non-disturbance or micro-disturbance online measurement.

[0014] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 This invention provides an online identification method for grid impedance of a grid-type SVG. Figure 2 A flowchart of an adaptive virtual synchronization control method for a network-type SVG provided in another embodiment of the present invention; Figure 3 Another embodiment of the present invention provides an adaptive switching control method for the operating mode of a network-type SVG. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of this invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0019] It should be noted that some exemplary embodiments are described as processes or methods depicted as flowcharts. Although the flowcharts describe the steps as sequential processes, many of these steps can be performed in parallel, concurrently, or simultaneously. Furthermore, the order of the steps can be rearranged. A process can be terminated when its operation is complete, but it may also have additional steps not included in the figures. A process can correspond to a method, function, procedure, subroutine, subroutine, etc.

[0020] (Example 1) This embodiment provides a method for online identification of grid impedance for a grid-type SVG. The grid-type SVG, or grid-type static var generator, refers to a static var generator employing grid-type control. A static var generator is a flexible AC transmission system device based on fully controlled power electronic devices (such as IGBTs), and its full English name is Static Var Generator, abbreviated as SVG. Unlike conventional grid-following SVGs that track grid voltage as a current source, grid-type SVGs, by simulating the operation mechanism of synchronous generators, can autonomously generate and adjust the amplitude and phase of their output voltage, thus acting as a controlled voltage source, providing active voltage support, inertia, and damping characteristics to the grid.

[0021] The core feature of the described online grid impedance identification method is that, while the grid-connected SVG is operating normally and performing its core regulation functions such as reactive power compensation and voltage support, this method, based on the aforementioned grid control capabilities, executes synchronously and in parallel. In other words, the identification function is deeply embedded in the main control system of the grid-connected SVG, utilizing its existing sensors, controllers, and power units to measure and estimate the equivalent impedance on the grid side in real time without interrupting or interfering with its main functions, thus achieving the integration of sensing and regulation.

[0022] like Figure 1 As shown in the figure, this embodiment provides a method for online identification of grid impedance for grid-type SVG, which may include the following steps: S100, without interrupting the normal adjustment function of the grid-type SVG and without relying on additional hardware, performs a complete impedance measurement operation to obtain the equivalent impedance parameters used to characterize the strength of the power frequency grid in real time.

[0023] The impedance measurement operation includes: applying or using an electrical disturbance as an excitation, simultaneously collecting voltage response data and current response data caused by the electrical disturbance at the grid connection point, and then calculating or estimating the equivalent impedance parameter based on the response data.

[0024] To ensure measurement accuracy, data acquisition must meet the following general technical requirements: Synchronization: Voltage and current sampling must be strictly synchronized, usually achieved by the same analog-to-digital converter module of the network-type SVG master controller or a channel with synchronous triggering function, to ensure that the electrical state is obtained at the same moment.

[0025] Sampling rate: The sampling frequency fs (fs = 1 / Ts) must satisfy the Nyquist sampling theorem and be significantly higher than the highest frequency component of the signal of interest, typically in the kilohertz range. Ts is the sampling period (a known fixed parameter) of the digital control system of the networked SVG. Typically, the sampling rate is in the kilohertz range.

[0026] Signal preprocessing: The acquired raw signal can first pass through an anti-aliasing filter (usually implemented in the ADC front-end hardware). In the digital domain, further filtering (such as low-pass and band-pass) can be performed to suppress high-frequency switching noise and background harmonics. To simplify subsequent processing of the power frequency fundamental component, the three-phase instantaneous values ​​are usually converted into components in a two-phase stationary coordinate system (αβ) or a synchronous rotating coordinate system (dq) through coordinate transformation (such as Clark transform or Park transform), for example, to obtain u α [k], i α [k] or u d [k], i d [k] and other sequences.

[0027] Specifically, the impedance measurement operation is implemented in any of the following ways: Method (a): Controlled disturbance injection measurement method The core of this method lies in the fact that, while achieving the main control objective, the grid-type SVG itself injects a preset, amplitude-limited reactive power disturbance signal into the power grid as an excitation, simultaneously measuring voltage and current response data. By analyzing the response data at the frequency of the disturbance signal, the equivalent impedance parameter is directly calculated. Specifically, this includes: 1. Generation and injection of disturbance signals: A specific reactive power disturbance signal ΔQ is superimposed on the original reactive power control reference value Qref of the grid-type SVG (usually derived from the output of its voltage regulation loop). inj The reference value for the total reactive power after superposition is Qref. total =Qref+ΔQ injAnd send it to the reactive power control channel of the SVG.

[0028] The disturbance signal ΔQ inj This is a preset, amplitude-limited reactive power disturbance signal. Its core design parameters include: Frequency f dist Choose a low-frequency band distinct from the mains frequency (50Hz / 60Hz) and its major integer harmonics, for example, in the range of 2Hz to 15Hz. This design helps avoid confusion with grid background harmonics and ensures that the SVG's current closed-loop controller still has good tracking performance at this frequency. The frequency f dist An example value is 5Hz or 10Hz.

[0029] Waveform: Preferably a sine wave or a pseudo-random binary sequence with wide frequency spectrum characteristics.

[0030] Amplitude: Strictly limited to 0.5% to 2% of the rated capacity Sn of the grid-type SVG. For example, for an SVG with a rated capacity Sn of 10 MVA, the disturbance amplitude |ΔQ inj The voltage range is from 50kVar to 200kVar. This design ensures that the voltage fluctuation at the grid connection point caused by disturbances is minimal (typically well below 1% of the nominal voltage), meeting grid power quality standards and enabling disturbance-free or minimal-disturbance measurements.

[0031] 2. Synchronous data acquisition and impedance calculation: Injecting perturbation signal ΔQ inj Simultaneously, utilizing the built-in voltage and current sensors of the grid-connected SVG, the instantaneous three-phase voltage sequence at the grid connection point is acquired with high precision and synchronously. abc (k) and the sequence of instantaneous three-phase currents output by SVG i abc (k). The acquisition window length should cover an integer number of cycles of the disturbance signal (e.g., for a 5Hz disturbance, acquire for 0.4 seconds, which is 2 cycles). u abc (k) and i abc (k) constitutes the voltage response data and current response data.

[0032] In this way, the actively injected signal ΔQ inj As a known excitation, the voltage and current responses generated are recorded synchronously, providing a data foundation for subsequent calculations. Subsequently, the equivalent impedance parameters are obtained based on the voltage and current response data, including the following sub-steps: S1001, Data preprocessing: Processing the synchronously acquired time-domain voltage signal u abc (t) and current signal i abc(t) Preprocessing is performed. This preprocessing includes applying a notch filter corresponding to the switching frequency of the SVG power unit to filter out or significantly attenuate the main high-frequency harmonic components generated by the SVG's own switching action, thereby improving the signal-to-noise ratio. The energy of the characteristic harmonic components is mainly concentrated at the switching frequency of the meshed SVG and its integer multiples.

[0033] S1002, Extraction of specific frequency components: Perform frequency domain analysis on the preprocessed time-domain signal to accurately extract the components caused by the injected perturbation ΔQ. inj The response component generated by the excitation. Specifically, it can be: Fast Fourier Transform (FFT): An FFT is performed on a data window covering an integer number of perturbation periods to obtain its discrete spectrum. Within this discrete spectrum, the frequency f corresponding to the known perturbation frequency is located. dist The corresponding frequency index is determined, and the complex result at that frequency index is read and used as the complex component ΔU(f) of the voltage response. dist ) and the complex component of current disturbance ΔI(f dist ), where ΔU(f dist ) and ΔI(f dist All of these are complex numbers, containing the amplitude and phase information of the signal at that frequency.

[0034] Bandpass filtering and correlation detection: using a center frequency of f dist Narrowband digital filters are used to filter the signal, or correlation detection methods are used to extract the in-phase and quadrature components at that frequency.

[0035] S1003, Impedance Calculation: According to Ohm's law, calculate the impedance at the disturbance frequency f. dist The equivalent impedance of the power grid under the following conditions is Zg(f dist ): Zg(f dist )=ΔU(f dist ) / ΔI(f dist )=Rg(f dist )+jXg(f dist ).

[0036] Among them, Rg(f dist ) and Xg(f dist ) are the equivalent resistance and reactance at the center frequency, respectively, and j is the imaginary unit.

[0037] Power frequency impedance mapping: Since the power grid impedance usually changes gradually in the low-frequency range, the frequency f can be mapped. dist The reactance Xg(f) dist It can be directly regarded as an estimate of the equivalent reactance parameter Xg for subsequent strength assessment. Alternatively, a small-amplitude frequency reduction can be performed based on a known power grid impedance frequency characteristic model to obtain a more accurate power frequency impedance value.

[0038] To reduce the impact of random noise, step S100 can be repeated multiple times (e.g., 3 times) to average the calculated set of impedance values ​​as the final output.

[0039] Method (b): Online identification of natural fluctuations This method does not require active disturbance injection. Instead, it treats the inherent natural power fluctuations in the power grid as continuous excitation, continuously collecting voltage and current fluctuation data caused by these fluctuations. An adaptive estimation algorithm (such as recursive least squares with a forgetting factor) is used to estimate the equivalent impedance parameters in real time from the continuous voltage and current fluctuation data stream. Specifically, this includes: 1. Data model establishment: (1) Under the fundamental frequency, the grid connection point of the grid-type SVG is equivalent to a Thevenin circuit. The discrete-time system equation of the Thevenin circuit can be expressed as: u[k]=Rg×i[k]+Lg×(di(t) / dt)+e[k].

[0040] Where u[k] and i[k] are the instantaneous components of the voltage and current at the grid connection point at the kth sampling time (e.g., the d-axis or α-axis components after coordinate transformation); Rg and Lg are the equivalent resistance and inductance of the grid side to be determined; e[k] is the equivalent internal potential of the grid.

[0041] (2) The continuous-time derivative term di(t) / dt is discretized using a first-order backward difference, i.e., at the k-th sampling time, (di(t) / dt)≈(i[k]-i[k-1]) / Ts. Simultaneously, the fundamental angular velocity ω is introduced. s (ω) s =2π×50 or 2π×60rad / s), which can convert the equivalent inductance Lg to be determined into Xg (Xg=ω s ×Lg), where Xg represents the equivalent reactance of the power grid at power frequency as seen from the grid connection point, i.e., the fundamental equivalent impedance of the power grid. After simplification, an approximate equation suitable for parameter identification is obtained: u[k]≈Rg×i[k]+(Xg / ω s )×(i[k]-i[k-1]) / Ts+e[k].

[0042] Further organize into a linear regression form: y[k]=φ T [k]×θ+ε[k].

[0043] in: y[k]=u[k], which is the system output observation value.

[0044] φ[k]=[i[k],(i[k]-i[k-1])] T , which is the observation data vector.

[0045] θ=[Rg,Xg / (ω s ×Ts)] T Let θ be the parameter vector to be identified. The equivalent reactance Xg of the power grid can be directly calculated from the identified θ.

[0046] ε[k] represents the model error or noise term at time k.

[0047] 2. Data Acquisition and Recursive Identification: (1) Continuous data acquisition: Using the built-in sensor of the grid-type SVG, the instantaneous values ​​of voltage u[k] and current i[k] at the grid connection point are continuously acquired to form a data stream. This data stream constitutes the voltage response data and current response data output in this step.

[0048] (2) Recursive Least Squares Algorithm: The recursive least squares algorithm with a forgetting factor is used to update the parameter estimate θ online. hat [k]. The algorithm's recursive steps are as follows: Initialize the parameter estimation vector θ hat [0] and covariance matrix P[0].

[0049] At each sampling time k, execute: Calculate the gain vector: K[k] = P[k-1] × φ[k] / (λ + φ) T [k]×P[k-1]×φ[k]); φ T [k] is the transpose of φ[k]; Update parameter estimation: θ hat [k]=θ hat [k-1]+K[k]×(y[k]-φ T [k]×θ hat [k-1]); Update the covariance matrix: P[k] = (IK[k] × φ) T [k])×P[k-1] / λ; Where λ is the forgetting factor, a constant between 0.95 and 0.9999, whose function is to give higher weight to recent data in the algorithm, thereby enabling it to track time-varying parameters. I is the identity matrix.

[0050] Through the above recursive calculation, the parameter vector θ to be identified (containing Xg information) can be continuously and in real time updated in each control cycle.

[0051] Method (a) is intuitive in principle and simple in calculation, and has high identification accuracy and anti-interference capability in statically stable power grids with low background harmonics. Method (b) is completely non-disruptive and can provide continuous impedance estimates, making it particularly suitable for dynamic power grids with large background fluctuations. Both provide complete and fully disclosed technical solutions for achieving uninterrupted online identification without additional hardware.

[0052] For subsequent processing, the three-phase instantaneous values ​​are typically converted into components in a two-phase stationary coordinate system (αβ) or a synchronously rotating coordinate system (dq) using coordinate transformations (such as Clark transformation or Park transformation). For example, converting to u... α [k],u β [k] and i α [k], i β [k], or further converted to u d [k], u q [k] and i d [k], i q [k]. This helps simplify subsequent processing of the power frequency fundamental component.

[0053] To further improve the long-term measurement accuracy and reliability of method (a) under complex power grid noise environments, this method can also introduce a closed-loop optimization step. This closed-loop optimization step includes: dynamically adjusting the amplitude and / or frequency of the subsequently injected reactive power disturbance signal based on the confidence evaluation index of the current or historical impedance identification results. The core of this closed-loop optimization step is that the result of each impedance identification is not only used as an output, but also as feedback information for optimizing the next excitation signal, forming an intelligent closed loop.

[0054] Specifically, after each step S110 is completed and the reactance estimate Xg is obtained, the following optimization process is executed: (1) Quality assessment: Calculate the confidence assessment index C for this measurement. index This indicator is used to quantify the reliability of the identification results. It can be obtained through any of the following quantifiable methods: Frequency domain signal-to-noise ratio: C index =20×log10(|ΔU(f dist )| / σ noise ), where σ noise Analysis can be performed using the perturbation frequency f dist Outside a narrow interval centered on (i.e., [f]) dist -△f,f dist The voltage or current spectrum energy (Δf, typically 0.5-1 Hz) is used to estimate σ. noise This represents the background noise power within that frequency band.

[0055] Statistical dispersion of results: If N independent and continuous rapid repeated measurements are performed within a complete measurement cycle (e.g., N=3), resulting in a set of reactance estimates {Xg(1), ..., Xg(N)}, then calculate the sample variance σX of this set of data and define C. index =1 / σX, the smaller the dispersion, the higher the confidence level.

[0056] (2) Decision-making and adjustment: C index Compared with the preset high quality threshold C high and low quality threshold C low Compare.

[0057] If C index <C low This indicates that the measurement was severely affected by noise interference, resulting in low confidence. The optimization logic will generate adjustment instructions to adaptively enhance the signal within the parameter constraints of the disturbance signal. Amplitude Adjustment: Within the upper limit of amplitude (2%Sn), adjust the amplitude of the next injected disturbance |ΔQ inj |Increase by one step (e.g., 10% of the current value or a fixed value).

[0058] Frequency adjustment: Within the frequency range (2Hz~15Hz), adjust the disturbance frequency f dist Make a small adjustment (e.g., ±1Hz) to try to avoid strong background harmonic interference at the current moment.

[0059] If C index >C high This indicates excellent measurement conditions and high confidence. The optimization logic can adopt a conservative or optimized strategy: maintain the current disturbance parameters unchanged, or, with a certain margin, attempt to slightly reduce the disturbance amplitude (e.g., reduce it by 5%) to further reduce reactive power disturbance to the power grid and achieve better power quality.

[0060] If C low ≤C index ≤C high This indicates that the measurement confidence level is within an acceptable range, and the optimization logic will maintain the current perturbation parameter (|ΔQ). inj |,f dist )constant.

[0061] The high confidence threshold C high With low confidence threshold C low The value of needs to be determined comprehensively based on the specific control system of the network-type SVG, the sensor accuracy, and the expected measurement performance requirements.

[0062] C high The setting: This typically corresponds to the typical confidence level that the system can achieve under good measurement conditions. For example, C can be...high C was calculated based on multiple measurements taken under ideal laboratory conditions or when background interference from the power grid was minimal. index 80% to 90% of the average.

[0063] C low The setting corresponds to the critical confidence level at which the measurement result is unacceptable and parameter optimization is necessary. It is typically set as the C level corresponding to when the measurement result error exceeds the allowable range (e.g., 10%) in a strong noise interference test. index Value, or set to C high Value is 30% to 50%.

[0064] Basis for setting: Those skilled in the art can determine the C based on the measured data or system simulation results from the initial debugging phase. high and C low Perform calibration.

[0065] (3) Parameter update and iteration The new perturbation parameter (|ΔQ) generated by the above decision inj | new f dist_new The new parameters are updated in the configuration register of the signal generator module. These new parameters will take effect during the impedance identification cycle of the next trigger mode (a), thereby achieving a closed-loop adaptive cycle of measurement-evaluation-optimization.

[0066] The beneficial effect of this closed-loop optimization step is that it enables the measurement system to possess environmental self-sensing and strategy self-adjustment capabilities. When faced with changing background harmonics or random noise, it can automatically find a dynamic optimal balance between injecting a sufficiently large amplitude to obtain a high signal-to-noise ratio and sufficiently small disturbance to the power grid to meet power quality requirements. Thus, it maintains high accuracy and high reliability in long-term operation, significantly improving the practicality and advancement of the method.

[0067] Furthermore, to improve the estimation reliability of method (b) in steady-state or low-excitation scenarios and to prevent the algorithm from diverging due to measurement noise, the online natural fluctuation identification method also includes an excitation level judgment step.

[0068] This step is executed before initiating or updating the recursive least squares algorithm. Its core is to determine whether the energy of the current fluctuation data exceeds a preset threshold. Only if it does so is the recursive least squares algorithm initiated or updated; otherwise, the current estimated equivalent impedance parameter value is maintained. Specifically, this includes: Step 1, Excitation Validity Judgment: Before executing the core identification algorithm, it is first determined whether the natural fluctuations of the current fluctuation data contain sufficient information. Statistical characteristics of the current fluctuation data within a preset time window are calculated as a quantitative indicator of the excitation level. These statistical characteristics can be the variance of the current sequence, the norm rate of change, or the root mean square value of its first-order difference, etc. If the value of the statistical characteristic exceeds a preset excitation threshold, it is determined that the current data contains valid excitation information, allowing the initiation or continuation of subsequent parameter update recursive operations; otherwise, it is determined that the current excitation is insufficient, parameter updates are paused, and the estimated equivalent impedance parameter value from the previous sampling time remains unchanged. This mechanism ensures that the identification algorithm updates only when the data signal-to-noise ratio is sufficiently high, thereby effectively suppressing the accumulation or divergence of estimation errors caused by background measurement noise dominating the signal when the power grid is in steady state and fluctuations are weak.

[0069] The preset excitation threshold is set comprehensively based on the measurement noise level, desired identification sensitivity, and update speed in the network-type SVG control system. Its purpose is to effectively distinguish between informative excitation caused by actual load or power fluctuations and background measurement noise.

[0070] In one exemplary embodiment, the excitation threshold can be set to 2 to 5 times the effective noise value of the current sensor in steady state. For example, if the effective noise value of the current measurement is approximately 0.5% of the rated current, the excitation threshold can be set to 1% to 2.5% of the corresponding change in rated current. In another exemplary embodiment, the preset excitation threshold can also be set to the point where, within a set time window, the root mean square value of the first difference (i[k]-i[k-1]) of the current sampling value sequence i[k] exceeds a certain threshold (e.g., 0.2% of the rated current). Those skilled in the art can calibrate and adjust the threshold according to the noise characteristics of the actual system.

[0071] Step 2, Recursive Least Squares (RLS) Identification: When a valid excitation is identified, the recursive calculation of the recursive least squares algorithm with a forgetting factor is initiated or continues. The input to this algorithm is a preprocessed discrete sequence of voltage and current (e.g., d-axis component u). d [k], i d [k]), the algorithm recursion process is as described above, including: calculating the gain vector and updating the parameter estimation vector θ. hat [k]=[Rg hat ;Xg hat / (ω s ×Ts)] T And update the covariance matrix. The forgetting factor λ is used to assign higher weights to recent data, enabling the algorithm to track time-varying parameters.

[0072] Step 3, Equivalent Reactance Parameter Extraction: Directly extract the scaled reactance parameter estimate θ2[k]=Xg from the parameter vector updated in real time by the RLS algorithm. hat [k] / (ω s ×Ts), and through Xg[k]=θ2[k]×ω s Based on the relationship ×Ts, the estimated value of the equivalent reactance parameter Xg[k] can be calculated in real time.

[0073] In this invention, through the two methods described above, a key equivalent reactance parameter value Xg, which has been processed to resist interference, can be output. This provides a direct and reliable input parameter for calculating the real-time equivalent short-circuit ratio (rtESCR), thus fully realizing the invention's objective of obtaining basic parameters of power grid strength online.

[0074] In this invention, the equivalent reactance parameter and its symbol Xg both refer to the same physical quantity, namely, the equivalent inductive reactance value at power frequency when viewed from the grid connection point of the grid-type SVG towards the grid side.

[0075] Furthermore, to significantly improve the robustness and estimation accuracy of method (b) under complex and variable power grid conditions, this method can also employ a strategy of parallel operation of multiple models and confidence fusion. Specifically, this includes: running at least two sets of recursive least squares identification models with different dynamic characteristics in parallel, with each model outputting intermediate estimates of the equivalent reactance parameters; calculating the instantaneous confidence weights of each intermediate estimate in real time; and weighting and fusing the intermediate estimates according to the instantaneous confidence weights to obtain the final output equivalent reactance parameters. The core idea of ​​this strategy is: instead of relying on the optimal parameter selection of a single model, it runs multiple complementary models in parallel and intelligently fuses their results.

[0076] The specific implementation steps are as follows: (1) Parallel initialization and execution of models: Under the framework of the recursive least squares algorithm, initialize M (e.g., M=2) identification models (Model1, Model2, ..., Model3) that run in parallel. M These models employ the same data preprocessing and system equations, but achieve complementary characteristics by setting different key algorithm parameters. Specifically, the forgetting factor λ1 of the first model is greater than the forgetting factor λ2 of the second model, resulting in higher accuracy for steady-state estimation of slowly changing parameters in the first model, and more agile tracking of rapid parameter changes in the second model. For example: Model 1: Set a large forgetting factor λ1 (e.g., 0.999). Due to its slow forgetting speed and long memory of historical data, it can provide a steady-state estimate with high accuracy and small variance when the grid impedance changes slowly, but it may lag in tracking sudden changes.

[0077] Model 2: Set a smaller forgetting factor λ2 (e.g., 0.99). It forgets data quickly and is more sensitive to new data, thus it can quickly track sudden changes in grid impedance, but the estimation results may fluctuate slightly more under steady-state conditions.

[0078] Each model r performs recursive calculations independently in each sampling period k, outputting its own parameter estimation vector θ. hat_r [k] and the intermediate estimate of the corresponding equivalent reactance parameter Xg r [k]. The value of r ranges from 1 to M.

[0079] (2) Instantaneous confidence assessment: The instantaneous confidence weight w of each model output is calculated in real time. r [k] includes: Calculate the prediction error e of model r at the current time k. r [k]:e r [k]=y[k]-φ T [k]×θ hat_r [k].

[0080] The covariance of this prediction error can be approximated as: S r [k]=φ T [k]×P r [k-1]×φ[k]+1 (assuming the measurement noise variance is 1).

[0081] The confidence weight of model r at the current time can be defined as: w r [k]=1 / (e_ r 2 [k] / S r [k]) represents the degree of agreement between the predicted data and the measured data of the model r. The smaller the error, the higher the weight.

[0082] Normalize the weights of all models: w r [k]=w r '[k] / Σ(w s '[k]), w s [k] represents the confidence weight of model s at the current time, where s ranges from 1 to M.

[0083] (3) Weighted fusion output: The intermediate estimates of each model are fused according to the instantaneous confidence weights to obtain the final output reactance value: Xg[k]=Σ(w r [k]×Xg r [k]), sum over r=1 to M.

[0084] The beneficial effects of this fusion mechanism are as follows: when the grid impedance is stable, the high-precision model (such as Model 1) naturally receives a higher weight, and the output tends to be its high-precision estimate; when the grid is disturbed and the impedance changes rapidly, the prediction error of the fast tracking model (such as Model 2) will decrease rapidly, and its weight will increase, thus enabling the fused output to respond quickly to changes. This method effectively overcomes the classic problem of difficulty in selecting single model parameters (such as the forgetting factor) and can provide near-optimal estimation performance in various dynamic scenarios.

[0085] S110, output the electrical equivalent reactance parameter as a basic parameter for real-time evaluation of power grid strength.

[0086] The equivalent reactance parameter Xg is output in scalar numerical form. The specific output method matches the excitation method used in S100: When using method (a) (actively injecting perturbation), a new Xg value can be output after each complete perturbation injection and measurement cycle, thus enabling periodic parameter updates.

[0087] When using method (b) (analyzing natural fluctuations), the continuous recursion of the recursive least squares algorithm can be relied upon to output a real-time updated sequence of Xg[k] values, with the sampling period Ts of the network-type SVG digital control system as the interval, thereby achieving continuous parameter refresh.

[0088] The output Xg value forms the key data interface for online impedance identification and upper-level adaptive control. Its direct application is that the upper-level controller (such as a virtual synchronous adaptive controller) receives this Xg value and, in conjunction with the synchronously acquired grid connection point voltage amplitude Upcc, calculates it according to the formula rtESCR ≈ (U 2 The real-time equivalent short-circuit ratio rtESCR is calculated using pcc / |Xg|) / Pn. Here, Pn represents the rated active power capacity of the local grid-type SVG, typically close to the rated apparent capacity Sn. The calculated rtESCR is a core indicator for quantifying the grid's support capacity, providing a direct decision-making basis for subsequent parameter adaptive adjustments or operating mode switching of the grid-type SVG.

[0089] Through steps S100 to S120, the online identification method of grid impedance of the present invention fully achieves its inventive purpose: during the grid-connected operation of the grid-type SVG, an accurate and reliable basic parameter of grid strength (Xg) is obtained online without disturbance, thereby laying a crucial sensing foundation for the intelligent and adaptive operation of the grid-type SVG.

[0090] Furthermore, the method is executed in a cycle of tens to hundreds of microseconds, synchronized with the power control cycle of the network-type SVG.

[0091] In this invention, the tens to hundreds of microseconds (e.g., typical 50μs, 100μs, or 250μs) are on the same order of magnitude as the switching cycle of the power devices and the core current and voltage closed-loop control cycle of the grid-type SVG. This ensures that the operation frequency of impedance identification matches the underlying power regulation dynamics of the grid-type SVG, enabling the capture of rapid changes in the grid state. Synchronization refers to the synchronization of the scheduling trigger of this identification task with the main control interrupt (or clock) of the grid-type SVG, ensuring that data acquisition, processing, and output are strictly aligned with the power control actions in time, avoiding timing discrepancies. Cyclic execution means that the steps (S100 to S120) of the method are encapsulated as a task and scheduled to run within the cycle. For mode (b), its RLS algorithm is recursively applied once per cycle; for mode (a), its excitation injection, data acquisition window management, and signal processing flow are also scheduled and managed within this periodic framework.

[0092] The advantages of synchronous loop execution are: It ensures the timeliness of data: it can provide the latest grid impedance information with a delay of milliseconds or even sub-milliseconds, meeting the requirements of upper-level adaptive control for fast response.

[0093] Deep integration has been achieved: the identification function is integrated as an inherent task module of the network-type SVG real-time control software, rather than an independent background or upper-level slow task, thus truly realizing the integrated design of perception and control.

[0094] This ensures the determinism of the system: a fixed execution cycle synchronized with the master controller facilitates the planning of system resources and real-time analysis, thereby improving the reliability and predictability of the entire control system.

[0095] This embodiment also provides a grid-type SVG, including an online impedance identification unit, which is configured to perform the online grid impedance identification method as described in the foregoing embodiments.

[0096] This mesh-based SVG is a physical embodiment of the aforementioned method invention, and its specific structure and integration method are as follows: Hardware carrier of the impedance online identification unit: The unit is physically integrated within the main controller of the meshed SVG. Its hardware foundation is the core processing unit of the main controller (such as a high-performance DSP, FPGA, or multi-core processor) and the inherent synchronous sampling ADC channel, sensor interface, and necessary memory of the meshed SVG system. This perfectly aligns with the method's characteristic of not relying on additional hardware.

[0097] Modular implementation of the impedance online identification unit: At the software or firmware level, the unit can be considered as consisting of a series of functional modules, which correspond to the steps of the method: Incentive Management Module: Responsible for generating and managing the disturbance signal ΔQ in mode (a). inj Alternatively, configure the RLS algorithm parameters in method (b).

[0098] Data acquisition and preprocessing module: controls the ADC to sample synchronously and performs preprocessing such as filtering and coordinate transformation on the raw voltage and current data.

[0099] Core algorithm processing module: Embeds and runs the proprietary signal processing algorithms, such as the FFT analysis module or the recursive least squares algorithm module, to perform impedance extraction calculations.

[0100] Output interface: The calculated equivalent reactance parameter Xg is provided in real time to the virtual synchronous machine control unit or system status management unit of the network-type SVG through the internal data bus or register.

[0101] Synergy with the overall SVG system: Input source: The voltage and current input of this unit come directly from the voltage transformer (PT) and current transformer (CT) built into the grid-type SVG for power control.

[0102] Output destination: The output Xg value is directly fed to the adaptive virtual synchronization control unit of the network SVG as a key input for calculating rtESCR and dynamically adjusting the virtual inertia (J) and damping (D) parameters accordingly.

[0103] Operation scheduling: The execution of this unit is managed by the unified real-time operating system (RTOS) or interrupt scheduler of the network-type SVG to ensure that it is synchronized with the power control task and executed cyclically in accordance with the aforementioned cycle.

[0104] By specifically configuring the method in a physical online impedance identification unit and clarifying the integration relationship between this unit and the grid-type SVG main controller and other functional units, this embodiment fully discloses how to manufacture and use an intelligent grid-type SVG product with real-time power grid strength sensing capability, thereby completing the complete support from method invention to product implementation.

[0105] (Example 2) This embodiment provides an adaptive virtual synchronization control method for a grid-type SVG. This embodiment and the aforementioned Embodiment 1 (online grid impedance identification method) belong to two levels under the same inventive concept: Embodiment 1 solved the fundamental problem of how to sense grid strength in real time, while this embodiment solves the core control problem of how to use the sensing results to achieve optimal adaptive control of controller parameters. The method described in this embodiment can directly call or integrate the method described in Embodiment 1 to obtain key basic parameters of grid strength.

[0106] like Figure 2 As shown, the adaptive virtual synchronization control method for network-type SVG provided in this embodiment may include the following steps: S200 acquires the electrical quantities of the SVG grid connection point in real time.

[0107] In this invention, electrical quantities refer to physical quantities used to characterize the electrical operating state of a grid connection point. Their core is key information that can directly or indirectly reflect system power, voltage, and frequency. Electrical quantities mainly include: Raw electrical quantity: The instantaneous value of the three-phase voltage at the grid connection point, u, is acquired in real time by the voltage and current sensors built into the grid-type SVG. a u b u c and the instantaneous value of three-phase current i a i b i c .

[0108] Derived electrical quantities: Key physical quantities obtained by real-time processing of original electrical quantities, such as: Fundamental component: The amplitude Upcc and phase θ of the fundamental positive sequence voltage, as well as the amplitude and phase of the fundamental positive sequence current, extracted from the three-phase voltage by a phase-locked loop or software phase-locked algorithm.

[0109] Power quantities: the calculated instantaneous three-phase active power P and instantaneous reactive power Q.

[0110] Transformed coordinate quantities: Components obtained by transforming the three-phase instantaneous values ​​using coordinate transformations such as Clark transformation and Park transformation, and placing them in a two-phase stationary coordinate system (αβ) or a synchronously rotating coordinate system (dq), such as u. d ,u q and i d i q These components are commonly used inputs for subsequent virtual synchronous machine control decoupling and calculation.

[0111] The acquisition of electrical quantities relies entirely on the sensing and sampling system inherent in the grid-type SVG for its core power control, requiring no additional hardware. Real-time acquisition refers to continuous and cyclical data sampling and updating at the same rate as the power control cycle of the grid-type SVG main controller (typically tens to hundreds of microseconds), ensuring the control system can respond to dynamic changes in the power grid.

[0112] The raw electrical quantities (three-phase voltage and current) obtained in this step are from the same source as the input of Example 1 (online identification method for grid impedance). In one implementation, the high-precision data synchronously acquired and preprocessed by the method of Example 1 can be directly shared or called, especially the fundamental components of voltage and current or specific frequency components used to calculate grid strength, thereby ensuring data consistency and timeliness and reducing redundant calculations in the system.

[0113] S210, based on the electrical quantities, the equivalent reactance parameters characterizing the strength of the power frequency grid are obtained through impedance measurement operations, and the real-time equivalent short-circuit ratio is calculated based on the equivalent reactance parameters as a grid strength index.

[0114] In this embodiment, the impedance measurement operation refers to a measurement or estimation process performed synchronously during the grid-connected operation of the SVG (Static Var Generator), used to obtain key parameters characterizing grid strength in real time. Specifically, it directly applies the method described in Embodiment 1, that is, obtaining the equivalent reactance parameter through any of the following methods: Method (a): Controlled disturbance injection measurement method: The grid-type SVG is controlled to inject a preset, amplitude-limited reactive power disturbance signal into the power grid as an excitation, and voltage response data and current response data are measured simultaneously. The equivalent impedance parameter is directly calculated by analyzing the response data at the frequency of the disturbance signal; or, Method (b): Online identification of natural fluctuations: Continuously collect voltage and current fluctuation data caused by natural power fluctuations in the power grid, and use an adaptive estimation algorithm to estimate the equivalent impedance parameter from the fluctuation data in real time.

[0115] Specifically, the method is implemented by calling or integrating the online identification method of power grid impedance as described in Example 1, and obtaining the equivalent reactance parameter Xg in real time based on the electrical quantities obtained in step S200.

[0116] After obtaining Xg, the real-time equivalent short-circuit ratio rtESCR is calculated by combining it with the fundamental voltage amplitude Upcc obtained from real-time measurement or calculation. The specific calculation formula is derived based on the definition of short-circuit ratio as follows: rtESCR=Ssc / Pn; Where Pn is the rated capacity of the local grid-connected SVG, which is usually close to the rated apparent capacity Sn. Ssc is the real-time short-circuit capacity of the grid as seen from the grid connection point. According to circuit principles, the short-circuit capacity Ssc and the grid equivalent reactance Xg approximately satisfy the following relationship: Ssc≈U 2 pcc / |Xg|.

[0117] Substituting the above formula into the definition, we obtain the expression for real-time computation: rtESCR≈(U 2 pcc / |Xg|) / Pn.

[0118] As shown in the formula, the value of rtESCR is directly proportional to Upcc / |Xg|. A larger rtESCR value indicates a smaller equivalent internal resistance of the power grid and stronger support capacity, meaning a stronger power grid; conversely, a smaller value indicates a weaker power grid. The voltage amplitude Upcc can be obtained by extracting the fundamental positive-sequence component from the acquired three-phase voltage using a phase-locked loop.

[0119] S220, based on the real-time equivalent short-circuit ratio, dynamically select the controller operation mode that the network-type SVG should currently adopt from at least two predefined virtual synchronous machine control modes.

[0120] This step is the core of control strategy decision-making based on grid strength sensing results. The at least two control modes are defined differently based on grid strength and the core control objectives of the grid-type SVG: The first operating mode (grid-connected current source mode) corresponds to weak grid scenarios. In this mode, the core objective of the controller is to prioritize providing significant virtual inertia support to enhance grid frequency stability and counteract power surges.

[0121] The second operating mode (grid-type voltage source mode) corresponds to a strong power grid scenario. In this mode, the core objective of the controller is to prioritize rapid and accurate reactive power compensation and voltage regulation, requiring the control system to have a fast dynamic response and good damping characteristics to suppress oscillations.

[0122] The dynamic selection is achieved by comparing the real-time equivalent short-circuit ratio (rtESCR) with a preset threshold. When the real-time equivalent short-circuit ratio is less than or equal to the first threshold, the first operating mode is selected; When the real-time equivalent short-circuit ratio is greater than or equal to the second threshold, the second operating mode is selected; Wherein, the first threshold is less than the second threshold.

[0123] In this embodiment, the first threshold is the preset weak grid threshold in Implementation 1, and the second threshold is the preset strong grid threshold in Implementation 1. That is, the preset value used to define the grid strength boundary in the aforementioned embodiment. Its specific value can be determined according to the grid regulations, simulation analysis or on-site debugging. For example, it can be set to 2 and 3 respectively.

[0124] Furthermore, when the real-time equivalent short-circuit ratio is greater than the first threshold and less than the second threshold, the grid strength is determined to be at a moderate level. In this case, it is not rigidly classified into any typical mode, but rather considered a smooth transition range, meaning the controller's operating mode is in a transitional state from the first operating mode to the second operating mode. Within this range, the control parameters will be smoothly interpolated and adjusted based on the continuous values ​​of rtESCR (see subsequent step S230 for details), thereby achieving a seamless and smooth transition of the operating mode and avoiding switching shocks.

[0125] S230, based on the selected controller operating mode and the specific value of the real-time equivalent short-circuit ratio, the values ​​of the virtual inertia parameter and the damping parameter are determined through preset mapping rules.

[0126] This step is the core algorithmic step of adaptive control, and its function is to accurately map the quantitative sensing of grid strength (rtESCR) to the optimal parameter combination (J, D) of the virtual synchronous machine controller. The mapping rule is implemented based on the following core principles and mathematical relationships: (1) Setting the reference parameters: Set the baseline virtual inertia parameter value J for the first operating mode. weak With reference damping parameter value D weak .

[0127] Set the baseline virtual inertia parameter value J for the second operating mode. strong With reference damping parameter value D strong .

[0128] Wherein, the virtual inertia parameter value configured for the first operating mode is greater than the virtual inertia parameter value configured for the second operating mode, and the damping parameter value configured for the second operating mode is greater than the damping parameter value configured for the first operating mode, that is, according to the control objectives of different modes, the reference value satisfies: J weak >J strong (Inertia is greater under weak networks), and usually D strong >D weak (Stronger damping under strong netting).

[0129] (2) Segmented continuous mapping rule: In areas with weak power grids: when rtESCR≤TH weak When J=J weak D=D weak .

[0130] Strong power grid areas: when rtESCR≥TH strong When J=J strong D=D strong .

[0131] Transition region: when TH weak <rtESCR<TH strong When the virtual inertia parameter J and damping parameter D are used, the values ​​of the virtual inertia parameter J and the damping parameter D change continuously and smoothly between their corresponding reference values ​​according to the preset interpolation function of rtESCR, so as to achieve a smooth transition of control characteristics. That is, when the controller is in the transition state from the first operating mode to the second operating mode, the value of the virtual inertia parameter changes continuously between the virtual inertia parameter value corresponding to the first operating mode and the virtual inertia parameter value corresponding to the second operating mode according to the preset interpolation function.

[0132] In a preferred embodiment, the preset interpolation function is a linear interpolation function to provide a concise and deterministic mapping relationship. For example, the virtual inertia parameter J is calculated using the following formula: J=J weak -(J weak -J strong )×[(rtESCR-TH weak ) / (TH strong -TH weak )).

[0133] The damping parameter D can be calculated using a linear interpolation function of a similar form, i.e.: D=D weak +(D strong -D weak )×[(rtESCR-TH weak ) / (TH strong -TH weak )).

[0134] In another preferred embodiment, to meet specific dynamic performance requirements, the interpolation function may also be nonlinear, such as a sigmoid curve. Such functions can provide a smoother or more abrupt rate of parameter change near the threshold, thereby fine-tuning the dynamic characteristics of the transition process.

[0135] Through step S230, this method achieves a precise and smooth mapping between continuously changing grid conditions and continuously adjustable control parameters. This not only avoids system shocks caused by parameter jumps, but also ensures that the grid-type SVG maintains optimal dynamic performance throughout its entire operating domain.

[0136] S240, update the virtual synchronous machine controller of the network-type SVG using the determined virtual inertia parameters and damping parameters, and execute control.

[0137] This step is the final execution stage of adaptive control. The optimal virtual inertia parameters and optimal damping parameters corresponding to the current control cycle, determined in step S230, are loaded in real time into the virtual synchronous machine (VSG) control algorithm of the networked SVG, replacing the original fixed parameters, thereby completing the dynamic reconstruction of the controller's internal model.

[0138] The virtual synchronous machine controller contains the core components simulating the operation of a synchronous generator, mainly including an active-frequency loop (rotor motion equation) and a voltage-reactive power loop (excitation control). This invention focuses on the adaptive J and D parameters, which embody the core characteristics of inertia and damping. The voltage-reactive power loop parameters can be kept fixed or finely adjusted collaboratively, but they are not the core of this invention. The specific operation for parameter updating is as follows: the optimal virtual inertia parameters and optimal damping parameters are loaded into the dynamic equations of the virtual synchronous machine's active-frequency loop. J·dΔω / dt+D·Δω=P ref -P meas +ΔP droop .

[0139] Where J is the virtual inertia parameter, its value directly determines the inertial delay of frequency changes in the power system supported by the grid-type SVG. A larger J value indicates a stronger ability of the system to resist power disturbances and maintain frequency stability, resulting in slower frequency changes. D is the damping parameter, its value directly determines the decay rate of power oscillations. A larger D value indicates a stronger ability to suppress power and frequency oscillations, resulting in faster system recovery to stability. Δω is the angular frequency deviation, P... ref P is the active power reference value. meas To measure the active power, ΔP droop This is the active-frequency droop control term, where d / dt represents the derivative with respect to time.

[0140] The virtual synchronous machine controller, after updating the above parameters, performs closed-loop calculations by combining the real-time collected grid connection point voltage and current with the calculated power: Internal reference value calculation: The controller calculates the internal electromotive force reference value Eref and its phase θ required to maintain its voltage source characteristics (where the phase θ is obtained by integrating the virtual rotor angular velocity).

[0141] Modulation and driving: Eref and θ are fed into the space vector pulse width modulation (SVPWM) or sinusoidal pulse width modulation (SPWM) module to generate PWM (pulse width modulation) signals to drive the fully controlled power switching devices (such as IGBTs) inside the network-type SVG.

[0142] Power output: The main power circuit of the grid-type SVG operates precisely according to the PWM signal, and finally outputs a three-phase AC voltage with adaptive inertia and damping characteristics at the grid connection point.

[0143] Through step S240, this method completes the entire closed-loop adaptive control from grid sensing to parameter decision-making, dynamic reconfiguration, and precise execution. Ultimately, the grid-based SVG can operate with dynamic characteristics optimally matched to the real-time grid strength, exhibiting a high-inertia stable support source under weak grid conditions and transforming into a fast-responding precision regulator under strong grid conditions, thereby achieving intelligent, adaptive, and proactive support for the grid.

[0144] Furthermore, the adaptive virtual synchronization control method (S200 to S240) provided in this embodiment is executed cyclically within a preset control cycle, enabling the virtual inertia parameters and damping parameters to dynamically and adaptively adjust as the grid strength changes. The preset control cycle is the power control cycle of a grid-type SVG (typically tens to hundreds of microseconds), which allows the virtual inertia parameters and damping parameters to track the dynamic changes in grid strength (rtESCR) at millisecond speeds, achieving real-time and continuous adaptive adjustment of controller parameters.

[0145] The method described in this embodiment, closely integrated with Embodiment 1 (Online Grid Impedance Identification Method), constitutes a complete perception-decision-execution closed-loop adaptive system. This system achieves a fully automated link from real-time grid state perception to intelligent control strategy decision-making, and then to online reconfiguration and execution of controller parameters. This fundamentally solves the core technical challenge pointed out in the background art: fixed-parameter virtual synchronous machines cannot maintain optimal performance under a wide range of grid strengths. Through this method, the grid-type SVG transforms from a device with fixed parameters into a grid adaptive regulator capable of intelligently adapting to environmental changes, thereby significantly improving the dynamic performance, stability margin, and overall support capability of the grid-type SVG in complex grid environments with a high proportion of renewable energy.

[0146] (Example 3) This embodiment provides an adaptive switching control method for the operating mode of a grid-connected SVG. This embodiment is a further deepening and specific application of the technical system of the aforementioned embodiments: its essence is to focus the adaptive virtual synchronization control method described in Embodiment 2 on the direct application scenario of switching between grid-following mode and grid-connected mode; at the same time, its implementation relies on the real-time grid strength sensing capability provided in Embodiment 1 to obtain the key basic parameters for mode switching decisions.

[0147] like Figure 3 As shown, the adaptive switching control method for the operating mode of a network-type SVG provided in this embodiment may include the following steps: S300, when the grid-type SVG is in the first operating mode, obtains the equivalent reactance parameters characterizing the strength of the power frequency grid in real time through impedance measurement operation, and calculates the real-time equivalent short-circuit ratio.

[0148] This step involves triggering the mode switch and obtaining the decision-making basis. Specifically, the online grid impedance identification method described in Example 1 is invoked to obtain the equivalent reactance parameter Xg in real time in an online and non-intrusive manner. Subsequently, combined with the grid connection point voltage Upcc, the equivalent reactance parameter Xg is obtained according to the formula rtESCR≈(U 2 The real-time equivalent short-circuit ratio rtESCR is calculated using pcc / |Xg|) / Pn. This indicator will serve as a direct quantitative basis for assessing the grid's support capacity and determining whether a switch in operating mode is necessary.

[0149] In this embodiment, the first operating mode is the grid-connected current source mode in Embodiment 2, and the second operating mode is the grid-connected voltage source mode (virtual synchronous machine mode) in Embodiment 2.

[0150] The impedance measurement operation can be performed based on any of the methods described in detail in Embodiment 1: Method (a): Controlled disturbance injection measurement method: The grid-type SVG is controlled to inject a preset, amplitude-limited reactive power disturbance signal into the power grid as an excitation, and voltage response data and current response data are measured simultaneously. The equivalent impedance parameter is directly calculated by analyzing the response data at the frequency of the disturbance signal; or, Method (b): Online identification of natural fluctuations: Continuously collect voltage and current fluctuation data caused by natural power fluctuations in the power grid, and use an adaptive estimation algorithm to estimate the equivalent impedance parameter in real time from the fluctuation data. S310, compare the real-time equivalent short-circuit ratio with a preset switching threshold. When the real-time equivalent short-circuit ratio is continuously lower than the switching threshold within a preset judgment time window, trigger the switching process from the first operating mode to the second operating mode.

[0151] This step involves intelligent decision-making and prevention of accidental triggering during mode switching. Its core lies in setting a switching criterion with anti-interference capabilities to avoid frequent or unnecessary mode switching due to instantaneous fluctuations in the power grid.

[0152] The preset switching threshold is essentially a critical value used to determine when the power grid strength weakens to the point where it is insufficient to stably support grid-connected operation and grid support functions need to be activated. This threshold (TH) switch ) can be set to the weak grid threshold (TH) in Example 2. weak The switching threshold can be the same as the standard threshold, or calibrated separately according to the stability requirements of a specific application scenario. For example, the preset switching threshold is usually set based on engineering experience with the short-circuit ratio (SCR), with a typical value in the range of 2 to 3. The specific value can be calibrated according to the specific grid strength, stability requirements, and control margin of the equipment itself at the grid-connected point of the SVG.

[0153] In this invention, the preset determination time window (T)window (For example, it can be from 100 milliseconds to 500 milliseconds).

[0154] In this invention, the condition for triggering switching is not that the instantaneous value of rtESCR is lower than a preset switching threshold, but rather that within the entire preset decision time window, the value of the rtESCR sampled value sequence obtained by the control system sampling period is lower than the preset switching threshold TH. switch The proportion of sampling points exceeds a preset proportion (e.g., 95%). This design can effectively filter out instantaneous fluctuations caused by load switching, fault disturbances, etc., and ensure that switching decisions are based on the stable trend of continuous weakening of grid strength, thereby avoiding false trips.

[0155] When the switching conditions are met, the switching process from the current operating mode (network-following mode) to the target operating mode (network-building mode) is formally triggered, and the subsequent parameter preset and switching execution phase begins.

[0156] Through the optimized design of this step, the triggering of mode switching has both sensitivity and reliability. It can respond in a timely manner when the power grid deteriorates, and can effectively suppress noise interference, thus ensuring the stability of system operation.

[0157] S320, in the switching process, based on the current real-time equivalent short-circuit ratio, the target virtual inertia parameter and target damping parameter applicable to the second operating mode are determined by a preset mapping rule.

[0158] This step is a crucial preparatory step for mode switching. Its core is to pre-calculate the controller parameters that should be used immediately after the grid-type SVG switches to the second operating mode, i.e., the grid-type mode, which are optimally matched with the current grid strength, before performing the physical switch. This allows for a smooth transition by adjusting the parameters first and then switching the mode.

[0159] The preset mapping rules inherit and apply the adaptive parameter mapping principle described in Example 2, and the specific configuration is as follows: The value of the target virtual inertia parameter J is inversely correlated with the value of rtESCR. This is because, in the upcoming grid configuration mode, the weaker the grid (the smaller the rtESCR), the larger the virtual inertia needs to be to enhance frequency stability.

[0160] The value of the target damping parameter D can be positively correlated with the value of rtESCR. This is because under a strong power grid (with a larger rtESCR), stronger damping is required to effectively suppress potential power oscillations.

[0161] To obtain a definite and smooth parameter variation, a linear interpolation function is used for calculation. Specifically, the target virtual inertia parameter J is determined by the following formula: J=J weak -(Jweak -J strong )×[(rtESCR-TH weak ) / (TH strong -TH weak )).

[0162] Where J is the target virtual inertia parameter, rtESCR is the real-time equivalent short-circuit ratio, and TH weak TH is the preset switching threshold. strong TH is the preset intensity threshold. strong >TH weak J weak and J strong They are respectively with TH weak and TH stron The preset virtual inertia parameter value corresponding to g.

[0163] The damping parameter D can be calculated using a linear interpolation function of the form D = D weak +(D strong -D weak )×[(rtESCR-TH weak ) / (TH strong -TH weak )]. D weak With D strong : These are the target damping reference values ​​(D) corresponding to weak and strong power grid conditions, respectively. weak <D strong ).

[0164] This step preheats the network mode with optimal control parameters before switching, which fundamentally avoids the risk of power surges or instability caused by mismatch between controller parameters and target mode in traditional switching. It is the core guarantee for achieving safe and smooth mode switching.

[0165] In this invention, the virtual inertia parameter and the target virtual inertia parameter refer to the same physical quantity, and the damping parameter and the target damping parameter refer to the same physical quantity, namely, the adjustable parameter used in the virtual synchronous machine controller to adjust the power oscillation decay rate. In different embodiments, depending on the context, they are referred to as the damping parameter adjusted in real time or the target damping parameter preset before mode switching.

[0166] S330, adjust the virtual synchronous machine control parameters of the network-type SVG to the target virtual inertia parameters and target damping parameters, and then perform a control mode switch to make the network-type SVG switch to the second operating mode.

[0167] This step, based on the optimal matching parameters calculated in step S320, performs a switch from the first operating mode (e.g., grid-connected current source mode) to the second operating mode (e.g., grid-connected voltage source mode, i.e., virtual synchronous machine mode). Specifically, it includes the following sub-steps: S3301, Controller parameter update: Update the virtual synchronous machine controller parameters of the network-type SVG to the target virtual inertia parameters and target damping parameters.

[0168] S3302, Controller Switching and Enabling: After completing the parameter update, first lock out the controller corresponding to the first operating mode (such as a current control loop based on grid voltage orientation), and then enable the virtual synchronous machine controller (such as a grid controller containing virtual rotor motion equations) whose parameters have been updated for the second operating mode. This step-by-step operation is designed to avoid control conflicts.

[0169] S3303, Smooth Transition of Control Signals: At the moment of switching, key control signals (especially voltage reference commands) are processed by ramp function or filtered by first-order inertial element to ensure a smooth transition from the current operating state to the initial state of the new mode, and to avoid power surges caused by sudden changes in control commands.

[0170] Furthermore, after S330 performs the switch to put the network-type SVG into the second operating mode, the method further includes: S340, continuously monitor the real-time equivalent short-circuit ratio. If the real-time equivalent short-circuit ratio rises and remains above a hysteresis threshold, trigger the process of switching back from the second operating mode to the first operating mode, wherein the hysteresis threshold is higher than the preset switching threshold.

[0171] After S330 is completed and the grid-type SVG is stably operating in the second operating mode (grid-type mode), it continuously and periodically monitors the grid status and specifically executes the following logic: Continuous monitoring: Continue to execute step S310, periodically acquire and calculate the real-time equivalent short-circuit ratio (rtESCR).

[0172] Hysteresis detection: During the operation of the mesh-type SVG in the second operating mode, when the value of rtESCR is detected to fall below the hysteresis threshold TH... hysteresis The state rose and exceeded TH hysteresis Then, a timer is started. If the timer reaches the preset monitoring duration T... hold Previously, rtESCR was consistently maintained at TH hysteresis If the above condition is met, the reverse handover condition is deemed satisfied; if the rtESCR falls back to TH during this process... hysteresis The timer will then be immediately reset to zero and monitoring will continue. holdThe setting needs to comprehensively consider the typical fluctuation period of the power grid and the dynamic response requirements of the system. Typically, T... hold It should be much larger than the transient fluctuation timescale commonly found in power grids (e.g., set to hundreds of milliseconds to several seconds) to ensure the reliability of the judgment.

[0173] Process Triggering: Based on the above determination, the control process of switching back from the second operating mode (network construction mode) to the first operating mode (network following mode) is automatically triggered. The reverse handover process includes a process similar to the aforementioned steps, which pre-calculates the target parameters based on the current rtESCR value and performs a safe handover to ensure the smoothness of the reverse handover.

[0174] This third embodiment is a systematic integration and advanced application of the key capabilities provided in embodiments one and two: Sensing Dependence: The start-up condition judgment and continuous monitoring of this method (such as steps S310 and S340) directly depend on the real-time equivalent short-circuit ratio (rtESCR) as its core input. Without accurate and rapid online impedance sensing, the strength of the power grid cannot be reliably assessed, and this method will lose its decision-making basis.

[0175] Control kernel: The optimal control parameters calculated and applied by this method for the network configuration mode before and after the switch (as in step S320) fully inherit and apply the adaptive mapping method of virtual synchronous machine parameters based on grid strength, as detailed in Example 2. Example 2 provides a control parameter preheating kernel for smooth switching and mode optimization of this method.

[0176] System Integration: This third embodiment integrates and schedules the perception results of the first embodiment with the decision rules of the second embodiment at a higher dimension of system operation mode switching. It clarifies the criteria for when to switch, the process of how to prepare before switching, and the specific operation steps for safely executing the switch. This forms a complete, autonomous, and safe system-level control strategy that goes from power grid state perception to controller parameter adaptation and finally achieves seamless switching of operation modes.

[0177] Therefore, Embodiment 1 and Embodiment 2 serve as the sensing foundation and control algorithm kernel of this embodiment, respectively, jointly supporting the system-level adaptive switching function implemented in Embodiment 3. These three elements progressively enhance each other, collectively constituting the complete technical solution provided by this invention that comprehensively improves the adaptability and support capabilities of grid-type SVG in power grids with varying strength.

[0178] This invention also provides an electronic device, including: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being configured to perform the method described in this invention.

[0179] This invention also provides a computer-readable storage medium storing computer-executable instructions for performing the methods described in this invention.

[0180] It should be understood that the various forms of processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution disclosed in this invention can be achieved, and this is not limited herein.

[0181] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for online identification of grid impedance for grid-type SVG, characterized in that, This method is performed during the grid-connected operation of the grid-type SVG to obtain fundamental parameters for assessing grid strength, and the method includes the following steps: S100, without interrupting the normal regulation function of the grid-type SVG and without relying on additional hardware, performs an impedance measurement operation to obtain an equivalent impedance parameter for characterizing the strength of the power frequency grid; the impedance measurement operation includes: applying or using an electrical disturbance as an excitation, synchronously collecting voltage response data and current response data caused by the electrical disturbance at the grid connection point, and then calculating or estimating the equivalent impedance parameter based on the response data. S110, output the equivalent impedance parameter as a basic parameter for real-time evaluation of power grid strength.

2. The method according to claim 1, characterized in that, The impedance measurement operation is performed in any of the following ways: (a) Controlled disturbance injection measurement method: The grid-type SVG is controlled to inject a preset reactive power disturbance signal with limited amplitude into the power grid as an excitation, and the voltage response data and current response data are measured simultaneously. The equivalent impedance parameter is directly calculated by analyzing the response data at the frequency of the disturbance signal. or, (b) Online identification method for natural fluctuations: continuously collect voltage and current fluctuation data caused by natural power fluctuations in the power grid, and use an adaptive estimation algorithm to estimate the equivalent impedance parameter in real time from the fluctuation data.

3. The method according to claim 2, characterized in that, When mode (a) is used, the reactive power disturbance signal is a sine wave or pseudo-random binary sequence with a frequency in the range of 2Hz to 15Hz, and the amplitude of the reactive power disturbance signal is limited to 0.5% to 2% of the rated capacity of the grid-type SVG.

4. The method according to claim 3, characterized in that, The method also includes a closed-loop optimization step: Based on the confidence evaluation index of the current or historical impedance identification results, the amplitude and / or frequency of the subsequently injected reactive power disturbance signal are dynamically adjusted.

5. The method according to claim 2, characterized in that, When method (b) is used, the adaptive estimation algorithm is a recursive least squares algorithm with a forgetting factor; the method further includes: Before initiating or updating the recursive least squares algorithm, it is determined whether the energy of the current fluctuation data exceeds a preset threshold. Only if it exceeds the threshold is the recursive least squares algorithm initiated or updated; otherwise, the current equivalent impedance parameter estimate is maintained.

6. The method according to claim 5, characterized in that, When method (b) is used, the adaptive estimation algorithm employs a multi-model parallel identification and fusion strategy, specifically including: At least two recursive least squares identification models with different dynamic characteristics are run in parallel, and each recursive least squares identification model outputs an intermediate estimate of the equivalent reactance parameter. Calculate the instantaneous confidence weights of each intermediate estimate in real time; Based on the instantaneous confidence weights, the intermediate estimates are weighted and fused to obtain the final output equivalent reactance parameters.

7. The method according to claim 5 or 6, characterized in that, The forgetting factor ranges from 0.95 to 0.

999.

8. The method according to claim 1, characterized in that, In the impedance measurement operation, before calculating or estimating the equivalent impedance parameters, the voltage response data and current response data are preprocessed. The preprocessing includes applying a notch filter corresponding to the switching frequency of the grid-type SVG to filter out the high-order harmonic components characteristically generated by the switching action of the power devices of the grid-type SVG.

9. An electronic device, characterized in that, Including processor and memory; The processor executes the steps of the method as described in any one of claims 1 to 8 by invoking programs or instructions stored in the memory.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium is used to store a program or instructions that cause a computer to perform the steps of the method as described in any one of claims 1 to 8.

Images