Transformer simulation test current optimization method based on dynamic impedance matching

Abstract

The invention relates to the technical field of power system testing, in particular to a transformer simulation test current optimization method based on dynamic impedance matching, which comprises the following steps: performing low-power scanning on a tested transformer to establish a nonlinear model capable of predicting harmonic waves; actively synthesizing the virtual impedance of the target power grid; and a feed-forward compensation current is generated by using a harmonic prediction model and is injected into the transformer after being superposed with a fundamental wave instruction. According to the method, accurate identification modeling is carried out on nonlinear characteristics of a tested unit, active synthesis of virtual impedance of a target power grid is combined, and harmonic compensation current is generated in real time by utilizing a feedforward control thought so as to optimize finally injected test current. According to the method, the problems of inaccurate power grid simulation and insufficient nonlinear response coupling of the transformer in a traditional test are solved, high-fidelity current injection is realized, real electrical behaviors of the transformer under various complex power grid working conditions can be accurately simulated, and the flexibility, accuracy and safety of the test are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of power system testing technology, specifically to a method for optimizing transformer simulation test current based on dynamic impedance matching. Background Technology

[0002] As a core component of the power grid, the reliability and stability of power transformers directly affect the safety of the entire power system. Therefore, conducting comprehensive and precise performance tests on transformers before they are put into operation, especially verifying their interaction behavior with power grids of different characteristics, is a crucial step.

[0003] Currently, there are two main approaches to testing the grid-connected performance of transformers. The first is to directly connect them to a real power grid or a high-power laboratory for testing. While direct, this method is extremely costly, has a long testing cycle, and carries significant safety risks. More importantly, it cannot flexibly alter grid characteristics and struggles to simulate the various complex operating conditions a transformer might encounter, such as weak grids or harmonic grid environments containing numerous power electronic devices. The second method is simulation analysis based on digital models. Traditional simulations often rely on simplified linear transformer models, neglecting key nonlinear characteristics such as core hysteresis and saturation effects. These characteristics are precisely the root causes of harmonic currents and grid resonance, leading to significant deviations between simulation results and actual conditions. Some improved testing methods attempt to use power hardware-in-the-loop (PHIL) technology to simulate the power grid through power converters. However, existing technology has limited ability to simulate the broadband impedance characteristics of the power grid and struggles to accurately predict and compensate for severe current distortions caused by the transformer's own nonlinearity. Ultimately, the injected test current waveform deviates significantly from the ideal target, affecting the accuracy and effectiveness of the test.

[0004] To address this, a method for optimizing transformer simulation test current based on dynamic impedance matching is proposed. Summary of the Invention

[0005] The purpose of this invention is to provide a transformer simulation test current optimization method based on dynamic impedance matching. By accurately identifying the nonlinear harmonic characteristics of the transformer under test and actively synthesizing an arbitrarily defined target power grid virtual impedance, and then combining feedforward compensation technology to generate and inject an optimized command current, this invention solves the technical problem in the prior art of insufficient coupling between the simulated power grid environment and the nonlinear response of the transformer, which leads to test current distortion and inaccurate test results.

[0006] To achieve the above objectives, the present invention provides the following technical solution: A method for optimizing transformer simulation test current based on dynamic impedance matching includes: An automated low-power scan is performed on the transformer under test to generate a characteristic fingerprint map of the unit under test that characterizes the broadband electrical properties. Based on the feature fingerprint of the tested unit, the parameters of the nonlinear electrical model are calibrated, and a dynamic nonlinear harmonic prediction map for forward-looking harmonic compensation is constructed. The target virtual impedance of the target power grid is defined based on a broadband impedance shaping matrix that records multiple target frequency points and the corresponding impedance amplitude and phase parameters of each point. The active impedance synthesizer is controlled to synthesize the target virtual impedance at the port of the transformer under test. The dynamic nonlinear harmonic prediction diagram is used in real time to generate a feedforward compensation current command, which is superimposed with the fundamental command to generate a final command current. The wide bandwidth power converter is controlled to inject the final command current into the transformer under test through the synthesized target virtual impedance.

[0007] Preferably, the step of generating the feature fingerprint map of the tested unit specifically includes: A broadband sweep frequency excitation signal covering multiple frequency points is injected into the transformer under test; the broadband sweep frequency excitation signal is a pseudo-random binary sequence signal; while injecting the broadband sweep frequency excitation signal, the voltage response signal and current signal at the port of the transformer under test are simultaneously acquired; the acquired voltage response signal and current signal are processed by Fourier transform to calculate the complex impedance value corresponding to each of the multiple frequency points, the complex impedance value including impedance amplitude and phase; the complex impedance values ​​corresponding to all frequency points are summarized to generate the characteristic fingerprint map of the unit under test.

[0008] Preferably, the nonlinear electrical model specifically includes: The nonlinear magnetization unit, based on a nonlinear function that can characterize the hysteresis loop and saturation effect, maps the magnetic flux linkage obtained by integrating the input voltage into a nonsinusoidal excitation current. A linear circuit unit, connected in parallel with the nonlinear magnetization unit, is used to simulate the DC resistance and main leakage inductance of the transformer winding; Frequency-dependent loss unit is used to simulate additional losses that vary with frequency caused by core eddy current effects and winding skin effects.

[0009] Preferably, the step of constructing the dynamic nonlinear harmonic prediction map specifically includes: The fundamental current reference signals with different amplitudes and phases are used as input excitations and loaded into the nonlinear electrical model with calibrated parameters. For each input fundamental current reference signal, the nonlinear electrical model is run to simulate and calculate the total output current waveform containing fundamental and harmonic components caused by the nonlinear characteristics of the model. Spectral analysis is performed on each simulated total output current waveform to separate the amplitude and phase information of each harmonic component. A mapping lookup table is established and stored, with the input fundamental current reference signal as the index and the amplitude and phase of the corresponding harmonic components as the content, thus forming the dynamic nonlinear harmonic prediction graph.

[0010] Preferably, the broadband impedance shaping matrix is ​​a database that pre-stores multiple power grid model parameter sets, wherein each power grid model parameter set contains a set of frequency points and corresponding impedance vectors for defining different power grid characteristics; the step of defining the target virtual impedance is to select a corresponding power grid model parameter set from the database according to the test requirements, and load the frequency and impedance vector data therein into the control loop of the active impedance synthesizer as the real-time control target for synthesizing the target virtual impedance.

[0011] Preferably, the step of generating a feedforward compensation current command in real time using the dynamic nonlinear harmonic prediction diagram and injecting it into the transformer under test specifically includes: In each control cycle, the amplitude and phase of each harmonic component are retrieved in real time from the dynamic nonlinear harmonic prediction diagram using the current fundamental command as the query index. Based on the amplitude and phase of each harmonic component found, the harmonic current waveform is synthesized in real time and its phase is reversed by 180 degrees to generate the feedforward compensation current command. The feedforward compensation current command and the fundamental frequency command are superimposed in the time domain to generate the final command current; The final command current is used as a real-time reference value and input to the closed-loop current controller of the wide-bandwidth power converter to drive the wide-bandwidth power converter to output through the synthesized target virtual impedance.

[0012] Preferably, it also includes an online stability monitoring and adaptive adjustment step, which includes: While injecting the final command current into the transformer under test, the actual voltage and current at its port are monitored in real time; based on the actual voltage and current, the actual input impedance of the transformer under test is calculated in real time; the synthesized target virtual impedance is compared with the actual input impedance to obtain the impedance ratio characterizing the stability margin of the system; when the impedance ratio approaches the preset instability boundary, the control loop parameters of the active impedance synthesizer are automatically adjusted.

[0013] Preferably, the closed-loop current controller of the wide bandwidth power converter is a proportional resonant controller; the proportional resonant controller includes a fundamental frequency resonant element and n harmonic frequency resonant elements, wherein the harmonic frequencies are preset integer multiples of the fundamental frequency, and are used to perform high-gain, zero-steady-state-error feedback tracking control on the fundamental and residual harmonic components in the final command current.

[0014] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention first obtains a unique characteristic fingerprint of the transformer under test through automated low-power broadband scanning, ensuring safety and efficiency. Based on this measured data, the parameters of the nonlinear electrical model are calibrated, enabling the construction of a highly customized dynamic nonlinear harmonic prediction map. This method abandons the reliance on high-risk, high-power tests or the use of general, inaccurate simplified models in traditional testing. It profoundly characterizes the nonlinear dynamics of specific equipment, such as hysteresis and saturation, over a wide frequency range, fundamentally solving the simulation inaccuracy problem caused by model mismatch.

[0015] 2. This invention utilizes a broadband impedance shaping matrix in conjunction with an active impedance synthesizer to actively synthesize arbitrarily complex target virtual grid impedances at the ports of the transformer under test. This overcomes the limitations of traditional testing methods where the grid environment is fixed and difficult to reproduce. Whether it is a robust grid, a weak grid, or a complex grid with high harmonics and low inertia containing a large number of new energy sources and power electronic equipment, this method can flexibly and accurately reproduce the scenario through software definition. This capability allows the transformer to undergo the most stringent and complex real grid conditions it may encounter throughout its entire life cycle in a laboratory environment, greatly expanding the breadth and depth of testing and significantly improving the engineering practical value and reference value of the test results.

[0016] 3. This invention utilizes a pre-constructed harmonic prediction diagram to generate a feedforward compensation current command in real time. This command can proactively and accurately cancel the harmonic components that will be generated by the transformer's own nonlinear characteristics. By superimposing this compensation command with the fundamental frequency command, the final current waveform applied to the synthesized virtual impedance is ensured to be highly pure, precise, and controllable, perfectly reproducing the target test conditions. This "prediction + compensation" feedforward control strategy, compared to traditional feedback control, has a faster response and more stable control, fundamentally solving the problem of test current distortion under nonlinear loads and complex impedance environments, ensuring high fidelity in the test process and reliability of the results. Attached Figure Description

[0017] Figure 1 This is a flowchart of the transformer simulation test current optimization method based on dynamic impedance matching proposed in this invention. Figure 2This is a schematic diagram illustrating the steps of the transformer simulation test current optimization method based on dynamic impedance matching proposed in this invention. Figure 3 This is a structural diagram of the dynamic nonlinear harmonic prediction diagram proposed in this invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Example 1 Please see Figures 1 to 3 This invention provides a method for optimizing transformer simulation test current based on dynamic impedance matching, the technical solution of which is as follows: A method for optimizing transformer simulation test current based on dynamic impedance matching, such as Figures 1-2 As shown, it includes: An automated low-power scan is performed on the transformer under test to generate a characteristic fingerprint map of the unit under test that characterizes the broadband electrical properties. Based on the feature fingerprint of the tested unit, the parameters of the nonlinear electrical model are calibrated, and a dynamic nonlinear harmonic prediction map for forward-looking harmonic compensation is constructed. The target virtual impedance of the target power grid is defined based on a broadband impedance shaping matrix that records multiple target frequency points and the corresponding impedance amplitude and phase parameters of each point. The active impedance synthesizer is controlled to synthesize the target virtual impedance at the port of the transformer under test. The dynamic nonlinear harmonic prediction diagram is used in real time to generate a feedforward compensation current command, which is superimposed with the fundamental command to generate a final command current. The wide bandwidth power converter is controlled to inject the final command current into the transformer under test through the synthesized target virtual impedance.

[0020] Furthermore, the step of generating the feature fingerprint map of the tested unit specifically includes: A broadband sweep frequency excitation signal covering multiple frequency points is injected into the transformer under test; the broadband sweep frequency excitation signal is a pseudo-random binary sequence signal; while injecting the broadband sweep frequency excitation signal, the voltage response signal and current signal at the port of the transformer under test are simultaneously acquired; the acquired voltage response signal and current signal are processed by Fourier transform to calculate the complex impedance value corresponding to each of the multiple frequency points, the complex impedance value including impedance amplitude and phase; the complex impedance values ​​corresponding to all frequency points are summarized to generate the characteristic fingerprint map of the unit under test.

[0021] The amplitude of the broadband swept-frequency excitation small signal is set between 0.1% and 1% of the rated voltage of the transformer under test. This range is designed to ensure that the transformer operates in the linear region to obtain accurate baseline impedance characteristics, while ensuring that the response signal has a sufficient signal-to-noise ratio for accurate measurement. The order and clock frequency of the pseudo-random binary sequence signal are set according to the target analysis frequency band (e.g., 50Hz to 5kHz) and the required frequency resolution to ensure that its power spectrum can uniformly and fully cover the entire target frequency band.

[0022] The synchronous acquisition is accomplished by a high-speed data acquisition device with a vertical resolution of at least 16 bits and a sampling rate of no less than 100 kS / s. The synchronization error between channels is less than 10 nanoseconds to ensure the measurement accuracy of voltage and current signals in amplitude and phase.

[0023] Before performing Fourier transform processing, the acquired time-domain voltage and current signal sequences are preprocessed using a Hanning window to effectively suppress spectral leakage caused by signal truncation, thereby improving the accuracy of subsequent complex impedance calculations. The final generated characteristic fingerprint of the unit under test is specifically a data lookup table indexed by frequency, storing the impedance amplitude and phase values ​​at corresponding frequency points, for direct retrieval in subsequent steps.

[0024] This invention ensures extremely high accuracy, reliability, and repeatability in the generation of the characteristic fingerprint map of the unit under test by clearly defining core implementation details such as the amplitude of the frequency sweep excitation signal, key indicators of the data acquisition system, and signal processing algorithms. This method not only guarantees effective identification of the transformer in the linear region through precise control of the excitation amplitude, but also minimizes measurement errors such as noise and spectral leakage by utilizing high-specification synchronous acquisition and windowed Fourier transform techniques.

[0025] Furthermore, the nonlinear electrical model specifically includes: The nonlinear magnetization unit, based on a nonlinear function that can characterize the hysteresis loop and saturation effect, maps the magnetic flux linkage obtained by integrating the input voltage into a nonsinusoidal excitation current. A linear circuit unit, connected in parallel with the nonlinear magnetization unit, is used to simulate the DC resistance and main leakage inductance of the transformer winding; Frequency-dependent loss unit is used to simulate additional losses that vary with frequency caused by core eddy current effects and winding skin effects.

[0026] The nonlinear function preferably adopts the Jiles-Atherton hysteresis model, which can accurately describe the hysteresis loop and saturation characteristics of magnetic materials through a set of parameters with clear physical meaning, thereby establishing an accurate nonlinear mapping relationship between magnetic flux linkage and excitation current.

[0027] The DC resistance and main leakage inductance in the linear circuit unit are specifically equivalent parameters viewed from the primary side of the transformer.

[0028] The frequency-dependent loss unit is specifically implemented by setting up m parallel resistor-inductor branches. Through optimization algorithms such as particle swarm optimization and genetic algorithms, the resistance and inductance parameters in each branch are identified and configured, allowing the total equivalent parallel resistance to accurately approximate the actual loss characteristics determined by eddy current and skin effects over a wide frequency range. This enables accurate simulation of frequency-dependent losses in time-domain simulations. In this embodiment, m can be 3-5.

[0029] This invention constructs a high-fidelity electrical model with a complete structure and clear mechanism. It not only accurately reproduces core nonlinear effects such as hysteresis and saturation, but also endows the model with clear physical meaning. Simultaneously, it finely simulates frequency-related losses through a parallel resistor-inductor branch network, overcoming the inaccuracy of traditional models over a wide frequency range. This high-precision, wide-bandwidth model provides a crucial foundation for subsequent accurate dynamic harmonic prediction and high-precision current compensation control.

[0030] Furthermore, such as Figure 3 As shown, the steps for constructing the dynamic nonlinear harmonic prediction map specifically include: The fundamental current reference signals with different amplitudes and phases are used as input excitations and loaded into the nonlinear electrical model with calibrated parameters. For each input fundamental current reference signal, the nonlinear electrical model is run to simulate and calculate the total output current waveform containing fundamental and harmonic components caused by the nonlinear characteristics of the model. Spectral analysis is performed on each simulated total output current waveform to separate the amplitude and phase information of each harmonic component. A mapping lookup table is established and stored, with the input fundamental current reference signal as the index and the amplitude and phase of the corresponding harmonic components as the content, thus forming the dynamic nonlinear harmonic prediction graph.

[0031] The fundamental current reference signals of different amplitudes cover a scanning range from zero to 120% of the rated current of the transformer under test. The scanning step size can be non-uniformly spaced. In regions where the current nonlinearity changes drastically (such as when it is close to the rated value), a smaller step size (such as 1% of the rated value) is used, and in regions where the nonlinearity is weak, a larger step size (such as 5% of the rated value) is used to achieve a balance between efficiency and accuracy.

[0032] The preferred harmonic components include the 3rd, 5th, 7th, 9th, 11th and 13th harmonic components, which have the most significant impact on the power quality of the power grid.

[0033] In practical applications, when the fundamental current command value queried in real time falls between the discrete index points of the lookup table, a multidimensional linear interpolation algorithm (e.g., bilinear interpolation for both amplitude and phase dimensions) is used to calculate the amplitude and phase of each harmonic component corresponding to the current command in real time, thereby ensuring the smoothness and high accuracy of harmonic compensation.

[0034] Each typical power grid model in the broadband impedance shaping matrix corresponds to an independently constructed and stored dynamic nonlinear harmonic prediction map. During testing, when a power grid model parameter set is selected from the database according to experimental requirements, the system automatically calls the dynamic harmonic prediction map specifically associated with the power grid model while synthesizing the target virtual impedance. In this way, the accuracy of harmonic prediction is further improved, making feedforward compensation more precise.

[0035] In the mapping stage, this invention employs a non-uniform step-size scanning strategy, which greatly reduces the amount of simulation computation and data storage requirements while ensuring the accuracy of key areas, thus improving modeling efficiency. In the application stage, by introducing a multidimensional linear interpolation algorithm, the step problem of discrete lookup tables is effectively solved, enabling smooth and accurate harmonic prediction for any real-time command. This ensures the real-time performance and high fidelity of the final feedforward compensation, making the solution more practical for engineering applications.

[0036] Furthermore, the broadband impedance shaping matrix is ​​a database that pre-stores multiple power grid model parameter sets, where each power grid model parameter set contains a set of frequency points and corresponding impedance vectors for defining different power grid characteristics; the step of defining the target virtual impedance is to select a corresponding power grid model parameter set from the database according to the experimental requirements, and load the frequency and impedance vector data therein into the control loop of the active impedance synthesizer as the real-time control target for synthesizing the target virtual impedance.

[0037] The active impedance synthesizer consists of a wide-bandwidth power converter and its digital controller. Its control loop achieves the synthesis of the target virtual impedance through a dynamic voltage adjustment method based on output current measurement.

[0038] Specifically, the controller measures the actual current flowing out of the power converter in real time. This real-time current value is fed into a programmable digital filter. The characteristics of this digital filter are set online using target frequency and impedance vector data loaded from a database, allowing it to mimic the impedance behavior of the target power grid. After filtering, a dedicated voltage adjustment signal is generated. The controller subtracts this voltage adjustment signal from the original voltage control target and uses this final result to control the power converter to produce the actual output voltage.

[0039] Whenever the output current changes, the output voltage will also be adjusted in the opposite direction with precision, so that from the outside, the electrical characteristics of the power converter's ports are just like a real power grid with a target impedance.

[0040] This method parameterizes complex power grid models and stores them in a database. During testing, the control objectives of the active impedance synthesizer can be reconstructed online simply by calling the software. This completely eliminates the constraints of traditional testing that relies on bulky, fixed physical components to simulate the power grid, giving the test environment great flexibility and reconfigurability. Therefore, it can conveniently, safely, and cost-effectively simulate various complex operating conditions, from strong to weak power grids, significantly improving the breadth, depth, and real-world relevance of transformer grid-connected performance testing.

[0041] Furthermore, the step of generating a feedforward compensation current command in real time using the dynamic nonlinear harmonic prediction diagram and injecting it into the transformer under test specifically includes: In each control cycle, the amplitude and phase of each harmonic component are retrieved in real time from the dynamic nonlinear harmonic prediction diagram using the current fundamental command as the query index. Based on the amplitude and phase of each harmonic component found, the harmonic current waveform is synthesized in real time and its phase is reversed by 180 degrees to generate the feedforward compensation current command. The feedforward compensation current command and the fundamental frequency command are superimposed in the time domain to generate the final command current; The final command current is used as a real-time reference value and input to the closed-loop current controller of the wide-bandwidth power converter to drive the wide-bandwidth power converter to output through the synthesized target virtual impedance.

[0042] The closed-loop current controller of the wide-bandwidth power converter specifically employs a proportional resonant controller. This controller includes a proportional element and multiple resonant elements connected in parallel. The center frequency of one resonant element is set to the fundamental frequency, while the center frequencies of the remaining resonant elements are set to the frequencies of the harmonics requiring compensation (e.g., the 3rd, 5th, and 7th harmonics). This structure allows the controller to have extremely high theoretical gain at these specific command frequencies, ensuring that the actual current waveform output by the power converter accurately reproduces every frequency component in the final command current, thus guaranteeing the final effectiveness of the feedforward compensation strategy.

[0043] The parameters of the closed-loop current controller are dynamically matched to the target virtual impedance. In the database of the broadband impedance shaping matrix, each power grid model parameter set not only contains the frequency points and impedance vectors defining the power grid characteristics, but also pre-stores a set of optimal proportional resonant controller parameters that match the power grid model. When a selected power grid model parameter set is loaded, the system not only loads the impedance vector data to the active impedance synthesizer, but also synchronously loads this set of optimal controller parameters to the closed-loop current controller. This ensures that when simulating any power grid characteristic, the closed-loop control system always maintains optimal stability margin and dynamic tracking performance.

[0044] This invention achieves ultimate optimization of the test current by combining forward-looking feedforward compensation with high-precision feedback control. This method not only generates feedforward compensation commands in real time through harmonic prediction maps to pre-cancele the nonlinear distortion of the transformer, but more importantly, it employs a dedicated proportional resonant (PR) controller as the core of the feedback execution. The PR controller can achieve zero steady-state error tracking at the fundamental and harmonic frequencies, ensuring that the complex command current can be reproduced with high fidelity by the power converter. This dual guarantee of "feedforward prediction" and "precise feedback execution" fundamentally ensures the waveform quality of the injected current and the final accuracy of the entire simulation test.

[0045] Furthermore, it also includes an online stability monitoring and adaptive adjustment step, which includes: While injecting the final command current into the transformer under test, the actual voltage and current at its port are monitored in real time; based on the actual voltage and current, the actual input impedance of the transformer under test is calculated in real time; the synthesized target virtual impedance is compared with the actual input impedance to obtain the impedance ratio characterizing the stability margin of the system; when the impedance ratio approaches the preset instability boundary, the control loop parameters of the active impedance synthesizer are automatically adjusted.

[0046] The step of calculating the actual input impedance in real time specifically involves performing a short-time Fourier transform on the acquired voltage and current signal sequences to obtain the amplitude and phase of the input impedance at the fundamental frequency and each key subharmonic frequency.

[0047] The impedance ratio is preferably defined, according to the simplified principle of the Nyquist stability criterion, as the ratio of the amplitude of the synthesized target virtual impedance to the amplitude of the actual input impedance at each of the aforementioned key frequency points.

[0048] The preset instability boundary is specifically a safety threshold. For example, when the impedance amplitude ratio calculated at any frequency point is greater than 0.7, it is determined that the system stability margin is insufficient, and the adjustment mechanism is triggered.

[0049] The specific strategy for automatic adjustment is as follows: when the impedance ratio is detected to exceed the instability boundary, the controller prioritizes reducing the gain and / or increasing the damping of the digital filter stage in the active impedance synthesizer used to synthesize the impedance at that frequency. This adjustment aims to proactively increase the stability margin of the system. Although it temporarily sacrifices the accuracy of the impedance simulation at that frequency, it effectively prevents system oscillations and ensures the safety and continuity of the experiment.

[0050] This invention significantly improves the robustness, safety, and intelligence of the test system by introducing an online stability monitoring and adaptive adjustment mechanism. This method calculates the impedance ratio at key frequencies in real time and compares it with a safety threshold, enabling proactive prediction of instability risks and preventing destructive oscillations. When approaching the instability boundary, the system automatically performs precise fine-tuning of the controller parameters at specific frequencies, sacrificing temporary local accuracy for overall system stability and test continuity. This intelligent adaptive safety strategy makes it possible to simulate high-risk conditions such as severe resonance, fundamentally ensuring equipment safety and test reliability.

[0051] Furthermore, the closed-loop current controller of the wide bandwidth power converter is a proportional resonant controller; the proportional resonant controller includes a fundamental frequency resonant element and n harmonic frequency resonant elements, wherein the harmonic frequencies are preset integer multiples of the fundamental frequency, and are used to perform high-gain, zero-steady-state-error feedback tracking control on the fundamental and residual harmonic components in the final command current.

[0052] The resonant elements of the n harmonic frequencies are specifically key harmonics that need to be precisely compensated according to the test requirements, such as the 3rd, 5th, 7th and 11th harmonics.

[0053] To accommodate the slight fluctuations that may occur in the fundamental frequency of the power grid during actual operation and to improve the robustness of the controller, the resonant circuit preferably adopts a non-ideal resonant controller structure. This structure introduces a damping term into the transfer function of the resonant circuit, causing it to form a finite gain peak with a specific bandwidth (e.g., 5 Hz) near the resonant frequency, rather than an ideal infinite gain, thereby ensuring high tracking accuracy even when there are slight frequency shifts.

[0054] The proportional gain of the proportional resonant controller and the resonant gain parameters of each resonant element are preferably tuned using a frequency response analysis method based on Bode plots, according to the mathematical model of the power converter and the controlled object composed of the synthesized virtual impedance. This ensures that the entire closed-loop system has sufficient phase and amplitude margins, guaranteeing its stable operation. When implemented in a digital controller, the controller is preferably discretized using a bilinear transform.

[0055] This invention constructs a closed-loop electrical control system that combines high precision and robustness through the refined design and systematic parameter tuning of a proportional resonant controller. The non-ideal resonant element employed, by rationally setting the bandwidth, effectively overcomes the sensitivity of ideal controllers to grid frequency fluctuations, significantly enhancing the system's real-world adaptability and robustness. Furthermore, parameter tuning using frequency domain analysis based on the system model ensures that the controller achieves high dynamic tracking performance while possessing sufficient stability margin, guaranteeing operational safety. This refined design enables high-precision multi-frequency current tracking control to move from theory to stable and reliable engineering practice.

[0056] This invention achieves high-precision optimization of test current through a comprehensive method integrating accurate identification, environmental simulation, and feedforward compensation. The method first accurately acquires the nonlinear harmonic "fingerprint" of the transformer under test and flexibly synthesizes the virtual impedance of any target power grid, constructing a highly realistic "device-grid" interactive environment. Based on this, its innovative feedforward compensation technology can proactively offset the current distortion generated by the transformer, ensuring the purity and accuracy of the final injected test current waveform. This fundamentally solves the distortion problem caused by nonlinearity and impedance coupling in traditional testing, significantly improving the accuracy and reliability of the simulation test.

[0057] Example 2 This embodiment aims to conduct high-fidelity simulation and safety testing of a specific type of distribution transformer in a simulated weak power grid environment in an industrial park containing a large number of cable lines and reactive power compensation capacitors, to simulate a specific subharmonic resonance phenomenon that may occur.

[0058] Select a 10kV / 400V, 800kVA SCB10 dry-type distribution transformer. From the broadband impedance shaping matrix database, select a representative industrial park weak current grid model. The electrical characteristics of this model are as follows: due to the presence of numerous long (capacitive) cables and parallel capacitor banks used for power factor correction, the grid exhibits a very high parallel resonant impedance peak near the 250Hz (5th harmonic) frequency point. This operating condition is highly susceptible to interaction with the 5th harmonic current generated by transformer saturation, leading to harmonic amplification and severe voltage distortion.

[0059] First, the 800kVA transformer was connected to the test system. By injecting a pseudo-random binary sequence small signal with an amplitude of 0.5% of the rated voltage, and using a Fourier transform algorithm configured with a Hanning window, the characteristic fingerprint map of the transformer under test in the frequency band of 50Hz to 2.5kHz was accurately plotted.

[0060] Based on the obtained characteristic fingerprint of the unit under test, the parameters of a nonlinear electrical model with a built-in Jiles-Atherton hysteresis model and multiple parallel RL loss branches are calibrated. Subsequently, through non-uniform step size scanning simulation, a dynamic nonlinear harmonic prediction diagram specifically for this transformer is established. This diagram can accurately predict the amplitude and phase of the 3rd, 5th, and 7th harmonic currents generated by the transformer under different fundamental current excitations.

[0061] The frequency-impedance data of the selected industrial park weak grid model were loaded into the controller of the active impedance synthesizer. The controller successfully synthesized a target virtual impedance containing a 250Hz high impedance peak at the transformer port using a dynamic voltage adjustment method based on output current measurement.

[0062] The test begins by gradually increasing the fundamental current command injected into the transformer. When the fundamental current increases to the point that the transformer core approaches saturation, a 5th harmonic current will naturally be generated. The dynamic nonlinear harmonic prediction graph predicts the amplitude and phase of the upcoming 5th harmonic current in real time and accurately based on the current fundamental current command.

[0063] The system immediately generates a 5th harmonic compensation current waveform that is equal in magnitude to the predicted value but opposite in phase (reversed by 180 degrees). This compensation current is superimposed on the fundamental frequency command to form the final command current, which is then precisely driven by the proportional resonant controller to inject into the transformer via the power converter.

[0064] Through the above operations, the 5th harmonic current generated by the transformer's own nonlinearity is precisely canceled out at the port by the compensation current injected by the system feedforward. Therefore, although the external virtual power grid has extremely high resonant impedance at the 5th harmonic frequency, the voltage at the transformer port is not distorted because no significant 5th harmonic current flows into the virtual power grid, and the system remains stable.

[0065] Throughout the high-risk resonance simulation test, the online stability monitoring and adaptive adjustment module runs continuously. It calculates the amplitude ratio of the system source impedance to the actual load impedance in real time using short-time Fourier transform, ensuring that this ratio remains below the safety threshold of 0.7. If any unexpected disturbance causes a decrease in the stability margin, the system immediately fine-tunes the gain of the impedance simulation near the resonance point, prioritizing equipment safety and preventing oscillations, thus providing a decisive safety guarantee for simulating harsh operating conditions.

[0066] This embodiment verifies that the present invention can accurately and stably reproduce the interaction process between a transformer and a specific resonant power grid while ensuring absolute safety, and can optimize system behavior through active compensation, providing unprecedented testing capabilities for evaluating the grid connection performance of transformers.

[0067] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for optimizing transformer simulation test current based on dynamic impedance matching, characterized in that, include: An automated low-power scan is performed on the transformer under test to generate a characteristic fingerprint map of the unit under test that characterizes the broadband electrical properties. Based on the feature fingerprint of the tested unit, the parameters of the nonlinear electrical model are calibrated, and a dynamic nonlinear harmonic prediction map for forward-looking harmonic compensation is constructed. The target virtual impedance of the target power grid is defined based on a broadband impedance shaping matrix that records multiple target frequency points and the corresponding impedance amplitude and phase parameters of each point. The active impedance synthesizer is controlled to synthesize the target virtual impedance at the port of the transformer under test. The dynamic nonlinear harmonic prediction diagram is used in real time to generate a feedforward compensation current command, which is superimposed with the fundamental command to generate a final command current. The wide bandwidth power converter is controlled to inject the final command current into the transformer under test through the synthesized target virtual impedance.

2. The method for optimizing transformer simulation test current based on dynamic impedance matching according to claim 1, characterized in that: The steps for generating the feature fingerprint of the tested unit specifically include: A broadband sweep frequency excitation signal covering multiple frequency points is injected into the transformer under test; the broadband sweep frequency excitation signal is a pseudo-random binary sequence signal; while injecting the broadband sweep frequency excitation signal, the voltage response signal and current signal at the port of the transformer under test are simultaneously acquired; the acquired voltage response signal and current signal are processed by Fourier transform to calculate the complex impedance value corresponding to each of the multiple frequency points, the complex impedance value including impedance amplitude and phase; the complex impedance values ​​corresponding to all frequency points are summarized to generate the characteristic fingerprint map of the unit under test.

3. The method for optimizing transformer simulation test current based on dynamic impedance matching according to claim 1, characterized in that, The nonlinear electrical model specifically includes: The nonlinear magnetization unit, based on a nonlinear function that can characterize the hysteresis loop and saturation effect, maps the magnetic flux linkage obtained by integrating the input voltage into a nonsinusoidal excitation current. A linear circuit unit, connected in parallel with the nonlinear magnetization unit, is used to simulate the DC resistance and main leakage inductance of the transformer winding; Frequency-dependent loss unit is used to simulate additional losses that vary with frequency caused by core eddy current effects and winding skin effects.

4. The method for optimizing transformer simulation test current based on dynamic impedance matching according to claim 3, characterized in that, The specific steps for constructing the dynamic nonlinear harmonic prediction graph include: The fundamental current reference signals with different amplitudes and phases are used as input excitations and loaded into the nonlinear electrical model with calibrated parameters. For each input fundamental current reference signal, the nonlinear electrical model is run to simulate and calculate the total output current waveform containing fundamental and harmonic components caused by the nonlinear characteristics of the model. Spectral analysis is performed on each simulated total output current waveform to separate the amplitude and phase information of each harmonic component. A mapping lookup table is established and stored, with the input fundamental current reference signal as the index and the amplitude and phase of the corresponding harmonic components as the content, thus forming the dynamic nonlinear harmonic prediction graph.

5. The method for optimizing transformer simulation test current based on dynamic impedance matching according to claim 1, characterized in that: The broadband impedance shaping matrix is ​​a database that pre-stores multiple power grid model parameter sets, where each power grid model parameter set contains a set of frequency points and corresponding impedance vectors for defining different power grid characteristics. The step of defining the target virtual impedance is to select a corresponding power grid model parameter set from the database according to the test requirements, and load the frequency and impedance vector data therein into the control loop of the active impedance synthesizer as the real-time control target for synthesizing the target virtual impedance.

6. The method for optimizing transformer simulation test current based on dynamic impedance matching according to claim 4, characterized in that: The steps of generating feedforward compensation current commands in real time using the dynamic nonlinear harmonic prediction diagram and injecting them into the transformer under test specifically include: In each control cycle, the amplitude and phase of each harmonic component are retrieved in real time from the dynamic nonlinear harmonic prediction diagram using the current fundamental command as the query index. Based on the amplitude and phase of each harmonic component found, the harmonic current waveform is synthesized in real time and its phase is reversed by 180 degrees to generate the feedforward compensation current command. The feedforward compensation current command and the fundamental frequency command are superimposed in the time domain to generate the final command current; The final command current is used as a real-time reference value and input to the closed-loop current controller of the wide-bandwidth power converter to drive the wide-bandwidth power converter to output through the synthesized target virtual impedance.

7. The method for optimizing transformer simulation test current based on dynamic impedance matching according to claim 1, characterized in that, It also includes an online stability monitoring and adaptive adjustment step, which includes: While injecting the final command current into the transformer under test, the actual voltage and current at its port are monitored in real time; based on the actual voltage and current, the actual input impedance of the transformer under test is calculated in real time; the synthesized target virtual impedance is compared with the actual input impedance to obtain the impedance ratio characterizing the stability margin of the system; when the impedance ratio approaches the preset instability boundary, the control loop parameters of the active impedance synthesizer are automatically adjusted.

8. The method for optimizing transformer simulation test current based on dynamic impedance matching according to claim 6, characterized in that: The closed-loop current controller of the wide bandwidth power converter is a proportional resonant controller; the proportional resonant controller includes a fundamental frequency resonant element and n harmonic frequency resonant elements, the harmonic frequencies being preset integer multiples of the fundamental frequency, used for high-gain, zero-steady-state-error feedback tracking control of the fundamental and residual harmonic components in the final command current.

Images