New energy grid-connected inverter parameter identification method fusing time-frequency domain dynamic characteristics

Abstract

The invention discloses a new energy grid-connected inverter parameter identification method fusing time-frequency domain dynamic characteristics, and the identification process is divided into three stages: a time domain identification stage, a frequency domain identification stage and a time-frequency domain characteristic verification stage. According to the method, a new energy grid-connected inverter impedance characteristic curve database is established by combining time domain and frequency domain operation characteristic curves of a new energy grid-connected inverter'white box model 'of a known control strategy, and the operation characteristic curve in the impedance characteristic curve database is compared with the operation characteristic curve of a new energy grid-connected inverter'black box model'. And the control mode conforming to the operation characteristics of the new energy grid-connected inverter is identified and obtained. Compared with an existing new energy inverter parameter identification method, the new energy inverter parameter identification method has the advantages that the modeling process of the new energy inverter is simplified, and the new energy inverter parameter identification efficiency is greatly improved; and the input of a new control strategy is supported, and a guarantee is provided for the identification range and precision of the impedance characteristic curve database.

Description

Technical Field

[0001] This invention relates to the field of parameter identification technology for grid-connected inverters for new energy, and specifically to a method for parameter identification of grid-connected inverters for new energy that integrates time-frequency domain dynamic characteristics. Background Technology

[0002] At present, conventional wind and solar power plants in the new power system have room for improvement in terms of poor stability and weak support capabilities. Clarifying the parameters and control modes of the grid-connected inverters of new energy is a prerequisite for studying the grid-connected stability of new energy power plants. However, due to reasons such as commercial secrets and intellectual property rights, manufacturers cannot provide accurate parameters and control modes of grid-connected inverters of new energy. There are multiple strategies in the steady-state power control and voltage ride-through control links of grid-connected inverters of new energy, and different models may adopt different control modes. Therefore, while identifying parameters, it is necessary to find the control mode that best matches the operating characteristics of the grid-connected inverter, as recorded in the literature [1]: Wang Xiaotong, Wang Tong, Deng Jun, et al. Control mode and parameter integrated identification strategy of electromechanical transient model of photovoltaic inverter [J]. Power System Technology, 2023, 47(09): 3547-3558. DOI: 10.13335 / j.1000-3673.pst.2022.2271. This requires identifying parameters and finding a control mode that matches the inverter's operating characteristics based on the measured data from the manufacturer's "black box" model of the new energy grid-connected inverter. Existing research mostly focuses on parameter identification by combining theoretical mathematical models with intelligent optimization algorithms, with limited analysis of actual manufacturer models and often using only time-domain or frequency-domain identification methods to identify the control mode and parameters. Summary of the Invention

[0003] To address the shortcomings of existing parameter identification technologies for renewable energy grid-connected inverters, this invention provides a parameter identification method for renewable energy grid-connected inverters that integrates time- and frequency-domain dynamic characteristics. The identification process consists of three stages: time-domain identification, frequency-domain identification, and time- and frequency-domain characteristic verification. This method establishes a database of impedance characteristic curves for renewable energy grid-connected inverters by combining the time- and frequency-domain operating characteristic curves of a "white-box model" of a renewable energy grid-connected inverter with known control strategies. The operating characteristic curves in this database are then compared with the operating characteristic curves of a "black-box model" of the renewable energy grid-connected inverter to identify control modes that conform to the operating characteristics of the renewable energy grid-connected inverter.

[0004] The technical solution adopted in this invention is as follows: A time-domain identification method for grid-connected inverters of new energy sources includes the following steps: A1: Based on the grid-connected inverter of new energy sources or its electromagnetic packaging model, high and low voltage ride-through fault conditions are set up through experiments or simulations to obtain the voltage of the grid-connected inverter of new energy sources throughout the fault response process.U Active power P reactive power Q Active current I p reactive current I q Key electrical quantities; A2: Based on the key electrical quantity data obtained under high and low voltage ride-through fault conditions, the following plots are drawn respectively. PU , QU , I p - U , I q - U Isomorphic characteristic curves; A3: After initially identifying the control mode and parameters of the new energy grid-connected inverter through the characteristic curve, set a continuous voltage change from 0 to 1.4 pu, simulate and observe whether the new energy output reaches saturation, and thus determine the limiting value of the new energy grid-connected inverter under high and low voltage ride-through fault conditions. A4: Extract local control loop features by combining time-domain waveforms, compare the response characteristics during the fault period, and determine the structure of the fault-crossing control; A5: By collecting the time-domain waveforms before and after the fault in a white-box model or an electromagnetic encapsulation model with known control modes under fault ride-through conditions, a fault ride-through morphology database is formed.

[0005] A6: Based on the measured data under fault conditions, the output results of the measured data and theoretical formulas can be fitted by the least squares method or intelligent algorithm based on the aforementioned general fault transient control strategy to identify the control parameters of the transient link, so as to preliminarily determine the possible fault transient control strategy and provide initial conditions for subsequent frequency domain identification.

[0006] In A1, the electromagnetic packaging model is as follows: Figure 6 The image shows an electromagnetic encapsulation model from a certain manufacturer. The modules within the red box represent the control parameters of the input inverter. After encapsulation, the specific control parameters include the voltage outer loop proportional-integral coefficient K. p_v K i_v Since all of these are unknown, it is also known as a black box model.

[0007] Voltage of new energy grid-connected inverters throughout the fault response process U Active power P reactive power Q Active current I p reactive current I q Key electrical quantities; as shown in Figures 3(a) and 3(b), these are the fault response curves of key electrical quantities throughout the entire process.

[0008] In A2, respectively, the following are drawn PU , QU , I p - U , I q - U Isomorphic characteristic curves; as follows: The control modes commonly used in fault ride-through conditions are usually either constant power or constant current. (1); (2); (3); (4); in: This represents the active power during the fault period; This represents the reactive power during the fault period; This represents the active current during the fault period; Represents the reactive current during the fault period; P 0 and Q 0 represents the initial active power and initial reactive power of the converter, respectively; K P and K Q These represent the active power control coefficient and the reactive power control coefficient, respectively. P SET and Q SET These represent the active power setpoint and the reactive power setpoint, respectively. IP 0 and IQ 0 represents the initial value of active current and the initial value of reactive current, respectively; IP SET and IQ SET These are the active current setting value and the reactive current setting value, respectively. K V This refers to the voltage control coefficient; K I This refers to the current control coefficient; V t Terminal voltage; V SET Voltage threshold; Equations (1) and (2) represent the principle of constant power mode, while equations (3) and (4) represent the principle of constant current mode.

[0009] In constant power mode, the power output of the renewable energy source during a fault is independent of the generator terminal voltage and depends only on the power control coefficient and the initial power. P0、 Q 0 and power setting value P SET , Q SET That is, it is only subject to power limiting constraints; In constant current mode, the power output of the new energy source during a fault is proportional to the terminal voltage or its change, and is also affected by the initial current value. IP 0、 IQ 0 and current setting value IP SET , IQ SET The impact.

[0010] In A3, the control mode and parameters of the new energy grid-connected inverter are initially identified through characteristic curves; Figures 3(a) and 3(b) show the characteristic curves at the fault moment. In Figure 3(a), the active power is maintained near the control parameter 0.875pu, and the active current decreases as the voltage increases, therefore a constant power mode is adopted. For the reactive power and reactive current curves, the reactive current decreases as the voltage increases, therefore a constant current mode is adopted.

[0011] By determining whether the output of new energy sources reaches saturation, the limiting value of the grid-connected inverter under high and low voltage ride-through fault conditions can be determined. For example... Figure 7 As shown, the voltage and reactive current change curves of the new energy inverter during the low voltage ride-through test show that the reactive current reaches the limit value of 1.05 when the voltage drops to 0.35Un.

[0012] In section A4, local control loop features are extracted by combining time-domain waveforms, and the response characteristics during the fault period are compared to determine the structure of the fault-crossing control, as detailed below: like Figure 8 As shown, the dynamic response of a grid-connected inverter after a fault can be divided into three stages: steady state before the fault, during the fault, and steady state after the fault, as well as two transition stages: transient state when the fault occurs and transient state when the fault recovers.

[0013] For constant power control, without an outer power loop, the power is directly given. When a high-low crossover occurs, the current signal of the inner loop changes rapidly. When an outer power loop exists, due to the regulation of the PI controller in the power loop, the change in the inner loop current lags behind that without an outer loop, and this can be reflected more intuitively in the time-domain waveform.

[0014] For constant voltage control, i.e., the terminal voltage of the new energy control unit, a significant characteristic is that the reactive power and reactive current of the system respond synchronously. Therefore, reactive power and current change synchronously before a fault occurs. However, under constant power control, power and current change independently, and their trajectories do not overlap.

[0015] This is used to determine the characteristics of the local control loop: whether there is an outer loop, constant voltage control, or constant current control, etc.

[0016] In A5, the white-box model refers to the unencapsulated control parameter module in A1, meaning all control parameters in the model are known. The electromagnetic encapsulation model with a known control mode refers to the encapsulated control parameter module in A1, where the specific control parameters are provided by the equipment manufacturer. In other words, all control parameters in the model are also known.

[0017] The time-domain waveforms before and after the fault during fault ride-through are used to form a fault ride-through morphology database. The fault ride-through morphology database consists of: (The database is structured as follows...) Figure 8 The dataset is formed by integrating the inverter fault dynamic response curves shown.

[0018] In A6, based on measured data under fault conditions, the output results of the measured data and theoretical formulas can be fitted using the least squares method or intelligent algorithm based on the aforementioned general fault transient control strategy. This enables the identification of control parameters for the transient process, thereby initially determining possible fault transient control strategies. Specifically, as follows: By employing the least squares method or intelligent algorithms to fit the output results of measured data and theoretical formulas, the control parameters of transient processes are identified. The curve with the smallest error between the fault ride-through mode database and the measured data is found to initially determine the control mode, such as constant power control or constant voltage control. The principle of the least squares method is as follows: (5); (6); In the above formula: This indicates the electromagnetic packaging model of the new energy inverter in Output measurement value at time; This represents the output measurement value of the white-box model of the new energy inverter at time t. This represents the residual between the electromagnetic packaging model and the white-box model of the new energy inverter at time t. N This represents the total number of sampling points at time t. J ( t ) is the objective function fitted by the least squares method.

[0019] A frequency domain identification method for grid-connected inverters for new energy sources includes the following steps: B1: Based on the new energy white-box model with different control strategies in the time-domain simulation model, frequency sweep is performed under the conditions of large and small output of new energy to obtain the frequency response curves of typical control modes in the 0-2000Hz wideband, and an impedance characteristic curve database is established. B2: Adjust the inner and outer loop control parameters, and through a large number of simulations and frequency sweep analyses, obtain the degree of influence of the inverter's control parameters on the impedance characteristic curve under different control strategies and operating points in the impedance characteristic curve database. B3: Define the impedance sensitivity to changes in parameter amplitude or phase. and phase sensitivity The indicators are used to form low, medium, and high frequency band influence markers, which facilitates the identification of impedance curves based on the impedance characteristic curve database.

[0020] B4: For actual grid-connected inverters or black-box models of new energy, the small-signal response characteristics of new energy units in the 0-2000Hz frequency band are first obtained by the perturbation method, and the perturbation voltage, current and its response current and voltage data within the specified frequency range are obtained by frequency sweeping. B5: Then, the impedance characteristic data curve of the new energy unit is obtained by FFT calculation, and the low frequency band is defined as 0-10Hz, the mid frequency band as 10-100Hz, and the high frequency band as 100-2000Hz. B6: Based on the impedance characteristic data curves obtained from B5, the amplitude and phase errors of the impedance characteristic data curves for each frequency band and the reference curves in the impedance characteristic curve database at different frequency bands are calculated based on the error index. B7: Obtain the control strategy corresponding to the impedance curve with the smallest error among all indicators, and use it as a candidate control strategy.

[0021] In B1, the white-box models for different control strategies are as follows: A white-box model is one where the control parameter module in A1 is unencapsulated. This means that all control parameters in the model are known, as is the specific control mode of the new energy model.

[0022] The control strategies include two control modes: constant power and constant voltage. Constant power control includes two types: one with a power outer loop and one without a power outer loop. Constant voltage control mainly refers to grid-type control, including two types: droop control and virtual synchronous machine control.

[0023] In B1, impedance characteristic curves under typical control modes within a 0-2000Hz wideband are obtained by frequency sweeping under both high and low power output conditions of the new energy source; the details are as follows: Simulations are performed by changing the frequency of the injected voltage or current signal, using a series of continuous frequencies to obtain the values ​​of three-phase voltage and three-phase current within a specified frequency range. For example... Figure 9 The figure shows the impedance characteristic curves of the new energy inverter obtained by frequency sweep under high and low output conditions in a wide frequency band.

[0024] In B2, the inner and outer loop control parameters include: such as the proportional coefficient and integral coefficient K of the inner current loop. p_i Ki_i Voltage outer loop proportional coefficient and integral coefficient K p_v K i_v The proportional and integral coefficients of the phase-locked loop, K p_pll K i_pll wait.

[0025] In B3, the impedance amplitude sensitivity to parameter changes or phase changes is defined. and phase sensitivity The specific indicators are as follows: (7); (8); In the above formula, The impedance amplitude of the new energy inverter; These are the inner and outer loop control parameters for the new energy inverter; The impedance amplitude of the new energy inverter; the impedance amplitude sensitivity and phase sensitivity indicators are arbitrary parameters. The physical meaning of the relative rate of change of impedance amplitude and phase is as follows: Indicates parameters An increase of 1% results in a 1% increase in impedance amplitude; Indicates parameters An increase of 1% increases the phase by 1 degree; and The parameters can be quantified and the key parameters that have a significant impact on impedance characteristics can be screened.

[0026] In B4, the black-box model is the new energy inverter model explained in A1, which encapsulates control parameters and unknown control modes. First, the small-signal response characteristics of the new energy unit in the 0-2000Hz frequency band are obtained using the perturbation method, as follows: The principle of the perturbation method is as follows... Figure 10 As shown: A small disturbance signal within a certain frequency range is injected at the grid connection point of the unit to obtain the small signal response data (voltage, current) of the unit in the specified frequency band, and then the impedance characteristics of the system are calculated by FFT processing.

[0027] The disturbance voltage, current and its response current and voltage data within a specified frequency range are obtained by frequency sweeping. Specifically, the simulation is performed by changing the frequency of the injected voltage or current signal, and the values ​​of the three-phase voltage and three-phase current within the specified frequency range are obtained by a series of continuous frequencies.

[0028] In B5, the broadband impedance data of the new energy unit is then calculated using FFT, as follows: FFT calculation is a fast Fourier decomposition algorithm. The voltage and current data obtained from the frequency sweep of the new energy inverter are then processed by FFT to obtain the following result: Figure 11 The impedance characteristic data curve is shown.

[0029] In section B6, the amplitude and phase error indices of each impedance characteristic data curve and each reference curve in the impedance characteristic curve database are calculated at different frequency bands; specifically as follows: ①: The formula for calculating the mean absolute error is: (9); In equation (7): K s , K c These represent the total number of simulation data within the calculation error interval and the total number of reference data in the database, respectively. , These represent the per-unit values ​​of the model simulation data and the per-unit values ​​of the reference data in the database, respectively; x = L , M , H These represent the low, mid, and high frequency bands, respectively. ②: The formula for calculating the maximum error is: (10); ③: The formula for calculating the weighted average absolute error is: (11); In equation (9): k 1. k 2. k 3 represents the average absolute error coefficients for the low, medium, and high frequency bands, respectively; s 1,L , s 1,M , s 1,H These represent the average absolute error values ​​for the three frequency bands, respectively.

[0030] In B7, the control strategy corresponding to the impedance curve that minimizes the error of each index is obtained, as follows: Based on the measured impedance characteristic data curves of the black-box model obtained from B5, the error calculation formula in B6 is used to compare them with reference curves in the impedance characteristic curve database. The specific control strategy represented by the impedance characteristic curve in the database with the smallest error compared to the measured curve is identified. Specific control modes (constant voltage control, constant power control, etc.) and specific control parameters (voltage outer loop proportional-integral coefficient, current inner loop proportional-integral coefficient, etc.) are extracted from the database and used as the specific control strategy for the measured black-box model impedance characteristic data curves.

[0031] A time-frequency domain verification method for new energy grid-connected inverters is proposed. Combining the impedance parameter sensitivity index calculated in step B3 above, the average error or the maximum error in each frequency band is analyzed. For larger maximum errors, the influence of measurement should be eliminated, and parameter adjustments should be made to see if the maximum error can be reduced. For larger average errors, parameters should be adjusted according to the influence law of impedance sensitivity, and repeated frequency sweep observations should be performed to optimize the impedance characteristic data curve. Alternatively, the impedance characteristic curve database can be expanded based on a new control strategy combination. After the frequency domain characteristics are basically satisfied, the simulation results of different operating conditions of the white-box model and black-box model of this type of new energy grid-connected inverter are compared, and secondary parameter optimization is performed based on the error situation.

[0032] Includes the following steps: C1: Combine impedance parameter sensitivity index to analyze the average error or maximum error of each frequency band.

[0033] C2: Based on candidate control strategies, simulations are performed under typical disturbance or fault conditions by calling white-box models; C3: Analyze the difference between the identification results in step B7 and the actual response characteristics of the grid-connected inverter, and fine-tune the white-box model parameters; C4: For inverters that do not perform well in identifying typical control strategies in the impedance characteristic curve database, further fine-tune the initial control parameters or obtain new control strategy combinations by combining frequency domain impedance characteristics and time domain simulation analysis.

[0034] In C1, the average error or maximum error of each frequency band is analyzed in conjunction with the impedance parameter sensitivity index, as detailed below: Based on the impedance parameter sensitivity index, analyze the parameters with higher sensitivity in the part with large error. If the maximum error is large, focus on eliminating the influence of measurement and adjusting the parameters to see if the maximum error can be eliminated or reduced. If the average error is large, adjust the parameters according to the influence law of impedance sensitivity and perform repeated frequency sweep observations to optimize the impedance curve.

[0035] In C2, the white-box model refers to the unencapsulated state of the control parameter module in A1. That is, all control parameters in the model are known, and the specific control mode of the new energy model is also known.

[0036] The simulation under typical disturbance or fault conditions using the white-box model is the same as in step A1. Set up high-voltage ride-through and low-voltage ride-through fault conditions, or add disturbances to obtain the characteristic curves of the corresponding parameters (active power, active current, reactive power, reactive current, etc.) through simulation. In C3, the difference between the identification result and the actual grid-connected inverter response characteristics is specifically identified through error analysis to find the part with larger errors.

[0037] Fine-tuning the white-box model parameters specifically involves adjusting known parameters in the white-box model that are highly sensitive to a significant portion of the error, such as the proportional-integral coefficient of the voltage outer loop and the proportional-integral coefficient of the current inner loop.

[0038] In C4, the initial control parameters are further fine-tuned or new control strategy combinations are obtained by combining frequency domain impedance characteristics and time domain simulation analysis. If the comparison results show a weighted average error for some frequency bands The following threshold conditions were not met: (12); Based on sensitivity indicators, parameters with high sensitivity to frequency bands with large errors in the white-box model can be found, such as the proportional-integral coefficient of the voltage outer loop and the proportional-integral coefficient of the current inner loop. By increasing or decreasing these parameters, frequency domain impedance characteristic analysis and time domain simulation analysis can be performed to ensure that the error meets the threshold range. The new combination of control parameters obtained after adjustment can be recorded as a new control strategy in the impedance characteristic curve database.

[0039] This invention provides a parameter identification method for new energy grid-connected inverters that integrates time-frequency domain dynamic characteristics. The technical effects are as follows: 1) In the frequency domain identification stage of this invention, impedance characteristic curves of new energy white-box models based on different typical control strategies are obtained by impedance scanning under different operating conditions, and an impedance characteristic curve database is established. Its advantage lies in associating control mode, operating mode, frequency band characteristics and impedance curves, realizing the structuring of complex nonlinear systems. Compared with existing new energy inverter parameter identification methods, it simplifies the modeling process of new energy inverters and greatly improves the efficiency of new energy inverter parameter identification. It supports the input of new control strategies, which provides a guarantee for the identification range and accuracy of the impedance characteristic curve database.

[0040] 2) In the frequency domain identification stage of this invention, the sensitivity of parameter changes to impedance amplitude or phase changes is defined. and phase sensitivity The advantage of this index lies in its ability to quantify the sensitivity of control parameters to impedance amplitude and phase, calculate sensitivity indices for low, medium, and high frequency bands, identify the control parameters that have the greatest impact on impedance characteristics in a specific frequency band, and improve parameter identification accuracy. For parts of the identification results with large errors, parameters with higher sensitivity can be adjusted first to improve parameter identification efficiency.

[0041] 3) In the time-frequency domain characteristic verification stage of this invention, black-box models whose identification parameter errors in the frequency domain identification stage do not meet the threshold are identified. Its advantage lies in the fact that based on the candidate control strategy, through the closed-loop mechanism of time-domain simulation verification and frequency domain error analysis, the initial control parameters are further fine-tuned, which can more accurately identify the control strategy of the new energy black-box model and improve the identification accuracy. After obtaining the new control strategy combination, it can be entered into the impedance characteristic curve database to improve the database and further enhance the applicability of the impedance characteristic database. Attached Figure Description

[0042] The present invention will be further described below with reference to the accompanying drawings and embodiments: Figure 1 This is a flowchart of the parameter identification method of the present invention.

[0043] Figure 2 This is a flowchart of the time-domain identification stage of the present invention.

[0044] Figure 3(a) shows a schematic diagram of the time-domain discrete analysis of the new energy control strategy. Figure one ; Figure 3(b) shows a schematic diagram of the time-domain discrete analysis of the new energy control strategy. Figure 2 .

[0045] Figure 4 This is a flowchart of the frequency domain identification stage of the present invention.

[0046] Figure 5 This is a flowchart of the time-frequency domain verification stage of the present invention.

[0047] Figure 6 This is a schematic diagram of an electromagnetic encapsulation model (black box model).

[0048] Figure 7 The voltage and reactive current variation curves of the new energy inverter during the low voltage ride-through test are shown.

[0049] Figure 8 This is the dynamic response curve of the grid-connected inverter after a fault.

[0050] Figure 9 The impedance characteristic curves of the new energy inverter are obtained by frequency sweeping under high and low output conditions, covering a wide frequency band.

[0051] Figure 10 This is a schematic diagram of the perturbation method.

[0052] Figure 11 This is the impedance characteristic data curve. Detailed Implementation

[0053] This invention proposes a structured modeling method for new energy grid-connected inverters that integrates time-domain and frequency-domain dynamic characteristics. By combining the analysis of time-domain dynamic response characteristics with frequency-domain impedance characteristics, high-precision collaborative identification of new energy grid-connected inverter parameters is achieved. The identification process consists of three stages: time-domain identification, frequency-domain identification, and time-frequency domain characteristic verification.

[0054] (1) In the time domain identification stage, the waveforms of a large number of new energy grid-connected inverter white box models or electromagnetic encapsulation models with known control modes are obtained by simulation, and a fault ride-through characteristic database is established. Then, the transient control strategy of the inverter is initially identified through the fault evolution mode, and the transient link parameters are obtained by fitting through optimization algorithm.

[0055] (2) In the frequency domain identification stage, the impedance characteristic curve is obtained by sweeping the frequency of a typical white box model, and the impedance characteristic is divided into three frequency bands: low, medium and high. An impedance characteristic curve database is established. Then, the amplitude and phase error index of the measured impedance curve and the reference impedance curve of each frequency band are calculated to match the impedance curves and obtain the best control structure as a candidate.

[0056] (3) In the time-frequency domain characteristic verification stage, the error between the converter using the candidate control strategy and the measured data under various operating conditions is tested, and the parameters are fine-tuned to optimize the effect. For control strategies that cannot be effectively matched with the database, control strategy combination or algorithm optimization should be carried out to further identify the converter. This method improves the identification efficiency and accuracy of new energy converters from a multi-time-frequency perspective by combining the time-domain and frequency-domain response characteristics of the converter, providing technical support for the simulation modeling and testing of new energy in large power grids.

[0057] I. Time Domain Identification: In professional electromechanical simulation software such as BPA and PSASP, the active power control methods of new energy grid-connected inverters in high-voltage and low-voltage ride-through modes are divided into constant active power control, constant active current control, low-voltage limit active current control, and active power recovery control (including slope control and exponential recovery); reactive power control methods are divided into constant reactive power control, constant reactive current control, and voltage-reactive power droop control. Therefore, time-domain identification mainly consists of two key processes: identifying the control mode and identifying the control parameters.

[0058] S1.1: Based on the grid-connected inverter of new energy or its electromagnetic packaging model, high and low voltage ride-through fault conditions are set through experiments or simulations, including symmetrical three-phase short circuit faults and asymmetrical single-phase ground faults, to obtain key electrical quantities such as voltage U, active power P, reactive power Q, active current Ip, and reactive current Iq of the grid-connected inverter of new energy during the entire fault response process.

[0059] S1.2: Based on the data collected under typical high-low fault ride-through conditions, characteristic curves such as PU, QU, Ip-U, and Iq-U are plotted respectively. The commonly used control modes during fault ride-through are usually constant power or constant current. For example, equations (1) and (2) are the principles of constant power mode, and equations (3) and (4) are the principles of constant current mode. After identifying and obtaining the control mode and parameters, the output of new energy can be simulated and observed to determine whether the output of new energy grid-connected inverter reaches saturation by setting a continuous voltage change from 0 to 1.4 pu.

[0060] S1.3: Not only can control parameters be identified using voltage and power changes during a fault, but local control loop characteristics can also be extracted by combining time-domain waveforms. For example, steady-state control modes include constant power (divided into two categories: with and without a power outer loop, i.e., grid-following type) mode and constant voltage mode (grid-forming type). During the transient period, fault ride-through control is employed.

[0061] 1) For constant power control, without an outer power loop, the power is directly given, and the current signal of the inner loop changes rapidly when a high-low current surge occurs. When an outer power loop exists, due to the regulation of the PI controller in the power loop, the change in the inner loop current lags behind that without an outer loop, and therefore can be reflected more intuitively in the time-domain waveform.

[0062] 2) For constant voltage control, i.e., the terminal voltage of the new energy control unit, a significant characteristic is that the reactive power and reactive current responses of the system are synchronized. Therefore, reactive power and current change synchronously before a fault occurs. Under constant power control, however, power and current change independently, and their trajectories do not overlap. Furthermore, comparing the response characteristics during a fault can reveal the structure of the fault ride-through control. For example, the simplest fault ride-through strategy calculates the power or current during the fault using a linear relationship. In this mode, the power and current responses after the fault are relatively fast. Alternatively, an outer-loop control strategy can be used, where the reference value of the current during the fault period in the inner loop is calculated from the outer-loop ride-through control.

[0063] By collecting the time-domain waveforms before and after the fault in a white-box model or an electromagnetic encapsulation model with known control modes under fault ride-through conditions, a fault ride-through morphology database can be formed. Then, based on the measured data under fault conditions, the output results of the measured data and theoretical formulas can be fitted using the least squares method or intelligent algorithm based on the aforementioned general fault transient control strategy to identify the control parameters of the transient link. However, this step is only used to preliminarily determine the possible control modes in order to provide initial conditions for subsequent frequency domain parameter identification.

[0064] II. Time Domain Identification: Time-domain identification methods only provide a preliminary understanding of the outer-loop control strategy, not the overall control information, and rely solely on high-low impedance characteristic simulation, making it difficult to accurately reflect the control strategy. Frequency-domain information typically covers low, mid, and high frequency bands, allowing for control strategy identification through the impedance characteristics of these three bands. The frequency-domain identification method proposed in this invention comprises the following steps: S2.1: Impedance Characteristic Database Establishment: Based on white-box models of different control strategies in the time-domain simulation model (excluding high-low throughput control, only steady-state control), including two control modes: constant power and constant voltage. Constant power control includes two types: power outer loop and no power outer loop. Constant voltage control mainly refers to grid-type control, mainly including droop control and virtual synchronous machine control. Based on the time-domain simulation model, frequency response curves under typical control modes in the 0-2000Hz frequency band are obtained by frequency sweeping under high and low power output conditions of new energy sources, and an impedance characteristic curve database is established. S2.2: Impedance Parameter Sensitivity Analysis: Adjusting the inner and outer loop control parameters, a frequency sweep was performed again based on the time-domain simulation model to obtain the impedance characteristic variation trends across low, medium, and high frequency bands under typical parameter ranges. Through extensive simulations and frequency sweep analyses, the influence of key parameters on the impedance characteristic curves under different control strategies and operating points in the impedance characteristic database was obtained. S2.3: For actual grid-connected inverters or black-box models of new energy sources, the small-signal response characteristics of the new energy unit in the 0-2000Hz frequency band are first obtained through the perturbation injection method. Perturbation voltage / current and its response voltage / current data within a specified frequency range are obtained through frequency scanning. Then, wideband impedance data of the new energy unit is calculated using FFT. The low-frequency band is defined as 0-10Hz, the mid-frequency band as 10-100Hz, and the high-frequency band as 100-2000Hz. Then, based on the obtained impedance curves under multiple operating conditions, they are compared with impedance curves in the database at the low, mid, and high frequency bands respectively. The comparison method is as follows: based on the three defined frequency bands, the amplitude and phase error indicators of the measured curve and the reference curve (the curve in the database is used as the reference curve) for each frequency band are calculated. The indicators can be divided into mean absolute error. s 1. Maximum error s 2 and weighted average absolute error s 3.

[0065] Based on the calculated error index, we find the one with the minimum weighted average absolute error and determine whether it meets the error threshold. s e If the error requirement is met, the corresponding control strategy in the database is output as a candidate control strategy. If the error requirement is not met, it means that the curve in the existing database cannot be matched well. The parameters can be further optimized by adjusting the control strategy based on the control strategy with smaller error value in the database through the time-frequency domain characteristic verification in the third stage.

[0066] III. Time-frequency domain characteristic verification: Time-frequency domain characteristic verification is based on candidate control strategies. It involves conducting simulations under typical disturbance or fault conditions using a white-box model, analyzing the differences between the identified model results and the actual controller response characteristics, and fine-tuning the white-box model parameters. For converters in the database where typical control strategies do not perform well, initial control parameters are further fine-tuned or new control strategy combinations are obtained by combining frequency-domain impedance characteristics and time-domain simulation analysis. Finally, the average error or the maximum error in each frequency band is analyzed based on the impedance parameter sensitivity index calculated in the above steps.

[0067] For cases with a large maximum error, the influence of measurement should be eliminated, and parameter adjustments should be made to see if the maximum error can be reduced. For cases with a large average error, the impedance curve can be optimized by adjusting parameters and conducting repeated frequency sweeps, based on the influence law of impedance sensitivity. Alternatively, a new control strategy combination can be used to expand the impedance characteristic database. After the frequency domain characteristics are basically satisfied, the simulation results of different operating conditions based on the white-box model and black-box model of this type of new energy grid-connected inverter should be compared, and secondary parameter optimization should be performed based on the error situation.

[0068] After the frequency domain characteristics are basically satisfied, the simulation results of the white-box model and the black-box model of the new energy grid-connected inverter under different operating conditions are compared to verify the consistency of the time domain simulation results. Based on the analysis of steady-state and transient results, the control parameters are further fine-tuned and optimized to make the time domain response characteristics of the two more matched.

[0069] IV. Implementation Examples: like Figure 1 The overall technical route of the parameter identification method for new energy grid-connected inverters that integrates time-domain and frequency-domain dynamic characteristics proposed in this invention mainly includes three stages: time-domain identification, frequency-domain identification, and time-frequency characteristic verification.

[0070] like Figure 2 This section outlines the specific technical roadmap for the time-domain identification phase. Based on a new energy grid-connected inverter or its electromagnetic encapsulation model, time-domain simulation is performed by setting a high-voltage ride-through condition to obtain the voltage throughout the fault response process. U Active power P reactive power Q Active current I p reactive current I q Key electrical quantities were measured, and the control mode and parameters were initially identified.

[0071] Equations (1) and (2) illustrate the principle of constant power control mode, where... P 0 and P SETThese represent the initial active power and the active power setpoint, respectively. Q 0 and Q SET These represent reactive power and reactive power setpoint, respectively. K P and K Q These represent the active power control coefficient and reactive power control coefficient. For constant power control, the active power during a fault is... P ref and reactive power Q ref Depends on the active power control coefficient K P and reactive power control coefficient K Q Initial active power P 0 and initial reactive power Q 0 and active power setpoint P SET and reactive power setpoint Q SET However, the only variable that has a major influence on its value is the active power control coefficient. K P Reactive power control coefficient K Q and merit settings P SET and reactive power setpoint Q SET .

[0072] (1); (2); Equations (3) and (4) illustrate the principle of constant current mode. In the equations... IP 0 and IQ 0 represents the initial value of active current and the initial value of reactive current, respectively. IP SET and IQ SET These are the active current setting value and the reactive current setting value, respectively. K V This is the voltage control coefficient. K I This is the current control coefficient. V t Terminal voltage, V set This refers to the voltage threshold. For constant current control, the fault primarily affects the active current. IP ref and reactive current IQ ref Size has voltage control coefficientK V Current control coefficient K I Active current setpoint IP SET Reactive current setting value IQ SET .

[0073] (3); (4); That is, to obtain the voltage throughout the entire fault response process. U Active power P reactive power Q Active current I p reactive current I q Draw the key electrical quantities separately. P - U , Q - U , I p - U , I q - U Characteristic curves are used to preliminarily identify the control mode of new energy grid-connected inverters or their electromagnetic encapsulation models through curve characteristics.

[0074] Figures 3(a) and 3(b) show scatter plots of active power, active current, reactive power, and reactive current of the photovoltaic inverter under symmetrical high-voltage surge conditions at high and low outputs, respectively. The scatter plot for the high-output symmetrical high-voltage surge condition shows that the active power of the wind turbine remains around 0.875 pu during the surge period, and the active current decreases as the voltage increases; therefore, constant power control is employed. For the reactive power and reactive current curves, the reactive current decreases as the voltage increases; therefore, constant current control is used. The scatter plot for the low-output symmetrical high-voltage surge condition shows that the active power of the wind turbine remains around 0.19 pu during the surge period, and the active current decreases as the voltage increases; therefore, a constant power control strategy is adopted. For the reactive current curve, the reactive current decreases as the voltage increases; therefore, constant current control is used. After identifying and obtaining the control mode and parameters, the output of the new energy source can be simulated and observed by setting a continuous voltage change from 0 to 1.4 pu, thereby determining the limiting value of the new energy grid-connected inverter under high and low load conditions.

[0075] By accumulating white-box models or electromagnetic encapsulation models with known control modes, and using time-domain waveforms before / after the fault, amplitude limiting values, and other known data under fault ride-through conditions, a fault ride-through morphology database can be formed. For the "black-box" model of new energy grid-connected inverters, based on the measured data under fault conditions, the least squares method or intelligent algorithm can be used to fit the output results of the measured data and theoretical formulas based on the aforementioned general fault transient control strategy, thereby identifying the control parameters of the transient link and initially determining possible steady-state and fault ride-through control strategies. However, this step is only used to initially judge possible control modes in order to provide initial conditions for subsequent frequency domain parameter identification.

[0076] like Figure 4 This document outlines the specific technical roadmap for the frequency domain identification phase. Based on a time-domain simulation model, frequency sweeps are performed under both high and low power output conditions of the new energy source using electromagnetic mode transient simulation or hardware-in-the-loop simulation. A small disturbance signal within the 0-2000Hz frequency band is injected at the unit's grid connection point using a voltage disturbance method. The disturbance voltage amplitude is 1%-5% of the rated voltage amplitude. Frequency response curves under typical control modes are obtained through FFT calculations, and an impedance characteristic curve database is established. Then, the inner and outer loop control parameters are adjusted, and the time-domain simulation model is swept again. Sensitivity analysis is performed on each impedance parameter. The analysis process defines the impedance amplitude sensitivity to parameter changes in impedance amplitude or phase. and phase sensitivity Indicators. These are obtained through extensive simulations and frequency sweep analyses of time-domain simulation models, determining the degree of influence of key parameters on the impedance characteristic curves under different control strategies and operating points within the impedance characteristic database. Impedance amplitude sensitivity to changes in impedance amplitude or phase. and phase sensitivity The formula for calculating the indicator is as follows: (7); (8); This indicator means any parameter. x The physical meaning of the relative rate of change of impedance with respect to the change of resistance is as follows: Indicates parameters x An increase of 1% results in a 1% increase in impedance amplitude; Indicates parameters x An increase of 1% increases the phase by 1 degree. After obtaining the sensitivity of the influence of key parameters of impedance curves under different control strategies and different operating points on the impedance characteristic curve in the impedance characteristic database, influence flags for low, medium, and high frequency ranges are formed. The low frequency range is defined as 0-10Hz, the medium frequency range as 10-100Hz, and the high frequency range as 100-2000Hz, which facilitates the subsequent identification of impedance curves based on the database.

[0077] For actual grid-connected inverters or black-box models of new energy units, a small disturbance signal with a voltage amplitude of 1%-5% of the rated voltage amplitude within the 0-2000Hz frequency band is first obtained using the disturbance injection method. Frequency scanning is then used to obtain data on the disturbance voltage and its response current components at each frequency point within the specified frequency range. Finally, wideband impedance data of the new energy unit is calculated using FFT. Based on the impedance curve obtained under specified operating conditions, it is compared with impedance curves in the database at low, medium, and high frequency bands. The comparison method is as follows: Based on the three frequency bands, the amplitude and phase error indices of the measurement curves and reference curves (the curves in the database are used as reference curves) for each frequency band are calculated. The indices can be divided into mean absolute error. s 1. Maximum error s 2 and weighted average absolute error s 3.

[0078] The formula for calculating the mean absolute error is: (9); In the formula K s , K c These represent the total number of simulation data points and the total number of reference data points in the database, respectively, within the calculation error interval. X s , X c These represent the per-unit values ​​of the model simulation data and the per-unit values ​​of the reference data in the database, respectively. x = L , M , H These represent the low, medium, and high frequency bands, respectively.

[0079] The formula for calculating the maximum error is: (10); In the formula X s , X c These represent the per-unit values ​​of the model simulation data for the electrical quantities to be calculated and the per-unit values ​​of the reference data in the database, respectively.

[0080] The formula for calculating the weighted average absolute error is: (11); In the formula, k 1. k 2. k 3 represents the average absolute error coefficient for the low, medium, and high frequency bands, respectively. s 1,L , s 1,M , s 1,H These represent the average absolute error values ​​for the three frequency bands, respectively.

[0081] Based on the calculated error index, we find the one with the minimum weighted average absolute error and determine whether it meets the error threshold. s e If the error requirement is met, the corresponding control strategy in the database is output as a candidate control strategy. If the error requirement is not met, it means that the curve in the existing database cannot be matched well. The parameters can be further optimized by adjusting the control strategy based on the control strategy with smaller error value in the database through the time-frequency domain characteristic verification in the third stage.

[0082] like Figure 5 This document outlines the specific technical roadmap for the time-frequency domain characteristic verification stage. Based on candidate control strategies, simulations under typical disturbance or fault conditions are conducted using a white-box model. The differences between the identified model results and the actual controller response characteristics are analyzed, and the white-box model parameters are fine-tuned. For converters in the database where typical control strategies do not perform well, initial control parameters are further fine-tuned or new control strategy combinations are obtained by combining frequency-domain impedance characteristics and time-domain simulation analysis. Finally, based on the impedance parameter sensitivity indices calculated in the above steps, the average error or the maximum error in each frequency band is analyzed.

[0083] Among them, the maximum error s A large error of 2 indicates that individual data points have significant errors, and that the impedance characteristic curve is greatly affected by factors such as measurement interference and parameter influences. In this case, the focus should be on eliminating measurement interference and adjusting parameters to see if the maximum error can be reduced. Mean Absolute Error s A large value of 1 indicates that the overall fitting effect of the control strategy in a certain segment is poor, suggesting that the database does not contain this type of control strategy or is greatly affected by parameters. In this case, the first step should be to adjust the parameters according to the influence of impedance sensitivity and perform repeated frequency sweeps to optimize the impedance characteristic curve. Alternatively, a new combination of control strategies can be obtained to expand the impedance characteristic database.

[0084] After the frequency domain characteristics are basically satisfied, the simulation results of the white-box model and the black-box model of the new energy grid-connected inverter under different operating conditions are compared to verify the consistency of the time domain simulation results. Based on the analysis of steady-state and transient results, the control parameters are further fine-tuned and optimized to make the time domain response characteristics of the two more matched.

Claims

1. A time-domain identification method for a new energy grid-connected inverter, characterized in that... Includes the following steps: A1: Based on the grid-connected inverter of new energy sources or its electromagnetic packaging model, high and low voltage ride-through fault conditions are set up through experiments or simulations to obtain the voltage of the grid-connected inverter of new energy sources throughout the fault response process. U Active power P reactive power Q Active current I p reactive current I q Key electrical quantities; A2: Based on the key electrical quantity data obtained under high and low voltage ride-through fault conditions, the following plots are drawn respectively. PU , QU , I p - U , I q - U Characteristic curve; A3: After initially identifying the control mode and parameters of the new energy grid-connected inverter through the characteristic curve, set a continuous voltage change from 0 to 1.4 pu, simulate and observe whether the new energy output reaches saturation, and thus determine the limiting value of the new energy grid-connected inverter under high and low voltage ride-through fault conditions. A4: Extract local control loop features by combining time-domain waveforms, compare the response characteristics during the fault period, and determine the structure of the fault-crossing control; A5: By collecting the time-domain waveforms before and after the fault in the white-box model or the electromagnetic encapsulation model with known control modes under the fault ride-through condition, a fault ride-through morphology database is formed. A6: Based on the measured data under fault conditions, the output results of the measured data and theoretical formulas can be fitted using the least squares method or intelligent algorithm based on the aforementioned general fault transient control strategy, so as to identify the control parameters of the transient link and preliminarily determine the possible fault transient control strategy.

2. The time-domain identification method for a new energy grid-connected inverter according to claim 1, characterized in that: In A2, respectively, the following are drawn PU , QU , I p - U , I q - U Characteristic curves; details are as follows: The control modes commonly used in fault ride-through conditions are usually either constant power or constant current. （1）； （2）； （3）； （4）； in: This represents the active power during the fault period; This represents the reactive power during the fault period; This represents the active current during the fault period; Represents the reactive current during the fault period; P 0 and Q 0 represents the initial active power and initial reactive power of the converter, respectively; K P and K Q These represent the active power control coefficient and the reactive power control coefficient, respectively. P SET and Q SET These represent the active power setpoint and the reactive power setpoint, respectively. IP 0 and IQ 0 represents the initial value of active current and the initial value of reactive current, respectively; IP SET and IQ SET These are the active current setting value and the reactive current setting value, respectively. K V This refers to the voltage control coefficient. K I This refers to the current control coefficient; V t Terminal voltage; V SET Voltage threshold; Equations (1) and (2) represent the principle of constant power mode, while equations (3) and (4) represent the principle of constant current mode. In constant power mode, the power output of the renewable energy source during a fault is independent of the generator terminal voltage and depends only on the power control coefficient and the initial power. P 0、 Q 0 and power setting value P SET , Q SET That is, it is only subject to power limiting constraints; In constant current mode, the power output of the new energy source during a fault is proportional to the terminal voltage or its change, and is also affected by the initial current value. IP 0、 IQ 0 and current setting value IP SET , IQ SET The impact.

3. The time-domain identification method for a new energy grid-connected inverter according to claim 2, characterized in that: In section A4, local control loop features are extracted by combining time-domain waveforms, and the response characteristics during the fault period are compared to determine the structure of the fault-crossing control, as detailed below: The dynamic response of a grid-connected inverter after a fault can be divided into three stages: steady state before the fault, during the fault, and steady state after the fault, as well as two transition stages: transient state when the fault occurs and transient state when the fault recovers. For constant power control, without an outer power loop, the power is directly given. When a high-low crossover occurs, the current signal of the inner loop changes rapidly. When an outer power loop exists, due to the regulation of the PI controller of the power loop, the change of the inner loop current lags behind that without an outer loop, which can be reflected intuitively in the time domain waveform. For constant voltage control, i.e., the voltage at the terminal of the new energy control unit, the reactive power and reactive current of the system respond synchronously. Therefore, reactive power and current change synchronously before a fault occurs. However, under constant power control, power and current change independently and their trajectories do not overlap. This is used to determine the characteristics of the local control loop: Whether there is an outer loop, constant voltage control, or constant current control.

4. The time-domain identification method for a new energy grid-connected inverter according to claim 3, characterized in that: In A6, based on measured data under fault conditions, and based on a general fault transient control strategy, the least squares method or intelligent algorithm is used to fit the output results of the measured data and theoretical formulas to identify the control parameters of the transient link, so as to initially determine possible fault transient control strategies: as follows: By fitting the output results of measured data and theoretical formulas using the least squares method or intelligent algorithms, the control parameters of the transient process are identified; the curve with the smallest error between the fault ride-through mode database and the measured data is found to preliminarily determine the control mode; the principle of the least squares method is as follows: （5）； （6）； In the above formula: This indicates the electromagnetic packaging model of the new energy inverter in Output measurement value at time; This represents the output measurement value of the white-box model of the new energy inverter at time t. This represents the residual between the electromagnetic packaging model and the white-box model of the new energy inverter at time t. N This represents the total number of sampling points at time t. J ( t ) is the objective function fitted by the least squares method.

5. A frequency domain identification method for new energy grid-connected inverters, characterized in that... Includes the following steps: B1: Based on the white-box model of different control strategies in the time-domain simulation model, frequency response curves under typical control modes in the 0-2000Hz wideband are obtained by frequency sweeping under the conditions of large and small power output of new energy, and an impedance characteristic curve database is established. B2: Adjust the inner and outer loop control parameters, and through a large number of simulations and frequency sweep analyses, obtain the degree of influence of the inverter's control parameters on the impedance characteristic curve under different control strategies and operating points in the impedance characteristic curve database. B3: Define the impedance sensitivity to changes in parameter amplitude or phase. and phase sensitivity The indicators are used to form low, medium and high frequency band influence markers, which facilitates the identification of impedance curves based on the impedance characteristic curve database. B4: For actual grid-connected inverters or black-box models of new energy, the small-signal response characteristics of new energy units in the 0-2000Hz frequency band are first obtained by the perturbation method, and the perturbation voltage, current and its response current and voltage data within the specified frequency range are obtained by frequency sweeping. B5: Then, the impedance characteristic data curve of the new energy unit is obtained by FFT calculation, and the low frequency band is defined as 0-10Hz, the mid frequency band as 10-100Hz, and the high frequency band as 100-2000Hz. B6: Based on the impedance characteristic data curves obtained from B5, the amplitude and phase errors of the impedance characteristic data curves for each frequency band and the reference curves in the impedance characteristic curve database at different frequency bands are calculated based on the error index. B7: Obtain the control strategy corresponding to the impedance curve with the smallest error among all indicators, and use it as a candidate control strategy.

6. The frequency domain identification method for a new energy grid-connected inverter according to claim 5, characterized in that: In B3, the impedance amplitude sensitivity to parameter changes or phase changes is defined. and phase sensitivity The specific indicators are as follows: （7）； （8）； In the above formula, The impedance amplitude of the new energy inverter; These are the inner and outer loop control parameters for the new energy inverter; The impedance amplitude of the new energy inverter; the impedance amplitude sensitivity and phase sensitivity indicators are arbitrary parameters. The physical meaning of the relative rate of change of impedance amplitude and phase is as follows: Indicates parameters An increase of 1% results in a 1% increase in impedance amplitude; Indicates parameters An increase of 1% increases the phase by 1 degree; and The parameters can be quantified and the key parameters that have a significant impact on impedance characteristics can be screened.

7. The frequency domain identification method for a new energy grid-connected inverter according to claim 6, characterized in that: In section B6, the amplitude and phase error indices of each impedance characteristic data curve and each reference curve in the impedance characteristic curve database are calculated at different frequency bands; specifically as follows: ①: The formula for calculating the mean absolute error is: （9）； In equation (7): K s , K c These represent the total number of simulation data within the calculation error interval and the total number of reference data in the database, respectively. , These represent the per-unit values ​​of the model simulation data and the per-unit values ​​of the reference data in the database, respectively; x = L , M , H These represent the low, mid, and high frequency bands, respectively. ②: The formula for calculating the maximum error is: （10）； ③: The formula for calculating the weighted average absolute error is: （11）； In equation (9): k 1. k 2. k 3 represents the average absolute error coefficients for the low, medium, and high frequency bands, respectively; σ 1,L , σ 1,M , σ 1,H These represent the average absolute error values ​​for the three frequency bands, respectively.

8. A time-frequency domain verification method for a new energy grid-connected inverter, characterized in that: Includes the following steps: C1: Combine impedance parameter sensitivity index to analyze the average error or maximum error of each frequency band; C2: Based on candidate control strategies, simulations are performed under typical disturbance or fault conditions by calling white-box models; C3: Analyze the difference between the identification results in step B7 and the actual response characteristics of the grid-connected inverter, and fine-tune the white-box model parameters; C4: For inverters that do not perform well in identifying typical control strategies in the impedance characteristic curve database, further fine-tune the initial control parameters or obtain new control strategy combinations by combining frequency domain impedance characteristics and time domain simulation analysis.

9. The time-frequency domain verification method for a new energy grid-connected inverter according to claim 8, characterized in that: In C1, the average error or maximum error of each frequency band is analyzed in conjunction with the impedance parameter sensitivity index, as detailed below: Based on the impedance parameter sensitivity index, analyze the parameters with higher sensitivity in the part with large error. If the maximum error is large, focus on eliminating the influence of measurement and adjusting the parameters to see if the maximum error can be eliminated or reduced. If the average error is large, adjust the parameters according to the influence law of impedance sensitivity and perform repeated frequency sweep observations to optimize the impedance curve.

10. The time-frequency domain verification method for a new energy grid-connected inverter according to claim 9, characterized in that: In C4, the initial control parameters are further fine-tuned or new control strategy combinations are obtained by combining frequency domain impedance characteristics and time domain simulation analysis. If the comparison results show a weighted average error for some frequency bands The following threshold conditions were not met: （12）； Based on the sensitivity index, the parameters with high sensitivity to the frequency band with large errors in the white-box model are found. By increasing or decreasing these parameters, frequency domain impedance characteristic analysis and time domain simulation analysis are performed to make the error meet the threshold range. The new combination of control parameters obtained after adjustment is recorded as a new control strategy in the impedance characteristic curve database.

Images