New energy broadband oscillation control system and method based on impedance form

Abstract

The invention provides a new energy broadband oscillation control system and method based on impedance morphology, and relates to the technical field of new energy grid connection, and the system comprises a data collection module, an impedance estimation module, a morphological feature extraction module, a morphological trend determination module, a coordination control decision module and a control execution module. The data acquisition module is used for acquiring real-time measurement signals of three-phase voltage and current of a grid-connected point, and outputting a synchronized electrical quantity sequence after the real-time measurement signals are subjected to discretization, time alignment, debiasing and filtering preprocessing; the impedance estimation module converts a time domain signal acquired by the data acquisition module into a wide-frequency-band voltage-current frequency domain component through short-time frequency domain transformation; by extracting multi-layer morphological characteristics such as the slope, the curvature and the peak-valley inflection point of the impedance curve, the traditional single-point index monitoring limitation is broken through, the structural change of the oscillation brewing stage can be captured, and the recognition opportunity is advanced to the precursor stage of'no obvious waveform characteristics but transformed structure '.

Description

Technical Field

[0001] This invention relates to the field of new energy grid connection technology, and in particular to a new energy broadband oscillation control system and method based on impedance mode. Background Technology

[0002] Broadband oscillation monitoring in new energy mainly relies on two types of technical means: one is impedance measurement and stability criterion methods, which obtain the voltage-current frequency domain response of the grid connection point through frequency sweeping, small disturbance injection, or step response, and judge the resonance risk based on indicators such as gain margin, phase margin, and impedance intersection. This method has been applied to laboratory test platforms and some on-line monitoring equipment at the field stations. The other is signal mode identification methods based on operating waveforms, which use wavelet transform, short-time Fourier transform, empirical mode decomposition (EMD), Prony's method, etc., to extract time-frequency features of formed or developing oscillation waveforms and identify the oscillation frequency and damping change trend.

[0003] However, with the increasing penetration of new energy sources, the integration of inverter clusters, and the emergence of weak grid conditions, the shortcomings of existing technologies are becoming increasingly apparent: First, focusing only on the amplitude or phase information at specific frequencies fails to capture the evolution trend of the overall geometric shape of the impedance curve, while wideband oscillations often originate from structural features such as increased local slope, curvature changes, and an increase in the number of peaks and valleys; Second, relying on frequency sweeping or long window identification results in insufficient real-time performance, making it difficult to adapt to scenarios of rapid impedance drift caused by inverter mode switching, energy storage start-up and shutdown, and DC power adjustment, easily leading to output lag and identification distortion; Third, modal identification methods require oscillations with sufficient energy. The waveform cannot detect impedance structure changes in the precursor stage of oscillation, resulting in an inherent defect of "post-event identification". Fourth, monitoring and control are separated, lacking an active control closed loop based on risk trends. Even if an anomaly is detected, manual decision-making is required, resulting in a slow response speed. Fifth, in complex scenarios such as multiple inverters in parallel and weak power grids, traditional equivalent models are difficult to reflect irregular impedance characteristics with multiple peaks, segments, and inflection points, leading to insufficient reliability of judgment results and frequent occurrences of "model is normal but field oscillation". Therefore, this invention proposes a new energy broadband oscillation control system and method based on impedance morphology to solve the problems existing in the prior art. Summary of the Invention

[0004] To address the aforementioned issues, this invention proposes a broadband oscillation control system and method for new energy based on impedance morphology. By extracting multi-layer morphological features such as the slope, curvature, and peak-valley inflection points of the impedance curve, it overcomes the limitations of traditional single-point indicator monitoring and can capture structural changes during the oscillation incubation stage. This advances the identification opportunity to the precursor stage of "no obvious waveform features but structural changes have occurred," thus solving the deficiency of existing technologies in being unable to identify early trends.

[0005] To achieve the objectives of this invention, the following technical solution is provided: a new energy broadband oscillation control system based on impedance morphology, comprising a data acquisition module, an impedance estimation module, a morphological feature extraction module, a morphological trend determination module, a coordinated control decision module, and a control execution module. The data acquisition module acquires real-time measurement signals of three-phase voltage and current at the grid connection point, and outputs a synchronized electrical quantity sequence after discretization, time alignment, bias removal, and filtering preprocessing. The impedance estimation module transforms the time-domain signal acquired by the data acquisition module into broadband voltage-current frequency-domain components through short-time frequency domain transformation, constructs a complex impedance sequence point by point, and uses a rolling window mechanism to realize real-time impedance updates and structural verification. The morphological feature extraction module extracts multi-layer morphological features of the complex impedance sequence to form a morphological description vector containing local structural information and full-band topological features.

[0006] The morphological trend determination module performs trend analysis based on morphological description vectors and historical data, and classifies risk levels through time difference, rate of change, continuity verification, abrupt change detection, and full-band fusion. The coordinated control decision module generates a multi-dimensional control vector including virtual damping, control loop gain, and power allocation based on the risk level and morphological error. The control execution module converts the control vector into equipment control commands through a smooth loading mechanism, realizes coordinated adjustment of multiple devices in the new energy power station, and outputs feedback to the data acquisition module to form a closed loop.

[0007] Further improvements are made in that the multi-layer morphological features of the morphological feature extraction module include: impedance structured sequence after frequency discretization, local slope features, local curvature features, peak and valley point and inflection point topological features, single-frequency point comprehensive morphological quantity, full-band morphological vector and smoothing processing results.

[0008] Further improvements are made in that the risk level of the morphological trend determination module includes three categories: slight shift, accelerated evolution, and structural deterioration, which are comprehensively determined based on the number of morphological shifts, the rate of change, the persistence, the amount of mutation, the full-band shift, and the distribution of peak and valley inflection points.

[0009] Further improvements are made in the following aspects: the control vector generation process of the coordinated control decision module includes: determining the control intervention intensity that matches the risk level, calculating the morphological error between the current morphological vector and the target morphological vector, weighted fusion of errors in each frequency band to obtain the virtual damping adjustment amount, determining the control loop gain adjustment amount based on the full-band offset, and selectively adding power adjustment amount according to the risk level.

[0010] Further improvements include: the smooth loading mechanism of the control execution module includes linear loading or exponential loading; multi-device collaborative adjustment adopts a distributed power correction strategy; adjustment weights are allocated according to device response speed; and built-in parameter boundary checks and rate limiting logic are included.

[0011] A broadband oscillation control method for new energy sources based on impedance morphology includes the following steps:

[0012] S1: Collect the three-phase voltage and current signals at the grid connection point, perform discretization, time alignment, bias removal and filtering preprocessing, and output the synchronized electrical quantity sequence through a sliding short window;

[0013] S2: Perform a short time-frequency domain transformation on the preprocessed time-domain signal, calculate the complex impedance point by frequency to construct a wide-band impedance spectrum, and ensure the continuity and reliability of the impedance curve through rolling window updates and structural checks.

[0014] S3: Extract multi-layer morphological features of the complex impedance sequence to form a morphological description vector;

[0015] S4: Based on morphological description vectors and historical data, trend analysis is performed to detect morphological shifts, rate of change, persistence and abrupt changes, and risk levels are classified by integrating full-band trend information;

[0016] S5: Generate multi-dimensional control vectors based on risk level and morphological error, and verify amplitude, rate and equipment capability;

[0017] S6: The control vector is converted into equipment control commands through a smooth loading mechanism and sent to the new energy power station equipment to realize multi-equipment coordinated adjustment. The electrical quantity signals after control are collected and steps S2-S6 are repeated to form a closed-loop vibration suppression.

[0018] Further improvements are made in the following aspects: In S2, the short-time frequency domain transformation adopts the short-time Fourier transform, and the spectral leakage is reduced by windowing. For frequency points where the current amplitude is lower than the threshold, the adjacent frequency points are interpolated instead.

[0019] Further improvements are made in the following: In S3, the multi-layer morphological feature extraction includes: discretizing the continuous impedance curve, calculating the slope and slope difference curvature of adjacent frequency points, identifying local peak and valley points and inflection points, fusing slope, curvature and peak features to construct a single-frequency comprehensive morphological quantity, splicing them to form a full-band morphological vector and smoothing it by neighborhood averaging.

[0020] Further improvements are made in the following aspects: In S4, the trend analysis includes: calculating the time difference of the morphological quantity to determine the offset direction, calculating the rate of change to identify the acceleration trend, smoothing through a time window to verify the continuity of change, setting a threshold to detect abrupt morphological changes, and weighted fusing the offset of the entire frequency band.

[0021] Further improvements are made in S6, where the smooth loading mechanism controls the parameter update rate through the loading coefficient, the multi-device collaborative adjustment allocates power adjustment weights according to the device response speed, and the closed-loop feedback achieves adaptive correction by iteratively updating the impedance shape.

[0022] The beneficial effects of this invention are as follows:

[0023] 1. This invention breaks through the limitations of traditional single-point index monitoring by extracting multi-layer morphological features such as the slope, curvature, and peak-valley inflection points of the impedance curve. It can capture structural changes in the oscillation incubation stage and advance the identification time to the precursor stage of "no obvious waveform features but structural changes have occurred", thus solving the defect of existing technologies that cannot identify early trends.

[0024] 2. This invention adopts a short-window impedance estimation and rolling update mechanism, which can achieve millisecond-level impedance shape updates without frequency sweeping excitation. Combined with trend-oriented control decision and automatic execution link, the response cycle of "monitoring-judgment-control" is shortened to the millisecond to second level, solving the problems of insufficient real-time performance and separation of monitoring and control in the prior art, and realizing the leap from "post-event remedy" to "pre-event suppression".

[0025] 3. This invention is based on the overall geometric behavior of impedance mode for judgment, without relying on a fixed equivalent model, and can adapt to complex scenarios such as multi-inverter clusters, weak power grids, and frequent switching of control strategies. Through multi-device collaborative adjustment and parameter smooth loading mechanism, it not only ensures the comprehensiveness of vibration suppression effect, but also avoids the disturbance of equipment by instantaneous parameter jumps, which significantly improves the safety and reliability of system operation. Attached Figure Description

[0026] Figure 1 This is a flowchart of the present invention;

[0027] Figure 2 This is a schematic diagram of the typical impedance amplitude change during the first 3 minutes of oscillation in Embodiment 2 of the present invention;

[0028] Figure 3 This is a schematic diagram of the impedance morphology change within 1.5 seconds after the control is activated according to Embodiment 2 of the present invention. Detailed Implementation

[0029] To enhance understanding of the present invention, the present invention will be further described in detail below with reference to embodiments. These embodiments are only used to explain the present invention and do not constitute a limitation on the scope of protection of the present invention.

[0030] Example 1

[0031] according to Figure 1 , 2As shown in Figure 3, this embodiment proposes a broadband oscillation control system and method for new energy based on impedance morphology. It consists of six parts: a data acquisition module, an impedance estimation module, a morphological feature extraction module, a morphological trend determination module, a coordinated control decision module, and a control execution module. During system operation, the data acquisition module first obtains real-time measurements of three-phase voltage and current from the grid connection point, including amplitude, phase, and synchronization time scale information. These data typically come from merging units, high-speed sampling devices, or grid-connected detection instruments, and undergo necessary filtering, alignment, and noise reduction before entering subsequent modules to ensure the effectiveness of subsequent frequency domain analysis. The acquisition module continuously provides updated data streams to the system and is the fundamental component of the entire system. After obtaining the basic measurements, the impedance estimation module performs short-time frequency domain analysis on these voltage and current sequences, forming voltage-current response results covering a wide frequency band, and further constructs an impedance sequence reflecting the current dynamic characteristics of the system. The morphological feature extraction module treats the impedance sequence as a curve distributed with frequency and extracts its local variation characteristics. Internally, the module compares information such as the amplitude of changes, extreme value positions, curve inflection points, and curve smoothness between different frequency points to obtain a set of characteristic quantities that can describe the overall shape of the impedance curve. These characteristics reflect the overall trend of the curve's geometric structure, indicating whether new resonance peaks have appeared, whether local steep regions are forming, or whether there is a shift in existing characteristics. These structural characteristics are uniformly output as a morphological description vector for subsequent analysis. After obtaining the morphological description vector, the morphological trend determination module combines short-term historical data to determine whether these morphological characteristics are evolving in an unfavorable direction. By comparing changes over continuous periods, it identifies typical patterns such as "morphological shift," "accelerated morphological deterioration," and "evolution from a single peak to multiple peaks," and classifies the risk according to the rate and duration of change to distinguish between ordinary fluctuations and trend anomalies that may lead to broadband oscillations. When the trend determination module outputs an anomaly signal, the coordinated control decision module selects the most suitable control scheme from a preset set of strategies based on the anomaly type and severity. When a sharpening of the mid-frequency structure is detected, the current control parameters of the wind turbine are adjusted first. When a new peak appears in the high-frequency band, the virtual damping or filtering enhancement strategy of the energy storage system is invoked first. If the low-frequency portion shifts upwards as a whole, the active and reactive power distribution can be adjusted in conjunction with the high-voltage direct current transmission or the power station's main control. The coordinated control decision module calculates the magnitude and direction of the control quantity based on the equipment's operating status and adjustable range, and generates control commands that can be directly sent to the field equipment. The control execution module is responsible for sending the above control commands to the corresponding equipment, such as the wind turbine controller, energy storage converter, multi-terminal DC controller, or static synchronous compensator device, through the field communication interface. During actual execution, this module reliably transmits parameter adjustment commands, mode correction commands, or adjustment instructions according to the equipment interface protocol.Through this process, the system can intervene before the oscillation fully forms, restoring the impedance at the grid connection point to a smoother and safer state, thereby achieving the purpose of active suppression.

[0032] In the entire system's operational chain, the data acquisition module plays the role of "transmitting the on-site electrical status into the calculation process as is." Since the operating conditions of new energy power plants often change rapidly and the sources of interference are quite diverse, the main task of this module is to ensure that the acquired voltage and current signals are consistent in time, reliable in amplitude and phase, and meet the basic requirements for subsequent impedance analysis.

[0033] On-site quantity acquisition and discretization:

[0034] The system acquires three-phase instantaneous voltage and current signals in real time at the grid connection point through voltage transformers and current transformers, denoted as u. a (t), u b (t), u c (t) and i a (t), i b (t), i c (t). Where t represents a continuous-time variable. For ease of subsequent digital processing, these continuous signals are sampled at a frequency f. s Convert to a discrete sequence. The sampling period is T. s =1 / f s After discretization, the voltage and current of phase a can be written as:

[0035] u a [n] = u a (nT s )

[0036] i a [n] = i a (nT s )

[0037] The discretization methods for phases b and c are the same.

[0038] Time alignment and synchronization:

[0039] To ensure the phase relationship of the three-phase quantities is not disrupted in subsequent frequency domain analysis, the data acquisition module maintains time alignment of multi-channel data as much as possible. In field stations with the necessary resources, a Precision Timing Protocol (PTP) such as IEEE 1588 or other synchronization mechanisms are used to ensure all sampling channels share the same time base. When slight offsets exist between different sampling channels, the system performs minor interpolation or patching within the sampling buffer to keep the three-phase sequences as synchronized as possible on the time axis. The data acquisition module prioritizes synchronization.

[0040] Pre-cleaning and basic filtration:

[0041] The original sampled signal contains some short-term spikes or weak random noise, such as anomalies caused by factors like circuit breaker operation, sudden load changes, or communication jitter. To avoid these individual samples having an excessive impact on the frequency domain results during backend calculations, the data acquisition module performs lightweight smoothing locally, such as subtracting the average value of the samples and reducing the DC bias in the sequence, making the amplitude more concentrated within the effective range.

[0042] When the voltage discrete sequence u[n] is debiased:

[0043]

[0044] Where N is the window length used to calculate the average value. This represents the processed voltage sequence.

[0045] Data window and transmission logic:

[0046] The data acquisition module maintains a short, sliding window over time to store the most recent sampling points. This short-window mechanism reduces system resource consumption and avoids interference from "old samples" in morphological judgment. The window update rhythm is consistent with the system's main loop, ensuring that each round of impedance estimation is performed on the latest sequence. Once a new sample sequence is ready, the module hands over the three-phase voltage and current sequences within the current window to the next module. The entire transfer process is performed in near real-time, allowing subsequent impedance estimations to closely follow changes in the station's operating status.

[0047] Connection logic with the rest of the system:

[0048] After completing the above actions, the data acquisition module hands over the three-phase voltage and current sequences in the current window to the impedance estimation module in the next stage. This is a continuous data push process: whenever the sampling window advances a small segment (consistent with the system's main loop cycle), the latest u[n] and i[n] sequences are directly sent to the impedance estimation link. Similarly, the data acquisition module provides the time stamp of the current window to the backend to maintain temporal continuity in the morphological feature and trend judgment modules. The subsequent impedance estimation module uses these sequences to calculate the frequency domain components of voltage and current at each frequency point, while the morphological feature extraction module further relies on these results to construct the geometric features of the curves. Thus, the data acquisition module determines the starting rhythm of the entire link. If this stage outputs stably, impedance estimation, morphological extraction, and trend analysis can form a consistent time link, thereby enabling the active control part to intervene at the correct time node.

[0049] The impedance estimation module acts as a "transformation hub" in the entire system, converting the voltage and current time-domain sequences provided by the data acquisition module into impedance curves that characterize the system's dynamic properties. Since broadband oscillations are not triggered by changes at a single frequency point, but rather by the co-evolution of structures across different frequency bands, the core of impedance estimation is not "accurately measuring the impedance at a specific frequency point," but rather realistically and continuously depicting the overall impedance pattern over a wide frequency range. This module obtains the instantaneous spectrum through time-frequency transformation over a short time window. This method can promptly capture local impedance drifts and structural abrupt changes in the context of rapidly changing operating conditions.

[0050] Preprocessing and window construction of time-domain signals:

[0051] Before starting the frequency domain analysis, the impedance estimation module performs some light and necessary adjustments to the sampled sequence to make the short-window transform more stable. For example, for a three-phase voltage sequence u... a (t), u b (t), u c (t) and current sequence i a (t), i b (t), i c (t) Normalization is performed to avoid unnecessary scaling of the frequency domain amplitude spectrum due to amplitude variations. To construct the short-time window used for the current analysis, the system selects the most recent sample segment, such as a sequence of length N. Light processing, such as windowing, is performed on the signals within the window to reduce spectral leakage and make the frequency domain results clearer.

[0052] After applying a window function w[n] to the voltage sequence, we can obtain:

[0053]

[0054] This process can significantly improve the stability of the subsequent short-time Fourier transform.

[0055] Short-time Fourier transform and frequency domain component acquisition:

[0056] After window function processing, the impedance estimation module performs a short-time Fourier transform on the voltage and current of each phase to obtain the instantaneous frequency domain components. Taking the voltage of phase a as an example, its frequency domain expression is denoted as U(ω), where ω is the angular frequency, satisfying ω=2πf, and f is the analysis frequency.

[0057] Short-time Fourier transform writing:

[0058]

[0059] Similarly, performing the same processing on the current sequence yields the following:

[0060]

[0061] In renewable energy power plants, the current waveform is more susceptible to inverter modulation than the voltage waveform, thus the current spectrum typically contains more high-frequency components. The impedance estimation module uses a short time window to suppress high-frequency spurious energy caused by sharp transitions when analyzing these components, making the spectrum more closely resemble the dynamic characteristics of the equipment itself.

[0062] Wideband impedance determination and complex impedance construction:

[0063] After obtaining the instantaneous frequency domain components of voltage and current, the impedance estimation module constructs a complex impedance at each angular frequency point. The impedance Z(ω) is expressed in complex form, with its real part reflecting the equivalent resistance and its imaginary part reflecting the equivalent reactance.

[0064] The impedance is defined as follows:

[0065]

[0066] Where Z(ω) is the complex impedance at a certain frequency, U(ω) is the frequency domain component of the voltage, and I(ω) is the frequency domain component of the current. This calculation is not performed once across the entire frequency range, but rather point-by-point at each angular frequency, thus forming a complete broadband impedance spectrum covering the entire frequency range from low to high frequencies. In practical engineering, to avoid impedance distortion caused by excessively low current amplitudes, the impedance estimation module uses a simple amplitude threshold judgment. If the current amplitude at a certain frequency is lower than the threshold, that point will be skipped or interpolated from an adjacent frequency point will be used instead to maintain the overall continuity of the impedance curve.

[0067] Impedance update mechanism under scrolling window:

[0068] The impedance estimation module relies on a short-window rolling update method. This mechanism is particularly suitable for scenarios where wind turbines, energy storage, and photovoltaic inverters frequently switch operating modes. Each time the acquisition window moves forward by a small step, the module re-completes the frequency domain transformation and impedance calculation. This ensures that the impedance curve is updated almost in milliseconds, promptly reflecting dynamic changes at the grid connection point. For example, when the inverter switches from active / reactive power control (PQ control) to voltage / frequency control (VF control), the low-frequency impedance changes immediately; when the energy storage system uses virtual damping, the imaginary part of the impedance adjusts significantly in the mid-frequency range. These changes can be reflected by the impedance estimation module within a few calculation cycles.

[0069] Ensuring the integrity of the impedance curve:

[0070] To ensure the subsequent morphological feature extraction module can reliably process the impedance curve, the impedance estimation module also performs several structural checks: Frequency point smoothing check: to prevent isolated jumps in frequency points that could disrupt the curve structure. Amplitude and phase continuity check: when a 180° phase jump occurs, phase unenvelopment is performed. Overall structural repeatability check: if the impedance change before and after different windows is too large, a light smoothing is performed to determine if it is caused by a short-term disturbance. These actions do not change the essential characteristics of the impedance; they simply ensure the curve is structurally continuous, providing a clean input for the next step of morphological extraction.

[0071] Connection logic with the rest of the system:

[0072] The resulting broadband complex impedance sequence Z(ω) is then fed into the morphological feature extraction module according to the update cycle. The subsequent impedance estimation module uses these sequences to calculate: the slope, curvature, number of peaks and troughs, inflection point positions, and the overall trend of the curve's shape over time as impedance changes with frequency. This structural information is more indicative of the brewing or approaching critical signs of broadband oscillations than simple impedance amplitude or phase. The impedance estimation module provides the system with a "time-continuous and frequency-complete" impedance trajectory, which the subsequent morphological feature module uses to determine whether the system has moved from the stability boundary to an unfavorable region.

[0073] The impedance morphology feature extraction module, based on the broadband complex impedance Z(ω) provided by the impedance estimation module, extracts morphological features layer by layer that reflect changes in system stability. Before entering broadband oscillation, the impedance curve of a new energy power electronic system often exhibits slight but structurally significant changes, such as local sharpening, peak drift, and decreased curve smoothness. Therefore, this module adopts a six-layer feature extraction system "from shallow to deep, from local to global," from the fundamental frequency structure to topological features, and then to the comprehensive morphological construction, to achieve a complete analysis of the impedance morphology.

[0074] Feature extraction systems at each level:

[0075] Layer 1: Frequency Discretization and Impedance Structuring

[0076] The impedance estimation module outputs a complex impedance function Z(ω) defined on a continuous frequency axis. For subsequent local and global analysis, a discrete angular frequency axis covering the required analysis range is first selected:

[0077] [ω1,ω2,…,ω] K ]

[0078] Where K is the number of frequency sampling points.

[0079] Impedance readings at each frequency point:

[0080] Zk =Z(ω) k )

[0081] The purpose of this layer is to convert the original continuous curve into a structured sequence, which facilitates subsequent geometric processing.

[0082] Layer 2: Local slope features

[0083] The slope of the impedance curve reflects whether a certain frequency band is "steepening." In the early stages of broadband oscillations, the impedance curve for a particular frequency band often appears to suddenly rise. By making a differential approximation over adjacent frequency points, the slope can be obtained:

[0084]

[0085] Among them, S k For the frequency point ω k The impedance slope (complex value) near Z; k+1 -Z k This represents the local change in impedance. A sustained increase in the slope amplitude often indicates that a control element is approaching an unstable region.

[0086] Layer 3: Local curvature features

[0087] Impedance curvature describes whether the slope is changing at an accelerating rate. Many precursors to oscillations are characterized by the formation of a noticeable bulge in the previously flat impedance curve, with the rate of change accelerating. Curvature is approximated using the slope difference method.

[0088] C k =S k+1 -S k

[0089] Where C k It is the curvature (complex value) at the k-th frequency point. The greater the curvature, the sharper the local structure, and it is also an important "precursor indicator" of broadband oscillation.

[0090] Layer 4: Peaks, Valleys, and Inflection Points

[0091] Besides local velocity and acceleration, the system is more concerned with whether the overall shape of the impedance curve has undergone "structural changes," such as the appearance of new local peaks, the drift of existing peaks to lower or higher frequencies, an increase in the number of valleys, or the emergence of bimodal or multimodal structures. These changes are closely related to the number of oscillation modes and the location of resonant points. Local peaks can be identified using the following logic:

[0092]

[0093] Among them, P k =1 represents ω k It is a local peak point.

[0094] In impedance morphology analysis, in addition to identifying local peaks, this invention also determines local valleys (local minimums) and inflection points. Valleys correspond to the "sunken sections" of the impedance curve, while inflection points correspond to the structural transitions where the shape changes from "convex" to "concave" or vice versa, which is crucial for understanding the overall geometry of the impedance curve.

[0095] A local trough refers to a frequency point where the impedance magnitude is significantly lower than that of its adjacent frequencies. Its definition is similar to that of a peak, but the comparison direction is reversed. For discrete frequency points ω... k If the following conditions are met:

[0096]

[0097] Then ω is called k This represents a local valley point. Where V... k =1 indicates the existence of a valley value. Valley value information in new energy systems is often related to inverter limit control, changes in filter structure, etc., and can reveal some stability risks deep in the control path.

[0098] Inflection points describe the location where a local change in the impedance curve shifts from "upward convexity" to "downward convexity" (i.e., from a convex shape to a concave shape, or vice versa), and are one of the key features of the impedance curve's topology. In the discrete case, inflection points can be identified by the sign of the slope change:

[0099]

[0100] Among them, I k =1 represents ω k The inflection point; S k The slope at the k-th frequency point (defined in the second layer of features); the condition "<0" means that the "bending direction" of the curve changes.

[0101] From an engineering perspective, the appearance of an inflection point signifies that the impedance's shape in that frequency band transitions from a "smooth segment" to a "rapidly changing segment," making it highly suitable as an early indicator of broadband oscillation risk. In practical morphological analysis, peak values ​​typically represent potential resonance points; trough values ​​typically represent potential energy absorption or control characteristic change zones; and inflection points typically represent the "turning point" of the curve structure. These three types of information collectively form the basis of impedance topology.

[0102] Layer 5: Construction of Single-Frequency Point Comprehensive Morphological Quantities

[0103] To avoid over-reliance on a single indicator in subsequent trend analysis, this invention integrates local features such as slope, curvature, and peak structure to construct a comprehensive morphological description of a single frequency point.

[0104] M(ω k )=α1|S k |+α2|Ck |+α3P k

[0105] Where M(ω) k ) represents the frequency point ω k The local composite morphological quantity; α1, α2, α3 are weights (determined by simulation or engineering calibration). This quantity reflects whether a potential resonant structure is forming in this frequency band.

[0106] Layer 6: Full-band morphological vector and smoothing processing

[0107] To allow the system to observe the overall structure of the impedance pattern across the entire frequency band, this module includes all (M(ω) K Concatenate them to form the shape vector at the current moment:

[0108] M t =[M(ω1),M(ω2),...,M(ω K )]

[0109] Among them, M t Let be the impedance shape vector at time t. To enhance the structure and noise resistance of the shape vector, a lightweight smoothing method, such as neighborhood averaging, is employed.

[0110]

[0111] Smoothed shape vector It is more suitable for trend analysis and risk assessment.

[0112] Connection logic with the rest of the system:

[0113] The impedance morphological feature extraction module ultimately outputs the following result: the complete morphological vector M at the current time. t Smoothed shape vector Key feature locations, such as peak drift direction and frequency bands where curvature is concentrated, are directly input into the next module, the morphology trend determination module. The latter will determine whether the impedance morphology is gradually shifting towards the "oscillation formation region" based on the time series.

[0114] The morphology trend determination module occupies a pivotal position in the entire system chain, entering from the "time dimension" to determine whether the impedance morphology exhibits directional changes. For example: Is the curvature of certain frequency bands gradually increasing? Are local peak values ​​slowly rising? Are there persistently shallowing troughs in the low-frequency band? This module employs a six-layer trend analysis structure, progressing from basic single-point changes to full-band offsets, and then to risk grading, achieving a progressively deeper trend determination process.

[0115] Pattern trend determination process:

[0116] Layer 1: Temporal Difference of Morphological Quantities

[0117] The goal of this layer is to determine whether the overall morphological quantity at a certain frequency point shows an "upward trend," that is, whether the impedance in that frequency region is shifting in a more acute and unstable direction.

[0118] For the frequency point ω k ,definition:

[0119] D t (ω k ) = M t (ω k )-M t-1 (ω k )

[0120] Where: M t (ω k ) represents the morphological composite quantity at time t (from step 3); D t (ω k The positive sign indicates the direction of the change in frequency (positive represents an increase, negative represents a decrease). If a key frequency band (such as the area near the characteristic frequency of the inverter control circuit) shows a continuous positive change, it usually means that the frequency band is becoming "sharper" or "steep".

[0121] Level 2: Rate of Change Judgment

[0122] Some frequency bands change in a positive direction but at a very slow rate, which is considered normal fluctuation; while the rate of change in other frequency bands gradually increases, often corresponding to the brewing of oscillations. Therefore, we further define the rate of change as follows:

[0123] V t (ω k ) = D t (ω k )-D t-1 (ω k )

[0124] Among them, V t (ω k V represents the rate of morphological change, measuring whether the change is accelerating. t (ω k When the value is continuously positive, it indicates that the rate of morphological deterioration in that frequency band is truly "accelerating." A gradual increase in curvature in the low-frequency band usually indicates that a low-frequency oscillation mode caused by a weak network is beginning to form.

[0125] Layer 3: Persistence Analysis

[0126] Impedance curves inevitably contain short-period disturbances, such as transient fluctuations, measurement noise, or communication jitter. To avoid these transient disturbances interfering with trend determination, this module uses a resmoothing window to determine whether the change is persistent. Definition:

[0127]

[0128] Where W is the length of the time window; This represents the average morphological shift over a period of time. If a certain frequency band... If the value remains positive continuously, it indicates that the morphological changes are trend-based, rather than just random.

[0129] Layer 4: Morphological mutation detection

[0130] Impedance characteristics do not always change slowly; sometimes, structural abrupt changes occur suddenly at a certain moment. For example: a new peak appears in a previously smooth segment; the peak height suddenly increases; a trough becomes significantly deeper within a cycle; an inflection point appears and rapidly spreads to both sides. To capture this "structural jump," the abrupt change is defined as:

[0131] J t (ω k )=|M t (ω k )-M t-1 (ω k )|

[0132] If J t If the frequency exceeds a certain empirical threshold (given by simulation calibration), it is considered to have experienced a morphological abrupt change and requires special attention. Frequency bands exhibiting abrupt changes are typically prioritized by active control modules.

[0133] Layer 5: Full-band fusion

[0134] Wideband oscillation is essentially the result of the coupling of characteristics across multiple frequency bands, and does not depend on a single frequency point. Therefore, this module fuses the trend information of all frequency bands. Define the full-band offset:

[0135]

[0136] Among them, T t This represents the overall offset of the current periodic impedance pattern; β k The weights for each frequency band are specified (critical frequency bands have higher weights). When multiple frequency bands simultaneously exhibit significant morphological shifts, and T... t When the change continues to increase, it indicates that the system as a whole is moving towards the edge of oscillation. This "overall shift" is more dangerous than a single-point change.

[0137] Level 6: Trend Level Classification

[0138] This module ultimately makes a comprehensive judgment based on the following multi-dimensional indicators: morphological shift D in a single frequency band. t The number of >0; the rate of increase V t Scope; The persistence; whether there are significant mutation points J t Full-band offset T t The strength of the signal; whether synchronization shift occurs across multiple frequency bands; whether peaks, troughs, and inflection points are concentrated in a specific frequency band. Three risk levels are given:

[0139] Risk Level I: Mild Shift: A Few Frequency Points D t >0;V t Mostly close to 0; no obvious mutations; T t At a low level, it indicates that the system is beginning to show slight structural shifts, which are often related to changes in operating points and disturbances in operating conditions.

[0140] Risk Level II is evolving rapidly: the morphological quantities of multiple key frequency bands continue to rise; V t It shows positive values ​​at multiple points (accelerated change); a few abrupt change points appear; the peak-valley structure gradually emerges; T t The system is clearly biased towards positive. At this point, the system is gradually approaching the oscillation boundary, which is a critical window for active control intervention.

[0141] Risk Level III: Structural Deterioration: Characteristic frequency bands show a significant increase; curvature concentration areas are significantly enhanced; multiple abrupt change points are identified; peak-valley structures appear densely; full-band shift T t The values ​​are high; multiple frequency bands are deteriorating synchronously. This trend indicates that the system is already in the "formation zone" of oscillation, and the next step may be to enter actual broadband oscillation.

[0142] Connection logic with the rest of the system:

[0143] The final output of the pattern trend determination module is: the trend direction at each frequency point; the upward speed and acceleration index D. t V t Change points and concentration ranges; Full-band offset T t Trend risk level (I, II, III). These results will be directly input into the next control decision module, which will automatically adjust impedance, virtual damping, control parameters, or power allocation according to the risk level to actively suppress broadband oscillations.

[0144] After identifying the morphological trend, the system can determine the current stage of impedance evolution: whether it's a slight shift, accelerating deterioration, or approaching the oscillation boundary. The coordinated control decision module integrates this trend-related information into a control and adjustment scheme with engineering feasibility. It doesn't simply send commands to a single device; instead, it establishes a coordinated adjustment logic across the entire renewable energy power plant—inverter control layer—grid-connected interface device—DC transmission system, enabling the impedance to be "moved" back to a stable region as a whole, rather than placing the entire vibration suppression responsibility on a single device. This invention introduces two fundamental concepts in the control decision-making process: first, the system's control behavior must match the strength of the morphological trend to avoid "over-adjustment" or "under-adjustment"; second, all control actions must revolve around "morphological error," aiming to bring the current impedance back close to an engineering-verified "safe state."

[0145] Coordination and control decision-making process:

[0146] In actual operation, the coordination control decision module first reads the risk level L given by the trend judgment module. t The permissible intensity of control interventions for this cycle is determined based on the risk level. This mapping relationship is represented by a function F(·):

[0147]

[0148] Among them, L t Three levels can be selected: 1, 2, and 3, corresponding to mild shift, accelerated evolution, and structural deterioration, respectively; U t This specifies the allowed control intensity level, essentially setting an "upper limit" for subsequent control quantities. During the engineering debugging phase, this function can be flexibly set based on simulation experience, historical data, and equipment capabilities.

[0149] After the control strength is determined, the coordinated control decision module further calculates the deviation between the current impedance shape and the target shape. To achieve this, this invention defines a target shape vector M. · It typically derives from two types of information: one is the "stable operating impedance profile" obtained from simulation; the other is the actual measured value during long-term stable operation of the station. The target profile represents "what impedance structure the system should recover to," rather than being set arbitrarily. Based on this, the profile error at each frequency point is defined as follows:

[0150] E t (ω k ) = M t (ω k )-M ★ (ω k )

[0151] This error reflects the degree to which the current impedance curve deviates from its stable form at different frequency bands.

[0152] After understanding the risk level and morphological error, the coordinated control decision module can construct the actual control quantity for the current cycle. To ensure that the control actions are both specific and systematic, this invention focuses key adjustment quantities in three directions: virtual damping, control loop gain, and power regulation. Virtual damping is used to quickly reduce the sharpening phenomenon in the mid-to-high frequency range, control gain adjustment is used to suppress the amplification of potential resonant modes, and power regulation releases local stress by changing the grid injection path. First, the virtual damping adjustment quantity ΔD is constructed. t It weights and sums the morphological errors of different frequency bands, thus generating an "overall vibration damping force" in the frequency domain:

[0153]

[0154] Among them, K D It is the damping adjustment coefficient, w k It is used to emphasize the weight of key frequency bands (for example, some inverters are particularly prone to oscillations in the 400–800Hz range, so this range has a greater weight).

[0155] Next, the control loop gain adjustment ΔG t Based on full-band morphological offset T t This is because gain adjustment is usually global and does not only affect a specific frequency band.

[0156] ΔG t =K G ·T t

[0157] Where K G T is the gain adjustment coefficient. t As defined in step 4, it represents the degree to which the current overall impedance configuration shifts in an unfavorable direction.

[0158] When the trend level is high, this invention adds a power regulation amount ΔP to the above control amount. t It operates on devices such as wind turbines or energy storage systems by fine-tuning power distribution, diverting some current paths away from high-risk frequency bands. This "diversion" adjustment method is often used in engineering to avoid excessively low local impedance or the formation of local resonance points.

[0159] The three control quantities mentioned above are combined into a single control vector in this invention:

[0160]

[0161] The coordinated control decision module limits the amplitude and rate of the vector and verifies equipment capabilities to ensure that the control action is effective without introducing additional instability. Subsequently, this set of control variables is sent to the corresponding controllers via the communication system, including the inverter master controller, STATCOM controller, energy storage converter, and hybrid DC control station, achieving true site-level coordinated active control. Finally, the new control parameters will be reflected in the impedance estimation of the next cycle. Through this compact closed loop, the system can continuously pull the impedance shape back to the stable region, forming a complete active vibration suppression link from "identification—judgment—adjustment".

[0162] Connection logic with the rest of the system:

[0163] The coordinated control decision module maintains strict timing and data coupling with the two key upstream and downstream components. The front-end impedance estimation, pattern extraction, and trend determination modules provide structured and quantifiable foundational data, preventing misjudgments in control decisions due to transient disturbances or local feature distortions. The back-end control execution module is responsible for applying the control vectors generated in this step to equipment such as wind power, photovoltaics, energy storage converters, STATCOM, and DC systems, forming a closed-loop real-time adjustment process. The control input is not simply "output and done," but rather enters the next cycle of data acquisition and impedance estimation, allowing the impedance changes caused by the new control actions to return to the front end of the data link, creating a stable and rapidly responsive loop for the entire system. Within this closed loop, there are no loose interfaces between the modules; instead, they operate synchronously through a highly coupled time window mechanism: the timestamp of data acquisition determines the calculation boundary of trend determination, and the result of trend determination defines the control intensity range of the coordinated control decision module. Subsequently, the control execution module applies specific control based on equipment capabilities and adjustment strategies, causing the system to re-enter the evaluation phase in the next sampling cycle. This compact loop not only ensures the real-time performance of control actions but also enables the entire system to rapidly capture and correct impedance pattern shifts on a millisecond to second timescale, achieving true active suppression of broadband oscillations. The system can intervene before oscillations fully form, rather than passively remedying them after obvious waveform anomalies. Ultimately, the coordinated control decision module, acting as a crucial bridge between upstream and downstream components, allows this invention to truly move from "detecting anomalies" to "correcting anomalies," thus forming a complete closed-loop logic that runs through the entire control chain of new energy power plants.

[0164] After the control decision module generates control vectors, these control quantities do not remain at the algorithm level. Instead, they must be distributed item by item to various devices such as wind turbine converters, energy storage converters, STATCOMs, grid-connected inverters, and hybrid DC controllers through a tightly organized execution chain to enable the system to truly function. The control execution module implements the control scheme into physical devices, determining the order of control actions, the speed of action, and the coordination methods between multiple devices.

[0165] The control execution module incorporates a smooth loading mechanism to transform the parameter increments provided by the decision module into slowly sized variables that the equipment can safely receive. For example, for the virtual damping adjustment ΔD... t The control execution module uses linear or exponential loading to proportionally push the new damping value in the target direction:

[0166] D new =D old +α D ·ΔD t

[0167] Where D old Given the current damping parameter, D new For the updated parameters, α D It is the damping loading factor (typically taken as 0.2 to 0.5).

[0168] Similarly, the control loop gain is updated incrementally to avoid sudden reductions in phase margin within the loop. The gain adjustment relationship is written as follows:

[0169] G new =G old +α G ·ΔG t

[0170] Among them, G old G is the current control loop gain. new For the updated control loop gain, ΔG t To control the loop gain adjustment, α G It is the gain loading factor (typically 0.1 to 0.3, used to maintain the dynamic stability of the control loop).

[0171] For power regulation that requires collaboration among multiple devices, this invention employs a distributed power correction strategy, whereby different devices share the total power adjustment amount ΔP according to their capabilities and response speeds. t The power adjustment amount for the i-th device is:

[0172] P i,new =P i,old +β i ·ΔP t

[0173] Among them, P i,old Let P be the current power of the i-th device. i,new For the updated power of the i-th device, β i Let be the weighting factor for the equipment, ensuring that the sum of the weights of all equipment equals 1.

[0174] This mechanism allows fast-responding energy storage, STATCOM, or modular multilevel converters to assume more regulation responsibility, while slower-responding wind turbines bear less weight, achieving true multi-device collaborative vibration suppression. After the above parameters are updated, the control execution module organizes these newly generated parameters into an "execution command vector" and distributes them according to device type, control domain, and priority.

[0175]

[0176] These commands are then transmitted to various power electronic devices via the station control layer and high-speed real-time links, enabling real-time updates of parameters such as virtual damping, controller gain, and power distribution within the devices. Simultaneously, the protection logic built into the control execution module performs boundary checks and rate limits on all parameter changes, ensuring that the control quantity never exceeds the device's capacity. For example, when the inverter is performing low-voltage ride-through or grid-connected switching, certain control channels are temporarily locked to avoid conflicts between control actions and protection strategies. After the control commands take effect within the device, the data acquisition in the next cycle immediately captures the new electrical quantities. The impedance estimation module regenerates the latest impedance curve Z(ω), and the shape extraction module generates a new shape vector M based on this. t The trend determination module also immediately assesses whether the new pattern is moving in a stable direction. Regardless of whether the control action fully achieves the expected result, it will be re-analyzed in the next cycle, forming the following continuous closed loop:

[0177]

[0178] This closed loop ensures that the system can "see what it has just done" and further correct the control inputs, achieving true active and stable control. Through this continuous iteration, gradual loading, and multi-device collaboration, the control execution module enables the invention to re-fix the system within a safe operating range before the wideband oscillation actually erupts, by real-time shaping of the impedance.

[0179] Connection logic with the rest of the system

[0180] The control execution module is the terminal node of the entire impedance morphology active control link and serves as the entry point for the "closed-loop return path." Each control command issued directly affects the data acquisition and impedance estimation in the next cycle. Therefore, it is not loosely coupled with all front-end modules but rather forms a synchronous collaboration under strict timing. Specifically, after the control execution module writes the updated control quantity into the device, the parameters of the device's internal current loop, voltage loop, phase-locked loop, and power control loop immediately take effect in the next sampling cycle. The new dynamic characteristics are directly reflected in the voltage and current waveforms and subsequently fully captured by the data acquisition module. The impedance estimation module generates a new Z(ω) sequence based on the new measurements, the morphology feature extraction module re-analyzes it, and the trend determination module judges whether the system has moved towards stability based on the new morphology. Each control execution is the starting point for the next "analysis-judgment-decision." The system continuously evaluates the effectiveness of the previous round of control actions. If it does not meet expectations, a new control scheme will be automatically provided in the next round; if the effect is significant, the control intensity will be gradually reduced, allowing the system to return to the most economical and natural operating state. The control execution module not only implements control actions but also provides the latest status feedback for the entire closed loop, enabling the invention to truly possess the capabilities of proactive adjustment, real-time correction, and rapid stabilization. Through this closed-loop collaboration, a complete link is constructed from "acquisition → impedance → morphology → trend → control decision → execution → re-acquisition". As the final step in the closed loop, the control execution module complements the front-end impedance estimation module, morphology extraction module, and trend determination module, ensuring that the entire system can still achieve accurate, rapid, and coordinated broadband oscillation proactive suppression even when facing weak power grids, strong power electronic coupling, and complex operating conditions.

[0181] This invention, for the first time, uses "impedance morphology characteristics" to uniformly express the dynamic behavior of new energy power stations in the mid-to-high frequency range. It no longer relies on a single frequency point or amplitude, but simultaneously considers multi-dimensional structural information such as slope, curvature, peak-to-valley points, inflection points, and overall trends, giving the system's dynamic characteristics higher sensitivity and robustness. This structured expression can accurately capture latent oscillation precursors, which is difficult to achieve with traditional amplitude-based indicators. The impedance estimation, morphology extraction, trend judgment, and coordinated control decision-making modules of this invention are not stacked in parallel, but are strictly coupled according to a time sequence, with each measurement flowing through a unified link. The modules use structured variables as inputs and outputs, forming a rigorous logical closed loop, giving the system a "first-response mechanism" that can identify dynamic deviations before oscillations occur. Unlike traditional methods that "compensate after voltage / frequency anomalies," this invention emphasizes intervention as soon as the trend worsens. It uses quantitative indicators such as trend velocity and trend acceleration to judge the risk level in advance and dynamically adjusts the control intensity according to the risk level, making control no longer a passive remedy but a proactive and appropriate adjustment. This control logic is particularly crucial in scenarios with weak power grids and multiple inverter coupling. The control decision module does not adjust parameters for only a single inverter, but can simultaneously act on multiple devices such as photovoltaic, energy storage, wind power, or flexible DC power. It coordinates the damping, current loop gain, and reactive / power distribution of different devices through unified morphological variables and trend indicators. Its core feature is that the control quantity comes from the same trend criterion, but the adjustment objects are diversified, making the entire power station behave as a "unified system" rather than multiple discrete independent inverters. The control execution module is designed with a parameter smoothing adjustment structure, allowing virtual damping, gain, and power commands to be applied gradually, avoiding disturbances to the internal loops of the devices caused by instantaneous large adjustments. Unlike the traditional method of "directly overriding control parameters," the structure of this invention ensures a continuous and smooth transition of the device control core, thereby improving the safety of long-term operation. The loop structure of "device output → data acquisition → impedance morphology update → trend calculation → control setpoint adjustment" in this invention is completely closed, enabling the system to automatically enter a self-correcting operation mode. Each control action is immediately reflected in the impedance morphology of the next cycle, allowing the system to "verify whether it has successfully controlled the device" and automatically weaken or strengthen the control based on the morphological offset. This self-calibrating structure is the core reason why this invention can maintain a stable vibration suppression effect over a long period of time. This invention constructs an active stability control system based on impedance characteristics, guided by trends, characterized by multi-device coordination, and guaranteed by a closed-loop structure. It achieves stable and reliable broadband oscillation suppression in weak power grids and environments with a high proportion of renewable energy connected to the grid.

[0182] Example 2

[0183] according to Figure 1 , 2As shown in Figure 3, this embodiment presents an engineering application example of a broadband oscillation active control method based on impedance morphology characteristics:

[0184] Scene and system composition:

[0185] This embodiment selects a real-world operating scenario of a 200MW wind farm connected to a ±320kV VSC-HVDC hybrid DC transmission system. The wind farm side includes 80 2.5MW grid-connected double-fed induction generators (DFIGs), and is equipped with a 20MW / 10MWh energy storage system and a 10Mvar STATCOM for dynamic reactive power and voltage support. The DC transmission end employs master-slave control, responsible for delivering constant power to the main grid.

[0186] Under weak network conditions (SCR=2.1), the system exhibits certain structural risks in both the high-frequency and mid-frequency impedance bands, especially when wind speed fluctuations are large (random fluctuations of 0.8–1.5 m / s). Several broadband oscillation precursors were observed at the wind farm in the 180–420 Hz frequency range. To verify the effectiveness of the method, it was deployed in the wind farm control system and connected to the control interfaces of the wind turbine, energy storage converter, STATCOM, and DC modular multilevel converter, forming a complete active vibration suppression link.

[0187] Implementation steps:

[0188] Data acquisition and impedance estimation:

[0189] Three-phase voltage and current signals were sampled at the grid connection point at a sampling rate of 12.8 kHz and a sampling window of 80 ms. Real-time wideband impedance curves were obtained through sequence impedance injection (small perturbation method + frequency domain least squares), with a frequency range of 0-1000 Hz and a resolution of 5 Hz. Figure 2 The typical impedance amplitude change in the first 3 minutes of oscillation (data from actual sampling) is shown below. Figure 2 It can be seen that the impedance rises rapidly above 300Hz, which may indicate the formation of a resonant structure.

[0190] Impedance morphology feature extraction and trend determination:

[0191] Based on the aforementioned impedance curve, morphological extraction (six-layer feature system) is performed to obtain the comprehensive morphological vector M at this moment. t Typical frequency band morphology parameters are shown in Table 1:

[0192] Table 1 Typical Frequency Band Morphology Quantities

[0193] Frequency (Hz) slope curvature Peak / Valley / Turn Characteristics <![CDATA[Integrated form quantity M(ω k ) <!-- 14 -->]]> 260 0.0041 0.0018 0 0.030 300 0.0062 0.0027 inflection point 0.052 340 0.0093 0.0051 peak 0.097 380 0.0148 0.0086 peak 0.153 420 0.0115 -0.0021 Peak followed by decline 0.128

[0194] V t (ω kThe morphological deterioration accelerates; the peak value shifts from 340Hz to 380Hz; the full-band offset T t =1.82 (significantly increased) The trend determination module classifies this state as Level II: accelerated evolution zone, with increased danger.

[0195] Control Decisions:

[0196] Based on the trend level (Level II), the decision module allows for moderate-intensity control interventions, providing the adjustment amount:

[0197] Virtual damping increment: ΔD t =+0.024

[0198] Control loop gain reduced: ΔG t =-0.015

[0199] Energy storage rapid power-up command (used to buffer high-frequency oscillation energy): ΔP t =+0.35MW

[0200] Then it enters the execution module.

[0201] Control execution and system response:

[0202] Actual parameters written:

[0203] The execution module issues instructions according to the smooth update mechanism:

[0204] D new =D old +0.4ΔD t =D old +0.0096

[0205] G new =G old +0.2ΔG t =G old -0.003

[0206] Energy storage output increased from 2.3MW to 2.65MW.

[0207] Impedance shape change within 1.5 seconds after control takes effect:

[0208] Within 1.5 seconds of the control taking effect, the newly added damping rapidly manifests in the mid-to-high frequency range, significantly improving the impedance curve, and the control effect is as follows: Figure 3 As shown in Table 2.

[0209] Table 2 shows the changes in impedance morphology within 1.5 seconds after the control takes effect.

[0210] Frequency (Hz) Before control 1.5 seconds after control 300 0.62 0.51 340 0.85 0.68 380 1.21 0.83 420 1.47 1.05

[0211] Morphological quantity M tThe corresponding decrease is approximately 32%-45%.

[0212] The trend assessment output changed from Level II to Level I, indicating a significant decrease in the risk of oscillation.

[0213] Evaluation of control effectiveness:

[0214] The high-frequency impedance peak decreased by 31%: the 380Hz peak value changed from 1.21 to 0.83, effectively moving it out of the danger zone.

[0215] Peak drift was blocked: the original trend of drifting towards higher frequencies (340→380→420Hz) stopped after control, and the curvature index was reduced by 55%.

[0216] Wideband oscillation blocking: Before using the control method of this invention, the system occasionally exhibited small oscillations of 20-40ms in a wide frequency band of 180-420Hz. After continuous monitoring for 15 minutes after control, the same characteristic did not reappear.

[0217] The energy storage capacity is only slightly adjusted by 0.35MW, and the control action is lightweight and efficient. This indicates that the present invention does not rely on "large-scale vibration suppression" but rather on a "morphological prediction + light adjustment" mechanism.

[0218] The control link is stable, with no reverse adjustment oscillations. The parameter loading mechanism effectively avoids loop oscillations or secondary disturbances.

[0219] Results and conclusions:

[0220] This embodiment clearly demonstrates that the "impedance morphology characteristics + trend determination + active control" method proposed in this invention can capture early structural changes before broadband oscillations are formed; the control quantity is generated based on morphological errors, which is highly targeted and gentle; the execution module can reliably, safely and smoothly apply adjustments; ultimately, the effects of impedance morphology recovery, peak value reduction and complete oscillation blocking are achieved; compared with the existing "post-event suppression", this invention significantly improves the foresight and initiative of system operation.

[0221] Example 3

[0222] according to Figure 1 , 2 As shown in Figure 3, this embodiment proposes an engineering application for active suppression of wideband oscillations dominated by energy storage:

[0223] System background:

[0224] This embodiment selects a real-world scenario where a 100MW / 200MWh energy storage power station and a 90MW photovoltaic power station are connected to a 220kV weak grid (SCR≈1.6). Due to the very limited grid-side short-circuit capacity, the control coupling between the energy storage inverter and the photovoltaic inverter is strong. Especially in the 200–600Hz range, the frequency response of the point of common coupling (PCC) impedance is prone to a bulging mid-frequency peak, a peak value that drifts with changes in operating conditions (320→380Hz), and an upward bending characteristic in the high-frequency band, making it a typical high-risk system with wideband oscillations.

[0225] Implementation steps:

[0226] Step 1: Real-time impedance measurement and data acquisition

[0227] During a rapid change in operating conditions from clear skies to cloud cover (energy storage output approximately 35MW), the system impedance (at some frequencies) was measured as shown in Table 3. The sampling rate was 20kHz, the impedance refresh period was 40ms, the frequency range was 0–800Hz, and the frequency resolution was 2Hz. It can be seen that the impedance increases exponentially starting from 320Hz.

[0228] Table 3 System Impedance (Partial Frequency Points)

[0229] f(Hz) Z(ω)|(pu) 240 0.42 280 0.55 320 0.71 360 0.94 400 1.28 440 1.61 480 1.73

[0230] Step 2: Shape Extraction (Slope, Curvature, Peaks and Valleys, and Trend Term)

[0231] The patented six-layer morphological system is adopted. The slope and curvature at some frequency points are as follows:

[0232] Table 4 Slope and Curvature of Some Frequency Points

[0233]

[0234] Step 3: Determine the shape offset and trend level

[0235] Define full-band offset: At a certain moment, the shape offset reaches T t =2.34 (far above the threshold of 1.2). The peak value drifted from 360Hz to 400Hz, and the average trend velocity term was positive. The average trend acceleration term was >0.0035. The trend level was determined to be: Level III, high risk.

[0236] Step 4: Proactive Control Decision

[0237] Based on Level III risk, the system allows for high-intensity control actions:

[0238] Virtual damping increment of energy storage inverter: ΔD t=+0.031

[0239] Photovoltaic inverter current loop gain reduced: ΔG PV =-0.018

[0240] Synchronous reduction of current loop gain in energy storage inverter: ΔG ES =-0.012

[0241] Energy storage active vibration damping power command: ΔP t =+0.9MW

[0242] Step 5: Control Execution

[0243] Virtual Damping Update

[0244] D new =D old +α D ·ΔD t

[0245] Where α D =0.35, then: D new =D old +0.01085.

[0246] Photovoltaic gain update

[0247] G PV,new =G PV,old +α G ·ΔG PV

[0248] Take α G =0.2, update size: G PV,new =G PV,old -0.0036

[0249] Energy storage power regulation

[0250] P t+1 =P t +β·ΔP t

[0251] It has strong energy storage response capability; taking β = 0.8, P t =35.0MW: P t+1 =35.0 + 0.72 = 35.72MW. Step 6: Impedance change after control takes effect.

[0252] Table 5 Changes in impedance morphology before and after control.

[0253] f(Hz) Before control After control 320 0.71 0.59 360 0.94 0.72 400 1.28 0.91 440 1.61 1.12 480 1.73 1.19

[0254] As shown in Table 5, the impedance peak decreased by 35-39%, the peak position shifted from 400Hz to 360Hz (peak shift reversed), the curvature decreased by 52%, the slope decreased by 47%, and the shape shift decreased to: T t+1 =0.93. The risk level has been reduced from Level III to Level I.

[0255] Summary of Results:

[0256] This embodiment demonstrates that the active control method based on impedance morphology characteristics proposed in this invention possesses significant foresight and vibration suppression capabilities in energy storage-dominated weak grid scenarios. The system can detect the continuous deterioration trend of impedance in the mid-to-high frequency bands before oscillations develop. Through minor adjustments to virtual damping, loop gain, and power distribution, the peak impedance in the key frequency band is reduced by 35%–39% within 1.5 seconds, the peak drift is reversed from 400Hz to 360Hz, and the curvature and slope decrease by more than 50%. The overall morphological shift rapidly falls back from the high-risk zone to a safe range. Subsequently, the impedance curve remains stable for the next 5 minutes, without further mid-to-high frequency spikes or drift. This fully verifies the real-time performance, stability, and system-level oscillation suppression effect of this invention in weak grid conditions and multi-power supply parallel scenarios.

[0257] This invention overcomes the limitations of traditional single-point index monitoring by extracting multi-layered morphological features such as the slope, curvature, and peak-valley inflection points of the impedance curve. It can capture structural changes during the oscillation incubation stage, advancing the identification time to the precursor stage where "there are no obvious waveform features but the structure has already changed," thus solving the deficiency of existing technologies in identifying early trends. Furthermore, this invention employs a short-window impedance estimation and rolling update mechanism, achieving millisecond-level impedance morphology updates without frequency sweep excitation. Combined with trend-oriented control decision-making and automatic execution links, it shortens the "monitoring-judgment-control" response cycle to milliseconds to seconds, solving the problems of insufficient real-time performance and the disconnect between monitoring and control in existing technologies, achieving a leap from "post-event remediation" to "early suppression." Simultaneously, this invention makes judgments based on the overall geometric behavior of the impedance morphology, without relying on a fixed equivalent model, making it adaptable to complex scenarios such as multi-inverter clusters, weak power grids, and frequent control strategy switching. Through multi-device collaborative adjustment and parameter smoothing loading mechanisms, it ensures the comprehensiveness of vibration suppression effects while avoiding disturbances to equipment caused by instantaneous parameter jumps, significantly improving the safety and reliability of system operation.

[0258] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A new energy broadband oscillation control system based on impedance morphology, comprising a data acquisition module, an impedance estimation module, a morphological feature extraction module, a morphological trend determination module, a coordinated control decision module, and a control execution module, characterized in that: The data acquisition module acquires real-time measurement signals of three-phase voltage and current at the grid connection point. After discretization, time alignment, bias removal, and filtering preprocessing, it outputs a synchronized electrical quantity sequence. The impedance estimation module transforms the time-domain signal acquired by the data acquisition module into wide-band voltage-current frequency-domain components through short-time frequency domain transformation, constructs a complex impedance sequence point by point, and uses a rolling window mechanism to realize real-time impedance updates and structural verification. The morphological feature extraction module extracts multi-layer morphological features of the complex impedance sequence to form a morphological description vector containing local structural information and full-band topological features. The morphological trend determination module performs trend analysis based on morphological description vectors and historical data, and classifies risk levels through time difference, rate of change, continuity verification, abrupt change detection, and full-band fusion. The coordinated control decision module generates a multi-dimensional control vector including virtual damping, control loop gain, and power allocation based on the risk level and morphological error. The control execution module converts the control vector into equipment control commands through a smooth loading mechanism, realizes coordinated adjustment of multiple devices in the new energy power station, and outputs feedback to the data acquisition module to form a closed loop.

2. The new energy broadband oscillation control system based on impedance mode according to claim 1, characterized in that: The multi-layer morphological features of the morphological feature extraction module include: impedance structured sequence after frequency discretization, local slope features, local curvature features, peak and valley point and inflection point topological features, single-frequency point comprehensive morphological quantity, full-band morphological vector and smoothing processing results.

3. The new energy broadband oscillation control system based on impedance mode according to claim 1, characterized in that: The risk levels of the morphological trend determination module include three categories: mild shift, accelerated evolution, and structural deterioration. The determination is based on a comprehensive assessment of the number of morphological shifts, the rate of change, the persistence, the amount of mutation, the total frequency band shift, and the distribution of peaks and valleys.

4. The new energy broadband oscillation control system based on impedance mode according to claim 1, characterized in that: The control vector generation process of the coordinated control decision module includes: determining the control intervention intensity that matches the risk level, calculating the morphological error between the current morphological vector and the target morphological vector, weighted fusing errors of each frequency band to obtain the virtual damping adjustment amount, determining the control loop gain adjustment amount based on the full-band offset, and selectively adding power adjustment amount according to the risk level.

5. A new energy broadband oscillation control system based on impedance mode according to claim 1, characterized in that: The smooth loading mechanism of the control execution module includes linear loading or exponential loading. The multi-device collaborative adjustment adopts a distributed power correction strategy, allocates adjustment weights according to the device response speed, and has built-in parameter boundary checks and rate limiting logic.

6. A new energy broadband oscillation control method based on impedance mode, employing the new energy broadband oscillation control system based on impedance mode described in any one of claims 1-5, characterized in that, Includes the following steps: S1: Collect the three-phase voltage and current signals at the grid connection point, perform discretization, time alignment, bias removal and filtering preprocessing, and output the synchronized electrical quantity sequence through a sliding short window; S2: Perform a short time-frequency domain transformation on the preprocessed time-domain signal, calculate the complex impedance point by frequency to construct a wide-band impedance spectrum, and ensure the continuity and reliability of the impedance curve through rolling window updates and structural checks. S3: Extract multi-layer morphological features of the complex impedance sequence to form a morphological description vector; S4: Based on morphological description vectors and historical data, trend analysis is performed to detect morphological shifts, rate of change, persistence and abrupt changes, and risk levels are classified by integrating full-band trend information; S5: Generate multi-dimensional control vectors based on risk level and morphological error, and verify amplitude, rate and equipment capability; S6: The control vector is converted into equipment control commands through a smooth loading mechanism and sent to the new energy power station equipment to realize multi-equipment coordinated adjustment. The electrical quantity signals after control are collected and steps S2-S6 are repeated to form a closed-loop vibration suppression.

7. The new energy broadband oscillation control method based on impedance morphology according to claim 6, characterized in that: In S2, the short-time frequency domain transformation adopts the short-time Fourier transform, and the spectral leakage is reduced by windowing. For frequency points where the current amplitude is lower than the threshold, the adjacent frequency points are interpolated and replaced.

8. The new energy broadband oscillation control method based on impedance mode according to claim 6, characterized in that: In S3, the multi-layer morphological feature extraction includes: discretizing the continuous impedance curve, calculating the slope and slope difference curvature of adjacent frequency points, identifying local peak and valley points and inflection points, fusing slope, curvature and peak features to construct a single-frequency comprehensive morphological quantity, splicing them to form a full-band morphological vector and smoothing it by neighborhood averaging.

9. The new energy broadband oscillation control method based on impedance mode according to claim 6, characterized in that: In S4, trend analysis includes: calculating the time difference of morphological quantities to determine the offset direction, calculating the rate of change to identify the acceleration trend, smoothing through a time window to verify the continuity of change, setting a threshold to detect abrupt morphological changes, and weighted fusing the offset across the entire frequency band.

10. A new energy broadband oscillation control method based on impedance morphology according to claim 6, characterized in that: In S6, the smooth loading mechanism controls the parameter update rate through the loading coefficient, the multi-device collaborative adjustment allocates power adjustment weights according to the device response speed, and the closed-loop feedback achieves adaptive correction by iteratively updating the impedance shape.

Images