Scene deduction method for high-proportion new energy power system adaptive to network construction technology application

Abstract

The invention discloses a high-proportion new energy power system scene deduction method adaptive to network construction technology application, and relates to the technical field of new energy power system scene deduction. The problem of control strategy verification distortion caused by data interaction mismatch in electromagnetic transient and electromechanical transient hybrid simulation is effectively solved; a prediction and correction coordination promotion strategy is introduced, and bidirectional refinement processing is performed on multiple time scale variables through the prediction and correction coordination promotion strategy, so that the fidelity of joint simulation data is remarkably improved; controller parameters are adaptively adjusted through a parameter positive definite loop, a digital twin model can gradually approach dynamic response characteristics of a real physical system through a closed-loop iteration mechanism, and finally reliable verification of a control strategy of the network-forming type new energy unit in a broadband oscillation scene is achieved.

Description

Technical Field

[0001] This invention relates to the field of new energy power system scenario simulation technology, specifically to a method for simulating scenarios of high-proportion new energy power systems that adapt to grid-type technology applications. Background Technology

[0002] To adapt grid-based technology for high-proportion renewable energy power systems, a dynamic simulation model is established to model the impact of renewable energy output fluctuations and load changes on the power grid. The simulation technique utilizes a multi-timescale analysis framework to evaluate the system's stability and controllability under different operating scenarios, thereby providing decision support for grid dispatch. Logically, this simulation process integrates the stochastic characteristics of renewable energy with the adaptive adjustment capabilities of grid-based technology. By iteratively optimizing and predicting system behavior, it identifies weak links and adjusts operating strategies, ultimately enhancing the power system's adaptability to high-proportion renewable energy integration.

[0003] Existing simulation techniques for high-proportion renewable energy power systems adapted to grid-connected technologies suffer from the following technical challenges: electromagnetic transient simulation and electromechanical transient simulation address microsecond-level fast dynamics and second-level slow dynamics, respectively. In hybrid simulations, the data interaction interface uses an approximate mapping method, leading to the smoothing or loss of high-frequency oscillation information during scale conversion. For example, when verifying the control strategy of grid-connected wind turbine generators to handle broadband oscillations, the rapid phase fluctuations of the grid voltage are simplified on the electromechanical transient side, making the feedback signal received by the controller unable to accurately represent the actual oscillation mode. This results in stability assessment results deviating from the actual system response, affecting the grid connection reliability of renewable energy units. Summary of the Invention

[0004] Based on the problems raised in the background technology, the purpose of this invention is to provide a simulation method for high-proportion new energy power system scenarios that is adapted to the application of grid-type technology. This method solves the problem of control strategy verification distortion of grid-type new energy units in broadband oscillation scenarios caused by data interaction mismatch at different time scales in the mixed simulation of electromagnetic transients and electromechanical transients.

[0005] This invention is achieved through the following technical solution:

[0006] This invention provides a method for extrapolating scenarios of high-proportion renewable energy power systems adapted to grid-type technologies, comprising the following steps:

[0007] Step 1: Collect multidimensional data of the power system, preprocess the multidimensional data, and generate an initial simulation database;

[0008] Step 2: Construct digital twin models of the electromagnetic transient model and the electromechanical transient model using the initial simulation database, and set dynamic interaction interfaces at the boundaries of the electromagnetic transient model and the electromechanical transient model;

[0009] Step 3: Obtain the impedance sensitivity coefficient through the digital twin model, and input the impedance sensitivity coefficient into the dynamic interactive interface for signal reconstruction weight adjustment;

[0010] Step 4: The dynamic interactive interface after signal reconstruction and weight adjustment adopts a prediction and correction coordinated advancement strategy to drive the electromagnetic transient model and the electromechanical transient model to perform joint simulation calculations and obtain the system time domain response data.

[0011] Step 5: Evaluate the system's time-domain response data. If the evaluation result does not meet the preset standard, generate optimized controller parameters and feed the optimized controller parameters back to the digital twin model, repeating steps 3 to 5.

[0012] In the above technical solution, it is first necessary to collect multidimensional data from the power system. This multidimensional data includes the power system's physical parameters, real-time operating data, and historical broadband oscillation event waveform data. Due to the diverse data sources and varying acquisition frequencies, rigorous preprocessing is required to integrate and form a structured initial simulation database with a unified time reference.

[0013] Then, based on the initial simulation database, digital twin models of electromagnetic transient models and electromechanical transient models are constructed, and dynamic interaction interfaces are set at the boundaries of electromagnetic transient models and electromechanical transient models. The core function of this dynamic interaction interface is to realize bidirectional conversion and transmission between electromagnetic transient data and electromechanical transient data, especially in the subsynchronous to supersynchronous frequency band where phase continuity needs to be maintained.

[0014] After the digital twin model is put into use, broadband impedance scanning analysis is initiated to obtain the impedance sensitivity coefficient, which reflects the potential resonance risk areas in each frequency band. The impedance sensitivity coefficient is fed back to the dynamic interactive interface in real time, and the signal reconstruction weights of each frequency band in the dynamic interactive interface are dynamically adjusted to prioritize ensuring the waveform fidelity of the risk frequency bands.

[0015] The dynamic interactive interface after signal reconstruction and weight adjustment adopts a prediction and correction coordinated advancement strategy for joint simulation. Based on the exchange results of the dynamic interactive interface, the predicted value and the correction value are used for iterative calculation. The policy progress is adaptively adjusted through a variable step size mechanism until the complete system time domain response data is output.

[0016] Transient stability assessment is performed on the system's time-domain response data. If the transient stability assessment result does not meet the preset standard, that is, the system has weak damping or negative damping characteristics, the control parameter tuning process is triggered to generate optimized controller parameters. The optimized controller parameters are fed back to the digital twin model, and the closed-loop iterative process from impedance scanning to stability assessment is restarted until the system damping characteristics meet the preset stable operation requirements.

[0017] In one optional embodiment, the multidimensional data is preprocessed, including the following steps:

[0018] A sliding time window is used to perform timestamp alignment processing on the multidimensional data;

[0019] Interpolation algorithms are used to process the timestamp-aligned multidimensional data to generate time-series aligned data.

[0020] The controller parameters of the grid-type new energy units in the time-series aligned data are normalized, and a parameter sensitivity sorting matrix is ​​established.

[0021] The key parameter set is obtained by filtering the time-series aligned data based on the parameter sensitivity sorting matrix;

[0022] Modal identification is performed on the historical broadband oscillation event waveform data in the time-aligned data, and the dominant oscillation frequency component and damping ratio are extracted;

[0023] The key parameter set, the dominant oscillation frequency components, and the damping ratio are integrated into a structured dataset with a unified time base to generate an initial simulation database.

[0024] In one optional embodiment, constructing digital twin models of the electromagnetic transient model and the electromechanical transient model using the initial simulation database includes the following steps:

[0025] The power grid topology and equipment parameters are read from the initial simulation database, and the power grid topology and equipment parameters are processed using node voltage equations and switching functions to establish an electromagnetic transient model;

[0026] The parameters of the power grid components are read from the initial simulation database, and the parameters are processed using the synchronous generator swing equation and switching function to establish an electromechanical transient model.

[0027] In one optional embodiment, a dynamic interaction interface is provided at the boundary between the electromagnetic transient model and the electromechanical transient model, including:

[0028] The initial parameters of the dynamic interactive interface are set using the dominant oscillation frequency component and the damping ratio; wherein, the initial parameters include the window function length of the fast Fourier transform and the impedance and frequency lookup table of the harmonic injection equivalent circuit, and the window function length is set to an integer multiple of the oscillation period based on the dominant oscillation frequency component.

[0029] In one optional embodiment, obtaining the impedance sensitivity coefficient through the digital twin model includes the following steps:

[0030] A pseudo-random disturbance sequence is generated based on the current operating point of the digital twin model. The pseudo-random disturbance sequence is injected into the power grid through the power converter in the digital twin model, and the current disturbance after injection into the power grid is obtained.

[0031] The voltage response of the grid connection point to the pseudo-random disturbance sequence is collected, and the cross-correlation function between the voltage response and the current disturbance is calculated using the correlation analysis method. The cross-correlation function is then converted to the frequency domain to obtain the frequency-varying impedance characteristics.

[0032] Resonance feature points are extracted from the frequency-varying impedance characteristics. The phase margin of each resonant frequency point is calculated based on the resonant feature points. Frequency bands with phase margins lower than a preset threshold are marked as potential resonant risk frequency bands.

[0033] An impedance sensitivity coefficient is generated based on the phase margin of the potential resonance risk frequency band.

[0034] In one optional embodiment, obtaining the impedance sensitivity coefficient through the digital twin model includes the following steps:

[0035] A pseudo-random disturbance sequence is generated based on the current operating point of the digital twin model. The pseudo-random disturbance sequence is injected into the power grid through the power converter in the digital twin model, and the current disturbance after injection into the power grid is obtained.

[0036] The voltage response of the grid connection point to the pseudo-random disturbance sequence is collected, and the cross-correlation function between the voltage response and the current disturbance is calculated using the correlation analysis method. The cross-correlation function is then converted to the frequency domain to obtain the frequency-varying impedance characteristics.

[0037] Resonance feature points are extracted from the frequency-varying impedance characteristics. The phase margin of each resonant frequency point is calculated based on the resonant feature points. Frequency bands with phase margins lower than a preset threshold are marked as potential resonant risk frequency bands.

[0038] An impedance sensitivity coefficient is generated based on the phase margin of the potential resonance risk frequency band.

[0039] In one optional embodiment, within each simulation step, the electromagnetic transient model and the electromechanical transient model perform preliminary data exchange through a dynamic interactive interface, including:

[0040] The electromagnetic transient model provides the electromechanical transient model with a power increment after Clarke transformation and low-pass filtering;

[0041] The electromechanical transient model provides the electromagnetic transient model with a voltage reference signal reconstructed by phase compensation and cubic spline interpolation.

[0042] In one optional embodiment, a predicted value and a corrected value are calculated based on the data after data exchange. The predicted value and the corrected value are then used for iterative calculation to correct the system state. This includes the following steps:

[0043] Step a: Calculate the predicted values ​​of the boundary variables by linear extrapolation of the power increment and the voltage reference signal;

[0044] Step b: Input the predicted values ​​of the boundary variables into the dynamic interaction interface to obtain the system state after the interaction between the electromagnetic transient model and the electromechanical transient model;

[0045] Step c: Based on the deviation between the system state after the interaction and the system state after the interaction, calculate the state variable correction value, and use the state variable correction value to correct the system state;

[0046] Step d: Repeat steps a through d until the deviation value is less than the preset tolerance.

[0047] In one optional embodiment, evaluating the system's time-domain response data includes the following steps:

[0048] Construct the system state-space equations based on the system's time-domain response data;

[0049] Based on the system state-space equations, construct a Lyapunov function and calculate the time derivative of the Lyapunov function along the simulation trajectory.

[0050] Based on the time derivative, the dominant Lyapunov index spectrum is extracted, and the damping of the system is evaluated according to the Lyapunov index spectrum to generate the evaluation result.

[0051] In one optional embodiment, a predicted value and a corrected value are calculated based on the data after data exchange. The predicted value and the corrected value are then used for iterative calculation to correct the system state. This includes the following steps:

[0052] Step a: Calculate the predicted values ​​of the boundary variables by linear extrapolation of the power increment and the voltage reference signal;

[0053] Step b: Input the predicted values ​​of the boundary variables into the dynamic interaction interface to obtain the system state after the interaction between the electromagnetic transient model and the electromechanical transient model;

[0054] Step c: Based on the deviation between the system state after the interaction and the system state after the interaction, calculate the state variable correction value, and use the state variable correction value to correct the system state;

[0055] Step d: Repeat steps a through d until the deviation value is less than the preset tolerance.

[0056] In one optional embodiment, evaluating the system's time-domain response data includes the following steps:

[0057] Construct the system state-space equations based on the system's time-domain response data;

[0058] Based on the system state-space equations, construct a Lyapunov function and calculate the time derivative of the Lyapunov function along the simulation trajectory.

[0059] Based on the time derivative, the dominant Lyapunov index spectrum is extracted, and the damping of the system is evaluated according to the Lyapunov index spectrum to generate the evaluation result.

[0060] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0061] 1. By constructing a dynamic interactive interface with adaptive weight adjustment capabilities, the problem of control strategy verification distortion caused by data interaction mismatch in the hybrid simulation of electromagnetic transients and electromechanical transients is effectively solved;

[0062] 2. The dynamic interactive interface dynamically optimizes the signal reconstruction weights based on the wideband impedance scanning analysis results, enabling the interface to maintain the phase continuity of the resonant risk frequency band during data conversion, thus overcoming the smoothing effect of existing fixed weight mapping on high-frequency oscillation information.

[0063] 3. By employing a coordinated prediction and correction strategy to perform bidirectional refinement of variables across multiple time scales, the fidelity of the joint simulation data is significantly improved.

[0064] 4. When the system exhibits weak damping or negative damping characteristics, the controller parameters are adaptively adjusted through a positive definite parameter loop, and the digital twin model is gradually made to approximate the dynamic response characteristics of the real physical system through a closed-loop iterative mechanism, so as to achieve reliable verification of the control strategy of grid-connected new energy units in wide-frequency oscillation scenarios. Attached Figure Description

[0065] To more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention and should not be considered as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort. In the drawings:

[0066] Figure 1 This is a flowchart illustrating the scenario simulation method for high-proportion new energy power system applications using adaptive grid-type technology provided in Embodiment 1 of the present invention. Detailed Implementation

[0067] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the embodiments and accompanying drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and are not intended to limit the present invention.

[0068] Example 1

[0069] Embodiment 1 of the present invention provides a method for extrapolating scenarios of high-proportion renewable energy power systems adapted to grid-type technology applications, such as... Figure 1 As shown, the deduction method includes the following steps:

[0070] Step 1: Collect multidimensional data of the power system, preprocess the multidimensional data, and generate an initial simulation database;

[0071] Step 2: Construct digital twin models of the electromagnetic transient model and the electromechanical transient model using the initial simulation database, and set dynamic interaction interfaces at the boundaries of the electromagnetic transient model and the electromechanical transient model;

[0072] Step 3: Obtain the impedance sensitivity coefficient through the digital twin model, and input the impedance sensitivity coefficient into the dynamic interactive interface for signal reconstruction weight adjustment;

[0073] Step 4: The dynamic interactive interface after signal reconstruction and weight adjustment adopts a prediction and correction coordinated advancement strategy to drive the electromagnetic transient model and the electromechanical transient model to perform joint simulation calculations and obtain the system time domain response data.

[0074] Step 5: Evaluate the system's time-domain response data. If the evaluation result does not meet the preset standard, generate optimized controller parameters and feed the optimized controller parameters back to the digital twin model, repeating steps 3 to 5.

[0075] It should be noted that this method first requires the collection of multidimensional data from the power system. This multidimensional data includes the power system's physical parameters, real-time operating data, and historical broadband oscillation event waveform data. Due to the diverse data sources and varying acquisition frequencies, rigorous preprocessing is required to integrate and form a structured initial simulation database with a unified time reference.

[0076] Then, based on the initial simulation database, digital twin models of electromagnetic transient models and electromechanical transient models are constructed, and dynamic interaction interfaces are set at the boundaries of electromagnetic transient models and electromechanical transient models. The core function of this dynamic interaction interface is to realize bidirectional conversion and transmission between electromagnetic transient data and electromechanical transient data, especially in the subsynchronous to supersynchronous frequency band where phase continuity needs to be maintained.

[0077] After the digital twin model is put into use, broadband impedance scanning analysis is initiated to obtain the impedance sensitivity coefficient, which reflects the potential resonance risk areas in each frequency band. The impedance sensitivity coefficient is fed back to the dynamic interactive interface in real time, and the signal reconstruction weights of each frequency band in the dynamic interactive interface are dynamically adjusted to prioritize ensuring the waveform fidelity of the risk frequency bands.

[0078] The dynamic interactive interface after signal reconstruction and weight adjustment adopts a prediction and correction coordinated advancement strategy for joint simulation. Based on the exchange results of the dynamic interactive interface, the predicted value and the correction value are used for iterative calculation. The policy progress is adaptively adjusted through a variable step size mechanism until the complete system time domain response data is output.

[0079] Transient stability assessment is performed on the system's time-domain response data. If the transient stability assessment result does not meet the preset standard, that is, the system has weak damping or negative damping characteristics, the control parameter tuning process is triggered to generate optimized controller parameters. The optimized controller parameters are fed back to the digital twin model, and the closed-loop iterative process from impedance scanning to stability assessment is restarted until the system damping characteristics meet the preset stable operation requirements.

[0080] It is important to emphasize that the core innovation of this method is the setting of a dynamic interactive interface. First, by constructing a dynamic interactive interface with adaptive weight adjustment capabilities, the problem of control strategy verification distortion caused by data interaction mismatch in the mixed simulation of electromagnetic transients and electromechanical transients is effectively solved. Second, the dynamic interactive interface dynamically optimizes the signal reconstruction weights based on the wideband impedance scan analysis results, enabling the interface to maintain the phase continuity of the resonant risk frequency band during data conversion, overcoming the smoothing effect of existing fixed weight mapping on high-frequency oscillation information.

[0081] Meanwhile, this method introduces a prediction and correction coordination strategy, which performs bidirectional fine-tuning of variables at multiple time scales through the prediction and correction coordination strategy, significantly improving the fidelity of the joint simulation data.

[0082] Finally, when the system exhibits weak damping or negative damping characteristics, the controller parameters are adaptively adjusted through a positive definite parameter loop, and the digital twin model is gradually made to approximate the dynamic response characteristics of the real physical system through a closed-loop iterative mechanism, ultimately achieving reliable verification of the control strategy of grid-connected new energy units in a wide-frequency oscillation scenario.

[0083] In one optional embodiment, the multidimensional data is preprocessed, including the following steps:

[0084] A sliding time window is used to perform timestamp alignment processing on the multidimensional data;

[0085] Interpolation algorithms are used to process the timestamp-aligned multidimensional data to generate time-series aligned data.

[0086] The controller parameters of the grid-type new energy units in the time-series aligned data are normalized, and a parameter sensitivity sorting matrix is ​​established.

[0087] The key parameter set is obtained by filtering the time-series aligned data based on the parameter sensitivity sorting matrix;

[0088] Modal identification is performed on the historical broadband oscillation event waveform data in the time-aligned data, and the dominant oscillation frequency component and damping ratio are extracted;

[0089] The key parameter set, the dominant oscillation frequency components, and the damping ratio are integrated into a structured dataset with a unified time base to generate an initial simulation database.

[0090] In this optional embodiment, a sliding time window algorithm is used to align the timestamps of the multi-source data streams, and an interpolation algorithm is used to unify the sampling frequency of the timestamp-aligned multi-source data in order to eliminate timing deviations.

[0091] For the controller parameters of grid-connected new energy units, normalization processing is required, and a parameter sensitivity ranking matrix is ​​established based on the normalized controller parameters. Then, based on the established parameter sensitivity ranking matrix, data that have a significant impact on broadband oscillation characteristics are selected from the time-series aligned data as the key parameter set.

[0092] Simultaneously, modal identification is performed on the waveform data of historical broadband oscillation events in the time-series aligned data, and characteristic quantities such as the dominant oscillation frequency component and damping ratio are extracted.

[0093] The data processed as described above are integrated to generate a structured dataset with a unified time reference, which serves as the initial simulation dataset.

[0094] Furthermore, physical parameters include: grid topology, controller parameters of grid-connected wind turbines or photovoltaic inverters, and load dynamic response characteristics; operational data includes: real-time data collected by the SCADA system and synchronous phasor data recorded by the PMU device.

[0095] During the data preprocessing stage, the width of the sliding time window needs to be dynamically adjusted according to the actual data sampling interval:

[0096] For example, in the scenario of the above data, where the SCADA system records 1 record per second and the PMU device records 100 records per second, the window length is set to 2 and 3 times the dominant oscillation period, respectively. The interpolation algorithm preferentially adopts cubic spline interpolation, which is superior to linear interpolation in maintaining waveform smoothness. The parameter sensitivity ranking matrix is ​​realized by calculating the eigenvalues ​​of the Jacobian matrix, and key parameters such as PLL bandwidth and current loop proportional gain are arranged in descending order of influence. The mode identification adopts the random subspace method to extract 5 and 10 dominant modes from the historical oscillation waveforms and record the distribution of each mode frequency in the range of 2Hz and 500Hz.

[0097] In one optional embodiment, constructing digital twin models of the electromagnetic transient model and the electromechanical transient model using the initial simulation database includes the following steps:

[0098] The power grid topology and equipment parameters are read from the initial simulation database, and the power grid topology and equipment parameters are processed using node voltage equations and switching functions to establish an electromagnetic transient model;

[0099] The parameters of the power grid components are read from the initial simulation database, and the parameters are processed using the synchronous generator swing equation and switching function to establish an electromechanical transient model.

[0100] In this optional embodiment, for grid-type new energy power stations with dense power electronic equipment, an electromagnetic transient model including the power semiconductor switching transient process is established using node voltage equations and switching functions; for the AC grid part mainly composed of existing synchronous machines, an electromechanical transient model is established using synchronous generator swing equations and network algebra equations.

[0101] Finally, digital twin models of the electromagnetic transient model and the electromechanical transient model are constructed.

[0102] Furthermore, constructing an electromagnetic transient model requires considering the Nyquist frequency corresponding to the IGBT switching frequency (typically 2kHz and 10kHz), and the switching function uses double-precision floating-point numbers to describe the on / off state.

[0103] When constructing electromechanical transient models, dynamic equivalents can be used for power grid areas that are far from new energy power plants, while retaining the internal impedance characteristics of the equivalent power source.

[0104] In one optional embodiment, a dynamic interaction interface is provided at the boundary between the electromagnetic transient model and the electromechanical transient model, including:

[0105] The initial parameters of the dynamic interactive interface are set using the dominant oscillation frequency component and the damping ratio; wherein, the initial parameters include the window function length of the fast Fourier transform and the impedance and frequency lookup table of the harmonic injection equivalent circuit, and the window function length is set to an integer multiple of the oscillation period based on the dominant oscillation frequency component.

[0106] In this optional embodiment, during the initialization of the dynamic interactive interface, the window function length of the fast Fourier transform needs to be set according to the dominant oscillation frequency component extracted in the preprocessing stage, and an impedance and frequency lookup table of the harmonic injection equivalent circuit needs to be established; wherein, the window function constant is usually set to an integer multiple of the oscillation period.

[0107] Furthermore, in the initialization parameters of the dynamic interactive interface, the Hanning window is selected as the window function type to reduce spectral leakage, and the impedance values ​​corresponding to the 3rd, 5th, and 7th characteristic harmonics of the harmonic injection equivalent circuit need to be preset. The boundary node setting location is usually selected at the grid connection point of the new energy power plant or the main power grid hub substation.

[0108] In one optional embodiment, obtaining the impedance sensitivity coefficient through the digital twin model includes the following steps:

[0109] A pseudo-random disturbance sequence is generated based on the current operating point of the digital twin model. The pseudo-random disturbance sequence is injected into the power grid through the power converter in the digital twin model, and the current disturbance after injection into the power grid is obtained.

[0110] The voltage response of the grid connection point to the pseudo-random disturbance sequence is collected, and the cross-correlation function between the voltage response and the current disturbance is calculated using the correlation analysis method. The cross-correlation function is then converted to the frequency domain to obtain the frequency-varying impedance characteristics.

[0111] Resonance feature points are extracted from the frequency-varying impedance characteristics. The phase margin of each resonant frequency point is calculated based on the resonant feature points. Frequency bands with phase margins lower than a preset threshold are marked as potential resonant risk frequency bands.

[0112] An impedance sensitivity coefficient is generated based on the phase margin of the potential resonance risk frequency band.

[0113] In this optional embodiment, after the digital twin model is put into operation, a wideband impedance scan analysis is started synchronously. A pseudo-random disturbance sequence with gradually changing amplitude is injected through the power converter in the digital twin model. The scan frequency band needs to cover the subsynchronous to supersynchronous range. The pseudo-random disturbance sequence is injected into the power grid, and the injected current disturbance is obtained.

[0114] The cross-correlation function between the grid-connected point voltage response and current disturbance is calculated using correlation analysis and converted into frequency-varying impedance characteristics. A pole identification algorithm is used to extract resonant characteristic points, including resonant peaks and valleys, from the frequency-varying impedance characteristics. The phase margin at each resonant frequency is calculated using the Nyquist stability criterion. Frequency bands below the safety threshold are marked as potentially risky frequency bands, and impedance sensitivity coefficients are generated. The impedance sensitivity matrix, composed of these coefficients, is fed back to the dynamic interactive interface in real time to dynamically adjust the signal reconstruction weights for each frequency band, prioritizing waveform fidelity and phase continuity in the risky frequency bands.

[0115] Furthermore, the amplitude of the pseudo-random disturbance sequence is controlled between 1% and 5% of the rated current to avoid excessive impact on the system. The correlation analysis delay parameters are set segmentally according to the scanning frequency: 10 cycles for the low-frequency band (<100Hz) and 50 cycles for the high-frequency band. The phase margin threshold is set according to the IEC standard: greater than 30° for the subsynchronous frequency band (10Hz and 40Hz) and greater than 15° for the supersynchronous frequency band (100Hz and 500Hz). When generating the impedance sensitivity matrix, frequency bands with a margin more than 10° below the threshold are assigned weight coefficients of 0.8 and 1.0, respectively, while marginal frequency bands (below the threshold by 5°) are assigned weights of 0.3 and 0.5.

[0116] In one optional embodiment, the dynamic interactive interface after signal reconstruction and weight adjustment adopts a prediction and correction coordinated advancement strategy to drive the electromagnetic transient model and the electromechanical transient model to perform joint simulation calculations, including the following steps:

[0117] Set the simulation step size for the electromagnetic transient model and the electromechanical transient model;

[0118] Within each simulation step, the electromagnetic transient model and the electromechanical transient model exchange data through a dynamic interactive interface;

[0119] Based on the data after data exchange, the predicted value and the corrected value are calculated, and the predicted value and the corrected value are used to perform iterative calculation. The system state is then corrected based on the results of the iterative calculation.

[0120] Repeatedly perform data exchange and iterative calculations until the simulation ends, and output the system's time-domain response data.

[0121] In this optional embodiment, an electromagnetic transient model and an electromechanical transient model are set up, specifically including: the electromagnetic transient model is solved using implicit integration with a simulation step size of microseconds, and the electromechanical transient model is solved using explicit integration with a simulation step size of milliseconds.

[0122] Within each simulation step, the trend of boundary variables is first predicted through linear extrapolation. Then, the electromagnetic transient model and the electromechanical transient model perform preliminary data exchange to obtain the predicted and corrected values ​​of the system state. Based on the data exchanged, iterative calculations are performed using the predicted and corrected values, and the system state is corrected using the results of the iterative calculations. The data exchange and iterative calculations are repeated until the simulation time ends. The simulation progress is adaptively adjusted through a variable step size mechanism until complete system time-domain response data is output.

[0123] Furthermore, the implicit integration method employs the trapezoidal integration rule, while the explicit integration method uses the fourth-order Runge and Kutta method. The data exchange period is set to the least common multiple of the two simulation step sizes; for example, when the electromagnetic transient step size is 50 μs and the electromechanical transient step size is 1 ms, the exchange period is set to 1 ms.

[0124] In one optional embodiment, within each simulation step, the electromagnetic transient model and the electromechanical transient model perform preliminary data exchange through a dynamic interactive interface, including:

[0125] The electromagnetic transient model provides the electromechanical transient model with a power increment after Clarke transformation and low-pass filtering;

[0126] The electromechanical transient model provides the electromagnetic transient model with a voltage reference signal reconstructed by phase compensation and cubic spline interpolation.

[0127] The power increment includes both active and reactive power increments. The power increment after Clarke transform needs to be filtered by a first-order inertial filter with a time constant of 20ms. The phase compensation of the voltage reference signal is dynamically calculated based on the real-time impedance phase difference. The cubic spline interpolation node interval is set to an integer multiple of the electromagnetic transient step size.

[0128] In one optional embodiment, a predicted value and a corrected value are calculated based on the data after data exchange. The predicted value and the corrected value are then used for iterative calculation to correct the system state. This includes the following steps:

[0129] Step a: Calculate the predicted values ​​of the boundary variables by linear extrapolation of the power increment and the voltage reference signal;

[0130] Step b: Input the predicted values ​​of the boundary variables into the dynamic interaction interface to obtain the system state after the interaction between the electromagnetic transient model and the electromechanical transient model;

[0131] Step c: Based on the deviation between the system state after the interaction and the system state after the interaction, calculate the state variable correction value, and use the state variable correction value to correct the system state;

[0132] Step d: Repeat steps a through d until the deviation value is less than the preset tolerance.

[0133] In one optional embodiment, evaluating the system's time-domain response data includes the following steps:

[0134] Construct the system state-space equations based on the system's time-domain response data;

[0135] Based on the system state-space equations, construct a Lyapunov function and calculate the time derivative of the Lyapunov function along the simulation trajectory.

[0136] Based on the time derivative, the dominant Lyapunov index spectrum is extracted, and the damping of the system is evaluated according to the Lyapunov index spectrum to generate the evaluation result.

[0137] In this optional embodiment, when the Lyapunov exponent spectrum shows that the system has weak damping or negative damping characteristics, it means that the evaluation result does not meet the preset standard. At this time, the adaptive tuning loop of the control parameters is triggered to generate optimized controller parameters.

[0138] Furthermore, when constructing the state-space equations, electromechanical transient simulation results are used for slow variables such as generator power angle and speed, while electromagnetic transient simulation results are used for fast variables such as inverter capacitor voltage and inductor current. The Lyapunov function construction includes system kinetic energy (generator rotor kinetic energy) and potential energy (network electromagnetic potential energy), and the time derivative is calculated using the central difference method. The dominant Lyapunov exponents are selected from the top three eigenvalues ​​with the largest real parts, and the judgment threshold is set to 0.1 (corresponding to a damping ratio of approximately 1.6%).

[0139] In one alternative embodiment, generating optimized controller parameters includes the following steps:

[0140] With the goal of maximizing the system phase margin, a sequential quadratic programming algorithm is used to solve for the parameter combination that satisfies the controller gain.

[0141] The weight coefficients for different operating scenarios are determined using a fuzzy inference system. Based on the weight coefficients, the parameter combination is corrected to obtain the optimized controller parameters.

[0142] In this optional embodiment, with the goal of maximizing the phase margin, a sequential quadratic programming algorithm is used to solve for the optimal combination of controller parameters that satisfies the constraints. Then, a fuzzy inference system is used to refine the parameters based on weight coefficients for different operating scenarios. The optimized parameters are updated in real time to the controller instance in the digital twin model, restarting the closed-loop iterative process from impedance scanning to stability assessment until the system damping characteristics meet the requirements for stable operation.

[0143] Furthermore, the constraints of the sequential quadratic programming algorithm include: the phase-locked loop bandwidth is limited to the range of 5Hz to 15Hz, and the proportional gain of the current loop does not exceed 150% of its nominal value. The input variables of the fuzzy inference system are the oscillation frequency and severity, and the output is a weighted coefficient of stability and response speed. For example, stability is prioritized for low-frequency oscillations (<30Hz), while response speed is prioritized for high-frequency oscillations (>100Hz). The parameter correction amplitude is limited to no more than 20% of the original value in each iteration.

[0144] When optimizing controller parameters, a gradual update strategy is adopted. The update ratio is set to 50% in the first iteration, and the update ratio is dynamically adjusted (10% and 100%) in subsequent iterations based on the degree of stability improvement. In addition to meeting stability requirements, the iteration termination condition includes a maximum iteration limit (usually 20 or 30 times) to prevent non-convergence. Boundary checks are performed on the controller parameters before each iteration to prevent them from exceeding the allowable range of the physical device.

[0145] Among them, the real-time data collected by the SCADA system needs to be filtered out for bad data, and outliers other than 3σ are removed using the Raida criterion; the synchronization phasor data recorded by the PMU device is checked for synchronization accuracy between phasor measurement units, and data with time scale error greater than 1 sampling period (10ms) needs to be re-aligned.

[0146] The parameters of the grid-type wind turbine controller should highlight the per-unit conversion relationship of the virtual inertia time constant (usually 2s and 6s) and damping coefficient (usually 0.5 and 2.0pu).

[0147] Boundary variable predictions are calculated using a second-order extrapolation formula. The prediction value at the current step size t is calculated based on historical data at times t and Δt, and t and 2Δt. The system state after interaction is obtained by solving the interface algebraic equations, with residual tolerances set to 1e and 4. State variable correction values ​​are calculated using the Newton-Raphson method, with the number of iterations limited to 10. Deviation tolerances are set according to simulation accuracy requirements, typically with voltage deviation less than 0.01 pu and phase deviation less than 0.1°.

[0148] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for scenario deduction of a high-proportion new energy power system adapted to a meshed network type technology, characterized in that, The method comprises the following steps: Step 1, collecting multi-dimensional data of a power system, preprocessing the multi-dimensional data, and generating an initial simulation database; Step 2, constructing a digital twin model of an electromagnetic transient model and an electromechanical transient model by using the initial simulation database, and setting a dynamic interaction interface at the boundary of the electromagnetic transient model and the electromechanical transient model; Step 3, obtaining an impedance sensitivity coefficient through the digital twin model, and inputting the impedance sensitivity coefficient into the dynamic interaction interface for signal reconstruction weight adjustment; Step 4, adopting a prediction and correction coordinated promotion strategy by the dynamic interaction interface after the signal reconstruction weight adjustment, driving the electromagnetic transient model and the electromechanical transient model to perform joint simulation calculation, and obtaining system time domain response data; Step 5, evaluating the system time domain response data, if the evaluation result does not satisfy a preset standard, generating an optimized controller parameter, and feeding back the optimized controller parameter to the digital twin model to repeat steps 3 to 5.

2. The adaptive network-configuration technology application high-proportion new energy power system scenario deduction method according to claim 1, characterized in that, The preprocessing of the multi-dimensional data comprises the following steps: performing timestamp alignment processing on the multi-dimensional data by using a sliding time window; processing the multi-dimensional data after the timestamp alignment processing by using an interpolation algorithm, and generating time sequence alignment data; performing normalization processing on a controller parameter of a networked new energy unit in the time sequence alignment data, and establishing a parameter sensitivity sorting matrix; screening a key parameter set from the time sequence alignment data according to the parameter sensitivity sorting matrix; performing modal identification on historical wide-frequency oscillation event waveform data in the time sequence alignment data, and extracting a dominant oscillation frequency component and a damping ratio; integrating the key parameter set, the dominant oscillation frequency component and the damping ratio into a structured data set with a unified time reference, and generating an initial simulation database.

3. The method according to claim 2, wherein, The construction of the digital twin model of the electromagnetic transient model and the electromechanical transient model by using the initial simulation database comprises the following steps: reading power grid topology structure and device parameters from the initial simulation database, processing the power grid topology structure and the device parameters by using a node voltage equation and a switch function, and establishing an electromagnetic transient model; reading power grid element parameters from the initial simulation database, processing the power grid element parameters by using a synchronous generator swing equation and a switch function, and establishing an electromechanical transient model.

4. The adaptive network-configuration technology application high-proportion new energy power system scenario deduction method according to claim 3, characterized in that, Setting a dynamic interaction interface at the boundary of the electromagnetic transient model and the electromechanical transient model comprises: setting initial parameters of the dynamic interaction interface by using the dominant oscillation frequency component and the damping ratio; wherein the initial parameters comprise a window function length of a fast Fourier transform and an impedance and frequency query table of a harmonic injection equivalent circuit, and the window function length is set as an integer multiple of an oscillation period according to the dominant oscillation frequency component.

5. The adaptive network-configuration technology application high-proportion new energy power system scenario deduction method according to claim 1, characterized in that, Obtaining an impedance sensitivity coefficient through the digital twin model comprises the following steps: generating a pseudo-random disturbance sequence according to a current operating point of the digital twin model, injecting the pseudo-random disturbance sequence into a power grid through a power converter in the digital twin model, and obtaining current disturbance after the injection into the power grid; Collecting voltage response of the grid point to the pseudo-random disturbance sequence, calculating cross-correlation function of the voltage response and the current disturbance by using correlation analysis method, converting the cross-correlation function to frequency domain to obtain frequency-varying impedance characteristics; Extracting resonance characteristic points from the frequency-varying impedance characteristics, calculating phase margin of each resonance frequency point according to the resonance characteristic points, and marking the frequency band with a phase margin lower than a preset threshold as a potential resonance risk frequency band; Generating an impedance sensitivity coefficient according to the phase margin of the potential resonance risk frequency band.

6. The adaptive network-configuration technology application high-proportion new energy power system scenario deduction method according to claim 1, characterized in that, The dynamic interaction interface adjusted by the signal reconstruction weight adopts a prediction and correction coordinated promotion strategy to drive the electromagnetic transient model and the electromechanical transient model to perform joint simulation calculation, including the following steps: Setting the simulation step length of the electromagnetic transient model and the electromechanical transient model; In each simulation step, the electromagnetic transient model and the electromechanical transient model perform data exchange through the dynamic interaction interface; Based on the data after data exchange, calculate the prediction value and the correction value, use the prediction value and the correction value for iterative calculation, and modify the system state through the result of iterative calculation; Repeat the data exchange and iterative calculation until the simulation time ends, and output the system time domain response data.

7. The adaptive network-configuration technology application high-proportion new energy power system scenario deduction method according to claim 6, characterized in that, In each simulation step, the electromagnetic transient model and the electromechanical transient model perform preliminary data exchange through the dynamic interaction interface, including: The electromagnetic transient model provides the power increment after the Clarke transformation and low-pass filtering to the electromechanical transient model; The electromechanical transient model provides the voltage reference signal after phase compensation and cubic spline interpolation reconstruction to the electromagnetic transient model.

8. The adaptive network-configuration technology application high-proportion new energy power system scenario deduction method according to claim 7, characterized in that, Based on the data after data exchange, calculate the prediction value and the correction value, use the prediction value and the correction value for iterative calculation, and modify the system state through the result of iterative calculation, including the following steps: Step a, calculate the boundary variable prediction value by linear extrapolation on the power increment and the voltage reference signal; Step b, input the boundary variable prediction value into the dynamic interaction interface to obtain the system state after interaction of the electromagnetic transient model and the electromechanical transient model; Step c, calculate the state variable correction value based on the deviation value between the system state after interaction and the system state after interaction, and modify the system state by using the state variable correction value; Step d, repeat steps a to d until the deviation value is less than a preset tolerance.

9. The adaptive network-constructing technology application high-proportion new energy power system scenario deduction method according to claim 1, characterized in that, Evaluating the system time domain response data, including the following steps: Construct a system state space equation according to the system time domain response data; Based on the system state space equation, construct a Lyapunov function and calculate the time derivative of the Lyapunov function along the simulation trajectory; Based on the time derivative, extract the dominant Lyapunov exponent spectrum, and perform damping evaluation on the system according to the Lyapunov exponent spectrum to generate an evaluation result.

10. The adaptive network-configuration technology application high-proportion new energy power system scenario deduction method according to claim 9, characterized in that, Evaluating the system time domain response data, including the following steps: Construct a system state space equation according to the system time domain response data; Based on the system state space equation, construct a Lyapunov function and calculate the time derivative of the Lyapunov function along the simulation trajectory; Based on the time derivative, a dominant Lyapunov exponent spectrum is extracted, and a damping evaluation is made on the system according to the Lyapunov exponent spectrum, and an evaluation result is generated.

Images